Economics of International Airline Transport PDF

THE ECONOMICS OF
INTERNATIONAL AIRLINE
TRANSPORT
ADVANCES IN AIRLINE
ECONOMICS
Series Editor: James Peoples
Recent Volumes:
Volume 1: Competition Policy and Anti-Trust, Darin Lee
Volume 2: The Economics of Airline Institutions, Operations and
Marketing, Darin Lee
Volume 3: Pricing Behavior and Non-Price Characteristics in the Airline
Industry, James Peoples
ADVANCES IN AIRLINE ECONOMICS VOLUME 4
THE ECONOMICS OF
INTERNATIONAL
AIRLINE TRANSPORT
EDITED BY
JAMES PEOPLES
University of Wisconsin-Milwaukee, WI, USA
United Kingdom North America Japan

India Malaysia China
Emerald Group Publishing Limited
Howard House, Wagon Lane, Bingley BD16 1WA, UK
First edition 2014
Copyright r 2014 Emerald Group Publishing Limited
Reprints and permission service

Contact: permissions@emeraldinsight.com
No part of this book may be reproduced, stored in a retrieval system, transmitted in

any form or by any means electronic, mechanical, photocopying, recording or
otherwise without either the prior written permission of the publisher or a licence
permitting restricted copying issued in the UK by The Copyright Licensing Agency
and in the USA by The Copyright Clearance Center. Any opinions expressed in the
chapters are those of the authors. Whilst Emerald makes every effort to ensure the
quality and accuracy of its content, Emerald makes no representation implied or
otherwise, as to the chapters’ suitability and application and disclaims any warranties,
express or implied, to their use.
British Library Cataloguing in Publication Data

A catalogue record for this book is available from the British Library
ISBN: 978-1-78350-639-2
ISSN: 2212-1609 (Series)
ISOQAR certified
Management System,
awarded to Emerald
for adherence to
Environmental
standard
ISO 14001:2004.
Certificate Number 1985

ISO 14001
CONTENTS
LIST OF CONTRIBUTORS vii
EDITORIAL BOARD MEMBERS xi
CHAPTER 1 INTRODUCTION AND OVERVIEW

James Peoples 1
CHAPTER 2 AIR TRANSPORT LIBERALIZATION

AND ITS EFFECTS ON AIRLINE COMPETITION AND
TRAFFIC GROWTH AN OVERVIEW
Xiaowen Fu and Tae Hoon Oum 11
CHAPTER 3 GOVERNMENT REGULATION OF

INTERNATIONAL AIR TRANSPORTATION
Darren Prokop 45
CHAPTER 4 SKIES WIDE SHUT AN ASSESSMENT

OF INTERNATIONAL AIR TRANSPORT
LIBERALIZATION
Pierre Latrille, Antonia Carzaniga and Marta Soprana 61
CHAPTER 5 INTERNATIONAL MERGERS AND

ACQUISITIONS IN THE AIRLINE INDUSTRY
James Nolan, Pamela Ritchie and John Rowcroft 127
CHAPTER 6 AIRFARES AND COMPETITION ON

INTERNATIONAL ROUTES
John Bitzan, Alice Kones and James Peoples 151
v
vi CONTENTS
CHAPTER 7 THE CHOICE OF AIRPORT, AIRLINE,

AND DEPARTURE DATE AND TIME: ESTIMATING
THE DEMAND FOR FLIGHTS
Diego Escobari and Cristhian Mellado 177
CHAPTER 8 AIR CARGO SERVICES AND THE

EXPORT FLOWS OF DEVELOPING COUNTRIES
Henry L. Vega 199
CHAPTER 9 AN ASSESSMENT OF THE CAUSAL

RELATIONSHIP BETWEEN AIR PASSENGER
TRAFFIC AND TRADE IN ASIA-PACIFIC
Elien Van De Vijver, Ben Derudder and Frank Witlox 235
CHAPTER 10 INTERNATIONAL AND NATIONAL

POLITICAL REGULATIONS OF AVIATION’S
CLIMATE IMPACT AND COST IMPACTS ON AIR
FREIGHT
Janina D. Scheelhaase 255
CHAPTER 11 PLANNING A COMPETITIVE

AEROTROPOLIS
John D. Kasarda and Stephen J. Appold 281
CHAPTER 12 AIRPORT AND AIRLINE

SUBSTITUTION EFFECTS IN MULTI-AIRPORT
MARKETS
Dan Mahoney and Wesley W. Wilson 309
CHAPTER 13 BUYER SUBSIDIES IN TWO-SIDED

MARKETS: EVIDENCE FROM ONLINE TRAVEL
AGENTS
Volodymyr Bilotkach and Nicholas G. Rupp 339
LIST OF CONTRIBUTORS
Stephen J. Appold Research Associate, Center for Air

Commerce, Kenan-Flagler Business
School, Chapel Hill, NC, USA
Volodymyr Bilotkach Senior Lecturer in Economics, Newcastle
University, Newcastle, UK
John Bitzan Associate Professor, Department of
Management and Marketing, North
Dakota State University, Fargo, ND, USA
Antonia Carzaniga Counsellor, World Trade Organization
(WTO), Geneva, Switzerland
Ben Derudder Professor, Geography Department, Ghent
University, Ghent, Belgium
Diego Escobari Assistant Professor, Department of
Economics and Finance, The University
of Texas Pan American, Edinburg,
TX, USA
Xiaowen Fu Senior Lecturer, Business School,
University of Sydney, Sydney, Australia
John D. Kasarda Director of Center for Air Commerce ,
Center for Air Commerce, Kenan-Flagler
Business School, Chapel Hill, NC, USA
Alice Kones Research Assistant, Department of
Economics, University of Wisconsin-
Milwaukee, Milwaukee, WI, USA
Pierre Latrille Counsellor, World Trade Organization
(WTO), Switzerland, Geneva
vii
viii LIST OF CONTRIBUTORS
Dan Mahoney Graduate Teaching Fellow, Department of

Economics, University of Oregon, Eugene,
OR, USA
Cristhian Mellado Assistant Professor, Department of
Economics, Universidad Catolica de la
Santisima Concepcion, Concepcion, Chile
and Research Assistant and PHD Student,
Department of Economics and Finance,
The University of Texas Pan American,
Edinburg, TX, USA
James Nolan Professor, Department of Bioresource
Policy, Business and Economics,
University of Saskatchewan, Saskatoon,
SK, Canada
Tae Hoon Oum UPS Foundation Chair Professor,
Sauder School of Business, University
of British Columbia, Vancouver, BC,
Canada
James Peoples Professor, Department of Economics,
University of Wisconsin-Milwaukee,
Milwaukee, WI, USA
Darren Prokop Professor of Logistics, Department of
Logistics, College of Business and Public
Policy, University of Alaska Anchorage,
Anchorage, AK, USA
Pamela Ritchie Dean, Faculty of Business and IT,
University of Ontario Institute of
Technology, Oshawa, ON, Canada
John Rowcroft Adjunct Professor, Faculty of Business and
IT, University of Ontario Institute of
Technology, Oshawa, ON, Canada
Nicholas G. Rupp Associate Professor of Economics,
Department of Economics, East Carolina
University, Greenville, NC, USA
List of Contributors ix
Janina D. Scheelhaase Head of Air Transport Economics,

German Aerospace Centre (DLR), Institute
of Air Transport and Airport Research,
Cologne, Germany
Marta Soprana Consultant, World Trade Organization
(WTO), Geneva, Switzerland
Elien Van De Vijver PHD Student, Department of Geography,
Ghent University, Ghent, Belgium
Henry L. Vega Research Fellow, Center for
Transportation, Policy, Operations, and
Logistics, School of Public Policy, George
Mason University, Arlington, VA, USA
Wesley W. Wilson Professor of Economics, Department of
Economics, University of Oregon, Eugene,
OR, USA
Frank Witlox Professor, Geography Department, Ghent
University, Ghent, Belgium
EDITORIAL BOARD MEMBERS
Volodymyr Bilotkach Theodore E. Keeler

Newcastle University University of California, Berkeley
John Bitzan B. Starr McMullen

North Dakota State University Oregon State University
Jan K. Brueckner Steven Morrison

University of California, Irvine Northeastern University
Kevin Cullinane Nicholas G. Rupp

Edinburgh Napier University East Carolina University
Martin Dresner Ian Savage

University of Maryland Northwestern University
David Gillen Wayne Talley

University of British Columbia Old Dominion University
Timothy J. Hazledine Wesley W. Wilson

University of Auckland University of Oregon
Marc Ivaldi
Université Toulouse 1 Capitole
xi
CHAPTER 1
INTRODUCTION AND OVERVIEW
James Peoples
Globalization has placed a premium on the ability of countries to create

opportunities for their domestic companies to compete successfully interna-
tionally. Easy access to low-cost transportation plays a vital role in enhan-
cing the international competitiveness of companies. Historically
companies overwhelmingly selected maritime shipping to transport cargo
overseas, especially the hauling of high-volume, low-price per-unit bulk
products such as oil, and grain.1 Rail and trucking services traditionally
dominated the transportation of cargo across international land borders.
Recent cost efficiency gains and technological advancements in air trans-
portation services present shippers with a viable alternative mode to tradi-
tional transportation services. For instance, fuel efficiency improvements of
aircrafts, flexible work rules, and the formation of carrier alliances contri-
bute to increasing use of air transportation to haul export goods. In addi-
tion, increasing global demand for high-value, low-volume products such
as electronic equipment further contribute to the viability of air transporta-
tion services as a low-cost, short-delivery-time alternative. Recent evidence
of nontrivial growth reported by the International Air Transport
Association (IATA) reveals a 3.9 percent increase in global freight trans-
ported by air carriers from October 2012 to October 2013.2 Passenger ser-
vice is also experiencing growth in air transportation services as
international passenger volume increased by 5.7 from September 2012 to
September 2013 (International Air Transport Association [IATA], 2013).3
The Economics of International Airline Transport

Advances in Airline Economics, Volume 4, 19
Copyright r 2014 by Emerald Group Publishing Limited
All rights of reproduction in any form reserved
ISSN: 2212-1609/doi:10.1108/S2212-160920140000004001
1
2 JAMES PEOPLES
Furthermore, growth in international air transportation of passengers for

this time frame is not limited to traditionally high-demand countries
located in North America and Europe. Growth is also occurring in devel-
oping countries located in Africa, Asia/Pacific, Latin America, and the
Middle East.4 Thus, air transportation service is truly a global phenomenon
generating significant demand from all geographic regions. Gaining greater
understanding of economic conditions facilitating this growing demand as
well as gaining further insight on the external benefits and costs associated
with international air transportation growth is critical for countries and
companies planning to become or to remain economically competitive in
the twenty-first century.
This volume of Advances in Airline Economics contributes to our under-
standing of the economics of international air transportation by providing
in depth analyses of changes in international air transportation’s legal
environment, analyses of airline companies market behavior in this new
environment, examination of trade growth in response to changing airline
market behavior, and examination of the external costs and benefits asso-
ciated with growth in international air transportation. Analyses are
grounded in theories established in transportation economics, and chapter
contributors employ innovative empirical techniques to examine how chan-
ging policies in international air transportation have influenced the growth
and development of the aviation industry as well as the growth of regional
and local economies.
LIBERALIZATION OF INTERNATIONAL AIR

TRANSPORTATION SERVICES
The first set of chapters describes the details of international agreements

liberalizing air transportation services and also reports the effect of these
agreements on international air transportation services. The initial chapter
by Xiaowen Fu and Tae Hoon Oum reviews major liberalizing airline ser-
vice agreements (ASA’s) that were established over the last twenty years.
These agreements include the US Open-Skies agreement with Canada, the
establishment of the Australia-New Zealand single aviation market, the
establishment of the EU single aviation market, the establishment of
the USEU Open Aviation Area (OAA) and ongoing liberalization propo-
sals between Korea and Japan, and the US and Japan. Their review reveals
that these liberalization packages contributed to increased international
Introduction and Overview 3
competition in the air transportation services market, increased flight fre-

quency, improved airline efficiency, increased traffic volumes and the estab-
lishment of new route services. These authors also examine the role of
liberalization policies on the relationship between airports and airlines.
Making this analysis is important in part because airports are an essential
resource for airline operations. The authors reveal that the potential for
vertical arrangement among airports and airlines in this less economically
restrictive business environment could pose challenges to competition and
consumer welfare enhancement. Fu and Oum conclude by explaining ana-
lytical, econometric and computational network methods used to model
the effects of liberalization in complex market conditions. Understanding
the attributes and short-comings of these models is necessary to help for-
mulate effective liberalization policy for continued growth in the interna-
tional air transportation services industry in concert with enhanced
consumer welfare for individuals using this service.
The succeeding chapter by Darren Prokop observes that while liberaliza-
tion policy has created a business environment encouraging competition
along international routes, many bilateral air agreements do not provide
companies complete freedom to service foreign countries. He highlights the
potential for further gains for cargo shippers using international air trans-
portation services by describing the concepts of the Freedoms of the Air
and Open Skies agreements. He explains that while bilateral Open Skies
agreements do allow head-to-head competition in import and export cargo,
a country’s air space may not be completely open to foreign competition.
Prohibiting the freedom to engage in an activity called cabotage signifi-
cantly limits foreign carriers’ operational freedom in host countries because
foreign carriers are not granted the right to pick-up and drop-off domestic
cargo within a host countries borders. Prokop provides evidence on the
welfare effects associated with the prohibition of cabotage by reporting on
the air cargo transfer operations at Ted Stevens Anchorage International
Airport. Unilateral liberalization allows foreign carriers to transfer cargo at
Ted Stevens Anchorage International Airport and then continue to ship
that cargo to another US destination. As Prokop notes, such activity is not
identical to cabotage but is fairly close, so examining economic outcomes
derived from unilaterail liberalization for international air transportation
service at the Ted Stevens Anchorage International Airport provides
important evidence of potential gains to shippers that might arise from eas-
ing restrictions on cabotage activity.
While both of the initial chapters detail the legal changes to bi- and
multilateral agreements that characterize the current policy shift to a less
4 JAMES PEOPLES
restrictive regulatory environment for international air transportation ser-

vices, these chapters also report that there is much more opportunity to
further promote competition through the relaxation of remaining regula-
tory restrictions on entry. The following chapter by Pierre Latrille,
Antonia Carzaniga and Marta Soprana uses a unique technique to quan-
tify the extent to which international air transportation services are truly
open to competition. Their technique allows for analysis of key market
access features of bilateral air service agreements such as setting of tariffs,
approval of cooperative agreements and approval of multiple designation
clauses. Their findings are consistent with observations from Chapters 2
and 3 as Latrille, Carzangina and Soprana demonstrate that although the
air transportation sector has experienced some liberalization over the past
few years, relaxation of restriction on international air transportation ser-
vices has been, overall, rather marginal.
AIR COMPANY PERFORMANCE AND BEHAVIOR IN A

MORE LIBERALIZED INTERNATIONAL
TRANSPORTATION REGULATORY ENVIRONMENT
Chapters 57 empirically examine company performance and behavior,

with the objective of providing information on changes in industry perfor-
mance and consumer welfare during the recent period of regulatory liberali-
zation. In Chapter 5 James Nolan, Pamela Ritchie, and John Rowcroft
observe that international merger and acquisition activity has increased
during the current period of easing of regulatory restrictions of interna-
tional air transportation services. Theory established in the industrial orga-
nization literature suggests the potential for efficiency gains arising from
international mergers and acquisitions. Such efficiency gains arise in part
from economies of scale and scope. Nolan, Ritchie, and Rowcroft use the
data envelop analysis (DEA) procedure to empirically examine relative
operating efficiencies between recently internationally merged airline com-
panies and other air companies. They also incorporate a step-wise model to
estimate an industry production function with the objective of using esti-
mation results to examine the relative production of the group of interna-
tional merged air carriers. Their findings reveal efficiency gains associated
with merger and acquisition activity among international airlines as an
additional source of benefits to the industry derived from the liberalization
of international air service transportation.
Consolidation of international air transportation services could raise

concerns regarding lessening of competition if newly merged air carriers
become dominant carriers along high density routes. Contestability theory
suggests that this should not be a problem. This theory explains that the
threat posed by potential competitors can act as an incentive to force the
dominant incumbent to charge competitive prices. While research empiri-
cally examining this theory for domestic markets do not provide strong
support for its predictions on fares and market competition, findings by
Morrison and Winston (1987) on domestic fares do support the notion of
imperfect contestability. They observe that the presence of a potential com-
petitor serving the destination or originating airport of a dominant carrier
is associated with lower fares on that route. In their chapter John Bitzan,
Alice Kones and James Peoples test whether imperfect contestability also
applies to fare-setting on international routes. Estimating a series of fare
equations, they find actual and potential competition are important deter-
minants of international airfares. These results are interpreted as indicating
that pricing behavior along US-international routes is consistent with the
theory of imperfect contestability.
With greater consolidation, even in the absence of potential competitors,
consumer demand elasticity limits the extent to which dominant carriers
can charge higher prices without losing significant revenue generating traf-
fic. In Chapter 7 Diego Escobari and Cristhian Mellado estimate passenger
demand for flights in an international travel market using an estimation
approach that allows analysis of passenger flight choices in response to
prices depending on the departing airport, the identity of the carrier, and
the departure date and time. The results reveal that consumers demand for
international air transportation, at least for the route observed in this chap-
ter, is not inelastic. They also find demand elasticity varies by carrier, time
of day service and departure dates. In general, the absence of inelastic con-
sumer demand does suggest that dominant carriers serving international
routes face a limit on their ability to engage in noncompetitive pricing.
INTERNATIONAL AIR TRANSPORTATION AS A

FACILITATOR OF INTERNATIONAL TRADE
The third set of chapters examine the significance of international air trans-
portation as a facilitator of international trade. In Chapter 8 Henry L.
Vega uses a gravity model of trade to measure the effects of air freight costs
6 JAMES PEOPLES
on export flows from developing countries to the US and to countries in

the EU. Given the increasing use of air transportation service to haul per-
ishable and exotic goods, as well as to haul high-value, low-volume high-
technology goods, Vega estimates the effect of freight rates on the export
of these groups of products via air transportation. His findings lend some
support to the notion that trade levels are suboptimal for developing coun-
tries due in part to high air transportation cost, but also cautions against
applying sweeping generalizations to all developing countries. For instance,
his findings on air transportation originating in developing countries reveal
less sensitivity to higher freight rates for perishable and exotic goods com-
pared to high-technology goods. Hence, policy-makers should adopt a
nuanced approach that considers the freight type when establishing regula-
tory reform of international air transportation services.
The succeeding chapter by Elien Van De Vijver, Ben Derudder and
Frank Witlox continues the analysis of international air transportation ser-
vices and international trade by empirically testing the hypothesis of co-
evolution for trade and air passenger traffic for Asia-Pacific countries and
for country pairs within the Asia-Pacific region. Using a time-series cross-
section Granger causality test, the authors consider four possible causality
scenarios. Their findings reveal significant differing patterns of causality
among countries. For instance, they do not find a causality link for the
most developed countries in the Asia-Pacific rim region. In contrast, they
report a significant causal relationship between air passenger service and
trade between more developed and less developed economies. Findings also
reveal a causal trade air transportation service relationship for Asia-Pacific
rim countries having adopted liberal air transportation policies.
While the preceding chapters indicate increasing passenger and shipper
demand for international air transportation generally evolves with trade,
providing such services creates an external cost to society in the form pollu-
tion emissions. In Chapter 10 Janina D. Sheelhaase reports several coun-
tries impose pollution emissions regulation using market-based measures
such as fees on local emission of carbon dioxide (CO2) and Nitrous Oxide
(NOX). These fees can potentially contribute to greater operating expenses
by increasing the cost of providing air transportation services. Sheelhaase
examines this issue by initially providing an overview of the current politi-
cal regulations on aviation’s climate relevant emissions in Europe,
Australia and New Zealand and of the planned regulations in other parts
of the world. She then presents an extensive survey of the literature on the
cost impacts of most of these regulations on air freight services. She
observes that cost impacts on air freight services due to emissions
regulations tend to be small, due in part to the airframe/engine combina-

tion used by air transportation companies. She concludes by providing sug-
gestions that would help air companies further reduce operating costs
attributable to emissions regulation.
THE ECONOMICS OF INTERNATIONAL AIRPORTS
Critical to airlines competing successfully in this era of increasingly liberal-

ized access to international air transportation routes is the availability and
efficiency of international airport operations. The last set of chapters exam-
ine operations at international airports and their effect on local economies,
as well as their influence on the pricing and demand for service from their
locations. The initial chapter by John D. Kasarda and Stephen J. Appold
examines the growth in the size and operation of international airports.
Given the nontrivial employment of local resources by these airports, the
authors identify these large facilities as central to the development of the
aerotropolis, which they define as an urban-sub-region whose infrastruc-
ture, land-use and economy are centered on an airport. All cities identified
by the authors as aerotropolises are major hubs of international transporta-
tion services.5 In this chapter, the authors present planning principles that
enhance aviation-enabled trade in goods and services through improve-
ments in airport area connectivity with other modes of transportation, and
optimal use of land near the airport.
The succeeding chapter by Dan Mahoney and Wesley W. Wilson reports
on the importance of international airports as a determinant of airline
choice and passenger activity. They observe the significant role played by
airports in passengers’ choice of flights. Airport usage fees and competitive
pricing from low cost competitors serving passengers from alternative air-
ports in the same city highlight the potential influence of airport in the
selection of airline services. The authors estimate an airline demand equa-
tion for air transportation services that are offered from six large US
metropolitan locations, and each of these locations are host to multiple
international airports serving a large number of passengers. Their results
reveal demand elasticities differ by airport characteristics. For instance,
customer demand is less responsive to price changes at the largest airport
in the local market. Significant among the authors’ findings is that in
response to price increases for flights departing from the customers’ origi-
nal airport choice, he/she is highly likely to exit the local market rather
8 JAMES PEOPLES
than choose to fly from an alternative airport in the same area. Such consu-
mer behavior has significant implications for the development of local
economies.
The final chapter in this volume examines how online travel agencies use
of subsidies influence competition among airline companies originating and
departing from international airports located in the US. In Chapter 13
Volodymyr Bilotkach and Nicholas G. Rupp explain the critical role online
travel agencies play as facilitators of trade between passengers and air trans-
portation service providers. For instance, by introducing a most-favored-
customer subsidy program on June 6, 2008, the online travel agency Orbitz
guaranteed the lowest fares to its customers. As part of their analysis
Bilotkach and Rupp examine the change in the number of carrier ticketing
options per route following the introduction of Orbitz’s subsidy program.
Using data on 50 major international serving airlines based in the United
States, the authors find an increased availability of ticketing options for cus-
tomer using Orbitz’s service as well as those using the online services of their
rivals. Bilotkach and Rupp interpret these findings as evidence that Orbitz’s
subsidy program promotes competition in the online travel agency market.
CONCLUDING COMMENTS
Economic theory suggests introduction of more liberal airline service agree-

ments should create a business environment that facilitates gains to ship-
pers, travelers, and local and regional economies. Gains arise from airline
companies charging competitive fares, and improving service for interna-
tional flights. Chapters in this volume explore several areas in which custo-
mers as well as local and regional economies have benefitted from the
relaxation of policies restricting air companies operations in foreign air
space. Findings reveal increased air traffic, growth in international trade,
enhanced operating efficiency, and increased flight options for flyers asso-
ciated with the growth of international air transportation services.
Contributions to this volume also expand the debate on the economic
effects of liberalization of international air transportation services by exam-
ining the role of airports and online ticket agencies as facilitators of inter-
national air trade.
While there is little disagreement regarding the benefits consumers and
economies derive from less regulation imposed on international air trans-
portation services, authors in this volume indicate potential sources of cost
to society associated with this policy shift. For instance, vertical mergers
between airlines and airports, and horizontal mergers and acquisitions and
alliances among foreign air carriers can potentially lessen competition. Still,
evidence of imperfect contestability, and consumer price sensitivity suggests
dominant carriers serving international routes are limited in their ability to
successfully benefit from practicing noncompetitive pricing. In fact, contri-
butors to this volume indicate there are substantial opportunities for even
greater international competition if future agreements authorize further
relaxation of restrictions on airline companies’ freedoms to operating in
foreign countries. Nonetheless, increasing demand for international air
transportation service from passengers and shippers suggests air transpor-
tation’s contribution to globalization will continue to grow.
NOTES
1. Ocean liners are also major international transporters of high-volume high-

price durable manufacturing products such as home appliances and automobiles.
2. Source: http://www.iata.org/whatwedo/Documents/economics/Freight-Analysis-
Oct-2013.pdf.
Hummels (2007) presents additional evidence of growing demand for air transporta-
tion services by shippers as he reports air shipment’s share of the value of imports
in the United States grew from 8.1 percent in 1970 to 31.5 percent by 2004, and air
shipment’s share of the value of US exports to non-North American countries
increased from 11.9 percent to 52.8 percent for the same observation period.
3. Source: http://www.iata.org/whatwedo/Documents/economics/Passenger-Analysis-
Sep-2013.pdf.
4. Africa, Asia/Pacific, Latin America and the Middle East experience growth in
the international passenger transported equaling 6.9, 8.5, 8.3, and 10.4 percent,
respectively, for the September 2012 to September 2013 time frame. Source: http://
www.iata.org/whatwedo/Documents/economics/Passenger-Analysis-Sep-2013.pdf.
In contrast growth rates for Europe and North America were only 3.4 and 3.1 per-
cent, respectively for the same time frame.
5. URL posting the list of aerotropolises worldwide is as follows: http://www.
aerotropolis.com/airportCities/about-the-aerotropolis.
REFERENCES
Hummels, D. (2007). Transportation costs and international trade in the second era of globali-
zation. Journal of Economic Perspectives, 21(3), 131154.
International Air Transport Association (IATA). (2013). Air Passenger Market Analysis.
Retrieved from http://www.iata.org/whatwedo/Documents/economics/Passenger-
Analysis-Sep-2013.pdf
Morrison, S., & Winston, C. (1987). Empirical implications and tests of the contestability
hypothesis. Journal of Law and Economics, 30, 5366.
CHAPTER 2
AIR TRANSPORT LIBERALIZATION

AND ITS EFFECTS ON AIRLINE
COMPETITION AND TRAFFIC
GROWTH AN OVERVIEW
Xiaowen Fu and Tae Hoon Oum
ABSTRACT
This chapter reviews the effects of air transport liberalization, and inves-
tigates the roles played by airport-airline vertical arrangements in liber-
alizing markets. Our investigation concludes that liberalization has led to
substantial economic and traffic growth. Such positive outcomes are
mainly due to increased competition and efficiency gains in the airline
industry, and positive externalities to the overall economy. Liberalization
allows airlines to optimize their networks, and thus may introduce sub-
stantial demand and financial uncertainty to airports. Vertical arrange-
ments between airlines and airports may offer a wide range of benefits to
the parties involved, yet such arrangements could also lead to airline
entry barriers which reduce the effects of liberalization. Three
approaches have been developed to model the effects of liberalization in
complex market conditions, which include the analytical, econometric

ISSN: 2212-1609/doi:10.1108/S2212-160920140000004000
11
12 XIAOWEN FU AND TAE HOON OUM
and computational network methods. These approaches should be selec-

tively utilized in policy studies on liberalization.
Keywords: Air liberalization; airport-airline vertical arrangements;
airline competition
JEL classifications: L510; L930; L980; L160
INTRODUCTION
Air transportation is becoming increasingly important to the world’s econ-

omy. It not only directly contributes to the production of passenger trans-
port and logistics services, but also provides essential inputs to other
sectors such as tourism and trade. In terms of value, about one third of the
world’s international trade is currently moved by air. However, despite
major breakthroughs in air transport liberalization since the early 1990s,1
most international markets remain subject to some types of regulation
under the 1944 Chicago Convention on international air transportation.
The International Civil Aviation Organization (International Civil
Aviation Organization (ICAO), 2007) estimated that by 2006, only about
31% of the country pairs with non-stop services had embraced liberaliza-
tion. The commercial rights of airlines involving international routes have
been governed by the bilateral air services agreements (ASAs) between each
pair of countries involved. Although ICAO and the World Trade
Organization (WTO) have attempted to devise a multilateral framework
for trade in air services similar to the General Agreement on Tariffs and
Trade (GATT) and General Agreement on Trade in Services (GATS), only
a few regional open-skies agreements have been reached thus far, the most
notable example being the EU single aviation market. Although ASEAN
countries have agreed to form a regional open sky by 2015, the extent to
which the existing regulations can be removed is unclear given the restric-
tive policies governing the current markets. Most liberalization packages
have been implemented by developed countries, notably the United States.
In 2003 there were 87 liberalized agreements involving 70 countries, 59 of
which were U.S. open-skies agreements. As of October 2012, there were
over 400 liberalized agreements involving 145 states. Of these, over 100
were U.S. open-skies agreements (ICAO, 2013). The negotiations toward
liberalizing ASAs usually involve a lengthy political process, even among
Air Transport Liberalization 13
countries with strong economic and political ties. One major challenge in
the process is the different expectations of stakeholders on the effects of
alternative liberalization policies.
This chapter attempts to review the major regulatory policy changes
that have taken place in liberalizing the aviation markets over the last
20 years, thus to facilitate the decision-making by regulators, industry
managers and academic researchers. In particular, this chapter provides:
(1) a review of the literature on the economic effects of past air transport
liberalization events; (2) an examination of the roles of airports in liberal-
ization, airport-airline vertical arrangements and their implications on
competition and welfare, and (3) a brief summary and a discussion of the
different modeling approaches used to analyze the effects of liberaliza-
tion. In an effort to provide a comprehensive review, this chapter draws
heavily on recent studies by Oum, Fu, and Zhang (2009), Fu, Oum, and
Zhang (2010), Fu and Zhang (2010), Li, Lam, Wong, and Fu (2010), Fu,
Dresner, and Oum (2011), Fu, Homsombat, and Oum (2011),
Homsombat, Lei, and Fu (2011), Adler, Fu, Oum, and Yu (2014), and
Yang, Zhang, and Fu (2013).
This chapter is organized as follows. The second section reviews the gen-
eral findings on air transport liberalization with a focus on traffic volume,
airline competition and social welfare. The third section reviews the roles
of airports in liberalization and the evolving pattern of airline-airport verti-
cal arrangements. The fourth section reviews the methodologies developed
to measure liberalization effects. The final section summarizes and con-
cludes the chapter.
THE ECONOMIC EFFECTS OF AIR TRANSPORT

LIBERALIZATION
Under the 1944 Chicago Convention on international air transportation,

each country has exclusive sovereignty over its airspace. The commercial
rights of airlines on international routes are governed by bilateral ASAs
between each country pair. The WTO Secretariat (World Trade
Organization (WTO), 2006) identified seven ASA regulations as relevant
openness indicators for scheduled air passenger services: (1) grant of rights
(air freedoms allowing airlines to provide services over designated markets),
(2) capacity clause (regulation of the volume of traffic, frequency of service
and/or aircraft types), (3) tariff approval (whether fares must be approved
before they are applied), (4) withholding (which defines the conditions for
the foreign carrier to operate, such as ownership and effective citizen con-
trol requirements), (5) designation (which governs the number of airlines
allowed to serve the market between two countries and on specific routes),
(6) statistics (requiring the exchange of operational statistics between coun-
tries or their airlines, and (7) cooperative arrangements (which regulate the
cooperative marketing agreements between airlines). After reviewing 2,299
ASAs in the ICAO and WTO databases, Piermartini and Rousova (2008)
indicated that the regulations used most frequently were those on pricing,
capacity, and cooperative arrangements. Based on a numerical presentation
of the seven types of regulation, the WTO Secretariat (WTO, 2006) created
the Air Liberalization Index according to expert weightings. Another
stream of studies, including those by Gonenc and Nicoletti (2000), Doove,
Gabbitas, Nguyen-Hong, and Owen (2001), and Piermartini and Rousova
(2008), constructed an index of air service liberalization using factor analy-
sis techniques. They found that the index reflected the degree of liberaliza-
tion, and concluded that the removal of restrictions on price and capacity
setting, cabotage freedoms and beyond rights were the most traffic-
enhancing measures in liberalization.
Regulation on the aviation industry was first introduced after World
War I with the aim to promote commercial aviation as an infant industry.
The Kelly Air Mail Act of 1925 was passed in the United States, allowing
the Post Office to subsidize private airmail carriage by awarding contracts
with payments exceeding airmail revenue. The Civil Aeronautics Board
(CAB) was created as a regulator by the Civil Aeronautical Act of 1938,
and over time began to regulate route entries, rate levels, rate structures,
subsidies and merger decisions (Borenstein & Rose, 2007; Caves, 1962;
Levine, 1965). These regulations led to limited competition and high fares
(Jordan, 1970; Keeler, 1972; Levine, 1987) in addition to excessive capacity
and insufficient demand. In the years prior to deregulation, the average
load factors fell below 50% (Borenstein & Rose, 2007). Regulations in the
international markets were even more restrictive, as they were determined
by the most restrictive regulation policies in the countries involved. ASAs
generally regulate the services (passenger, cargo) and routes to be operated
between two countries, and stipulate fare-setting mechanisms, capacities
and frequencies. They usually specify which airlines have the rights to fly
on each route, and determine the capacities that can be provided by each.
Most countries impose very restrictive regulations on international travel in
terms of destinations, market entry, frequency, capacity, route allocation
and fare levels.
Although all of these regulations on domestic routes and international

markets led to limited competition and depressed traffic volumes, airlines
were unable to gain substantial profits: with airfare regulated, airlines com-
peted in quality (e.g., premium cabin services, modern and large aircrafts).
However, fares remained high and traffic volume did not grow quickly. In
addition, the worker unions of pilots, cabin crews and air traffic controllers
gained substantial bargaining power in salary negotiations, which further
pushed up airlines’ input prices. In 1978, the United States deregulated its
domestic inter-state markets,2 which led to increased competition, substan-
tially reduced fares and increased traffic volume. The 1979 U.S.
International Air Transportation Competition Promotion Act (IATCPA)
was passed the following year, emphasizing the government’s determina-
tion to promote liberalized ASAs with foreign countries.
The Effects of Major Liberalization Events
Whereas the tone of liberalization was set with the passage of the IATCPA
in 1979, the world’s first open-skies agreement was signed between the
Netherlands and the United States in 1992. The agreement gave both coun-
tries unrestricted landing rights at each other’s airports, and the United
States granted antitrust immunity to the alliance between Northwest
Airlines and KLM Royal Dutch Airlines. Although the aviation markets
between the two countries were already fairly liberalized before the agree-
ment, the open-skies agreement nevertheless significantly increased traffic
in the years that followed. Several major liberalization events have taken
place since, including:
• the U.S. open-skies agreement with Canada;
• the establishment of the Australia-New Zealand single aviation market;
• the establishment of the EU single aviation market;
• the establishment of the U.S.EU Open Aviation Area (OAA); and
• other ongoing and proposed liberalization packages between Korea and
Japan and Japan and the United States, and ASEAN open-skies
agreements.
On February 24, 1995, the governments of Canada and the United
States signed an “Open Skies” Agreement that allowed both Canadian
and American airlines to establish direct links between any pair of cities
located on either side of the border. The agreement authorized any
Canadian or U.S. airline to offer trans-border services without restricting
fares, flight frequencies or aircraft types. The number of Canadian cities

with trans-border air services increased from 21 in 1994 to 27 in 2005,
and the number of trans-border routes nearly doubled from 90 to 171.
The capacity of scheduled airline services increased by 25% in the first
year, and the average trans-border seat capacity per day increased by
about 49%, from 28,217 in 1994 to 41,968 in 2005. The number of air
carriers providing trans-border services nearly doubled from 11 carriers
in 1994 to 20 in 2005. The load factors of the main carriers rose by 10%
between 1994 and 2000. As worries existed that Canadian carriers could
not compete with their larger, more efficient U.S. rivals, the entry of U.S.
carriers into major Canadian destinations was phased in gradually.
However, Canadian carriers did well after liberalization. Air Canada
offered more than 1,200 code share flights throughout the United States
per week, and strengthened its hubs in Toronto and Vancouver. The
labor productivity of the major airlines grew by 18% during the period,
as measured in passenger kilometers per employee. Further, the growth
in trans-border air service made significant contributions to Canada’s
economy. The open-skies agreement generated an estimated 4,500 addi-
tional direct jobs per year, contributing almost $300 million in GDP.
Furthermore, the new air services stimulated other economic sectors such
as the tourism and export industries. It is estimated that a fifth freedom
passenger service offered by a U.S. carrier could generate as many as 105
direct aviation jobs, and about 1,300 direct jobs in the tourism industry
(Vancouver Airport Authority, 2000).
The “single aviation market” between Australia and New Zealand was
first negotiated in 1992, completed in 1996 and became fully operational in
2000 (Vowles & Tierney, 2007). The agreement allowed Australian and
New Zealand airlines to operate across the Tasman and beyond to third
countries without restriction, which liberalized air traffic between the two
countries to airlines from other countries. There were over 4.6 million pas-
sengers in the market in 2004, up from nearly 3.3 million passengers in
2000, the year the single aviation market was first fully implemented. The
entry of low-cost carriers such as Virgin Blue (later renamed Virgin
Australia) and other foreign carriers triggered sharp competition and sub-
stantial traffic growth. The number of passengers in the market increased
by over 25% during the 15 months following Virgin Blue’s announced
entry. The Single Aviation Market Agreement also provided low-cost car-
riers with the opportunity to serve some smaller destinations, creating a
new network of direct international connections that did not previously
exist.
The EU Single Aviation Market was created in 1992 and comprised the
12 member states of the European Economic Community (EEC).
Following the creation of the EU in 1993, the number of member states in
the single aviation market increased to 27 in 2007. Full cabotage rights
became effective on April 1, 1997 as a result of the three major liberaliza-
tion packages finished that year. The single aviation market evolved into a
wider European Common Aviation Area (ECAA) comprising 35 states in
2006. Among the significant effects of the single aviation market were the
development of low-cost airline services and dramatic increases in competi-
tion and consumer choices. Intra-EU routes with more than two carriers
increased by 385% between 1992 and 2007. The number of cross-border
intra-EU routes increased by 220% during the same period. Air travel in
Europe tripled between 1980 and 2000 (Schipper, Rietveld, & Nijkamp,
2002). In 2006, the EU air transport industry carried over 730 million pas-
sengers, 480 million of whom were carried within the EU.
Following the 2002 judgment by the European Court of Justice (ECJ)
on ASAs with the United States, the Council of the EU conferred on the
European Commission a mandate to negotiate a comprehensive ASA on
behalf of every member state with the United States. This ASA involved
the creation of an OAA between the two territories, the creation of “com-
mon” aviation areas (i.e., the integration of the EU’s neighboring states
into a single aviation market) with Morocco and the countries of the
Western Balkans in 2004, Ukraine in 2006 and Jordan in 2007 and the crea-
tion of an OAA with Canada in 2007. On March 2, 2007, the United States
and EU concluded a comprehensive air transport agreement involving all
27 EU countries. The agreement, which came into effect on March 30,
2008, extended the open-skies principles to 11 EU countries such as
Greece, Ireland, Spain and the United Kingdom, with which the United
States had restrictive agreements or none at all. In addition to the open-
skies agreements between the United States and the 27 member states of
the EU, the agreement allowed U.S. investors to participate as minority
shareholders (up to 49.9%) in EU-majority-owned airlines. This effectively
allowed U.S. investors to hold minority shares of state-owned firms.
European investors could take up to a 25% legislated cap on voting equity
and a 25%-minus-one-share regulatory cap on non-voting equity. A
second-stage agreement was concluded in 2010 that expanded U.S.EU
cooperation to environmental areas such as noise and emissions. The
United States was concerned about the proliferation of night flight curfews
at EU airports that could affect express delivery carriers such as DHL,
FedEx and UPS. In 2011, the agreement was extended to include Iceland
and Norway. It was estimated (Booz Allen Hamilton, 2007) that the
U.S.EU open-skies agreement would generate 9.6 million additional pas-
sengers annually during the first 5 years, and lead to substantial gains in
consumer surplus, additional employment and efficiency gains in the avia-
tion industry.
In general, all of these liberalization packages led to increased competi-
tion, decreased average fares, increased frequency, improved load factor
and airline efficiency and increased traffic volumes and new route services.
Such findings are consistent with an InterVISTAS (2006) study that
reviewed all of the major liberalization events until the early 2000s.
However, the costs and benefits have not been evenly distributed among
the countries and firms involved. Airline profitability has neither improved
with deregulation and liberalization (Goetz & Vowles, 2009). Some small
airports have seen reductions in direct services. Thompson (2002) con-
cluded that liberalization encouraged a proliferation of new entrant airlines
exploiting the opportunities offered by minor airports. However, deregula-
tion may also increase the market power of dominant airlines, which could
limit the market access of small airports. An updated investigation into the
European market by Dobruszkes (2009) concluded that competition gener-
ally increased following liberalization, but mostly benefited passengers in
large European cities and peripheral regions that received major tourist
flows from big cities in Western Europe. Whereas the benefits of liberaliza-
tion have certainly not been evenly distributed, there appears to be general
agreement that the overall effects have been positive (Button, 2009). Much
of the benefit has apparently comprised gains in consumer surplus, and
other industries and sectors using aviation as a major input.
Dynamics of Liberalization: Increased Competition and

Improved Efficiency
There are two key driving factors behind the positive outcomes of liberali-
zation. First, liberalization removes constraints on pricing, route entry, ser-
vice capacity and cooperative arrangements among alliance members. This
allows airlines to operate more efficiently, which leads to decreased costs
and increased service quality in terms of flight frequency, frequent flyer
programs and the like. Second, competition increases after liberalization
and deregulation, which forces down market prices and stimulates traffic
volume. Over time, only the most efficient and innovative firms are able to
survive, and thus the industry’s overall productivity increases. With the
increased competition brought on by deregulation and liberalization, much

of the efficiency gains have been passed on to consumers in the form of
lower prices. Table 1 compares the changes in the prices of air travel verses
other goods and services in the United States from 1978 to 2006. Air ser-
vices experienced some of the lowest nominal price increases, that is, only
1.51.6 times the price of 1978. The International Air Transport
Association (IATA) (2008) claimed that real yields fell by almost 50% since
air transport market liberalization and deregulation began in the late
1970s. During the period of 1974 to 2011, world GDP growth mostly fluc-
tuated between 2% and 4% over the years. In comparison, airlines’ real
yields in constant prices declined from about US$1.60/ATK to US$0.8/
ATK. Economic growth and declining yields were major factors contribut-
ing to traffic growth.
The prosperity of low-cost carriers (LCCs) has been a major driver for
increased competition and falling airline yields. Indeed, the so-called
“Southwest effect” has become well known. The entry of Southwest
Airlines into the market introduced a sharp increase in traffic volume and a
significant fall in fares on routes where, or close to where, the famous LCC
airline operated (Dresner, Lin, & Windle, 1996; Morrison, 2001; Richards,
1996; US DOT, 1993; Windle & Dresner, 1995). It would have been impos-
sible for LCCs to compete freely in regulated markets with incumbent
Table 1. Price Changes of Air Travel Versus Other Goods and Services.
ITEM-U.S. Good or Service Unit 1978 1990 2006 Growth
College tuition: public Year $688 $1,908 $5,836 8.5 ×

College tuition: private Year $2,958 $9,340 $22,218 7.5 ×
Prescription drugs Index 61.6 181.7 363.9 5.9 ×
New single-family home Home $55,700 $122,900 $246,500 4.4 ×
New vehicle Vehicle $6,470 $15,900 $28,450 4.4 ×
Unleaded gasoline Gallon $0.67 $1.16 $2.59 3.9 ×
CPI (urban-all items) CPI-U 65.2 130.6 201.6 3.1 ×
Movie ticket Ticket $2.34 $4.22 $6.55 2.8 ×
First-class postage Stamp $0.15 $0.25 $0.39 2.6 ×
Whole milk Index 81.0 124.4 181.6 2.2 ×
Grade-A large eggs Dozen $0.82 $1.01 $1.31 1.6 ×
Air travel: international Mile 7.49¢ 10.83¢ 11.85¢ 1.6 ×
Air travel: domestic Mile 8.49¢ 13.43¢ 13.00¢ 1.5 ×
Television Index 101.8 74.6 22.3 0.2 ×
Sources: General Accountability Office (GAO, 2008), Airline Industry: Potential Mergers and
Acquisitions Driven by Financial and Competitive Pressures, GAO-08-845, July 31, 2008.
airlines. The effects of LCCs have been most apparent in Europe, where
the prosperity of LCCs has led to substantial traffic increases in many sec-
ondary airports. Many military airports in Western Europe have been con-
verted for civil aviation purposes in recent decades. These airports provide
simple services and do not always have convenient ground transportation.
However, the associated airport charges are low and there is no congestion,
making them ideal for LCC services after liberalization. In contrast,
Dobruszkes (2009) found that traditional European airlines, especially the
majors (Air France, British Airways, Lufthansa, and KLM), did not benefit
directly from the liberalization of European airspace to operate flights not
centered on their country of origin. In fact, their contributions to the usage
of the fifth to ninth air freedoms in Europe amount to less than 1% each.
These carriers make greater use of the fifth to ninth freedoms outside
Europe, particularly on long-haul flights to the Far East that involve stop-
overs. In Europe, these carriers remain strongly rooted in their national
centers. The importance of LCCs to liberalization can also be testified in
the Canadian market, where cross-border LCC services have been limited.
Elwakil, Windle, and Dresner (2013) estimated that about 5 million trave-
lers avoid flying on trans-border routes every year. Instead, they choose to
use American airports that are close to the U.S.Canada border and cross
the border via surface transport. Elwakil et al. (2013) argued that such
“demand leakage” occurs mainly due to the lack of LCC competition.
Other contributing factors may include the alliance memberships of United
States and Canadian carriers, and the antitrust immunity granted for price
setting and scheduling.
The traffic stimulation effects of LCCs are becoming increasingly impor-
tant in Asia. Unlike developed markets such as North America and
Europe, where LCCs focus on secondary airports, LCCs have gained a sig-
nificant market share at hub airports. Homsombat, Fu, and Agachai (2010)
benchmarked the key performances of the major aviation hubs in
Southeast Asia, and concluded that the healthy development of hub air-
ports in the region occurred largely due to economic growth, market liber-
alization and the development of LCCs. This is most apparent in the
aviation market in Malaysia, which were dominated by Malaysian Airlines
but has experienced rapid market expansion with the growth of a successful
LCC AirAsia. Total passenger volumes doubled in Malaysia during
20002010, much faster than in its neighboring countries. This was largely
due to the growth of AirAsia, which contributed to Malaysia’s liberaliza-
tion policy. For example, Malaysia restricted the Singapore-Kuala Lumpur
route for years to protect Malaysian Airlines. In 2007, the Malaysian
government decided to allow AirAsia and Tiger Airways (an LCC from
Singapore) to operate on the route. The liberalization policy contributed to
AirAsia’s outstanding growth. Whereas the amount of passengers carried
by Malaysian Airlines dropped sharply from 18 million to 13 million from
2005 to 2010, AirAsia’s traffic volume increased from 4 million to 15 mil-
lion. As shown in Tables 2 and 3, AirAsia’s growth also benefited Kuala
Lumpur International Airport. AirAsia has increasingly gained in its mar-
ket share at the airport, where Malaysian Airlines once controlled 70% of
the market. As a result, the HHI concentration index at Kuala Lumpur
International Airport decreased from over 5,000 in 2002 to about 2,500 in
2010, figures comparable with the other three benchmarked airports.
Efficiency studies have verified the effects of liberalization and competi-
tion on airline productivity. Oum and Yu (1998) found that after deregula-
tion, many U.S. carriers achieved global leadership in terms of cost
competitiveness. Fethi, Jackson, and Weyman-Jones (2000) found EU liber-
alization to significantly improve airline efficiency. Inglada, Rey,
Rodrıguez-Alvarez, and Coto-Millan (2006) compared the economic and
technical efficiency of international air transport companies during
19962000 with their estimation of a stochastic frontier cost function and a
stochastic frontier production function. They found evidence of the benefits
of increasing competition to make Asian carriers more efficient. Forsyth,
Table 2. Traffic Volumes and Number of Airlines Serving the Airports.

HKG KUL SIN BKK
2002
Passenger volume 33,451,466 15,936,882 27,374,329 30,484,781
Average daily aircraft movements 567 350 479 542
Number of airlines serving the airport 63 37 62 78
2005
2010
Source: Homsombat et al. (2011). SIN Singapore Changi International Airport, KUL
Kuala Lumpur International Airport, BKK Bangkok Suvarnabhumi International Airport,
HKG Hong Kong International Airport.
Table 3. Market Shares of Dominant Carriers in the Four Airports.

Airport No. 1 Carrier No. 2 Carrier No. 3 Carrier Top 3 Carriers
2002 Market Share of Scheduled Seats

HKG 31.66% 10.68% 8.56% CX/KA/CI
KUL 68.59% 5.90% 3.48% MH/SQ/CX
SIN 43.95% 6.48% 4.78% SQ/QF/MH
BKK 50.44% 4.08% 3.79% TG/CX/PG
HKG 32.37% 13.52% 7.21% CX/KA/MU
KUL 57.39% 17.73% 3.42% MH/AK/SQ
SIN 43.22% 5.52% 3.90% SQ/QF/CX
BKK 39.71% 5.69% 4.33% TG/PG/FD
HKG 35.39% 13.79% 5.87% CX/KA/CI
KUL 35.20% 34.82% 5.53% MH/AK/D7
SIN 35.29% 6.51% 6.28% SQ/MI/TR
BKK 40.93% 11.20% 5.56% TG/FD/PG
Source: Homsombat et al. (2011). Airline codes are: AK (AirAsia), CI (China Airlines), CI
(China Airlines), CX (Cathay Pacific Airways), D7 (AirAsia X), FD (Thai AirAsia), KA
(Dragonair), MH (Malaysia Airlines), MI (SilkAir), MU (China Eastern Airlines), PG
(Bangkok Airways), QF (Qantas Airways), SQ (Singapore Airlines), TG (Thai Airways
International), and TR (Tiger Airways).
Hill, and Trengrove (1986), Encaoua (1991), Oum and Yu (1995), Oum, Fu,
and Yu (2005), Inglada, Coto-Millan, and Rodriguez-Alvarez (1999) and
Homsombat et al. (2010) also offered consistent findings. The increased
competition introduced by liberalization and deregulation has clearly forced
airline management to improve operational efficiency and adopt new tech-
nologies and innovations. Work unions have been pressed to work under
more flexible arrangements. Over time, the least efficient airlines must leave
the market, leading to increased efficiency across the industry.
Air Transport Liberalization and the Overall Economy
There is a two-way relationship between air transportation and the overall

economy. It is well known that air transport and logistics are “derived”
demands usually purchased as inputs or intermediate products for the con-
sumption/production of other services. The effects of economic growth on
air traffic volume growth are usually measured by “income elasticity,” the
value of which is estimated at 12 for most aviation markets. That is,
ceteris paribus, a 1% growth in income or GDP leads to a 12% percent
increase in air traffic volume. Oum et al. (2009) argued that income elasti-
city had been generally overestimated. Using the historical traffic growth
pattern and major traffic forecasts made by Airbus, Boeing and the IATA,
the authors computed “residual” traffic growth after removing the traffic
volume increase due to economic growth.3 Even if income elasticities are
assumed to be 1.5 for developing markets and 1.2 for mature markets,
which are at the low end of the existing estimates, residual growth would
be nearly zero or negative. That is, the income effects would explain all of
the traffic volume growth in every aviation market. Other factors such as
quality improvements (e.g., increased frequency, improved networks, better
frequent flyer programs), liberalization (e.g., the removal of constraints on
capacity, pricing, route entry and LCC competition), new technology and
better productivity (e.g., online booking, self-check-in, new generation of
aircrafts) would have no effect on traffic growth. This is clearly unreason-
able, and that the aforementioned factors (quality, liberalization, new tech-
nology, and productivity improvements) have not been well controlled in
income elasticity estimations. Oum et al. (2009) thus hypothesized that the
income elasticity for developed markets should be about 1, and that the
income elasticity for developing markets should be about 1.3.4
As in any other industry, growth in the aviation sector leads to higher
economic activities such as increased employment, tax revenue and produc-
tion outputs in related industries. The IATA noted that air transport
directly employs 4 million people worldwide and generates $400 billion in
output. In addition, the efficiency and quality improvements in air passen-
ger services contribute to growth in the hotel and tourism sectors. That is,
the aviation sector has a significant positive effect on other industries and
the overall economy. Button, Lall, Stough, and Trice (1999) found that the
presence of a hub airport increases high-tech employment by an average of
12,000 jobs in a region. Irwin and Kasarda (1991) found that expanding
the airline network serving a region had a significant positive effect on local
employment, especially in the service sector. As one may expect, many stu-
dies have found clear evidence that air liberalization benefits the tourism
industry. Shaw (1982) noted that following aviation deregulation in the
United States, many airlines experienced poor financial performance, which
led to some instability in aviation services. However, if other exogenous
factors were considered, deregulation should not be blamed as the main
cause; the tourism industry had benefited, and thus should favor further
liberalization. Similar findings related to the positive effects on tourism
have been identified for other markets such as the EU (European Low
Fares Airlines Association, 2004; European Union Committee of the
Regions, 2004; Leidner, 2004), Ireland (Sørenson & Dukes, 2005), Korea
(Chung & Whang, 2011), Malta (Graham & Dennis, 2010), Italy (Donzelli,
2010), and Spain (Rey, Myro, & Galera, 2011). In many cases, the prosper-
ity of LCCs after liberalization has been the main driving factor of the
growths in traffic volume and tourism.
On the cargo side, air transport also contributes significantly to the over-
all economy. In terms of value, one third of international merchandise
trades are delivered by air. Air cargo is ideal for high-value and perishable
goods. As for passenger traffic, the real yield for cargo has been declining
over time. Swan (2007) found that since 1970, both the price and produc-
tion costs of air travel have been declining at about 1% annually. Harrigan
(2005) estimated that the relative cost of air transport declined by 40%
between 1990 and 2004. Hummels (2006) found that the elasticity of air
shipping costs in terms of distance declined dramatically, from 0.43 in 1974
to 0.045 in 2004. Doubling of the shipping distance caused a 43% increase
in air shipping costs in 1974, but only a 4.5% increase in air shipping costs
in 2004. As a result, the average distance of air shipment is becoming
longer, and the average ocean shipment is becoming shorter. The efficiency
and quality improvements of air transportation have also promoted trade
and economic growth. One study by Aizenman (2004) argued that air ship-
ping may be an effective way of handling international demand volatility,
as air cargo delivery allows firms to wait until the realization of demand
shocks before deciding on the quantities to sell. Recent studies have found
that a 10% increase in time decreases bilateral trade volumes by 58%
(Djankov, Freund, & Pham, 2005; Hausman, Lee, & Subramanian, 2005).
In addition, as shipments are growing in value and becoming lighter over
time, the ad valorem cost of airfreight, i.e., the transport cost required to
move $1 of goods, is also decreasing. This has made air cargo an affordable
option for an increasing number of products.
LIBERALIZATION AND THE EVOLVING

RELATIONSHIPS BETWEEN AIRPORTS AND
AIRLINES
In contrast to the rich literature related to the effects of liberalization on

airline competition, traffic volume and economic growth, few studies have
investigated the roles played by airports in liberalization. In fact, even

within the aviation industry, there are conflicting views on optimal airport
policy. Whereas airlines demand more regulation on airports, airports call
for deregulation and liberalization. For example, Director General of
Airports Council International (ACI) Robert Aaronson argued that greater
flexibility and freedom would help airports fulfill their roles under liberali-
zation (Aaronson, 2003). However, the IATA (2007) made the following
argument: “There are significant potential benefits from greater liberaliza-
tion of the airline industry. There are also significant potential benefits
from more effective economic regulation of airports and air navigation ser-
vice providers (ANSPs).” These conflicting views reflect the complexity
involved in understanding the roles of airports in liberalization.
There is a two-way relationship between air service liberalization and
the airport industry. Without an efficient and competitive airport sector,
the benefits of air service liberalization cannot be fully realized. However,
whereas liberalization is generally beneficial to the airport industry in terms
of traffic increases and new business opportunities, it may also introduce
new challenges. The influence of airports on liberalization may include the
following. First, liberalization gives airlines the freedom to serve interna-
tional markets. Without the availability of airport slots and other airside
capacities, this potential benefit could not be realized. However, different
types of regulation (no regulation and deregulation may be considered one
type of regulation) introduce different incentives for airports to invest in
capacity. Second, dominance at an airport gives incumbent airlines sub-
stantial market power, which may serve as a barrier to entrant airlines and
thus decrease the competitive effects of liberalization. Third, airports often
have substantial market power, especially to local traffic. Such market
power may hinder the ability of airlines to offer quality service at competi-
tive rates. Meanwhile, liberalization may also introduce new challenges to
airports in the forms of revenue variability, the loss of strong hub carriers
and traffic volume reductions.
These challenges have led airlines and airports to enter into various ver-
tical arrangements. Starkie (2008) argued that contracts between airlines
and airports mark a major innovation in contemporary civil aviation, and
have served as catalysts to the market transformation in Europe since the
liberalization of aviation in the 1990s. Liberalization and the subsequent
growth of LCCs have led to increased competition between airports,
thereby promoting formal, specific contracts between airports and airlines.
Such contracts have led to a fundamental change in airport competition in
that airlines are now encouraged to establish operating bases at airports.
Although numerous studies have investigated issues such as airport reg-

ulation, capacity investments, environment control, privatization and pro-
ductivity, few have considered the relationship between air service
liberalization and airport management. In this section, we provide a brief
summary of the related findings and a tentative agenda for future research.
Airports as a Barrier of Entry and Source of Airline Market Power
Airports provide essential inputs to airlines such as landing/takeoff slots,

gate and terminal spaces and parking and refueling services. If an airline
could not obtain sufficient supplies of these inputs at reasonable costs, it
would not be able to compete effectively. Morrison and Winston (2000)
and Dresner, Windle, and Yao (2002), among others, found empirical evi-
dence that a dominant airline’s control over key airport facilities such as
slots and gates is likely to impose significant entry barriers for other poten-
tial competitors. The liberalization of the international market promotes
traffic growth in both intra-continental and intercontinental markets. Most
of the international air traffic occurs to/from/between these hub airports.
According to Airbus (2007), 50% of the world’s 100 fastest-growing city
pairs are among the 32 global hub cities, and almost all of the remaining
city pairs have a hub at one end or the other. This has important implica-
tions for air liberalization. First, the effectiveness of liberalization may be
significantly reduced if airport capacities were not available. Second, even
if competition were enhanced via increased services and competitor
options, many of the welfare gains would be offset by the losses caused by
congestion delays and other forms of service quality deterioration. Major
airport expansion projects in metropolitan regions are extremely expensive
and difficult due to the large capital investment needed and increasingly
stringent environmental review requirements. Because a significant propor-
tion of routes involve at least one hub airport, any service deterioration at
a hub airport has a detrimental effect on the overall network. Such negative
effects are very damaging to full-service airlines that use hub-and-spoke
networks. Airport slot allocation has been part of the liberalization nego-
tiation in markets involving certain congested airports such as Tokyo’s
Narita International Airport, Shanghai Hongqiao International Airport
and Beijing Capital International Airport on an ad hoc basis. Airports
have yet to become formally involved in liberalization talks.
Airport dominance can be a major source of an airline’s market power.
Dominance at an airport allows a carrier to achieve a substantially higher
markup above cost, a benefit that the literature has referred to as the “hub
premium” (Borenstein, 1989). Hub premiums have been found to range
from well below 10% (Lee & Prado, 2005; Morrison & Winston, 1995) to
around 20% (GAO, 1989, 1990; Lijesen, Rietveld, & Nijkamp, 2001), a
very significant advantage considering the low profit margin in the airline
sector. Meanwhile, Fu, Lijesen, and Oum (2006) and Fu et al. (2011)
argued that airports usually have substantial market power over local traf-
fic due to the extremely low price elasticity of the demand for landing and
takeoff slots with respect to airport charges. Therefore, the full benefits of
liberalization cannot be realized in the absence of a competitive, efficient
airport sector that provides high quality services to the aviation industry.
Starkie (2008) argued that the U.K. airport industry is competitive because
most of its airports are within a 2-hour drive from each other. However,
most of the regions in the rest of the world do not have such a luxury.
Thus, there is a need to introduce competition to the airport sector, or to
maintain effective economic regulation.
Demand and Financial Uncertainty for Airports in the

Liberalizing Market
Liberalization could introduce new challenges to airports. The liberaliza-

tion of the capital market and ownership control allows airlines to effec-
tively control/acquire carriers in other countries. The increased competition
introduced by liberalization makes such overseas expansion not only possi-
ble but also necessary. Such changes in the market structure lead to
dynamic changes in airline competition and network configuration. As a
result, airport traffic may experience significant uncertainty and fluctuation.
This pattern has already been observed in the deregulated U.S. domestic
market. De Neufville and Barber (1991) analyzed the traffic volatility for
the 38 largest U.S. airports during 19681988 and concluded that traffic
volatility clearly increased after the 1978 deregulation. Overall, volatility
did not diminish over time, and did not depend on the sizes of the airports
(i.e., volatility values may be considered homoscedastic). They further
argued that many factors contribute to increased airport traffic volatility.
The increasing usage of hub-and-spoke networks implies that a significant
proportion of connecting traffic could be easily routed via alternative hubs.
Because the aviation market is becoming more dynamic, airlines must
experiment with new operation and competition strategies. Airport entry
and exit have become easier, and airline mergers and alliances have led to
route and network reconfigurations. The increased volatility of traffic

volume and financial performance have clearly introduced major challenges
to airport planning and operation, as investments in the airport sector are
lumpy and cannot be changed within a short time (Oum & Zhang, 1990).
One major reason for airlines to form international alliances (i.e., Star
Alliance, Oneworld, SkyTeam) is to serve oversea markets where market
access is unavailable or network coverage is incomplete. However, with lib-
eralization, airlines may choose to either offer services by themselves or
merge with other airlines. Merged airlines may consolidate their operations
and networks, causing some airports to lose strong hub carriers. Fu et al.
(2011) argued that while full-service network airlines usually set up multiple
hub networks, it is not economical for an airline to have more than one
hub in a region, and thus a merged airline may decide to consolidate hubs
in a region. This would be a major challenge to hub airports that rely heav-
ily on a dominant carrier’s operations. For example, after the merger
between Air France and KLM, the Dutch government was concerned that
the merged airline may decrease the network coverage out of KLM’s hub
at Amsterdam (AMS). AMS is about 270 miles from Paris, too close to act
successfully as a dual hub. The Dutch government thus imposed the condi-
tion that the combined AF-KLM should maintain a minimum of 42 major
international key destinations from Amsterdam for at least 5 years after
the merger. Because AMS has long been an established major hub in
Europe, it has been able to maintain its hub status thus far. However, for
other regional hubs, the risk of losing a hub carrier may be a real problem.
In 2013, Tennessee politicians dubbed Delta’s decision to de-hub Memphis
as a betrayal of its promise that service from Memphis would not diminish
after its merger with Northwest in 2008. Meanwhile, Delta is building up
gateway operations at Seattle to strengthen its Asian network, as Asian
markets such as China, Japan and Korea are becoming increasingly liberal-
ized. Whereas the ASKs Delta deployed from Seattle increased by nearly
10% between October 2012 and June 2013, they fell nearly 40% in
Memphis during the same period (CAPA, 2013).
As an increasing number of international markets are liberalized and
restrictions on capital investments and foreign ownership are gradually
removed, some truly “global” airlines/airline groups may be formed. These
carriers may optimize their merged/consolidated networks in a way similar
to that of the North American legacy carriers after the deregulation of the
United States domestic market in 1978. However, airports are operating in
an environment different from that of three decades ago. Many airports
are privatized/commercialized, operate more like businesses and are free
from government subsidy. As a result, these airports proactively participate

in liberalization to maximize their financial return or secure their long-term
prosperity. Many have chosen to enter into vertical arrangements with their
airlines.
Airline and Airport Vertical Arrangement and

Implications for Liberalization
Starkie (2008) argued that the liberalization of aviation has increased the
risk of airport assets being stranded due to the opportunistic behavior of
airlines, which are free to change routes and switch airports at will after lib-
eralization. Consequently, airports have the incentive to establish long-
term contracts with airlines to achieve a better risk balance. Fu et al. (2011)
reviewed the vertical arrangements observed in the aviation industry and
classified them into the following categories:
• Signatory airlines of an airport: Airports award so-called “signatory air-
line” status to carriers that sign master use-and-lease agreements. Those
airlines become guarantors of the airport’s finances. The main contribu-
tion from signatory airlines is guaranteed service and usage commitment,
which allow the airport to decrease financing costs when securing long-
term loans.
• Airlines own or control airport facilities: Some airlines hold shares in air-
ports or directly control airport facilities. Such direct control allows car-
riers to optimize the terminal operations, and to share the revenue
generated from concession services.
• Long-term usage contracts: Airlines and airports can secure their coop-
eration via long-term contracts. For example, long-term, exclusive-use
lease agreements have been used extensively at large hub airports, award-
ing incumbent carriers the right to the facilities regardless of the actual
usage. Some LCCs have also organized long-term contracts with airports
to lock in favorable terms. Long-term contracts can also be beneficial to
airports. They encourage airlines to make long-term investments and to
develop more extensive networks, thus securing airport traffic in the long
run.
• Airports issuing revenue bonds to airlines: Many airports now choose to
issue special facility revenue bonds (SFRBs) to finance specific capital
improvement programs. Airports retain the asset ownership, but transfer
the right for exclusive usage and financial obligation to the sponsor
airline.
• Revenue sharing between airports and airlines: Airports have increasingly

relied on concession services to bring in more revenue. Many airports
began sharing their revenues with airlines to induce the airlines to bring
in more passengers by internalizing the positive externality of traffic
volume on concession service. On the airline side, certain carriers
demand a share of airport revenue as a condition upon initiating services
at these airports.
• Other agreements: Some airports enter into exclusive or non-exclusive
agreements with airlines by offering favorable treatments such as dis-
counted landing charges or even subsidies for new services. In Australia
and New Zealand, where airports are subject to the so-called “light-
handed” regulation, airports and airlines can negotiate prices and terms
for customized services.
Airports and airlines can apparently achieve various objectives via alter-
native vertical arrangements. However, as airports provide an essential
input for airlines, such vertical arrangements have raised competition con-
cerns from regulators and attracted academic investigation. Fu and Zhang
(2010), Zhang, Fu, and Yang (2010), Yang, Zhang, and Fu (2013) studied
the implications of revenue sharing between airports and airlines on compe-
tition and welfare. They concluded that while such practices may improve
welfare, they may also negatively affect airline competition. In particular,
their studies concluded that whether an airport is subject to competition is
critical to the welfare consequences. Barbot (2011) considered several
airport-airline vertical agreement scenarios, and found that while consu-
mers would be better off in general, it may still be of their interests to regu-
late certain types of contract. D’Alfonso and Nastasi (2012) extended
Barbot’s (2011) study and found mixed policy implications.
In general, recent studies on airline-airport vertical arrangement have
identified some “synergistic” effects on the airlines and airports involved.
However, the welfare and competition effects are complicated and depen-
dent on many market conditions. For “trans-border” liberalization on
intercontinental services, access to major hub airports is a must because
they are the only viable destinations. These airports often face capacity lim-
itations and have dominant network carriers. In such cases, liberalization
does not lead to a traffic increase unless airport access (e.g., slot allocation)
is also ensured for foreign carriers. Indeed, capacity shortages at Tokyo’s
two airports have greatly limited their roles as international hubs and their
capacities to realize the benefits of liberalization. Thus far, regulators have
left it to the airlines to work out airport access. For example, airlines have
long traded/swapped slots at London Heathrow Airport. In another case,

Delta relied on its SkyTeam partners, China Eastern Airlines and China
Southern, to obtain airport slots at convenient times for its Detroit-Beijing
and Detroit-Shanghai Pudong services.5 In such cases, dominant airlines at
hub airports clearly exert significant influence over market outcomes, and
thus may easily adopt anti-competitive strategies. However, in markets
where “beyond rights” are granted (i.e., freedoms are granted in addition
to first to fourth freedoms to airlines in OD countries and carriers in a third
country), a strong hub carrier is clearly beneficial. For example, Dubai
clearly benefited from having a strong hub carrier (i.e., Emirates) in its
competition with alternative transit hubs in the region, such as Abu Dhabi
and Doha. In such cases, close cooperation between hub airports and their
dominant carriers is likely to be pro-competition and enhance welfare.
In short to medium markets, the introduction of LCC competition may
partly solve airport access issues. Many LCCs operate point-to-point net-
works and thus do not have to use slots at peak times for flight connec-
tions. Their low prices also allow them to use less-convenient secondary
airports. For example, LCCs have used Gatwick and Stansted extensively
for London service as alternatives to Heathrow. Because AirAsia could not
obtain landing slots at Beijing Capital International Airport at any conve-
nient time, it thus chose to use nearby Tianjin Binhai International Airport
as a substitute. In such cases, cooperation between LCCs and airports are
likely to be pro-competition and facilitate liberalization.
Overall, the right policies to adopt on the airport sector and airline-
airport vertical arrangements depend greatly on the specific market under
consideration. Oum et al. (2009) argued that because markets within
Europe and North America are already fairly deregulated, much of the
future growth will come from intercontinental markets, which usually
involve major hub airports. Capacity constraints in those airports limit air-
line competition. In addition, established network carriers in those airports
have a strong incentive to work with the airports to strengthen their airport
dominance. These hub airports are also likely to cooperate with their
respective dominant carriers to secure future revenue and financial support.
As a result, liberalization cannot be effective unless policymakers make
sure that the airport industry becomes sufficiently competitive and efficient.
Oum et al. (2009) further suggested the following measures to regulators.
• Promote the construction and conversion of non-exclusive airport facil-

ities. This would provide potential entrants with essential airport facil-
ities, making it possible for them to compete with incumbent carriers.
• Promote competition among closely located airports.

• Monitor exclusive contracts between airlines and airports. While vertical
cooperation between airlines and airports can be beneficial to society, it
may also harm competition in the airline market. Therefore, there may
be a need to monitor exclusive contracts/agreements between airlines and
airports.
• Encourage competition between airline networks using different airports
in the same region. Airlines have an incentive to dominate their hub air-
ports. Thus, it is important to encourage competition from the other air-
lines using adjacent airports in the region.
Based on the reviews in this section, we can conclude that the availabil-
ity of adequate, efficient and competitive airport services is a necessary con-
dition for air service liberalization to deliver any material gains. Airport
dominance has long been a source of market power for hub carriers, and
airports themselves hold significant market power, especially over local
traffic. Therefore, the important roles of airports should be recognized in
liberalization initiatives. Where airport capacity is a constraint, the provi-
sion of landing slots to foreign entrants may be included in liberalization
negotiation. Regulators may step in instead of totally relying on airlines to
work out slot allocation independently. Meanwhile, liberalization intro-
duces significant uncertainty to airport revenue and demand. Airports may
lose hub carriers due to airline mergers/consolidations. As a result, priva-
tized/commercialized airports should and will proactively participate in lib-
eralization to maximize their financial returns or secure long-term usage.
Many have chosen to form vertical arrangements with their airlines. The
effects of such vertical arrangements on competition and welfare are com-
plex and dependent on market structure. Thus far, few studies have system-
atically investigated the effects of airport operation on air service
liberalization. The best policy for airport operation and management
remains unclear, and certainly calls for more advanced study.
MEASURING AND MODELING THE EFFECTS OF

LIBERALIZATION
Over the last three decades, deregulation and liberalization have become
worldwide trends, encouraging open-skies agreements or liberalized ASAs
in which restrictions on price, seat capacity and entry are removed or
relaxed. As shown in Section 2, many studies have found that deregulation

and liberalization introduce significant welfare gains and economic growth.
However, since the signing of the first open-skies agreements between the
Netherlands and United States in 1992, most international markets have
remained subject to some of sort of regulation. The negotiations held to lib-
eralize ASAs have usually involved a lengthy political process. One major
challenge in such a process is the different expectations of stakeholders
toward the effects of alternative liberalization policies. Most of the liberali-
zation studies have comprised ex post evaluations based on observed mar-
ket data. These studies have offered valuable insights into the effects of
particular policies that have already been adopted. However, they have not
provided immediate guidance to governments for formulating their own
policies ex ante, as each country has a unique aviation industry in terms of
its home carrier competitiveness, input prices, domestic market size and
competition, the geographic locations of its hub airports and the availabil-
ity of alternative modes of transport. As Button (2009) pointed out,
whereas liberalization is generally good to the aviation industry and coun-
tries involved, the benefits and costs are not uniformly distributed. For
example, Lau, Lei, Fu, and Ng (2012) studied the liberalization of air
transportation across the Taiwan Strait to the region’s aviation industries.
They concluded that liberalization substantially benefits the airports and
airlines in mainland China and Taiwan. In general, Taiwanese airports and
airlines have benefited more from liberalization compared with airports
and airlines in the mainland and Hong Kong. Such an asymmetric effect
occurs due to the larger size of the mainland Chinese aviation market,
which allows Taiwanese airlines to exploit network-related benefits.
Any liberalization agreement involves inputs and endorsements from
regulators, airlines, airports and related industries (i.e., tourism and logis-
tics sectors) in each affected country. Each agent’s opinion on liberalization
is mostly determined by its assessments of the benefits/costs it faces rather
than the overall economy. Governments and policymakers must evaluate
alternative policies so that the best solution can be adopted. Therefore,
there is a need to develop quantitative models that one can use to quantify
and predict market outcomes under alternative liberalization policies.
Adler et al. (2014) reviewed recent quantitative studies on airline compe-
tition and network rivalries and classified the modeling methods into three
approaches, including the analytical, econometric and computational net-
work approaches. The analytical approach typically models airline compe-
tition over a single origin-destination (OD) pair or simplified/stylized
networks, such that closed-form solutions may be obtained. Sample studies
include those on airline network rivalry and price competition (Adler,

2005; Adler & Berechman, 2001; Adler & Hashai, 2005; Brueckner,
Dyer, & Spiller, 1992; Hendricks, Piccione, & Tan, 1997, 1999; Oum,
Zhang, & Zhang, 1995; Zhang, 1996; Zhang & Wei, 1993), airport competi-
tion and pricing (Fu et al., 2006; Pels, Nijkamp, & Rietveld, 2001, 2003) and
airline scheduling (Brueckner, 2004; Schipper, Nijkamp, & Rietveld, 2007).
These analytical models are mathematically tractable. However, they cannot
be used to directly study large-scale, complex networks. Gillen, Harris, and
Oum (2002) used this approach to develop an economic model in which the
market outcomes were simulated over alternative liberalization policies.
However, as with most analytical studies, their model solved one OD pair at
a time, and thus failed to explicitly consider network effects. In addition, to
achieve modeling tractability, assumptions such as homogenous service and
symmetric costs/demands have often been imposed, and their influence has
often been difficult to predict. This approach provides strong theoretical jus-
tification for liberalization, and may be used to simulate the effects of the
progressive liberalization of particular routes. However, it faces great limita-
tions in quantifying the outcome of any major liberalization event.
When detailed data are available, the econometric approach can be
applied to estimate dynamic oligopoly models. Sample studies on the avia-
tion industry include those by Berry (1990, 1992), Berry, Carnell, and
Spiller (2006), Lederman (2007), Aguirregabiria and Ho (2012), and Fu,
Oum, and Yan (2014). Because parameters are empirically estimated for
airlines that compete over actual aviation networks, these models reflect
the reality in the market with minimal regularity assumptions. Therefore,
they could be ideal modeling tools for practical policy simulations.
However, econometric estimation requires extremely rich and detailed data,
which are usually not available. As a result, few of the liberalization studies
using this approach have been made available in the public domain. More
importantly, as there are no data on potential entrant airlines until they
actually serve the routes under investigation, it is not straightforward to
use such models to investigate the effects of liberalization ex ante.
Subjective assumptions must be introduced in policy simulations.
The computational network approach follows modeling principles simi-
lar to those of the analytical approach. However, with the integration of
optimization modules leading to computable solutions, this approach is
capable of handling relatively realistic airline networks. Sample studies
include those of Hansen (1990), Hong and Harker (1992), Dobson and
Lederer (1993), Lederer and Nambimadom (1998), Adler (2001, 2005), Yan
and Wang (2001), Hsu and Wen (2003), Yang (2008), Li et al. (2010), and
Adler, Pels, and Nash (2010, 2014). Among these, two studies were
designed to evaluate the effects of liberalization on traffic volume and eco-
nomic well-being. Li et al. (2010) developed a model for optimizing the
allocation of additional routes in a liberalizing airline market, where air-
port capacity may be constrained. Their model was useful in modeling gov-
ernment policy and airline competition in a progressively liberalized
aviation market. However, although the model considered airport capacity
constraints, the quasi-variational inequality solution adopted in the study
implied that every airline receives the equivalent marginal value from air-
port slots in equilibrium, which is only applicable in competitive markets.
In reality, airlines either buy/sell a limited number of slots (e.g., the case of
London Heathrow Airport) or mostly “grandfathered” airport slots under
the principal of precedence. Adler et al. (2014) developed an integrated
model over large-scale aviation markets and inter-modal networks while
explicitly considering airport constraints. More importantly, the model’s
parameters were mostly calibrated/estimated using data observed in the
markets. However, the study had some major practical and theoretical lim-
itations. Due to data availability, some of its key parameters were cali-
brated based on previous studies rather than empirically estimated. In
addition, only the largest representative airlines were considered in the
model due to the large network involved. In addition, Adler et al. (2014)
did not analytically prove the existence and uniqueness of the Nash
Equilibrium. While this could be addressed practically by performing more
simulations using different initial values, it remains a theoretical challenge
to be addressed.
Each approach has its strengths and weaknesses in evaluating the effects
of liberalization and deregulation. Subjective assumptions are almost inevi-
table, even for markets with rich industry data. Therefore, a reality check
and a good understanding of the aviation market are musts for modelers
and decision makers. After all, precisely predicting a market equilibrium
that includes new entrants must involve some uncertainty. Although these
quantitative studies are currently the best options available, decision
makers must keep their limitations in mind.
SUMMARY AND CONCLUSION
Although numerous studies have found that air service liberalization has
produced positive economic gains and social welfare improvements over
the past three decades, most international aviation markets are not covered
by open-skies agreements. Air transport remains one of the most regulated
industries in international market compared with other service sectors such
as trade, banking, telecommunication, insurance and retail. Even the cur-
rent liberalizations are often partial and incomplete, applying restrictive
regulations to foreign ownership, beyond rights and route entry. There is a
need for more advanced studies of air service liberalization. This chapter
provides an overview of the liberalization literature. Our review leads to
the following conclusions.
There is strong evidence, in our view beyond reasonable doubt, that air
service liberalization introduces substantial economic benefits to the coun-
tries involved. In the airline industry, liberalization has led to increased air-
line competition, decreased average fares, increased frequency, improved
load factor and airline productivity, increased traffic volumes and new
route services. A more competitive and efficient airline industry not only
leads to increased employment and economic output in the aviation sector
itself, but also benefits other related industries such as tourism, logistics
and trade. As a result, the overall economy benefits from air transport lib-
eralization. However, the costs and benefits are not evenly distributed
among the countries and firms involved. In particular, the profitability of
airlines has not improved in general. Some less-competitive airlines may
eventually leave the market. Certain airports and communities may see a
reduction in direct services. Nevertheless, compared with the gains of the
overall economy, such losses tend to be small. A more competitive and
dynamic market often promotes business innovation and helps competitive
start-up firms. It is usually difficult to predict which airline/airport will
emerge victorious. Carriers in an emerging economy (e.g., AirAsia in
Malaysia) could prevail, and new hubs may emerge as strong competitors
(e.g., Dubai and Kuala Lumpur).
There is a two-way relationship between air service liberalization and
the airport industry. The availability of adequate, efficient and competitive
airport services is a necessary condition for air service liberalization to deli-
ver any material gains. Airport dominance has long been a source of mar-
ket power for hub carriers, and airports themselves hold significant market
power, especially over local traffic. Therefore, the important roles of air-
ports should be recognized in liberalization initiatives. Where airport capa-
city is a constraint, the provision of landing slots for foreign entrants may
be included in the liberalization process. However, liberalization introduces
significant uncertainty to airport revenue and demand. Some airports may
lose hub carriers due to airline mergers/consolidations. As a result,
privatized/commercialized airports should and will proactively participate

in liberalization to maximize their financial returns or secure their long-
term business. Many have chosen to form vertical arrangements with their
airlines. The effects of such vertical arrangements on competition and wel-
fare are complex and dependent on market structures. Few studies have
systematically investigated optimal airport policies within a liberalization
framework. More regulatory attention and additional academic studies are
required.
Most studies on air transport liberalization have comprised either ex
post evaluations or qualitative theoretical models. While they offer very
valuable insights into the effects of a particular policy, they do not provide
analytical tools that quantify/forecast the market outcomes of alternative
policies ex ante. In general, three quantitative modeling methods could be
used to evaluate air transport liberalization, including the analytical
approach, econometric approach and computational network approach.
The analytical approach typically models airline competition over a single
OD pair or simplified/stylized networks, and thus is mathematically tract-
able. However, it is usually difficult to study large-scale, complex networks.
When detailed data are available, the econometric approach could be
applied to estimate dynamic oligopoly models. As parameters are empiri-
cally estimated for airlines that compete over actual aviation networks,
they could be ideal modeling tools for practical policy simulation.
However, econometric estimation requires extremely rich and detailed data,
which are usually not available. The computational network approach fol-
lows modeling principles similar to those of the analytical approach. With
the integration of optimization modules leading to computable solutions,
this approach is capable of handling relatively realistic airline networks.
However, model calibration problems persist, and the existence and
uniqueness of the Nash Equilibrium may not be proven. Overall, each
approach has its strengths and weaknesses, and thus a reality check and a
good understanding of the aviation market are musts for modelers and
decision makers.
Every country faces unique problems when it comes to air transport lib-
eralization. Comprehensive studies on the specific markets under investiga-
tion are therefore required in addition to advances in the general modeling
techniques for air transport competition and regulation. However, given
the strong evidence advanced by numerous studies, we believe that liberali-
zation is beneficial to the countries involved in the long term. The biggest
challenges to liberalization are often political will, foreign policy and resis-
tance from vested interests. In such cases, a more practical approach for
governments is to push forward with liberalization in sequential stages with

a definite road map. Periodical reviews can be conducted, as they not only
provide updated assessments to the decision makers, but also maintain
pressure on those who are against liberalization. Indeed, in the United
States, any airline or airport that is against liberalization must submit a
report explaining why the planned liberalization would harm public inter-
ests. Any government that remains hesitant in implementing air service lib-
eralization should consider following such an approach.
NOTES
1. Deregulation of the domestic markets took place earlier. Domestic markets in

the United States were fully deregulated in 1978. The world’s first open-skies agree-
ment was signed in 1992 between the Netherlands and the United States.
2. In the United States, whereas the CAB regulated inter-state routes as a federal
agency, intra-state markets were largely under the control of the individual states.
In general, higher degrees of competition and innovation were allowed on these
intra-state routes, and new start-up airlines such as Southwest were able to intro-
duce new business models and discount fares. As a result, the average yields on
intra-state routes were much lower than those on the inter-state routes, and CAB
deregulation was promoted.
3. That is, Calculated Residual Growth = Actual (Predicted) Traffic
GrowthGDP Growth × Income Elasticity.
4. However, one should be aware that if the quality and liberalization effects are
not explicitly controlled, using the overestimated income elasticity may provide a
better data fit than the “true” income elasticity in forecasting future demand.
5. http://centreforaviation.com/analysis/delta-air-lines-improves-china-position-
with-better-slots-from-partners-china-eastern southern-106837.
ACKNOWLEDGMENT
The authors gratefully acknowledge the research grant support of Social

Science and Humanities Research Council of Canada.
REFERENCES
Aaronson, R. (2003). What liberalization means for airports. Presentation at the 2003
ATConf/5 Conference, seminar prior to the ICAO Worldwide Air Transport
Conference.
Adler, N. (2001). Competition in a deregulated air transportation market. European Journal of

Operational Research, 129(2), 337345.
Adler, N. (2005). The effect of competition on the choice of an optimal network in a liberalized
aviation market with an application to Western Europe. Transportation Science, 39(1),
5872.
Adler, N., & Berechman, J. (2001). Evaluating optimal multi-hub networks in a deregulated
aviation market with an application to Western Europe. Transportation Research Part
A, 35(5), 373390.
Adler, N., Fu, X., Oum, T. H., & Yu, C. (2014). Air transport liberalization and airport slot
allocation: The case of the Northeast Asian transport market. Transportation Research
Part A, 62, 3–19.
Adler, N., & Hashai, N. (2005). Effect of open skies in the middle eastern region.
Transportation Research Part A, 39(10), 878894.
Adler, N., Pels, E., & Nash, C. (2010). High-speed rail and air transport competition: Game
engineering as tool for cost-benefit analysis. Transportation Research Part B, 44,
812833.
Aguirregabiria, V., & Ho, C.-Y. (2012). A dynamic oligopoly game of the US airline industry:
Estimation and policy experiments. Journal of Econometrics, 168(1), 156173.
Aizenman, J. (2004). Endogenous pricing to market and financing cost. Journal of Monetary
Economics, 51(4), 691712.
Barbot, C. (2011). Vertical contracts between airports and airlines: Is there a trade-off between
welfare and competitiveness? Journal of Transport Economics and Policy, 45(2), 277–
302.
Berry, S. (1990). Airport presence as product differentiation. American Economic Review, 80,
394399.
Berry, S. (1992). Estimation of a model of entry in the airline industry. Econometrica, 60,
889917.
Berry, S., Carnall, M., & Spiller, P. (2006). Airline hubbing, costs and demand. In D. Lee
(Ed.), Competition policy and anti-trust (Vol. 1, pp. 183214). Advances in Airline
Economics. Bingley, UK: Emerald Group Publishing Limited. Available at http://www.
amazon.com/Advances-Airline-Economics-Volume-Competition/dp/0444518436
Booz Allen Hamilton Ltd. (2007). The economic impacts of an open aviation area between the
EU and the US: A report prepared for Directorate General Energy and Transport of
European Commission.
Borenstein, S. (1989). Hubs and high fares: Dominance and market power in the U.S. airline
industry. RAND Journal of Economics, 20, 344365.
Borenstein S., & Rose N. L. (2007). How airline markets work … or do they? Regulatory reform
in the airline industry. NBER Working Paper 13452.
Brueckner, J. K. (2004). Network structure and airline scheduling. Journal of Industrial
Economics, 52(2), 291312.
Brueckner, J. K., Dyer, N. J., & Spiller, P. T. (1992). Fare determination in airline hub-and-
spoke networks. RAND Journal of Economics, 23, 309333.
Button, K. (2009). The impact of US-EU “open skies” agreement on airline market structures
and airline networks. Journal of Air Transport Management, 15, 5971.
Button, K., Lall, S., Stough, R., & Trice, M. (1999). High technology employment and hub
airports. Journal of Air Transport Management, 5, 5359.
CAPA. (2013). Delta leverages its strength in Seattle and Alaska Airlines partnership to bolster
service to Asia. Aviation Analysis, 2013.
Caves, R. (1962). Air transport and its regulators: An industry study. Cambridge, MA: Harvard
University Press.
Chung, J. Y., & Whang, T. (2011). The impact of low cost carriers on Korean Island tourism.
Journal of Transport Geography, 19, 13351340.
D’Alfonso, T., & Nastasi, A. (2012). Vertical relations in the air transport industry: A facility-
rivalry game. Transportation Research Part E, 48, 993–1008.
De Neufville, R., & Barber, J. (1991). Deregulation induced volatility of airport traffic.
Transportation Planning and Technology, 16(2), 117128.
Djankov, S., Freund, C., & Pham, C. S. (2005). Trading on time. World Bank Policy Research
Working Paper 3909.
Dobruszkes, F. (2009). Does liberalization of air transport imply increasing competition?
Lessons from the European case. Transport Policy, 16, 2939.
Dobson, G., & Lederer, P. J. (1993). Airline scheduling and routing in a hub-and-spoke sys-
tem. Transportation Science, 27(3), 281297.
Donzelli, M. (2010). The effect of low-cost air transportation on the local economy evidence
from Southern Italy. Journal of Air Transport Management, 16(3), 121126.
Doove, S., Gabbitas, O., Nguyen-Hong, D., & Owen, J. (2001). Price effects of regulation:
International air passenger transport, telecommunications and electricity supply.
Productivity Commission Staff Research Paper, AusInfo, Canberra October.
Dresner, M., Lin, J., & Windle, R. (1996). The impact of low-cost carriers on airport and route
competition. Journal of Transport Economics and Policy, 30, 309328.
Dresner, M., Windle, R., & Yao, Y. (2002). Airport barriers to entry in the US. Journal of
Transport Economics and Policy, 36(2), 389405.
Elwakil, O. S., Windle, R., & Dresner, M. (2013). Transborder demand leakage and the US-
Canadian air passenger market. Transportation Research Part E, 57, 4557.
Encaoua, D. (1991). Liberalising European airlines: Cost and factor productivity evidence.
International Journal of Industrial Organization, 9, 109124.
European Low Fares Airlines Association. (2004). Liberalisation of European air
transport: The benefits of low fares airlines to consumers, airports, regions and the
environment.
European Union Committee of the Regions. (2004). Own-initiative opinion on low cost air-
lines and territorial development. Opinion of the Committee of the Regions on the
“Third Report on Economic and Social Cohesion” presented to the 55th Plenary
Session, Official Journal of the European Union, Brussels.
Fethi, M. D., Jackson, P. M., & Wayman-Jones, T. G. (2000). Measuring efficiency of
European airlines: An application of DEA and Tobit analysis. Discussion Paper,
Efficiency and Productivity Research Unit, University of Leicester, UK.
Forsyth, P. J., Hill, R. D., & Trengrove, C. D. (1986). Measuring airline efficiency. Fiscal
Studies, 7, 6181.
Fu, X., Dresner, M., & Oum, T. H. (2011). Effects of transport service differentiation in the
U.S. domestic airline market. Transportation Research Part E, 47(3), 297305.
Fu, X., Homsombat, W., & Oum, T. H. (2011). Airport-airline vertical relationships, their
effects and regulatory policy implications. Journal of Air Transport Management, 17,
347353.
Fu, X., Lijesen, M., & Oum, T. H. (2006). An analysis of airport pricing and regulation in the
presence of competition between full service airlines and low cost carriers. Journal of
Transport Economics and Policy, 40(3), 425447.
Fu, X., Oum, T. H., & Yan, J. (2014). An analysis of travel demand in Japan’s inter-city mar-
ket: Empirical estimation and policy simulation. Journal of Transport Economics and
Policy, 48(1), 97113.
Fu, X., Oum, T. H., & Zhang, A. (2010). Air transport liberalization and its impacts on airline
competition and air passenger traffic. Transportation Journal, 49(4), 2441.
Fu, X., & Zhang, A. (2010). Effects of airport concession revenue sharing on competition and
social welfare. Journal of Transport Economics and Policy, 44(2), 119138.
General Accounting Office (GAO). (1989). Barriers to competition in the airline industry. T-
RCED-89-65, Published September 20, 1989.
General Accounting Office (GAO). (1990). Airline competition: Industry operating and market-
ing practices limit market entry. RCED-90-147, Published August 29, 1990.
General Accountability Office (GAO). (2008). Airline industry: Potential mergers and acquisi-
tions driven by financial and competitive pressures. GAO-08-845, July 31, 2008.
Gillen, D., Harris, R., & Oum, T. H. (2002). Measuring the economic effects of bilateral liber-
alization air transport. Transportation Research Part E: Logistics and Transportation
Review, 38(34), 155174.
Goetz, R. A., & Vowles, M. T. (2009). The good, the bad, and the ugly: 30 years of US airline
deregulation. Journal of Transport Geography, 17, 251263.
Gonenc, R., & Nicoletti, G. (2000). Regulation, market structure and performance in air passen-
ger transportation. OECD Economics Department Working Paper No (p. 254). Paris:
OECD.
Graham, A., & Dennis, N. (2010). The impact of low cost airline operations to Malta. Journal
of Air Transport Management, 16, 127136.
Hansen, M. (1990). Airline competition in a hub-dominated environment: An application of
non-cooperative game theory. Transportation Research Part B, 24(1), 2743.
Harrigan, J. (2005). Airplanes and comparative advantage. NBER Working Paper No. 11688.
Hausman, W. H., Lee, L. L., & Subramanian, U. (2005, October). Global logistics services, sup-
ply chain metrics and bilateral trade patterns. World Bank Policy Research Working
Paper 3773.
Hendricks, K., Piccione, M., & Tan, G. F. (1997). Entry and exit in hub-spoke networks.
RAND Journal of Economics, 28(2), 291303.
Hendricks, K., Piccione, M., & Tan, G. F. (1999). Equilibria in networks. Econometrica, 67(6),
14071434.
Homsombat, W., Fu, X., & Agachai, S. (2010). Policy implications of airline performance indi-
cators: Analysis of major North American airlines. Transportation Research Record,
2177, 4148.
Homsombat, W., Lei, Z., & Fu, X. (2011). Development status and prospects for aviation
hubs A comparative study of the major airports in South-East Asia. Singapore
Economic Review, 56(4), 573591.
Hong, S., & Harker, P. T. (1992). Air traffic network equilibrium: Toward frequency, price
and slot priority analysis. Transportation Research Part B, 26(4), 307323.
Hsu, C., & Wen, Y. (2003). Determining flight frequencies on an airline network with demand
supply interactions. Transportation Research Part E: Logistics and Transportation
Review, 39(6), 417441.
Hummels, D. (2006). Global trends in trade and transportation. ECMT-OECD Paper pre-
sented at the 17th International Symposium on Theory and Practice in Transport
Economics and Policy, Berlin, October 2527, 2006.
IATA. (2007). Airline liberalisation Lessons from other industries on the impact of remov-
ing operational, ownership and control restrictions. IATA Economic Briefing No. 7.
IATA. (2008). The impact of recession on air traffic volumes. Economic Briefing.
Inglada, V., Coto-Millan, P., & Rodriguez-Alvarez, A. (1999). Economic and technical effi-
ciency in the world air industry. International Journal of Transport Economics, 26(2),
219236.
Inglada, V., Rey, B., Rodrıguez-Alvarez, A., & Coto-Millan, P. (2006). Liberalisation
and efficiency in international air transport. Transportation Research Part A, 40,
95105.
International Civil Aviation Organization (ICAO). (2007). Overview of trends and develop-
ments in international air transport. Retrieved from http://www.icao.int/icao/en/atb/
epm/Ecp/OverviewTrends.pdf
International Civil Aviation Organization (ICAO). (2013). Liberalization of market access.
Working Paper presented by the United States of America, ATConf/6-WP/60.
InterVISTAS. (2006). The economic impact of air service liberalisation. Retrieved from http://
www.intervistas.com/4/reports/2006-06-07_EconomicImpactOfAirServiceLiberalization_
FinalReport.pdf
Irwin, M., & Kasarda, J. (1991). Air passenger linkages and employment growth in U.S.
metropolitan areas. American Sociological Review, 56, 524537.
Jordan, W. A. (1970). Airline regulation in America: Effects and imperfections. Baltimore:
Johns Hopkins Press.
Keeler, T. E. (1972). Airline regulation and market performance. Bell Journal of Economics
and Management Science, 3, 399424.
Lau, Y. Y., Lei, Z., Fu, X., & Ng, A. (2012). The implications of the re-establishment of direct
links across the Taiwan Strait on the aviation industries in Greater China. Research in
Transportation Economics, 35, 312.
Lederer, P. J., & Nambimadom, R. S. (1998). Airline network design. Operations Research,
46(6), 785804.
Lederman, M. (2007). Do enhancements to loyalty programs affect demand? The impact of
international frequent flyer partnerships on domestic airline demand. RAND Journal of
Economics, 38(4), 11341158.
Leidner, R. (2004). A multi-sector with dynamic markets structures, developments and impor-
tance for Europe’s economy. Report prepared for the Enterprise DG (Unit D.3) of the
European Commission.
Levine, M. E. (1965). Is regulation necessary? California air transportation and national regu-
latory policy. The Yale Law Journal, 74, 14161447.
Levine, M. E. (1987). Airline competition in deregulated markets: Theory, firm strategies and
public policy. Yale Journal on Regulation, 4, 283344.
Li, Z. C., Lam, H. K., Wong, S. C., & Fu, X. (2010). Optimal route allocation in a liberalizing
airline market. Transportation Research Part B, 44(7), 886902.
Lijesen, M., Rietveld, P., & Nijkamp, P. (2001). Hub premiums in European civil aviation.
Transport Policy, 8(3), 193–199.
Morrison, S. A. (2001). Actual, adjacent, and potential competition: Estimating the full effect
of Southwest Airlines. Journal of Transport Economics and Policy, 35, 239256.
Morrison, S. A., & Winston, C. (2000). The remaining role for government policy in the
deregulated airline industry. In S. Peltzman & C. Winston (Eds.), Deregulation of
Network Industries. Washington, DC: The Brookings Institution.
Oum, T. H., Fu, X., & Yu, C. (2005). New evidences on airline efficiency and yields: A com-
parative analysis of major North American air carriers and its implications. Transport
Policy, 12, 153164.
Oum, T. H., Fu, X., & Zhang, A. (2009). Air transport liberalization and its impacts on airline
competition and air passenger traffic. International Transport Forum 2009, Forum
Papers No. 4.
Oum, T. H., & Yu, C. (1995). A productivity comparison of the world’s major airlines.
Journal of Air Transport Management, 2, 181195.
Oum, T. H., & Yu, C. (1998). Winning airlines: Productivity and cost competitiveness of the
world’s major airlines. New York, NY: Kluwer Academic Press. Retrieved from http://
www.wkap.nl/prod/b/0-7923-8010-X
Oum, T. H., & Zhang, Y. (1990). Airport pricing: Congestion tolls, lumpy investment, and
cost recovery. Journal of Public Economics, 43, 353374.
Oum, T. H., Zhang, A., & Zhang, Y. (1995). Airline network rivalry. Canadian Journal of
Economics, 28, 836857.
Pels, E., Nijkamp, P., & Rietveld, P. (2001). Airport choice in a multiple airport region: A case
study for the San Francisco bay area. Regional Studies, 35(1), 19.
Pels, E., Nijkamp, P., & Rietveld, P. (2003). Access to and competition between airports: A
case study for the San Francisco bay area. Transportation Research Part A, 37(1),
7183.
Piermartini, R., & Rousova, L. (2008). Liberalization of air transport services and passenger
traffic. World Trade Organization Economic Research and Statistics Division, Staff
Working Paper ERSD-200806.
Rey, B., Myro, R., & Galera, A. (2011). Effect of low-cost airlines on tourism in Spain
A dynamic panel data model. Journal of Air Transport Management, 17, 163167.
Richards, K. (1996). The effects of Southwest Airlines on U.S. airline markets. Research in
Transportation Economics, 4, 33–47.
Schipper, Y., Rietveld, P., & Nijkamp, P. (2002). European airline reform: An empirical
welfare analysis. Journal of Transport Economics and Policy, 36, 189–209.
Schipper, Y., Nijkamp, P., & Rietveld, P. (2007). Deregulation and welfare in airline markets:
An analysis of frequency equilibria. European Journal of Operational Research, 178(1),
194206.
Shaw, S. (1982). Airline deregulation and the tourist industry. Tourism Management, 3(1),
4051.
Sørenson, F., & Dukes, A. (2005). EU/US air transport agreement Potential impacts on
Ireland. Chambers of Commerce of Ireland, Dublin.
Starkie, D. (2008). The airport industry in a competitive environment: A United Kingdom per-
spective. OECD/ITF Discussion Paper No. 200815.
Swan, W. (2007). Misunderstandings about airline growth. Journal of Air Transport
Management, 13, 38.
Thompson, I. B. (2002). Air transport liberalisation and the development of third level airports
in France. Journal of Transport Geography, 10, 273285.
US DOT (1993). The airline deregulation evolution continues: The Southwest effect. Office of
Aviation Analysis, US Department of Transportation.
Vancouver Airport Authority. (2000). YVR Impact Study (March, 2000).
Vowles, T. M., & Tierney, S. (2007). The geographic impact of “open skies” policies on trans-
Tasman air passenger service. Asia Pacific Viewpoint, 48(3), 344354.
Windle, R., & Dresner, M. (1995). The short and long run effects of entry on US domestic air
routes. Transportation Journal, 35, 1425.
World Trade Organization (WTO). (2006). Second review of the air transport annex:
Developments in the air transport sector (part two) Quantitative air services agree-
ments review (QUASAR), Volumes I and II. Note by the Secretariat, document S/C/
W/270/Add.1.
Yan, S., & Wang, C. (2001). The planning of aircraft routes and flight frequencies in an airline
network operations. Journal of Advanced Transportation, 35, 3346.
Yang, G., Zhang, A., & Fu, X. (2013). Determinants of airport-airline vertical arrangements
analytical results and empirical evidence. Paper presented at the 2013 ATRS
Conference, Bergamo, Italy.
Yang, T. H. (2008). Airline network design problem with different airport capacity constraints.
Transportmetrica, 4(1), 3349.
Zhang, A. (1996). An analysis of fortress hubs in network-based markets. Journal of Transport
Economics and Policy, 30, 293308.
Zhang, A., Fu, X., & Yang, G. (2010). Revenue sharing with multiple airlines and airports.
Transportation Research Part B, 44(89), 944959.
Zhang, A., & Wei, X. (1993). Competition in airline networks: The case of constant elasticity
demands. Economics Letters, 42, 253259.
CHAPTER 3
GOVERNMENT REGULATION OF
INTERNATIONAL AIR
TRANSPORTATION
Darren Prokop
ABSTRACT
This chapter examines the unique regulatory environment that trans-

border air carriers work within. Using a U.S. perspective the concept of
the bilateral air agreement is outlined and discussed. These agreements
form the basis for how two countries decide to share their airspaces
among their air carriers. The trend has been toward more liberal
approaches. To explain this trend the concepts of the Freedoms of the
Air and Open Skies are discussed. Other liberalization programs are also
discussed; specifically, co-terminalization and cabotage. Finally, the air
cargo transfer operations at Ted Stevens Anchorage International
Airport are used as an example to highlight a rare example of unilateral
liberalization on the part of the United States.
Keywords: Air transportation; economics of regulation; bilateral air
agreements
JEL classifications: L93; L51

ISSN: 2212-1609/doi:10.1108/S2212-160920140000004002
45
46 DARREN PROKOP
INTRODUCTION: AIR TRANSPORTATION

REGULATION IN CONTEXT
Every mode of commercial transportation is subject to some degree of gov-

ernment regulation. These regulations apply to the configuration and usage
of the conveyance and the qualifications and duties of operators and sup-
porting workers. They may also apply to what services can be provided
and how the transportation network is designed. Finally, government regu-
lation may also apply to how services are priced, what type of insurance
the carrier must hold, etc. In an international context, transportation regu-
lation is further complicated because origin and destination of the carrier
are in two different countries. On top of this, the countries traversed in the
performance of service may or may not be signatories to international
agreements applicable to that mode of transport. Nevertheless, each mode
of transport is subject to different types of government regulation. As it
turns out, air transportation is subject to a unique body of regulations.
A foreign-flagged vessel is transporting containers of manufactured
items from a port in Asia to the Port of Los Angeles. A truck is delivering
a load of automobile parts from Canada to a manufacturing plant in
Detroit. A train is delivering tankers laden with crude oil from Mexico to
refineries in the southern United States. As long as the foreign-based ocean
vessel, truck and locomotive (and rolling stock) meet domestic safety and
configuration requirements, the conveyances have freedom of entry and
exit (including chosen point of entry and chosen point of exit).
Government regulation over the transport network is very limited.
Suppose, instead, that the transport in question was natural gas in pipeline
into the United States from Mexico or Canada. In this case, the two gov-
ernments involved in origin and destination of the pipeline would have to
reach an agreement before a company (or consortium of companies) could
lay a trans-border pipeline. Of course, a pipeline, once built, is inflexible in
its routing so this sounds quite distinct from trans-border ocean vessel,
truck, and rail transport. As it turns out, trans-border air transport shares
this trait with the pipeline mode. In this sense, air transport is unique
among the other types of engine-based transportation.
Suppose Air France wishes to fly from Paris to Chicago, directly. British
Airways wishes to fly from London to Mexico City and overfly the United
States in the process. AeroMexico wishes to fly from Mexico City to
Madrid with a stop in Miami. Air Canada wishes to fly to Chicago and
then on to another city within the United States. Obviously, the air carrier
Government Regulation of International Air Transportation 47
in question would need to make sure it has acquired the necessary landing
slots at the chosen airports (and at the desired times) in order to fulfill its
route. Less obvious, however, is that any such route could only be contem-
plated after the governments in the origin and destination have first
reached an agreement on how to treat each country’s respective airlines
within their airspaces. Without a so-called bilateral air agreement in place,
all talk about commercial airlines entering this or that foreign city (these
many times) and leaving foreign airspace for home or some third country is
premature (and quite illegal if undertaken). Thus, just like for trans-border
pipeline operations, government-to-government negotiations decide the
context in which trans-border air transportation will take place.
This chapter will outline the unique regulatory environment that trans-
border air carriers work within as follows. A history of bilateral air agree-
ments will be provided and set in the context of competition within trans-
border air carrier markets. The nature of trans-border routes are further
clarified by providing an overview of the concepts known as Freedoms of
the Air and Open Skies. The trend toward market liberalization will also be
explained through defining activities known as co-terminalization and
cabotage. Finally, the air cargo transfer operations at Ted Stevens
Anchorage International Airport are used as an example to highlight a rare
example of unilateral liberalization on the part of the United States.
BILATERAL AIR AGREEMENTS

As of 2012 the United States had bilateral air agreements in place with 130
countries (within which are the 27 members of the European Union).
Numerous countries around the world are represented. While each bilateral
agreement is different, they do follow a template as will be discussed.
The commercial airline industry is truly a global one. Practically every
major city in every country in the world is connected to some air carrier’s
network. As such, every major city can reach every other major city
throughout the world via air transportation. This might suggest that com-
mercial airlines have been able to knit this vast collection of interconnected
networks of routes due to market forces alone. Actually, the truth is that
the knitting which has taken place has been conditioned by what govern-
ments allowed the carriers in terms of (if we can continue with the analogy)
how much yarn they are allotted and the stitching pattern they can use.
Despite the appearance of a completely interconnected global network of
48 DARREN PROKOP
routes, not every option is available to a commercial air carrier (and, con-
versely, that commercial air carrier may be protected from competition by
a foreign commercial air carrier). How restricted or how open a country’s
airspace is relative to foreign air carriers’ is indicated by the structure of
the bilateral air agreements it negotiates and the partners with which it
chooses to negotiate. Suffice it to say that no country allows foreign com-
mercial air carriers to compete head-to-head with its domestic industry
along domestic routes. Airspace is often looked upon as a strategic military
asset and business asset worthy of some degree of protection from foreign
entrants.
Unlike other modes of transport, commercial air carriers’ operational
planning is preceded by the regulations as set out in bilateral air agree-
ments. Basically, two governments reach an agreement as to what the com-
mercial air carriers may and may not do when in the partner’s airspace
and, sometimes, what it may or may not do before entering and/or depart-
ing that airspace. In other words, bilateral air agreements can address the
activities which take place in third-party countries. In the United States,
the Department of State takes the lead in negotiating trans-border air
transport policy with other countries. The Departments of Transportation
and of Commerce are involved too but only on an advisory level (as set out
in the United States Code 49 U.S.C. §40105).
The Conference on International Civil Aviation (1944), also known as the
“Chicago Convention,” gives sovereignty to every federal government over
its own airspace. The implications are: (1) use of airspace by foreign air-
craft (private or commercial) is subject to domestic government approval
and (2) commercial airlines cannot negotiate any agreement that involves
two or more distinct airspaces (such as an interlining agreement which is
common in other trans-border modes of transportation). The Chicago
Convention marks the setting of a legal framework for today’s interna-
tional commercial air travel.1 As such all bilateral air agreements in force
today share a common language that derived from the 1944 agreement.
The governing body for the Chicago Convention is known as the
International Civil Aviation Organization (ICAO) and is a part of the
apparatus of the United Nations. While the ICAO oversaw 57 countries in
1944, today the number is 190. While the Chicago Convention tried to
overcome the limitations of the Paris Convention (1919), the members
could not set aside political differences in order to devise a common eco-
nomic framework. The differences lay in terms of how operating rights into
and out of member countries, capacity utilization, and pricing policies
would be made. Should such policies be made by the ICAO (the U.K.’s
view at the time)? Or should they be made by market forces prevalent in

each country (the United States’ view at the time)? (Marchick & Newman,
2002, p. 445) These arguments still exist today.
The first bilateral air agreement between the United States and the
United Kingdom, using the ICAO’s protocols, came to be called Bermuda
I (since it was signed there in 1946). It would be superseded by Bermuda II
in 1977. Bermuda I became a template for other bilaterals. What distin-
guished it was its focus on three things: (1) capacity; (2) designation; and
(3) pricing. Also, this bilateral provided for ex post facto negotiation of
allowable capacity brought into each other’s airspaces. Presumably, these
discussions would center on how best to meet changes to consumer
demand, while at the same time balancing the interests of both countries’
air carrier industry. Designation meant that each country could name speci-
fic carriers which would be able to fly the agreed-to trans-border routes.
The agreement might allow the naming of one carrier or, perhaps, multiple
ones. Finally, pricing policy was such that the host country could consult
with the foreign carrier if the former was not satisfied with the latter’s
inbound and/or outbound air fares. If, however, an understanding is not
reached, then the host country may unilaterally adjust the fare.2 Bermuda
II came about in reaction to the U.K.’s displeasure with Bermuda I. The
major difference was the concern with over capacity on North Atlantic
routes; and, in that regard, the right of designation was allowed there but
phased out elsewhere. The formula for designation was set out in Annex 2
of the Bermuda II agreement.
The U.S. approach to bilateral air agreements was more market-based
than that of its U.K. partner. As the United States became to implement
deregulation of it domestic air carrier industry, this spirit also entered its
views on bilaterals. The new template would be known as Post-1977
Agreements. Here there would be no ex post facto revision of capacity
allowed into each partner’s airspace. Governments would be less able to
interfere once the bilateral is in place. Instead the carriers would let the
market itself determine if there was room enough for them along desig-
nated routes. Basically, no host country could unilaterally limit the volume
of traffic and the frequency of service provided by a partner’s carrier
(except for technical, not commercial, reasons). In terms of designation,
Post-1977 Agreements allowed the partner to name multiple air carriers.
This, in effect, allowed for more competition along the routes covered in
the bilateral. Finally, regarding pricing, a country could only unilaterally
change a partner carrier’s air fare if the flight in question originates in the
host country. In other cases, a disputed fare may stay in place unless and
50 DARREN PROKOP
until both partners agree to adjust it. This concept is known as double dis-
approval or mutual disapproval and contrasts with the effective veto one
partner has in the Bermuda I template (meaning double approval is
required to overcome a host country’s dispute with a partner’s carrier).
This more liberal view of bilateral air agreements coincided with deregula-
tion of the U.S. airline industry in 1978.
The Civil Aeronautics Board (CAB) was abolished by the Airline
Deregulation Act (1978). From 1938 to 1978 the CAB would regulate
domestic air carriers’ fares, routes and entry of new carriers into the mar-
ketplace. With deregulation, market forces would determine the nature of
these items. This would shake-up the industry in a variety of ways. Carriers
would work to lower their costs. Some start-ups would market themselves
as low-cost and keep frills to a minimum (e.g., Southwest Airlines). On the
other hand, national carriers with their monopoly over certain routes
would work to divide up the market into specific hubs. Dominating the
traffic into and out of specific airports allowed the national carriers to in
effect blunt some of the competitive forces they were facing in the early
years of deregulation. As some national carriers went out of business, it
simply allowed a handful of carriers to dominate transcontinental and
trans-border flights. So, despite the intent of deregulation to act as an
incentive to control costs, most national carriers to this day are wrestling
with cost control. On the other hand, the U.S. government pressed ahead
with more liberal approaches to bilateral negotiations. This would culmi-
nate into a concept known as “Open Skies” which will be discussed in
detail below.
FREEDOMS OF THE AIR
In an attempt to create a common set of terms when negotiating bilateral

air agreements the Chicago Convention established what has come to be
known as the first five freedoms of air. The first two are:
1st Freedom:
This is the right to fly over another country without landing (i.e.,
the freedom of peaceful transit).
2nd Freedom:
This is the right to make only a technical stop in another country for
repairs and/or refueling (i.e., the freedom to make non-traffic stops).
It should be noted that most countries within the Chicago

Convention grant these two freedoms to foreign carriers through
the International Air Services Transit Agreement. Out of the 190
members today 120 have adopted this agreement.3 The other three
of the five freedoms are:
3rd Freedom:
This is the right to carry revenue passengers/freight for drop-off in
a foreign country (e.g., from country A to B).
4th Freedom:
This is the right to carry revenue passengers/freight from a foreign
country for drop-off in the air carrier’s home country (e.g., from
country B to A).
5th Freedom:
This is the right to carry revenue passengers/freight between two or
more foreign countries as long as the flight originates and/or termi-
nates in the air carrier’s home country. An example of this would
be, in a three country setting, A to B to C or C to B to A.
Beyond the five freedoms of the air other freedoms would be
devised with each successive one implying more liberalized commer-
cial air travel granted to partner countries.
6th Freedom:
This is a combination of the 4th and 3rd Freedoms. The air car-
rier transports revenue passengers/freight via its home country
which is used as a layover point or drop-off/pick-up point. An
example of this in a three country setting would be B to A to C
or C to A to B.
There are two interpretations of 6th Freedom. Since the traffic
entered the foreign country (say C) from the home country (A), it
would be 3rd Freedom traffic; but since the point of origin was
another foreign country (in this case B), it would be 5th Freedom
traffic. Bilateral partners would have to agree on these points.
7th Freedom:
This is a variation of the 5th Freedom. An air carrier operates
entirely outside of its home country and transports revenue passen-
gers/freight between foreign countries. An example of this, in a
three country setting would be B to C to D or any combination of
these with the domicile at (A).
52 DARREN PROKOP
8th Freedom:
This is known as cabotage or, more specifically, consecutive cabo-
tage. An air carrier transports revenue passengers/freight between
two or more points within a particular foreign country and the
flight originates or terminates in home country (A), i.e., building
on the 5th Freedom or building on the 7th Freedom in different
countries.
A variation (sometimes called the 9th Freedom) involves just
cabotage traffic as part of the scheduled trip meaning all business
revenue from the flight is earned entirely within the foreign country.
This is also known as stand-alone cabotage.
Currently, none of the United States’ bilateral air agreements mention the
6th or 8th Freedoms. However, several of them provide for the 7th Freedom
in all-cargo operations. This helps air cargo integrators (i.e., carriers that
have both air and ground fleets thus offering courier services) like FedEx and
UPS to design routes around the world without having to include their
United States domicile as part of the route. In this way, service may remain
within key regions like the Far East.
AIR CARGO LIBERALIZATION
While bilateral air agreements set the foundation upon which air carriers
may design their trans-border routes, the rise of air cargo integrators in the
1970s (e.g., FedEx and UPS) has meant that the world has become an ever
more connected set of cargo routes. Couriered air freight is no longer con-
sidered a luxury; indeed, consumers are willing to pay vendors of books,
personal computers, clothing, etc. for one-three day delivery to their busi-
ness/residence from any distribution center anywhere in the world.
Liberalization of air cargo activities (within the confines of a bilateral)
serves to encourage competition and thereby lower the costs of such ser-
vice. This section discusses three key areas in air cargo liberalization.
Open Skies
A bilateral air agreement which allows for Open Skies is one which is cer-
tainly more liberal than Bermuda I or Bermuda II. Of course, it is impor-
tant to not read too much into the term. Open Skies does not mean that a
country’s airspace is completely open to carriers from its bilateral partner,

with their competing head-to-head with domestic carriers for domestic
cargo carriage. Doing so would imply they could engage in an activity
called cabotage (which will be discussed below).4 However, Open Skies
agreements do allow for head-to-head competition in import and export
cargo. Open Skies grants carriers of the bilateral partner: (1) entry into any
route of the country in question; (2) unlimited capacity and frequency
along those routes; (3) unlimited 5th Freedom (meaning arrival from or
departure to third countries); and (4) double disapproval of prices. So any
carrier from a partner country can fly into any domestic airport (assuming
it acquired the landing slot) with as much capacity and frequency as it can
secure. As of 2013, the United States had 110 Open Skies agreements out-
standing (see Open Skies Partners). Actually, the number is really 84 since
the 27-member European Union has been negotiating with the United
States as a single entity since 2008. The first Open Skies agreement was
signed between the United States and the Netherlands in 1992. From 1995
onward, the United States would form one or more Open Skies agreements
every year. The most recent is with Guyana (in force as of March 2013). It
is interesting to note that the United States and Mexico do not have an
Open Skies agreement in place, while the United States does have one
(since 1995) with its other NAFTA partner, Canada. Finally, 66 out of the
84 Open Skies agreements allow for 7th Freedom on all-cargo flights.
The United States’ current stance on bilateral negotiations is to seek
Open Skies. Of course, an Open Skies agreement (which is intended to lib-
eralize air traffic) is not a guarantee for partner carriers to be granted anti-
trust immunity (which, of course, serves to reward anti-competitive
behavior). Foreign countries have often sought this as a point of leverage
with the United States. Having antitrust immunity makes it easier for for-
eign carriers to set up strategic alliances or full-blown mergers with U.S.
carriers. Indeed, the first (1992) Open Skies agreement between the United
States and the Netherlands granted antitrust immunity to Northwest
Airlines and KLM. The two carriers could coordinate air fares even if it
were to the detriment of competing carriers along North Atlantic routes.5
The U.S.Japan Open Skies agreement of 2010 preceded a granting of
antitrust immunity to American Airlines with Japan Airlines and United
Airlines with All Nippon Airways.
The U.S. Department of Transportation’s (DOT) consideration of anti-
trust immunity and international alliance agreements has two steps. “First,
DOT determines whether an agreement ‘substantially reduces or eliminates
competition.’ If it does, then DOT must disapprove it unless DOT finds
54 DARREN PROKOP
that the agreement ‘is necessary to meet a serious transportation need or to

achieve important public benefits’ and there is no less anti-competitive
alternative. 49 U.S.C. §41309(b). If DOT approves an anti-competitive
agreement on those grounds, then it must exempt it from the antitrust
laws. 49 U.S.C. §41308(c). Second, if DOT finds that an agreement does
not reduce or eliminate competition and is consistent with the public inter-
est, then DOT must approve it. But exemption from the antitrust laws is
authorized only if it is required by the public interest; even then immunity
is authorized only ‘to the extent necessary to allow the person to proceed
with the transaction specifically approved by the order and with any trans-
action necessarily contemplated by the order.’ 49 U.S.C. §§41309(b) and
41308(b). DOT’s review thus encompasses both a competitive analysis of
the transaction and public interest considerations.” (Gillespie & Richard,
2011, pp. 56).
Thus, the term Open Skies is a bit of a misnomer. The result of Open
Skies agreements, while opening more routes to foreign competition, might
lead to some of them being dominated by one or two carriers.
Co-Terminalization
Co-terminalization is a form of air transport whereby a foreign carrier may

take-off and land at more than one airport within the host country. An
Open Skies agreement typically allows for such an operation.6 However,
the host country’s airports would have to be part of the same scheduled
route, meaning any and all host country airports preceding the final airport
in the host country are used as lay-over points. Passengers and cargo may
be allowed to be picked-up or dropped-off at these host country airports
but not both. A foreign carrier picking-up and dropping off passengers
and/or cargo in the host country would be performing cabotage (discussed
below) and which is not allowed unless specifically agreed-to in the bilateral
air agreement. This is why the co-terminalization operation is part of the
same route throughout. If it were not, then it would appear that the foreign
carrier is offering a distinct domestic route (with unique flight number)
from point A to B within the host country. To avoid confusion on the part
of potential passengers and cargo shippers, a single route and number for a
flight originating and/or terminating in a foreign country is necessary.
The U.S.Japan bilateral air agreement provides for Open Skies but not
7th Freedom in all-cargo operations. However, co-terminalization in all-
cargo operations is allowed. Japan designates two carriers to be incumbents
in JapanU.S. cargo flights and the United States designates three. These
incumbents face no restrictions on frequency and capacity. Each is allowed
to co-terminalize. Japan may designate one other carrier (considered a non-
incumbent) which will have the right to provide a frequency of 18 flights
per week and co-terminalize up to eight points in the United States. Of
course, Japan must inform the U.S. State Department of any and all co-
teminalized airports by this carrier at least 60 days before the service can
commence. The United States is allowed to designate one non-incumbent
as well; but it has the option to co-terminalize up to three points in Japan.
Cabotage
In terms of gauging how open a country is to foreign carrier operations,

cabotage represents the highest level of openness. Basically, a foreign car-
rier would have the same operational freedom as a domestic carrier. It is
for this reason that the granting of such an option is rare. Cabotage derives
from the French word caboter meaning coasting or to move from
coast-to-coast. Thus, the word has its origin in the ocean vessel mode. But
today, the word applies to any mode of transport and may be defined as:
transportation of domestic passengers/cargo by a foreign carrier
(i.e., operator and/or conveyance) from one point on domestic territory to
another. In this case, “territory” refers to land, water and airspace. Foreign
carriers transporting cargo into and out of a country are not engaged in
cabotage; but if they should pick-up and drop-off domestic cargo along the
way, this constitutes cabotage.
Transportation especially air transportation is often considered to
be a strategic industry deserving of protection from foreign competition.
The European Union (EU) has allowed 8th Freedom cabotage among its
27 members since 1997. Of course, the EU has established an economic
union and is attempting to progress toward a political union as well. This
relationship is not to be confused with free trade agreements which repre-
sent a lesser degree of economic integration. If the EU is striving to be a
so-called “United States of Europe” then 8th Freedom among its states
should not be surprising. After all, the issue of cabotage among U.S. states
never comes up because each is part of the same federal territory and the
air industry is subject to federal control and not individual state control.
It is the case that the U.S. Congress and/or the President may grant tem-
porary waivers to foreign carriers allowing them to engage in cabotage.
One case for a waiver is if something needs to be transported but there is
56 DARREN PROKOP
no available domestic carrier which is capable of hauling the load, for

example, when Volga-Dnepr Airlines provides charter services using the
large Antonov AN-124 freighter. These planes have flown oil rigs and space
launchers and are attractive for their hinged nose cones allowing for front
loading of heavy and bulky cargo. In the late 1990s, the company did per-
form some cabotage operations in the United States though these were
apparently without waivers being granted (Donoghue, 1999).
Air Cargo Transfer Operations in Anchorage
As this chapter has noted, the United States has wished to see more liberal-
ized, market-based bilateral air agreements. The United States has also
been at the forefront since World War II in negotiating free trade agree-
ments with a variety of countries. The United States, however, is not
known for unilateral liberalization of international trade. In fact, in the
ocean vessel sector, the United States has the world’s most stringent cabo-
tage regulations (Prokop, 2014). As noted above, the United States does
not allow for cabotage in any of its bilateral air agreements.
Despite the United States’ view toward reciprocity in trade, the air cargo
transfer operations allowed at Ted Stevens Anchorage International
Airport (abbreviated ANC) provide a unique example of unilateral trade
liberalization on the part of the United States. The history leading up to
this policy and an extensive treatment of the rules are found in Prokop
(2002, 2008). This section will provide a basic overview of the operations in
the context of the material discussed so far.
Basically, when a foreign air cargo carrier arrives at ANC, part or all
of its cargo may be transferred belly-to-belly from one airplane to
another. The other airplane may be another plane in the carrier’s fleet, a
domestic carrier, or a different foreign carrier. This operation is allowed
no matter if the foreign airplane is inbound to or outbound from the
United States. It is important to note that the transfer of air cargo must
take place between two airplanes that are already at ANC, meaning that
the cargo cannot be laid over.7 On the surface, this operation, when it
involves inbound U.S. traffic, appears to be cabotage; after all, it allows
a foreign carrier to pick-up air cargo in the United States (i.e., at ANC)
and then proceed to another airport in the contiguous United States.
Again, the United States has taken a liberal attitude regarding ANC and
does not consider the move to be cabotage, since the cargo did not really
lay-over.
An obvious benefit from the air cargo transfer at ANC is in taking

advantage of economies of fleet size. In effect, the interlining of air cargo
effectively creates more routes from a given number of airplanes. For
example, consider a Korean Air cargo flight from Incheon to Los
Angeles and a Japan Airlines cargo flight from Tokyo to New York
City. In both cases, the flights will land in ANC for refueling. Now sup-
pose they are dispatched such that they share layover time. If the
Korean Air (Japan Airlines) flight had cargo which was destined for
New York City (Los Angeles), it makes sense to transfer the cargo
between them. In effect, the two airplanes are offering four routes at
minimal cost relative to the extra revenue to be earned on more timely
and direct delivery.
The air cargo transfer operations at ANC make sense, because it is along
the Great Circle route for U.S.Asia air cargo. The economic trade-off
between carrying cost-incurring fuel and revenue-earning cargo is that
most carriers prefer to land at ANC and refuel. Currently, about 80% of
U.SAsia cargo by volume transits through ANC.
CONCLUSIONS
This chapter has demonstrated how air carriers’ plans within trans-border
markets are conditioned by government-to-government negotiations. It is a
unique regulatory environment when compared to other modes of trans-
portation. The bilateral air agreement is the outcome of these negotiations
and sets options and boundaries for what international air carriers may do.
The trend since World War II has been to liberalize air carrier activity.
This trend was discussed in the context of the so-called Freedoms of the
Air and in the context of Open Skies agreements. The history of the United
States, taken in terms of its relationships with its bilateral partners, was dis-
cussed in detail since the United States was the driving force in commercial
air liberalization. The concepts of co-terminalization and cabotage were
also discussed in order to illustrate the nature of any further liberalization
efforts. Of course, the lead has been taken at Ted Stevens Anchorage
International Airport because of the specific air cargo transfer options
allowed there. Such unilateral trade liberalization on the part of the United
States is a rare phenomenon. But, as was discussed, there is a still a long
way to go before it can be said that the skies are truly open to the benefit
of all international air carriers.
58 DARREN PROKOP
NOTES
1. The Chicago Convention finds its seeds in the “Paris Convention” or the
Convention Relating to the Regulation of Aerial Navigation (1919). Government
sovereignty over its own airspace was guaranteed among the signatories and each
member country’s nationally-registered aircraft were granted innocent passage over
the others’ airspace. Several countries were involved in the negotiations with the
major ones being: the United States, United Kingdom, France, Italy, Japan, and
China. Oddly enough, the United States signed but did not implement the agree-
ment, because it came out of the League of Nations Treaty which the U.S. Senate
chose not to ratify.
2. Of course, the host country is required to adjust the fare to a level no higher
than the lowest rate charged by its own airline(s) with similar service over the same
points. See Bermuda I, Article 10, Section G.
3. Canada is currently not a member of this agreement. “Canada signed the
Agreement on 10 February 1945 and deposited an instrument of acceptance thereto
on the same date with the Government of the United States of America. On
12 November 1986, a notice of denunciation of the Agreement by the Government
of Canada was received by the Government of the United States of America which
was to have taken effect on 12 November 1987. However, this notice was revoked
by a note dated 10 November 1987. By a second note dated 10 November 1987, the
Government of Canada gave a new notice of withdrawal from the Agreement,
which took effect on 10 November 1988.” (International Air Services Transit
Agreement). Of course, Canada’s proximity to the large U.S. market, its expansive
size, and having three oceans on its other sides gives it strong negotiating power
when it comes to such a unique airspace.
4. In addition to prohibiting cabotage, the United States also places restrictions
of foreign ownership of domestic airlines. Currently, the restriction is: 25% of vot-
ing equity; one-third of board at maximum; but the chairman of the board must be
a U.S. citizen (Hsu & Chang, 2005).
5. The U.S. Department of Transportation (DOT) has the authority to immunize
air carriers from U.S. antitrust laws regarding their trans-border operations (under
49 U.S.C. §41308§41309). Congress has not granted DOT the power to grant
immunity to carriers regarding solely domestics operations. For a list of the carriers
granted antitrust immunity within the world’s largest air carrier alliances, see
Gillespie and Richard (2011; table 1, p. 22).
6. Currently, the U.S.-Dominican Republic bilateral air agreement does not pro-
vide for Open Skies (though the two countries did negotiate for it in 1999, the
Dominican Republic has failed to ratify those terms). However, both countries have
made accommodations for co-terminalization on specific routes. For example, a
carrier from the Dominican Republic may select two intermediate U.S. airports on
a route taking it from Miami to New York. The selection must be relayed to the
U.S. Department of State at least 60 days prior to start of service. A U.S. carrier
has parallel options and requirements on routes into the Dominican Republic (with
the two airports being in Santo Domingo and Puerto Plata).
7. For a discussion of how these operations might be enhanced through allowing
a layover of cargo, see Prokop (2011).
REFERENCES
Donoghue, J. A. (1999). Little big airline shop. Air Transport World, 36(11), 9395.
Gillespie, W., & Richard, O. M. (2011). Antitrust immunity and international airline alliances
(pp. 127). Washington, DC: Economic Analysis Group, Antitrust Division, U.S.
Department of Justice. Retrieved from http://www.justice.gov/atr/public/eag/267513.
htm.
Hsu, C. J., & Chang, Y. C. (2005). The influences of airline ownership rules on aviation poli-
cies and carriers’ strategies. Eastern Asia Society for Transportation Studies, 5,
557569.
International Air Services Transit Agreement. ICAO Website. Retrieved from http://www.
icao.int/secretariat/legal/List%20of%20Parties/Transit_EN.pdf. Accessed on April 2,
2013.
Marchick, D., & Newman, D. S. (2002). Overview: Policy and the law of international civil
aviation. In D. Jenkins (Ed.), Handbook of airline economics (2nd ed., pp. 443454).
New York, NY: McGraw-Hill.
Open Skies Partners. U.S. Department of State. Retrieved from http://www.state.gov/e/eb/rls/
othr/ata/114805.htm. Accessed on April 19, 2013.
Prokop, D. (2002). The logistics of air cargo co-mingling at Ted Stevens Anchorage
International Airport. Journal of Air Transport Management, 8(2), 109114.
Prokop, D. (2008). Air cargo’s cutting edge: Transfer flexibility in Anchorage. Canadian
Journal of Transportation, 1(2), 99110.
Prokop, D. (2011). Freight Transport in Alaska: An Economic “Tipping Point”. Canadian
Transportation Research Forum. Proceedings of the 46th Annual Conference.
pp. 749756.
Prokop, D. (2014). International transportation management. In D. Prokop (Ed.), The busi-
ness of transportation. Westport, CT: Praeger Publishers.
FURTHER READING
Air Service Rights in U.S. International Air Transport Agreements. (2011). Air Service Rights
in U.S. International Air Transport Agreements. Washington, DC: Air Transport
Association of America, Inc.
International Civil Aviation Organization. (2013). ICAO Website. Retrieved from http://www.
icao.int/Pages/default.aspx. Accessed on April 2, 2013.
CHAPTER 4
SKIES WIDE SHUT AN

ASSESSMENT OF INTERNATIONAL
AIR TRANSPORT LIBERALIZATION$
Pierre Latrille, Antonia Carzaniga and

Marta Soprana
ABSTRACT
In spite of the extensive literature on the regulation of air transport

services, until the development of the Quantitative Air Services
Agreements Review (QUASAR) methodology no systematic review
existed of the degree of liberalization granted through air services
agreements. The chapter lays out QUASARs key features, and pre-
sents the main results its application has generated. It then elaborates
on how the methodology could be further refined and extended to
other segments of the air transport industry yet uncovered. Based on
QUASAR, the chapter critically evaluates some commonly held beliefs
about the liberalization of international passenger transport and then
moves on to explore the technical feasibility of creating a liberal mul-
tilateral regime for air transport services. QUASAR has demonstrated
$
The views expressed in this chapter are personal and do not bind either WTO
Members or the WTO Secretariat.

ISSN: 2212-1609/doi:10.1108/S2212-160920140000004003
61
62 PIERRE LATRILLE ET AL.
that, although the air transport sector has experienced some liberaliza-
tion over the past few years, this has been, overall, rather marginal.
The skies are not truly open.
Keywords: International Trade Agreements; openness; air
transportation
JEL classifications: F140; F130; L930
INTRODUCTION
International air transport is surprising in several regards. First, govern-

ments play an extensive role in setting commercial policies. Second, pub-
licly available economic data are rather scant. Thirdly, regulatory data are
difficult to obtain, in spite of the stringent transparency obligations applic-
able to a sector that relies on intergovernmental treaties that are subject to
legal and constitutional publication obligations.
Furthermore, the sector is ‘special’ in that it is the only service sector
that is not currently covered by the disciplines of World Trade
Organization (WTO). This exclusion is, however, subject to a review proce-
dure to consider the possible further application of the rules of the WTO’s
General Agreement on Trade in Services (GATS) to the sector.
As part of the process, developments in the sector are meant to be
assessed, including on the basis of background research by the WTO
Secretariat. Yet, at the outset of these research efforts, it proved impossible
to find any information assessing the degree of openness of the bilateral
agreements regulating the sector and, a fortiori, to gauge the average
degree of openness of countries’ aviation policies. Besides literary descrip-
tions of the US ‘open skies’ policy, the US internal deregulation, the intra-
EU deregulation and the Trans-Tasman liberalization, very little other
information existed: the rest of the world appeared, simply, not to exist.
However, a promising source of data was subsequently identified in the
International Civil Aviation (ICAO) World Air Services Agreements (WASA)
database. WASA contains all the Air Services Agreements (ASAs) that ICAO
contracting parties have registered with the organization pursuant to an obli-
gation set up by the Chicago Convention.1 However, the information con-
tained in the WASA had not been truly systematically exploited until then.
Against this background, Pierre Latrille and Antonia Carzaniga initially
developed the Quantitative Air Services Agreements Review (QUASAR)
Skies Wide Shut 63
methodology. In spite of its shortcomings, it has proven a useful instrument

of analysis. We will describe its characteristics and the results it has allowed
us to obtain in the second section. We will then elaborate on how it could
be refined in the third section, and, in the fourth section, how it might be
extended to other segments of air transport yet uncovered. In the fifth sec-
tion, we will discuss how liberalized international passenger transport is
and examine the feasibility of creating a liberal multilateral regime for air
transport services. Some concluding observations will be presented in the
sixth section.
MAPPING THE SKY: QUASAR
Background
As discussed, the desire to systematically assess the degree of liberalization

introduced by ASAs lies behind the development of QUASAR. In
QUASAR, agreements are evaluated both individually and collectively for
each country, by combining their regulatory features with passenger traffic
figures to assess a country’s overall aviation policy.
The QUASAR methodology was initially developed in 2006, applied
to 2005 data and distributed in paper format (the ‘original QUASAR’).2
It was subsequently disseminated on-line through the Air Services
Agreements Projector (ASAP), an application that not only displays rele-
vant ASA data but also charts corresponding information on a world
map.3 ASAP was initially released for QUASAR 2005 data.4 An update of
the application was subsequently released displaying 2011 data.5 An
upgraded version of ASAP, enabling the comparison of 2005 and 2011
data and including information about plurilateral air services agreements,
is scheduled for release late in 2014. The sections that follow will describe
the main elements making up QUASAR, by focusing first on the regula-
tory features of ASAs that were considered and then concentrating on the
traffic data used.
The Regulatory Features
As discussed, there is no single comprehensive information source about

ASAs. However, in May 2006 ICAO first published electronically the 2005
WASA database. The WASA contains codified summaries of the main pro-
visions of the bilateral ASAs that contracting parties of ICAO have regis-
tered with the Organization. The database has been subsequently updated
and expanded to include, in addition to the codified summaries, also the
text of the agreements.6 Since ICAO contracting states do not always com-
ply with their notification obligations (not all ASAs are notified, nor are
confidential memoranda accompanying those agreements), the agreements
contained in WASA do not give a complete picture of all bilateral agree-
ments in force. The WASA database also contains a number of outdated
agreements. Cases in point are the bilateral agreements concluded between
EC Member States (which were superseded by the Single Aviation Market
in 1993), which were excluded from the analysis. In total, 1,970 agreements
were assessed in the original QUASAR.
Regardless, WASA is by far the best and most homogeneous dataset on
bilateral air agreements. We therefore based the original QUASAR on
WASA, but began to expand its scope in subsequent updates. We pursued
different avenues, but most particularly we relied on the information sup-
plied by WTO Members in the course of the review of the exclusion of air
transport services from the scope of the GATS, the data collected through
the WTO’s Trade Policy Review Mechanism (when this was complete)7
and whatever information we could obtain through web searches. We thus
managed to expand the original sample of 1,970 agreements to 2,139 in the
revised 2005 sample, and 2,224 for the 2011 set, covering 184, 186 and 188
countries, respectively. The 2011 ASAP covers over 70 per cent of interna-
tional scheduled traffic.
The ASAs thus collected were coded in respect of a number of key mar-
ket access features with regard to international air passenger transport.
Specifically, among the numerous provisions in ASAs, the features deemed
to be of particular significance for market access that were retained were:
designation, withholding,8 tariffs, capacity, traffic rights, exchange of statis-
tics and co-operative arrangements. Table A.1 provides a description of
each feature and its variants. Weights were attributed to the various var-
iants of each of these provisions.9 These weights denote the marginal con-
tribution of each provision in liberalizing the air transport sector. Both the
features selected and the weights assigned were determined in consultation
with a panel of aviation professionals, government experts, international
civil servants and academics.10
On this basis, we computed an index, the Air Liberalization Index
(ALI). The ALI summarizes the various features of an ASA in a single
value, by assigning a weight to each provision considered. The value of the
Skies Wide Shut 65
ALI ranges between zero, for very restrictive ASAs, and fifty, for very lib-
eral ones such as a full common market between the parties.
We developed four different weighting methodologies to compute the
ALI, to take into account different countries’ situations. In addition to a
‘standard’ methodology, we devised one that gives more weight to fifth
freedom traffic rights (‘5th + ’, which might for instance be pertinent for
remotely located countries like Australia and New Zealand), one assigning
relatively more importance to liberal withholding/ownership provisions
(‘OWN + ’, which would be relevant for countries whose ‘national’ airline
is not nationally owned, such as Switzerland), and one putting more weight
on multiple designation provisions (‘DES + ’, which might be of significance
for countries that have more than one international airline, such as the
United Kingdom, the United States, China or Brazil). Table 1 illustrates
the features selected and weights assigned to compute the ALI according to
the four weighting systems.
Three elements in particular played a part in the expert panel’s decision
on the allocation of weights. First, it was agreed that, in order to determine
the maximum value of the ALI, an ‘ideal’ ASA, composed of the most lib-
eral variant for each market access feature, be used as a reference, rather
than the ‘liberal’ agreement most frequently encountered in reality. Second,
it was acknowledged that, as given ASA features tend be found in combi-
nation, their weight should be determined both in their joint as well as indi-
vidual impact. As will be discussed in more detail in the third section
below, for instance, restrictive features tend to be present together in
ASAs. Indeed, contrary to assertions that there is ‘an infinite variety’ of
ASAs, our analysis has shown that agreements with just seven combina-
tions of features account for roughly two-thirds of ASAP traffic. Third, it
was accepted that different views existed about the relative importance of
three features (i.e. fifth freedom traffic rights, the withholding/ownership
clause and the designation clause), and that these be reflected in the alterna-
tive weighting methodologies.
We also extended the application of the QUASAR methodology beyond
bilateral ASAs, to plurilateral agreements. Since the 1980s, there has been a
proliferation of plurilateral ASAs, whose relationship with pre-existing
bilateral ASA is not always clear. Nevertheless, plurilateral ASAs lend
themselves to the same assessment of openness through the application,
mutatis mutandis, of the methodology used for bilaterals.
In terms of information on plurilateral agreements, in the original
QUASAR we had relied on a list that had been drawn up by ICAO for the
Global Symposium on Air Transport Liberalization held in Dubai in
Table 1. QUASAR Weighing Methods.
66
Element Air Liberalization Index
Standard 5th + OWN + DES +
Grant of rights
Fifth freedom 6 12 5 5.5
Seventh freedom 6 5 5 5.5
Cabotage 6 5 5 5.5
Capacity
Pre-determination 0 0 0 0
‘Other restrictive’ 2 1.5 1.5 1.5
Bermuda I 4 3.5 3.5 3.5
‘Other liberal’ 6 5 5 5.5
Free determination 8 7 7 7.5
Tariffs
Dual approval 0 0 0 0
Country of origin 3 2.5 2.5 2.5
Dual disapproval 6 5 5 5.5
Zone pricing 8 4 7 3.5 7 3.5 7.5 3.5
5.5 5 5 5
PIERRE LATRILLE ET AL.

7 6 6 6.5
Free pricing 8 7 7 7.5
Withholding
Substantial ownership and effective control 0 0 0 0
Community of interest 4 3.5 7 3.5
Principal place of business 8 7 14 7.5
Designation
Single designation 0 0 0 0
Multiple designation 4 3.5 3.5 7.5
Skies Wide Shut
Statistics
Exchange of statistics 0 0 0 0
No exchange of statistics 1 1 1 1
Co-operative arrangements
Not allowed 0 0 0 0
Allowed 3 2.5 2.5 2.5
Total 50 50 50 50
Note: For all ALI variants, when the Zone Pricing provision is accompanied by another tariff clause provision (i.e. Dual approval, Dual
Disapproval or Country of Origin) the ALI weight is calculated as the average between the weights of the two provisions.
67
September 2006 (see International Civil Aviation Organization, 2006). We

subsequently reviewed this list to comprise only plurilaterals that were in
force at the end of 2005 for which the regulatory information was com-
plete. We then updated it in 2011 on the basis of the information available
with ICAO,11 complemented by information on the Multilateral Open
Skies Agreement for Member States of the Latin American Civil Aviation
Commission (LACAC)12 and fully standardized with the methodology
used to code bilateral ASAs.13
Although we applied the QUASAR methodology to plurilateral agree-
ments, we kept the analysis separate as in many instances there were doubts
about the effective implementation of plurilateral ASAs.14 Thus, whenever
air traffic relations between two individual countries were governed by
both a bilateral agreement and one (or more) plurilateral agreements, in
ASAP we provided information on all potentially applicable regimes. In
the calculation of the overall openness of a country’s aviation policy (see
the description of the WALI below), however, we restricted our analysis to
bilateral ASAs.
Traffic Data
Having calculated the degree of liberalization of individual ASAs via the

ALI, we then constructed an index to assess a country’s overall aviation
policy. It would of course have been possible to calculate a simple average
of the ALIs for all the agreements concluded by the country concerned.
But this would have given agreements concluded between aviation partners
with significant air traffic links the same importance as to agreements
between aviation partners without any air traffic. We therefore decided to
weigh the ALIs by the passenger traffic covered under the relevant agree-
ment, and compute a Weighted Air Liberalization Index (WALI).15 For
example, for country A, which has signed three bilateral agreements with
B, C and D, the WALI would be calculated as follows:
ðTA−B ×ALIASAðA−BÞ ÞþðTA−C ×ALIASAðA−CÞ ÞþðTA−D ×ALIASAðA−DÞ Þ

WALIA =
TA−B þTA−C þTA−D
where T denotes the traffic covered by the ASA in question.

Traffic data for the calculation of the WALI are scheduled passenger
traffic data, which were generously provided by the International Air
Skies Wide Shut 69
Transport Association (IATA). Although not an exact match for the

underpinning regulatory situation between the two countries concerned
(see below), the country-pair scheduled passenger traffic statistics provided
by IATA are by far the most comprehensive set of data available.
The IATA country-pair statistics are based on initial origin and final
destination of passengers, as determined by the ticket, rather than being
based on the on-flight origin and destination data. This means that the
IATA data will attribute to a given country-pair all passengers whose jour-
ney started and ended in the relevant countries, including those passengers
who did so indirectly, via a third country. A passenger travelling from the
United States to Burkina Faso under a single reservation will most likely
transit via France. IATA statistics will count this passenger as having flown
from the United States to Burkina Faso. In reality, however, the itinerary
is not governed by the USBurkina Faso ASA, but, rather, by the
USFrance and FranceBurkina Faso ones. As such, the traffic data do
not always correspond exactly to the ambit of a given bilateral agreement.
Yet, this limitation is essentially relevant only for thin routes without
direct traffic which, therefore, have a marginal statistical impact.16 In any
event, comprehensive statistics corresponding to the precise ambit of bilat-
eral agreements are either not publicly available to their full extent (as is
the case with ICAO OFOD statistics, which would have been the best-
suited data series but present serious reporting and confidentiality con-
straints17) or are just simply unaffordable (as would seem to be the case for
the Official Airline Guide data series).
Improvements in 2011
As discussed, the QUASAR methodology was originally applied to 2005

data. An update was produced with 2011 information and results were pub-
lished on-line through the ASAP application. However, the 2005 and 2011
datasets were not immediately comparable. They represented a ‘photogra-
phy’ of the situation in 2005, taken in 2006, and of the situation in 2011,
taken in 2012.
We therefore proceeded to review the 2005 dataset on the basis of the
ASA information for that year available to us in 2012. We amended the
2005 data with the newly acquired information about agreements that,
although already in force in 2005, had not been captured by the ICAO
WASA database at that time and, hence, not accounted for in our ‘original’
2005 computations. We thus added information on 177 newly discovered
agreements and on 397 agreements which had been amended before

1 January 2006. We additionally corrected 93 agreements, and excluded 9
from our sample.18 We performed the same exercise with the information
about plurilateral agreements, by revising the 2005 dataset on the basis of
the information available to us in 2012. We thus obtained two fully com-
parable sets, on the basis of which to assess the global evolution of aviation
policy between 2005 and 2011. The results of our analysis are presented in
the sections below.
Exploiting QUASAR
This part outlines some salient findings that emerge from the application of
the QUASAR methodology. We report exclusively on the results obtained
using the ‘standard’ weighing method,19 and compare the situation in 2005
(as revised in 2012) with that in 2011. Table A.2 details the results
obtained. Given that these findings are based on information that will be
made available through the ASAP application, we will be making use of
the term ASAP, as well as QUASAR, in our descriptions. This part is
divided into two sections: the first section describes the salient findings of
the analysis of bilateral agreements, whereas the second provides an over-
view of the situation with regard to plurilateral ASAs.
Bilateral ASAs
The Datasets. Let us start with a description of the data sample. ASAP
captures a total of 2,139 ASAs (out of the unknown total quantity) that
were in force at the end of 2005 and were concluded by 186 countries.20 In
terms of traffic, this accounts for 357 million international passengers.21 By
comparison, the 2011 dataset comprises 2,224 agreements (a 4 per cent
increase compared to the 2005 sample) concluded by 188 countries. More
specifically, between 2005 and 2011, 86 new agreements were signed and 56
were amended. The remaining 2082 ASAs did not undergo any changes
during the period under examination.
Total international passenger traffic increased by around 50 per cent
between 2005 and 2011, from around 496 million to 730 million passengers.
The share of that traffic accounted for by the agreements captured in
ASAP remained fairly stable, however, from 72.2 per cent in 2005 to 72.5
per cent in 2011, when the ASA dataset accounts for a total of nearly 530
million passengers. What this suggests is that, in spite of the 86 ‘brand new’
agreements that we were able to include to our sample in 2011, using traffic
Skies Wide Shut 71
as a very crude proxy, we were still missing about a quarter of all bilateral
ASAs.
Traffic Concentration. An analysis of the traffic covered by bilateral ASAs

reveals that, while in 2005 there were 67 agreements covering over one mil-
lion passengers (just over 3 per cent of the total), these have increased to
116 by 2011 (just above 5 per cent). These ‘heavy’ agreements involved
39 parties in 2005, which have grown to 56 by 2011. While they already
accounted for over half of the ASAP traffic in 2005, this share has further
increased to over 60 per cent by 2011.
Although more agreements have surpassed the one million passenger
mark in 2011, it would appear that the traffic-heaviest amongst them have
grown relatively less that average. Indeed, in 2011 the first 100 and 200
ASAs ranked by traffic cover globally a lower share of both total ASAs
and total ASAP traffic than in 2005 (Table 2). They also involve fewer
parties.
Table 2. Agreements by Traffic Ranking.

2005 2011
All agreements
Total ASAs 2,139 2,224
Total ASAP traffic 357,570,684 529,835,872
Total parties 186 188
Agreements covering over one million passengers
Total ASAs 67 116
Total traffic 197,578,550 324,451,316
Percentage of total ASAs 3.1 5.2
Percentage of total ASAP traffic 55.3 61.2
Total parties 39 56
First 100 agreements ranked by traffic
Total ASAs 100 100
Total traffic 225,070,935 307,408,608
Total parties 50 47
First 200 agreements ranked by traffic
Total ASAs 200 200
Total traffic 273,736,687 388,137,949
Total parties 79 76
Overall Openness of Bilateral ASAs. Overall, traffic weighted bilateral

agreements have become slightly more liberal over time. The WALI for the
entire ASAP sample has increased from 15.1 in 2005 to 15.3 in 2011. High-
traffic agreements appear to be relatively more liberal. In 2011, the first 100
agreements record a WALI of 17.6, whereas WALI for the first 200 is 16.9.
This mirrors the situation in 2005, where the high-traffic agreements were
relatively more open. Indeed, compared to 2005, while the WALI for the
first 100 agreements has remained unchanged from 2005 to 2011, the index
for the first 200 ASAs has increased from 16.7 to 16.9.
In order to put these figures into perspective, they need to be compared
to the ALIs for three archetypal ASAs. First, ‘traditional ASAs’, which
include third and fourth freedom traffic rights, pre-determination of capa-
city, dual approval of tariffs, substantial ownership and effective control,
single designation, exchange of statistics and no provision for co-operative
arrangements, score an ALI of zero. Second, ‘classical open skies’ agree-
ments, which comprise third, fourth and fifth freedom traffic rights, free
determination of capacity, free pricing, substantial ownership and effective
control, multidesignation, no exchange of statistics and provision for co-
operative arrangements, register an ALI of 30. Finally, ‘more than open
skies’ agreements, which are made up of third, fourth, fifth, seventh free-
dom and cabotage traffic rights, free determination of capacity, free pri-
cing, principal place of business (PPoB), multidesignation, no exchange of
statistics and provision for co-operative arrangements, record the highest
possible ALI score, 50. Against this background, the WALI registered by
high-traffic agreements might at first glance appear somewhat surprising.
Given that the parties to ‘classical open skies’ and ‘more than open skies’
agreements include the largest aviation markets (e.g. United States,
European Union, Singapore, Australia, New Zealand) one might have
expected the high-traffic ASAs to be substantially more open, in terms of
their WALIs, than smaller traffic ASAs.
However, a closer analysis provides at least two sets of explanations for
the above results. First, the outcome is affected by the absence of intra-EU
traffic, which would have scored a much higher ALI and weighed 254 mil-
lion passengers of intra-EU traffic in 2011. The exclusion of intra-EU traf-
fic lowers the WALI of the largest ASAs considerably. Second, for reasons
that will be explained below, plurilateral agreements are not factored in the
computation of the WALI. As a result, for instance, all the agreements
between EU Member States and the United States, some of which account
for very large passenger numbers, are all computed on the basis of their
bilateral ALIs, rather than the universally higher plurilateral ALI. Third,
Skies Wide Shut 73
a number of the agreements concluded by some of the ‘traffic heavy-

weights’ are not, as yet, governed by the most liberal provisions.22
Finally, when it comes to countries’ individual situations with regard to
the WALI situations vary, although an explanation for the way in which
the individual WALIs have changed is not always immediately available. In
2011, 118 countries have registered an increased or unchanged WALI com-
pared to 2005, whereas 70 have recorded a decrease. This is most probably
a result of the dramatic effects of the financial and economic crisis that
started in 2008.
Ali Ranges. In the six-year period under consideration, ASAs appear to

have evolved towards more openness, as indicated by the small but positive
shift towards more liberal ALI ranges. As shown in Chart 1, over 80 per
cent of ASAs have an ALI that ranges between 0 and 14, although this
share has slightly decreased from the 85.5 per cent registered in 2005.
Compared to 2005, significant changes have been registered for the most
liberal agreements. As a share of the total number of agreements, ASAs
with an ALI falling in the 2030 and 3150 ranges have increased, respec-
tively, by around 50 and over 30 per cent. This compares with an increase
of only 2 per cent for agreements with an ALI falling within the most
restrictive 04 range.
In terms of traffic accounted for by agreements falling within different
ALI ranges, the most liberal range has exhibited the highest increase.
However, the agreements falling within the most restrictive two ranges
have recorded the second and third highest increases. When looked at in
relation to the change in the number of agreements falling within those
ranges, these increases are even more significant. It is difficult to provide a
satisfactory explanation for these results. Clearly, the fact that these are
ranges of ALI values, rather than individual ALI values, may blur the pic-
ture somewhat. But the fact remains that, taken at face value, these
figures would suggest that traffic has increased relatively more under the
more restrictive agreements.
When it comes to regional specificities, the majority of agreements con-
cluded by countries in Africa and the Commonwealth of Independent
States (CIS) region has an ALI that ranges between 5 and 9. Within Asia
and Oceania, Europe, the Middle East and South and Central America and
the Caribbean (‘South America’ hereinafter) most ASAs fall within the
1014 ALI range. In the case of North America, instead, the largest num-
ber of agreements falls within the most liberal 2650 ALI range. Between
2005 and 2011, agreements falling in the 2025 ALI range have recorded
Chart 1. Agreements by ALI Ranges.
the most significant increases in all regions. The CIS region has experienced
a very significant decrease in ASAs falling within the most restrictive 04
ALI range.
In terms of traffic coverage, agreements in the CIS region falling within
the 04 range account for the highest percentage of traffic, representing
nearly 40 per cent of all traffic, while only representing just over 10 per
cent of the region’s ASAs. Similarly, the ASAs falling within the 2650
ALI range account for the smallest percentage of agreements of the region
but the highest percentage of traffic in both South America and Asia and
Oceania.
Market Access Features Prevalent in Bilateral Agreements. This section pre-

sents the results obtained for the individual market access features of bilat-
eral ASAs. In general, restrictive market access features still largely
Skies Wide Shut 75
dominate ASAs, and most particularly the substantial ownership and effec-
tive control clause, dual approval of tariffs, pre-determination of capacity
and absence of an explicit permission to enter into co-operative
agreements.
a. Traffic rights
A closer look at ASAs shows that, with regards to traffic rights, the
overall picture has not changed between 2005 and 2011: seventh free-
dom and cabotage are still very rare occurrences. Fifth freedom traffic
rights, on the other hand, are granted in two-thirds of all agreements
covering almost 80 per cent of passenger traffic. The only exception is in
the CIS region, where ASAs granting fifth freedom traffic rights account
for only 35 per cent of agreements and a roughly similar share of the
traffic. ASAs granting 5th freedom rights have increased the most in
South America followed by the Middle East between 2005 and 2011.
Seventh freedom rights are granted only in five agreements in 2011, either
in combination with fifth freedom traffic rights, cabotage or both, but never
as a stand-alone freedom.23 They cover a very marginal portion of passen-
ger traffic (0.2 per cent in 2011). Seventh freedom rights are found only
in ASAs signed by countries in Europe and Asia and Oceania. Cabotage
is granted only in eight agreements, concluded by countries in only
three regions: Asia and Oceania, Europe, and South America. Agreements
granting cabotage cover just over 1 per cent of passenger traffic.
b. Designation
In terms of designation, ASAs have displayed a small but significant
change towards the most liberal form, multiple designation. Agreements
providing for multiple designation account for almost 60 per cent of all
ASAs and over 80 per cent of passenger traffic in 2011. The multiple desig-
nation clause is most often found in ASAs signed by countries in Asia and
Oceania, Europe, Middle East and the American continent. In Africa, by
contrast, single designation is the most frequent designation clause.
ASAs containing multiple designation provision have increased by
nearly 8 per cent between 2005 and 2011, compared with an over one per
cent decrease for agreements with single designation clauses. In terms of
traffic covered by ASAs with multiple designation, no significant changes
have occurred between 2005 and 2011. All regions have registered an
increase in agreements with multiple designation, and most significantly
so in the Middle East, most likely reflecting the emergence of new national
airlines in the region. ASAs with single designation, instead, have
increased most considerably in the CIS region.
c. Withholding
Notwithstanding the overriding role played by the most restrictive
withholding clause, that is the requirement of substantial ownership and
effective control (SOEC), ASAs have slowly moved towards more lib-
eral withholding forms between 2005 and 2011. Although the SOEC
provision still dominates, accounting for almost 90 per cent of agree-
ments and passenger traffic, between 2005 and 2011 the total number of
ASAs containing a SOEC clause has remained stable. Conversely, the
most significant increase has been registered for agreements containing
the Community of Interest (CoI) clause, albeit it from a very low start-
ing point. In 2011, agreements with this clause represent only 1.4 per
cent of all agreements and 0.2 per cent of passenger traffic.
In each geographic region, the SOEC clause accounts for the highest
number of agreements, ranging from nearly 95 per cent in Africa to just
below 80 per cent in South America. ASAs containing this clause have
increased the most between 2005 and 2011 in the Middle East and Asia
and Oceania.
CoI, on the other hand, can be found more frequently in Europe,
where it accounts for 1.9 per cent of all agreements. No doubt, this
reflects the impact of the November 2002 European Court of Justice
judgements. These ruled that any bilateral agreement by an EU Member
State was illegal under EU law if it did not contain a withholding/own-
ership ‘Community clause’ to cover the possibility of designating airlines
of other EU Member States and resulted in the re-negotiation of the
withholding clause in the EU Member States’ ASAs in this sense. It also
needs to be acknowledged that, as will be discussed in Section
“Liberalization Through The Withholding/Ownership Clause”, these
figures are likely to underestimate the occurrence of CoI clauses in bilat-
eral ASAs concluded by EU countries.
The most liberal withholding criterion, the PPoB clause, has regis-
tered close to a 17 per cent increase in terms of ASAs from 2005 to
2011, accounting for over 8 per cent of total ASAs in 2011 and nearly
9 per cent of traffic. Total traffic covered by ASAs containing the PPoB
clause has increased almost 40 per cent, the least significant increase
amongst the three withholding clauses. Asia and Oceania, as well as
Europe, account for the highest number of ASAs containing the PPoB
clause. PPoB, as well as CoI, are almost never found in African ASAs.
Asia and Oceania and South America are the regions where ASAs con-
taining the PPoB clause have increased the most between 2005 and
2011, nearly doubling in number. ASAs for which there is no available
Skies Wide Shut 77
information on the withholding clause have increased both in terms of

number of ASAs and traffic coverage, accounting for nearly 4 per cent
of ASAs and over 3 per cent of total traffic in 2011.
d. Pricing
Similarly to the findings on the withholding clause, an in-depth analysis
of tariff clauses shows that, notwithstanding the predominance of ‘dual
approval’ clauses, ASAs have shown a small but significant tendency
towards more liberal forms between 2005 and 2011. Dual approval, the
most restrictive tariff clause, accounts for almost 80 per cent of all ASAs
and over 60 per cent of total traffic in 2011. At the other extreme, ‘free
pricing’, the most liberal clause, accounts for less than 3 per cent of all
ASAs and around 5 per cent of passenger traffic. However, from 2005
and 2011, ASAs containing a dual approval provision grew by only 0.2
per cent, compared to an increase of nearly 70 per cent for ASAs contain-
ing a free pricing provision. ‘Zone pricing’ provisions have registered the
most significant increase in terms of number of ASAs.
These changes are also reflected in the increased share of ASAs and
passenger traffic containing free pricing and zone pricing clauses. On the
other hand, ASAs containing dual approval have seen their share of total
ASAs decrease. Between 2005 and 2011, the semi-liberal ‘country of ori-
gin’ and ‘dual disapproval’ clauses have also registered an increase in the
share of total ASAs. In almost all regions, ASAs containing a dual
approval clause account between 70 per cent and 90 per cent of all agree-
ments. The only exception is North America, where ASAs containing a
dual approval clause only account for 45 per cent of agreements. For all
regions, dual disapproval is the second most frequent clause, except for
South America where country of origin comes second, followed by dual
disapproval. Dual approval has not changed significantly between 2005
and 2011, whereas dual disapproval has increased in all regions, especially
in Africa and North America.
Zone pricing is virtually inexistent in Africa, the CIS region and the
Middle East. It is most frequent in North America and Asia and Oceania,
where it has registered the highest increase: seven times more ASAs con-
tain zone pricing clauses in 2011 compared to 2005. Free pricing, which is
the least common clause across all regions, is mostly contained in agree-
ments signed by countries in Asia and Oceania, whereas it is almost
absent from ASAs concluded by countries in Europe and the CIS region.
Country of origin, instead, is most commonly found in South America,
where it has also increased most significantly. Around 5 per cent of agree-
ments do not contain information about the applicable pricing clause.
e. Capacity
Also with regard to the capacity clause, findings show that ASAs tend
towards more liberal clauses in 2011 compared to 2005. In fact, from
2005 to 2011 the number of agreements containing the most restrictive
clause, the pre-determination of capacity, has increased by only 3.5 per
cent, whereas agreements providing for the most liberal ‘free determina-
tion’ clause have increased by nearly 36 per cent. In parallel, the increase
in passenger traffic accounted for by free determination ASAs is higher
than that accounted for by pre-determination ASAs.
In absolute terms, however, the most restrictive, pre-determination
clause still dominates, accounting for almost 70 per cent of all agree-
ments and almost half of total passenger traffic in 2011. By contrast,
free determination clauses account for about 8 per cent of all agreements
and one fourth of total traffic. North America is the only region where
ASAs containing a free determination clause are more frequent than
those containing a pre-determination clause. In all other regions, pre-
determination accounts for the majority of all ASAs, ranging between
60 per cent of all agreements in South America and 87 per cent in the
CIS region. However, ASAs with free determination provisions have
grown more than those with pre-determination across all regions, with
the greatest increases registered in Africa and CIS region.
The semi-liberal Bermuda I is the second more frequent capacity
clause in terms of number of agreements, and third most frequent in
terms of passenger traffic. It accounts for 15 per cent of all agreements
and 16 per cent of total passenger traffic. Bermuda I clauses are rela-
tively more frequent in South America, Europe and Asia and Oceania,
and least frequent in the CIS region. ASAs with a Bermuda I clause
have registered a decrease in almost all regions, and especially so in the
American continent.
Finally, fewer than 7 per cent of ASAs provide for some other form
of capacity provision, with no significant change since 2005. Almost 90
per cent of these agreements, covering less than 9 per cent of traffic, can
be classified as ‘other restrictive’. No information is available on the
capacity clause only for less than 3 per cent of agreements, representing
less than one per cent of passenger traffic.
f. Co-operation
Between 2005 and 2011, the co-operative arrangement clause has
become slightly more prevalent in ASAs. Although agreements allowing
for co-operative arrangements represent just over 16 per cent of total
ASAs in 2011, they have grown by over 30 per cent between 2005 and
Skies Wide Shut 79
2011, compared to the unchanged situation for ASAs not including co-
operation clauses. In terms of passenger traffic, ASAs allowing for co-
operative arrangements have registered a 60 per cent growth. With the
exception of North America, where ASAs containing co-operative
arrangements represent over 60 per cent of all ASAs, these are the min-
ority across all other regions.
g. Statistics
In terms of exchange of statistics, no significant change has occurred
between 2005 and 2011. The absence of an exchange of statistics clause
is a sign of a more liberal regime, as it reflects an intention not to moni-
tor closely the evolution of capacity and tariffs. Agreements that do not
require statistics to be exchanged, however, account for only one fourth
of total agreements and one-third of total passenger traffic, showing no
significant change since 2005. Regardless, these agreements have
increased at a slightly faster rate than those which provide for an
exchange of statistics. Even if there is some regional variation, with
ASAs providing for statistics to be exchanged accounting for over 80
per cent of agreements in Asia and Oceania, but around 65 per cent in
the American continent and the CIS region, this provision is dominant
across all regions.
Concluding Observations. The application of the QUASAR methodology

allows us to draw the following conclusions about the overall situation of
bilateral ASAs in 2011, and how this compares to 2005. First, traffic is
highly concentrated: 9 per cent of ASAs, involving 76 parties, cover around
three-fourth of the total scheduled international passenger traffic captured
by ASAP in 2011.
Second, between 2005 and 2011, ASAs have generally become more lib-
eral, even though only slightly so. The overall WALI in fact, has increased
from 15.1 in 2005 to 15.3 in 2011. On a 0 to 50 scale, ASA remain, overall,
fairly restrictive. High-traffic bilateral agreements are only a little more lib-
eral than the average: in 2011, the first 100 agreements ranked by traffic
record a WALI of 17.6.
Third, restrictive market access features, such as dual approval of tariffs,
pre-determination of capacity and a withholding clause based on substan-
tial ownership and effective control, are still largely prevalent in ASAs.
However, although these restrictive features are found in the large majority
of agreements, in relative terms they have lost some importance between
2005 and 2011 as their more liberal variants have generally increased at
much faster rates.
Plurilateral ASAs
Similarly to the approach taken with bilateral agreements, we applied the
QUASAR methodology also to plurilateral ASAs. In terms of datasets,
four new agreements were signed between the start of 2006 and the end of
2011, namely the ECAA, EURO-MED, US-EU and LACAC agreements,
and as such were added to the 2011 sample. Furthermore, it is important to
note that the membership of three agreements has changed between 2005
and 2011. Specifically, the Andean Pact and COMESA registered a with-
drawal of a member (Venezuela and Angola, respectively), whereas the
MALIAT increased its membership with the addition of the Cook Islands
and Mongolia.
In comparison to the ‘original’ paper version of QUASAR dealing with
plurilaterals,24 the coding of these agreements was also fully standardized
and rendered entirely compatible with the methodology used to code bilat-
eral ASAs. With respect to those plurilateral agreements that foresee that
the individual parties may assume different levels of obligations (such as
the MALIAT or the LACAC), the coding has been done on the basis
of the level of obligations that are applicable to all of the parties.
The traffic covered by the plurilateral agreements retained in ASAP has
been calculated by adding up the traffic for all bilateral relations involved
and related it to total scheduled international traffic (496 million passen-
gers). An analysis of the traffic covered by plurilateral agreements shows
that, while in 2005 the 14 agreements covered by ASAP accounted for just
over 12 per cent of total international scheduled passenger traffic, the 18
ASAs included in ASAP for 2011 have nearly doubled this share, represent-
ing around a quarter of total traffic.
The high share of traffic covered by plurilateral agreements needs to be
interpreted with caution. There are at least two caveats. First, there is a fair
degree of uncertainty regarding the effective ratification of several of these
agreements. Second, even for those plurilateral agreements that have entered
into force, the degree of effective implementation would seem, according to
both operators and the parties concerned, to be extremely variable.
These caveats need to be borne in mind also when considering the degree
of openness provided for by plurilateral agreements and its evolution
between 2005 and 2011. Indeed, in 2005 plurilaterals already registered a
significantly higher level of openness than bilateral ASAs, with an overall
WALI of 34.4. This further increased by 2011, when the general WALI
measures 37.1.
In the six-year period under consideration, plurilaterals appear to have
indeed evolved towards more openness, as indicated by the positive shift
Skies Wide Shut 81
towards more liberal ALI ranges. In 2005, 93 per cent of plurilaterals regis-
tered an ALI falling in the 2650 range, with the reminder falling in the
1014 ALI range. By 2011, this liberal bias has been further consolidated,
with nearly 95 per cent of all agreements fitting in the 2650 ALI range.
What liberal features are found in plurilaterals, at least on paper? First
and foremost, multiple designation, which is universally present, and fifth
freedom traffic rights, which are granted in nearly 95 per cent of all agree-
ments. The capacity clause is also largely based on free determination (in
nearly 90 per cent of cases), with free pricing, permission to enter into co-
operative arrangements and no requirement that statistics be exchanged
found in two-thirds of all agreements. Around one-third of plurilaterals
grant seventh-freedom traffic rights and one-tenth permit cabotage, and by
far the most common withholding clause is the CoI clause, which is present
in half of all agreements, followed by the PPoB one, which is found in just
over a quarter. Substantial ownership and effective control, conversely, is
required in less than 17 per cent of all plurilateral agreements in 2011.
In sum, plurilateral agreements, if effectively implemented, and if effec-
tively overriding the generally much more restrictive regime provided for
by any applicable bilateral ASAs between any two of the parties concerned,
would have a decisive liberalizing impact on a quarter of the traffic.
REFINING QUASAR
This section discusses a number of suggestions for further refining the

QUASAR methodology. They are limited to the sectoral segment cover by
QUASAR that is international scheduled passenger transport. Extension to
other segments of the sector (i.e. international scheduled and non-scheduled
all-cargo traffic; international non-scheduled passenger traffic; domestic
scheduled and non-scheduled cargo and passenger traffic) will be discussed
in the fourth section.
Traffic Data
The traffic data used by QUASAR comes from the IATA mileage set and
measures the number of passengers having travelled on a given city-pair re-
aggregated at the country level, for 2005 and 2011. It could be refined along
three possible directions: the time dimension, the nature of the data and the
correction of the built-in biases. Starting with the time dimension, the first
obvious improvement would be the incorporation of traffic data for all
available years. This would provide a more continuous image of the evolu-
tion of traffic. It would also potentially match the regulatory dataset if, as
will be discussed below, yearly historical series were also to be reconsti-
tuted. This is dependent on the application of the same methodology for
data collection. An initial exploration of the question would seem to indi-
cate that we could not use traffic data further back than 1995.25 This lim-
itation is not so constraining for economists, for whom nearly twenty years
should be a long enough period to draw meaningful lessons.
The second possible refinement would involve using better data than the
simple number of passengers having travelled between two countries. The
number of passengers is indeed a very crude proxy for at least two reasons.
First, the profitability of long-, medium- and short-haul segments is very
different, at least for legacy airlines. Long-haul segments are generally prof-
itable, if not very profitable. By contrast, short- and medium-haul flights
appear to be much more challenging from a profitability standpoint.
However, even when those segments are loss-making, notably in periods of
economic downturn and weak demand, large legacy airlines tend to main-
tain them as they act as feeders for their long-haul segments. In this regard,
using passenger-kilometre figures, rather than passenger numbers, would
constitute a first improvement to the traffic data, as the value of long-haul
traffic for airlines is higher than that of short- and medium-haul traffic, at
least on a per capita basis.
Second, using passenger flows implies giving the same weight to long-
haul first-class passengers having paid US$5000 for their trip and economy-
class passengers having paid US$500 for the same trip. The profit margin
for the airline is clearly not the same for these two kinds of passengers.
Indeed, in spite of the absence of precise data, it is generally claimed that,
at least on long-haul segments, between 20 per cent and 30 per cent of pas-
sengers account for 80 per cent of the airline’s revenues. Hence price/
revenue data, rather than volume data (be it as passengers or passenger-
kilometres) would reflect much more accurately the reality of traffic.
Revenue implicitly factors in distance, on the one hand, as well as, on the
other, the fares paid and the number of passengers who paid them.
There would appear to be a gap in governments’ intervention in the sec-
tor. So far as we know,26 when negotiating ASAs, governments determine
capacities (i.e. total number of weekly flights by given aircraft types), and
hence passenger numbers, ‘in bulk’, as aggregate totals. They do not
negotiate how many first, business, or economy-class passengers will be
transported on a given aircraft. This gives airlines a certain degree of flex-
ibility. If load factors rise above 80 per cent but additional necessary
Skies Wide Shut 83
capacity is not granted by the partner country, airlines would be free to

modify their seating arrangements, for instance by adding more
profitable first and business-class seats. In other words, the bilateral state-
to-state negotiations of ASAs are about the total number of passengers,
matching the current data basis of QUASAR, whereas, for airlines, one
passenger is not necessarily like another. However, since QUASAR is
about the openness of state-to-state agreements, passenger numbers, as
imperfect a dataset as it may be, remains a very acceptable proxy.
Third, the traffic dataset could be improved by correcting its built-in
biases. As discussed in Section “Traffic Data”, the IATA country-pair traffic
dataset is not an exact match for the underpinning regulatory situation
between the two countries concerned. The only possible way of correcting
this bias, albeit only partially, may be to acquire data on actually operated
flights, from airline data providers such as OAG, which would enable the
identification of instances where direct flights do not exist between two coun-
tries parties to a bilateral ASA. It would still be debatable, however, whether
the correction of the dataset enabled by this information would be worth the
added effort.
Regulatory Data
Regulatory data could be improved in at least two ways. First, it could be

extended along two directions, namely time and geographical scope.
Second, it could be refined through a deepened analysis of capacity. Each
of these options is discussed below.
Extending the Regulatory Data

Thus far, WALIs have been computed exclusively for 2005 and 2011, because
traffic statistics were available only for these two years. However, it would
theoretically be possible to compute WALIs on a yearly basis starting from
1995.27 The resulting image would, however, still be partial, as the ‘gaps’ in
the regulatory information would clearly need to be filled simultaneously.
Some historical information on ASAs is available in ICAO’s WASA, but, as
with the rest of the WASA dataset, it is by no means complete.
In addition to deepening the time dimension of ASAs, it would also be
desirable to widen its geographical scope. As discussed, ICAO’s contract-
ing parties do not notify all of their ASAs to ICAO. Although ICAO has
been making an effort to incorporate un-notified agreements in its WASA
database, it remains largely incomplete.28 As indicated in Section “The
Regulatory Features”, even by complementing the WASA information

with data obtained from WTO Members and through the WTO’s TPRs, a
back-of-the-envelope calculation would suggest that in 2011 around a
quarter of all bilateral ASAs are still missing. It would thus seem that
only a publicly available ‘wiki-style’ system, whereby any interested party
could input information about the conclusion or amendment of an ASA,
might improve the situation.
Refining the Regulatory Data: Towards an ‘Aero-Policy Comparator’

The regulatory features taken into consideration by QUASAR do not
include two essential elements of ASAs: the routes that may be served and
the capacity (in terms of aircraft types and number of weekly flights) that
can be deployed on those routes. At first glance, this looks like a major
handicap, because what airlines and governments negotiate are, in most
instances, precisely city-pairs and routes,29 and additional capacity on exist-
ing or new city-pairs, rather than tariff, ownership or code-sharing clauses.
However, the reasons for this omission are linked to fairly significant infor-
mational, conceptual and statistical difficulties.
In the WASA database, the information about routes is sometimes con-
tained in a ‘table of routes’ in the codified summaries of ASAs.
Unfortunately, however, this information is not only incomplete, but it
also proved difficult to decipher when available.30 In the case of capacity,
there is an additional transparency problem, as capacity allowances are
often ‘hidden’ in the confidential sections of ASAs and are therefore neither
published nor notified to ICAO.31
At any rate, it would have proven difficult to incorporate information
about routes and capacity into QUASAR, which is an absolute scoring
methodology. The market access features considered by QUASAR are
attributed points depending of their implied degree of restrictiveness, out of
a pre-determined total maximum number of points for their fully liberal
variant. Routes and capacity do not easily fit into such a scoring approach.
Except for full open skies agreements,32 in fact, it would have been neces-
sary to decide on the relative restrictiveness of granting only limited,
selected routes. The same is a fortiori true when capacity is added to the
picture. How restrictive is the granting of three flights a week for 200-seat
aircrafts between Helsinki and Beijing? Does this same amount of capacity
amount to greater restrictiveness for a FrankfurtBeijing route or a
ZurichBeijing one, and by how much? Would flying to Shanghai be worth
more than going to Guangdong? These are some of the answers that we
Skies Wide Shut 85
would have had to find an answer to in order to properly incorporate

routes and capacity into QUASAR.
In light of these difficulties, we were forced to ignore information on
routes and capacity. Instead, we relied on the clauses agreed to in the ASA,
primarily the capacity clause but also the traffic rights granted, the designa-
tion, code-sharing and the exchange of statistics clauses, which all have,
albeit to various degrees, a capacity/routes dimension, since they give
concurring clues on countries’ capacity policy. Capacity clauses factored
in QUASAR are procedural modalities for determining capacity.
Theoretically, they have no bearing on capacity itself. It would be possible,
for instance, to envisage a situation with a very restrictive pre-
determination clause in conjunction with very generous actual capacity
granted, largely exceeding demand. However, in reality, restrictive ASA
features tend to be highly correlated with each other.33 A restrictive capa-
city clause will generally be associated with tight limits on capacity. The
same goes for tariffs (double approval and high tariffs), grant of rights
(absence of fifth freedom accompanied by absence of seventh, eighth and
ninth freedoms), an exchange of statistics clause (which clearly reflects a
desire to control capacity and tariffs) and, to a lesser extent, the absence of
a co-operation clause and the presence of a substantial ownership and
effective control withholding clause. There are clearly identified sets of
recurrent features in ASAs, and agreements that are procedurally restrictive
would usually correspond to agreements that are also quantitatively
restrictive.
Regardless, there would seem to be a potential, albeit indirect, method
of factoring routes and capacity in QUASAR through a system of relative
comparisons of city-pairs and country-pairs exhibiting similar features,
which we call the ‘aero-policy comparator’. Such comparisons would be
undertaken on the basis of information about the relative importance of
the two international airports and the capacity granted on the relevant city-
pairs. Imagine that country A (say, India) had five international airports
from which long-haul destinations are served, and that country B (say,
Germany) also had five such airports, each accounting for a share of each
country’s total long-haul international traffic.34 As the fictitious example
below illustrates, airports would be charted on a matrix containing contain
two pieces of information: the product of the two airports’ shares in rele-
vant traffic, which provides a conventional but relatively meaningful indi-
cator of the relative value of each potential route; and the capacity
effectively granted for each city-pair.35
Delhi (0.6) Mumbai (0.25) Kolkata (0.1) Chennai (0.03) Hyderabad (0.02)
Frankfurt (0.6) 0.36 0.15 0.06 0.018 0.012

‘Open ended’ 3 flights/week 2 flights/week ‘None’ ‘None’
Berlin (0.15) … … … … …
Munich (0.15) … … … … …
Hamburg (0.08) … … … … …
Rostock (0.02) … … … … …
This information would become meaningful when juxtaposed with

the same information for comparable country-pairs, whether involving
one of the two parties (e.g. IndiaUnited Kingdom, IndiaFrance, or
GermanyIndonesia, GermanyChina) or only third parties (e.g. United
KingdomPakistan, FranceIndonesia). The selection of ‘comparable’
country-pairs would be based on a battery of criteria, relating either the
each country’s economic, demographic and geographical characteristics
(such as per capita GDP, population size, country size, total international
long-haul traffic) or to the specific country-pair (e.g. distance, common lan-
guage, historical ties), along the lines of the gravity models used by trade
economists.36
Although such an ‘aero-policy comparator’ would not amount to a full
integration of routes and capacity into the QUASAR methodology, when
examined next to relevant traffic information it would enable a more
informed assessment of whether given route and capacity limitations are
‘biting’ and what their impact is.
EXTENDING QUASAR
This section discusses the feasibility of extending QUASAR to segments of

the air transport industry beyond scheduled international passenger trans-
port. Specifically, the air transport sector can be catalogued by mixing three
dimensions:
1. The organization of the traffic, that is whether it is scheduled or non-

scheduled.
2. The geographical scope of the traffic, that is whether it is domestic or
international.
3. The type of cargo carried, that is all-cargo traffic versus passenger traffic
(with cargo in the belly).
Skies Wide Shut 87
Combining these three dimensions results in eight possible segments for

the sector. Only one, that is scheduled international passenger transport, is
covered by QUASAR as it currently exists.
Table 3 outlines the feasibility of extending QUASAR to the seven seg-
ments of the air transport industry it does not yet cover. In particular, it
addresses the question of the economic importance of the various segments
and the availability of traffic and regulatory data.37
As Table 3 illustrates, two segments appear the most promising for an
extension of QUASAR. First, international scheduled all-cargo traffic.
Although this segment raises complex challenges, notably in terms of traffic
data collection, it is important not only economically but also from a regu-
latory perspective, as it constitutes a ‘prototype’ of what a liberalized inter-
national passenger traffic industry might look like. Indeed, as fifth and
seventh freedom points are relatively frequent for all-cargo transport, there
is more third-party competition in this segment. With seventh freedom
rights in particular, nationally flagged-airlines cease to serve exclusively the
import-export trade of the nation concerned and start being deployed on
‘cross-trades’.38 As a result, bilateral restrictions can be largely circum-
vented through ‘hubbing’, and as such tend to have less impact on carriers’
economic decisions, while shippers’ interests appear to carry a certain
weight in national policy-making.
Second, domestic scheduled passenger. This traffic segment may also, to
a certain extent, be seen as a precursor for the international one. Because
domestic deregulation is relatively easier to agree and quicker to imple-
ment, it often precedes international liberalization. This has been clearly
the case in the United States and the European Union. It is also a way for
national airlines to learn how to operate in a more competitive environ-
ment and, at least on large domestic markets, it would appear to affect the
scope and price of the services offered fairly similarly to international liber-
alization. Domestic deregulation has also triggered the emergence of low-
cost carriers, by far the most significant development in the aviation indus-
try over the last thirty years. Against this background, it would be useful to
set, via QUASAR, a common conceptual framework to compare the effects
of US and EU domestic deregulation and thus help predict the effect of
similar on-going processes in India, Indonesia, Japan, Korea and China, to
name just a few markets. For this segment, the availability of traffic data
should not pose serious difficulties, and the conceptual challenges that
indisputably exist should not prove insurmountable either.
At the other extreme, the least feasible candidates for a QUASAR exten-
sion are international non-scheduled passenger and cargo transportation,
Table 3. Applying QUASAR to Different Segments of the Air Transport Industry.
88
Traffic Segments Significance Traffic Data Possibly Relevant Regulatory Data
Availability Regulatory Features Availability
International scheduled Already covered by QUASAR in 2005 and 2011

passenger transport
International non- A declining and blurred No overarching industry Market access granted by To our knowledge, there
scheduled passenger economic model. organization the destination- is no publicly
transport Not necessarily comparable to IATA country, theoretically available information,
addressed by (IACA is only on the basis of a non- not even in literary
regulatory regional) discriminatory, form.
frameworks (e.g. EU Possible (untested) data generally applicable A questionnaire to the
Single Aviation sources: ACI, OAG, policy, but in reality individual civil
Market). ICAO, UN/WTO possibly on case-by- aviation authorities
Still notable on the Would probably imply case authorizations. would seem the only
leisure market the need for a In the absence of practical way of
(e.g. North-South, in questionnaire relevant information, collecting information
the European Union, addressed to civil restrictions, data on
in the United States, aviation authorities. routes and capacity
for pilgrimages) granted and tariffs
No global figures charged should be

collected partner-by-
partner
International scheduled all- Accounts for one-third IATA data do not Based on our experience, No global enquiry
cargo transport of total air cargo distinguish between carriers disagree on available
traffic (two-thirds are belly and all-cargo which market access Would probably imply
in the belly), and tonnages features matter and working with IATA,
probably more for Express carriers’ traffic which do not TIACA, GEA and
certain destinations is not covered by Capacity (number of selected carriers to
(Kenya, Colombia, IATA data, except to flights, tonnage, build a database of
Far East to the EU the extent where seasonal restrictions, market access
and United States) express carriers use types of aircraft features.
Skies Wide Shut
third-party belly allowed) clearly The WASA database
carrier capacity matters contains some cargo-
The GEA does not Fifth and, even more so, only provisions but
communicate seventh freedom rights they do not seem to
figures about its are of essence (there be detailed and
members’ traffic, are probably no exhaustive enough to
even in an aggregated eighth and ninth underpin a fully-
manner freedom rights fledged database
Possible (untested) data granted).
sources: TIACA, There do not seem to be
ACI, customs’ data price restrictions
Curfews and noise are
difficult questions
(economic restrictions
but not trade
restrictions, especially
if applied on a non-
discriminatory basis)
International non- The dividing line with To our knowledge There is no literature Challenging to obtain
scheduled all-cargo scheduled all-cargo traffic data for this available about the
transport traffic is difficult to segment is not restrictions faced by
draw, in particular for recorded, either at these operators
seasonal traffic worldwide level or
Seems important for even at the national
heavy lift sector, level
difficult to assess for
the rest
Domestic scheduled Relevant in mid-sized IATA and OAG The regulatory features To our knowledge, there
passenger transport and large countries datasets cover this for this type of traffic is no systematic
(especially if intra-EU type of traffic are, to some extent, inventory of
traffic is considered to comparable to those restrictions faced by
89
be domestic traffic) retained in QUASAR domestic traffic
Table 3. (Continued )
90
Traffic Segments Significance Traffic Data Possibly Relevant Regulatory Data
Availability Regulatory Features Availability
Capacities, routes and carriers, but this

prices were controlled should be relatively
by civil aviation easy to document by
authorities, and in searching for national
some cases they still legislations or via a
are. questionnaire
The policy towards start-
ups would have to be
taken into
consideration
Freedoms of the air are
meaningless
There is a partial overlap
with QUASAR in the
rare cases where
cabotage is allowed
for foreign air carriers

Domestic non-scheduled Probably marginal Probably only available Beyond safety As for the international
passenger transport except in very large at the local level considerations, from a segments, decisions
countries (e.g. the trade viewpoint seem to be taken on a
cruise packages in the restrictions are most case-by-case basis and
United States), likely linked to the risk their categorization
although this of head-to-head seems, a priori,
hypothesis would need competition with the difficult
to be tested. national carrier
Domestic scheduled all- Important only in large Challenging to obtain Applicable restrictions Challenging to obtain
cargo transport countries should normally be
similar to those in the
international scheduled
Skies Wide Shut
all-cargo segment.
However, it is difficult
to imagine a state
hampering the
development of its
own air transport
sector unless
intermodal or
environmental
considerations come
into play.
Domestic non-scheduled Probably marginal Probably not recorded In the absence of further Probably irrelevant
all-cargo transport beyond the heavy lift at all information, it is
sector difficult to imagine a
state imposing market
access restrictions in
this sector
Notes:
IACA: International Air Charter Association.
ACI: Airports Council International.
OAG: Official Airline Guide.
UN/WTO: World Tourism Organization.
TIACA: The International Air Cargo Association.
GEA: Global Express Association.
91
essentially because of lack of proper traffic data. Finally, domestic

non-scheduled passenger and all-cargo transport and domestic scheduled
all-cargo transport appear to be rather irrelevant, both because they are
economically insignificant in most markets and also because they are
devoid of key regulatory features from which to draw meaningful lessons.
TESTING THE MYSTERIES OF AIR TRANSPORT

LIBERALIZATION
Air transport services liberalization seems to be a constantly moving target.

As will be illustrated in the sections that follow, the analysis of the 2005
and 2011 ASA data undertaken via QUASAR, as well as the (rather lim-
ited) air transport literature would seem to suggest that:
• liberalization does not happen the way it is said to;
• some liberalization does happen, but very slowly and with an apparent
glass ceiling;
• progressive multilateral liberalization could, but is unlikely to happen.
A refinement and extension of QUASAR along the lines described in the
third and fourth sections would provide a useful tool to further analyse
and shed light on these three seeming mysteries of air transport services
liberalization.
Five ‘Myths’?
How liberalized is the aviation sector? This section will attempt to reply to
this question. Using, wherever possible, currently available QUASAR data,
it will try to shed some light on five commonly held beliefs about air trans-
port services liberalization. Specifically, it will try to answer the following
questions: How correct is it to say that air transport liberalization exists
per se? Does liberalization occur through fifth freedom traffic rights? Does
it occur through seventh freedom traffic rights? What is the link between
liberalization and the withholding/ownership clause? Do plurilateral ASAs
lead to liberalization?
Air Transport ‘Liberalization’ per se

Liberalization, in international trade parlance, means the opening of com-
petition to foreign third parties, in most instances on a non-discriminatory
Skies Wide Shut 93
basis.39 When applied to the aviation context, this concept of liberalization

does not seem to hold. This may appear paradoxical in an industry that
associates ‘open skies’ agreement with the notion of liberalization.
However, there is no single definition of the ‘open skies’ concept. The
United States uses the term to designate ASAs with a substantial ownership
and effective control clause, fifth freedom rights but no seventh freedom
rights, no limitation on routes, tariffs or capacity. Other countries (such
Australia and New Zealand) use the term to refer to ASAs that include the
PPoB clause and grant seventh freedom and, in certain instances, cabotage
rights. More importantly, regardless of the definition chosen, it needs to be
acknowledged that classical, US-inspired ‘open skies’ agreements40 remain
purely bilateral and introduce no third-party, ‘free for all’ competition. It
may be argued that third-party competition is established through fifth
freedom traffic rights. However, as will be explained in the section below in
greater detail, this happens very marginally.
Thus, in spite of their designation, ‘open skies’ agreements do not consti-
tute liberalization per se, but rather a simple bilateral easing of some regu-
latory constraints such as states’ controls on prices, capacity and routes.
More precisely, ‘open skies’ agreements provide at least for the following
liberal elements:
• Multidesignation, that is no limits on the number of carriers allowed

from each of the parties. However, a multidesignation clause is of inter-
est only to a handful of countries, as countries as economically impor-
tant as France and Germany only have one major international carrier.
• Open-ended granting of routes for third and fourth freedom rights. This
does indeed allow for new routes to be opened up, but only for the car-
riers of the two parties concerned, thus providing scope for potential col-
lusion, rather than necessarily more competition.
• Free determination of tariffs. While this provides the carriers of the two
parties with the possibility of competing on prices, it is puzzling that
complaints about bilateral price dumping seem to be never raised. This
may be attributed to the fact that private sector collusion has replaced
inter-state collusion, or, more likely, that tariff controls have become
irrelevant since the advent of yield management techniques. Since no two
passengers pay the same price for comparable seats on the same flight,
there is no point in trying to agree on a fixed set of tariffs.
• Free determination of capacity. The right to compete on quantities and
market share is indeed introduced, but still exclusively in purely bilateral
context.41
Out of these, only multidesignation allows additional airlines to compete

on the bilateral routes, and this is in practice relevant only for a handful of
countries that have several international carriers (basically the United
States, the United Kingdom, India, China, Korea, Japan and Russia, but
not Germany and France, for instance). For all other countries, however, a
classical ‘open skies’ agreement does not imply a break-up the duopoly, but
removes restrictions for the two airlines concerned in terms of business
decisions, such as routes to be served, capacities to be deployed and prices
to be set. While the two carriers are indeed free to use this operating auton-
omy to compete with each other, they are also in a position to exploit the
advantages procured by their duopoly situation.
Forms of international third-party competition do exist, however, with
‘true’ common markets and sixth freedom airlines. Common aviation mar-
kets seem, indeed, to be the only form of liberalization per se. However,
although they are quite numerous on paper, they seem to be rather rare in
practice42 and their significance in terms of liberalization is likely to be
small. The same is true for third-party competition involving sixth freedom
airlines. For geographical reasons, this is essentially limited to long-haul
routes between Europe and the Asia-Pacific region. Indeed, as QUASAR
has allowed us to demonstrate, ASAs are on average actually fairly restric-
tive. The analysis would thus seem to indicate that aviation liberalization is
generally more a myth than a reality.
Liberalization Through Fifth Freedom Traffic Rights

Fifth freedom traffic rights are nowadays a relatively common feature of
bilateral agreements. ASAP 2011 data show that this feature is present in
two-thirds of the agreements, covering 80 per cent of the traffic. At face
value, this would seem a significant step towards liberalization, as fifth free-
dom rights amount to the introduction of third-party competition on a
given country-pair (country A airline(s) on a country B to country C seg-
ment, e.g. British Airways on Singapore-Australia). However, one may
wonder if fifth freedom clauses are pervasive because they are, in reality,
harmless in terms of the degree of competition that they introduce. From a
legal perspective, the granting of fifth freedom rights in one bilateral agree-
ment does not grant its effective exercise, be it intermediary, behind and/or
beyond. For an airline of country A to operate fifth freedom traffic
between country B and C it is not sufficient that the agreement between
A and B grants fifth freedom rights via, beyond or behind a point in coun-
try C: it is also necessary that the agreement between A and C stipulates
that right.
Skies Wide Shut 95
QUASAR at present does not automatically factor in that combination

of agreements through a specific index. In the future, QUASAR could be
reviewed in this sense to allow, for instance, the identification of situations
where fifth freedom is ‘locked’ (i.e. absent from both agreements con-
cerned), ‘half-locked’ (i.e. granted by one of the two agreements concerned
but not by the other) or ‘unlocked’ (i.e. present in both agreements). It
could also allow an assessment of the extent to which any fifth freedom
traffic is effectively ‘unlocked’: a limited or an open-ended number of
points, limitations on capacity and/or tariffs. As two-thirds of the agree-
ments, accounting for 80 per cent of the traffic, grant fifth freedom rights,
it is reasonable to suppose that many, if not most, situations, at least in
traffic terms, fifth freedom is indeed ‘unlocked’.
Such an effort would be worth undertaking only if statistics on actually
operated flights showed a flurry of fifth freedom traffic. However, this does
not seem to be the case, except in two situations. First, when the serving of
an intermediate point is essential to the profitability of a very long-haul
flight (typically fifth freedom rights in Singapore for BA and QANTAS on
London to Sydney). Second, when the exploitation of a close-by beyond
point after a very long-haul trip brings marginal profits (e.g. Emirates fly-
ing from Sydney to Auckland after a Dubai to Sydney flight), also known
as ‘complexing’, that is using the bilateral partner’s airport as a distant hub
to serve less important airports (e.g. Air France using Bangkok to serve
Hanoi, Vientiane, Phnom Penh and Ho Chi Minh City).
Beyond those two cases, the effective use of fifth freedom rights appears
extremely marginal. It seems to be replaced by code-shared flights operated
by the alliance partner or by direct point-to-point flights. Additional work
on this issue through a detailed comparison (country-pair by country-pair)
of effective flights and QUASAR fifth freedom rights would be necessary in
order to verify the significance and extent of this phenomenon.
Liberalization Through Seventh Freedom Traffic Rights

The observations regarding fifth freedom rights are, a fortiori, true for
seventh freedom rights. Seventh freedom is ‘true’ third-party competition,
as it does not require the third-party airline to have flown first from its
country of origin to reach the first end of the segment it is intending to
serve. The effective exercise of seventh freedom rights, however, necessi-
tates the same combination of agreements as fifth freedom. Thus, also in
this case, QUASAR would need to be reviewed to factor in these combina-
tions. However, this effort would not seem worthwhile presently, because
seventh freedom rights can only be found in five agreements, out of over
2200, contained in the ASAP 2011 dataset. It is thus very likely that in
most, if not all, of those cases the exercise of seventh freedom is ‘half-
locked’ by the ‘veto’ of the third parties concerned. Access to real flight
data (OAG type) would allow for a more detailed analysis on the effective
exercise of these rights. Seventh freedom rights also exist in seven plurilat-
eral agreements out of 18 (ACAC, COMESA; ECAA; the defunct but
never denounced and fascinating IATA44; USCH; EUUS; and the
optional protocol of the MALIAT). Access to OAG-type of data on real
flights would be necessary in order to establish with certainty if and how
these rights are exercised.
Seventh freedom rights can also be found in the EU single aviation
market alongside eighth and ninth freedom, and are effectively exercised in
that context. For this reason, intra-EU traffic has been considered as
domestic, rather than international, and consequently excluded from the
scope of QUASAR. It would be interesting to assess if the allowance of
seventh freedom has added competition to existing city-pairs or created
brand new services, and if its impact has been limited to the point-to-point/
low-cost segment or has also reached the legacy and hub traffic segment.43
Within the EU, seventh freedom rights seem limited to short-haul traffic.
Even in the case of the EUUS plurilateral agreement, which has allowed
the exercise of ‘one sided’ seventh freedom competition through the CoI
ownership clause (i.e. Air France and Lufthansa flying from London to the
United States), this attenuated form of seventh freedom has not materia-
lized in practice. The reasons remain somewhat mysterious: insurmoun-
table barriers to entry? Lack of demand? Problems of long-haul traffic
organization (hubbing, feedering and corresponding need for slots)?
Customers’ habit of using their own national airlines? Regardless, the lack
of effective implementation of ‘one sided’ seventh freedom rights across the
single largest transatlantic market puts into serious doubt the feasibility of
such seventh freedom competition elsewhere.
Finally, several ASAs grant seventh freedom rights to all-cargo flights.
However, as explained in the second section, the current scope of
QUASAR does not extend to all-cargo information. Therefore, no sys-
tematic analysis of the number, coverage and characteristics of ASAs with
an all-cargo seventh freedom clause has been undertaken. A revised and
extended version of QUASAR would certainly prove to be very helpful in
undertaking this type of analysis. In conclusion, seventh freedom rights in
international scheduled passenger traffic are practically inexistent on paper
and are really not exercised. They seem to be a completely abstract concept,
and their contribution to liberalization would appear null.
Skies Wide Shut 97
Liberalization Through the Withholding/Ownership Clause

The term ‘ownership’ is often a source of semantic confusion in aviation, as
it is used often indistinctly to designate three interrelated but different con-
cepts: the regime on the establishment/acquisition of national carriers, the
designation policy of ones’ own carriers in a bilateral agreement context,
and the withholding clause on the basis of which a bilateral partner’s airline
may be denied access in the same bilateral agreement context.44
Theoretically speaking, there are no constraints on the level of foreign
ownership allowed for the establishment of new airlines, the acquisition of
a pre-existing private airlines or the privatization of an existing publicly
owned airline. There are indeed examples of legislations allowing full or
majority foreign ownership, Australia constituting a case in point with
Virgin Australia (when it was majority UK-owned) and Tiger Airways,
which is Singapore-owned. However, these cases remain rare, as evident by
the partial inventories made by ICAO, IATA and the WTO.45
This scarceness contrasts with most other industrial and services sectors,
where partial or even full foreign ownership is allowed, including in sectors
such as telecommunications and maritime transport that could be dubbed
as ‘strategic’ the same way air transport often is.46 The reason for restrict-
ing foreign ownership in aviation, beyond its purely accidental origin,47
cannot be attributed solely to the prestige attached to the national airline
or its strategic character, a notion that is gradually regressing with privati-
zations. It essentially stems from a very practical constraint. In order to fly
out of its host country, a foreign-owned airline must be not only be desig-
nated by the host country in the relevant bilateral agreement, but it must
also be accepted by the other party to that agreement.
In 2011, in nearly 90 per cent of agreements, accounting for the same
share of total traffic, the withholding clause imposed that the airlines of
both partners be substantially owned and effectively controlled (SOEC) by
the respective nationals of each party. While the interpretation of SOEC is
liable of many variations, the concept clearly does not cover a fully or
majority foreign-owned airline, nor a foreign airline in full management
control with a local sleeping partner, with face-value majority ownership.
Therefore, to continue flying to countries requiring SOEC, such foreign-
owned and/or controlled airlines must either benefit from a fragile unilateral
waiver or obtain from the two governments concerned the re-negotiation of
the withholding clause and the adoption of a ‘laxer’ clause, such as
‘Community of Interest’ (CoI) or ‘Principal Place of Business’ (PPoB).48
The SOEC clause is thus a major deterrent to the international consoli-
dation of airlines through full acquisition or mergers. Transnational
consolidation in the aviation sector has indeed remained the exception

rather than the rule. Foreign involvement does exist but, in most instances,
remains under the 50 per cent threshold and takes non-equity forms such
as management contracts and franchises. These ownership restrictions were
not a problem when airlines were practically all publicly owned, financed
and guaranteed. However, whereas progressive privatization has required
airlines to find private sources of financing, notably for the purchase of air-
crafts, they are confined to national sources of private funding. In develop-
ing countries, but also in small developed countries, those sources are
usually limited. Moreover, in all countries the aviation sector, a tradition-
ally cyclical and low-return sector, competes for funding with more attrac-
tive industries. Consequently, one of the end-results of the SOEC clause is
to strap airlines of much necessary cash.
International takeovers in aviation have so far maintained a certain spe-
cificity compared to international takeovers in other sectors. They have
never really led to full integration, unlike in the maritime transport sector.49
Leaving aside the case of multinational airlines (e.g. SAS, Air-Afrique,
Gulf Air), the main transnational aviation mergers, such as Air
FranceKLM, LufthansaAustrian Airlines and LufthansaSwiss, have
resulted in the survival of two brands, two management structures, two net-
works and two fleets. Those allegedly individual enterprises are in fact close
to the ‘holding model’ adopted by British Airways and Iberia for their
International Aviation Group (IAG).50
The cash-strapping effect of the pervasiveness of the SOEC clause most
likely explains why some airlines, like United Airlines in the United States,
have advocated the relaxation of ownership restrictions, both at the
national and international level. Unfortunately, efforts to relax the inter-
pretation of the existing rules were unsuccessful at both levels. At the
national level, attempts by the US administration were rebuffed by
Congress (see, for instance, Aviationpros, 2007). Internationally, endea-
vours led by IATA, which in 2008 produced the ‘Istanbul Declaration’
advocating the relaxation of the withholding clause, were also unsuccessful.
The representatives of 15 countries had signed that Declaration,51 and a
few months later, a draft ‘OWNCO Convention’ had even been circulated
and discussed among the group. Soon thereafter, however, all discussion
had stopped, possibly due to more pressing concerns for the sector stem-
ming from the financial crisis. Parallel discussions in ICAO did not bring
results either.
Some academics have tried, incidentally using QUASAR (see Jomini,
Achard, & Rupp, 2009), to estimate the potential impact of the entry into
Skies Wide Shut 99
force of the OWNCO Convention. However, this has proven to be a diffi-

cult exercise, as the liberalization of the withholding clause has a less direct
effect on traffic than the liberalization of capacity and routes. Regardless, if
adopted, the Convention would have certainly allowed more transnational
consolidation, more cross-national equity infusions and the emergence of
bigger and probably stronger actors, able to deploy more capacity and
exploit synergies more intensively than alliances, but also better positioned
to control competition and prices.
QUASAR has allowed us to establish that, between 2005 and 2011, the
number of agreements and the share of traffic covered by the SOEC clause
have both regressed. This is largely a consequence of the implementation of
a 2002 judgement by the European Court of Justice (ECJ) that has resulted
in the re-negotiation of the ASAs concluded by EU Member States to
ensure that they included a CoI withholding clause. The adoption of
the CoI clause by EU Member States was not meant as a liberalization
tool nor was it inspired by a liberalization drive.52 It only ensured that
EU Member States conformed to their intra-EU non-discrimination
obligations.53
The European Union has managed to impose the CoI clause to most of
its partners following the 2002 ECJ judgement.54 This move would have
resulted in an improvement of the ALI of EU Member States, but it has
unfortunately proven impossible to account for it in ASAP, as the neces-
sary information about which bilateral agreements have been ‘adapted’ and
now include the CoI clause is not available.55 Only those amended agree-
ments that have been notified to ICAO have been included in the 2011
ASAP dataset. Additionally, it has to be remembered that the EU
‘Community clause’ is asymmetric, and its effects are confined to the EU
side, so that for the other party the withholding clause may continue to be
based on SOEC.
Still, it is fair to note that, even for an actor as big and powerful as the
European Union, the replacement of the SOEC clause with a CoI one has
been difficult to negotiate, especially with other big and powerful actors. It
took years to negotiate it the United States, and it is still absent from the
ASA with Russia, for example, which creates legal uncertainty for the ser-
vices provided by Austrian Airlines and KLM to that country. There seems
to be a negotiating price to be paid for giving legal certainty to a de facto
situation.
Besides the EU and other countries with multinational airlines, there
does not seem at present scope for further expansion of the CoI clause. The
demise of some small- and medium-sized countries’ carriers might
nonetheless prompt the revival of the OWNCO idea of a wide ‘CoI’ club,
or even a PPoB ‘club’, as a way of securing foreing equity infusions.
However, back in 2005 the PPoB clause covered less than 10 per cent of the
traffic, mostly in Asia. Hong Kong and Macau (because of mainland inves-
tors) and Australia and New Zealand were its main supporters. This pic-
ture has not changed dramatically in 2011, when only 8 per cent of the
agreements, accounting for almost 9 per cent of the traffic, include the
PPoB clause.56
The main reason behind the relative lack of success of the PPoB clause,
which is otherwise commonplace in market economies under the name of
‘substantive business operations’, is reportedly the fear of unfair competi-
tion through substandard social and safety norms by so-called ‘flags of con-
venience’. To our knowledge there is no available literature on how this
phenomenon could threaten the aviation sector, but we discuss it in some
detail, and compare it to the situation in the maritime transport sector, in
Appendix B.
Liberalization Through Plurilateral Agreements

As QUASAR has demonstrated, plurilateral ASAs are, on paper at least,
significantly more liberal than bilateral agreements. The ALI of most pluri-
lateral ASAs is at least double that of the bilateral ASAs signed by the
countries concerned. In 2005 and 2011, the average ALI of plurilateral
ASAs stood at 31.4 and 33.0, respectively, as compared to the average ALI
of bilaterals of 9.6 in 2005 and 10.1 in 2011,57 and in most instances, these
agreements scored at least as much as US ‘open skies’ agreements. In 2011,
they involved 154 countries, amounting to almost 80 per cent of ICAOs
190 contracting parties58 and most, if not all, WTO members.
Thus, the suggestion has been made that merging plurilateral ASAs on
the basis of any common liberal features would offer a possible future liber-
alization scenario. How easy would their ‘multilateralization’ be? Most
likely, very difficult, for two main reasons. First, the fact that a given coun-
try has liberalized towards certain partners in a plurilateral context does
not imply that it may be willing to do so with every other partner. For
instance, countries might be ready to open up their markets up for greater
competition from relatively inefficient airlines originating from their pluri-
lateral partners, but not be so favourably inclined towards more competi-
tive airlines from other countries. Or they may treat neighbours and
countries with which they share historical ties differently from partners
with whom they have less obvious ties. A generalized multilateralization
would seem more unlikely in aviation than in a trade context, for instance,
Skies Wide Shut 101
as the balance of concessions in ASAs is always assessed in a narrow bilat-

eral, reciprocal context, whereas trade negotiators tend to think more in
Most Favoured Nation terms.
Second, plurilateral ASAs are somehow ‘limited’. Four issues stand out
in particular: traffic coverage, applied regimes, commercial exploitation
and template of the agreements. With regard to traffic coverage, in spite
of the numerous countries involved, plurilaterals cover just a fraction of
international traffic (12.4 per cent in 2005, 25.6 per cent in 2011). In addi-
tion, in view of their essentially regional nature and of the relatively small
size of the typical parties involved, the traffic covered by a hypothetical
merger would in many instances concern thin, if not inexistent, routes, traf-
fic-wise.
As for the regime effectively applied, there are two sets of issues. First,
not only do multiple plurilateral agreements at times overlap, but some
also overlap with (more restrictive) bilateral agreements, thereby potentially
nullifying the liberalizing effect of their more liberal plurilateral provisions.
The EgyptSudan case provides a case in point. Three different plurilateral
rules (Yamoussoukro, Comesa and ACAC) supposedly apply to this
country-pair, not to mention a fourth set of rules that might be provided
by an EgyptSudan bilateral agreement.59 Second, the extent to which
these agreements have been ratified or have entered into force for each indi-
vidual party concerned is often unclear.60
As regards the commercial exploitation of these agreements, our analysis
does not allow, for the time being, to refute the suspicion that many pluri-
lateral agreements simply might be ‘scraps of paper’.61 In order to assess if
and how the liberal opportunities offered by plurilateral agreements are
effectively exercised, it would be necessary to use OAG-type data in combi-
nation with the QUASAR information for each of these plurilateral agree-
ments. And if, indeed, such an analysis revealed that plurilateral openings
were not actually exploited by airlines, we would still be left wondering if
this was due to absence of commercial viability or of effective implementa-
tion by the plurilateral signatories.
Finally, even if plurilaterals were effectively exploited by the relevant
airlines, their features are too divergent to allow a minimum common
denominator to be easily established. Although superficially similar, for
several plurilaterals it is difficult to find a common denominator, as differ-
ent rules seem to apply to different parties within the same agreement (e.g.
exceptions, optional protocols and the like). Against this background, a
progressive liberalization through the merging of the various plurilateral
agreements looks rather unlikely.
Where Does Liberalization Happen?
On the basis of QUASAR, we have been able to establish that the air trans-
port sector has experienced some liberalization over the past few years, but
also that this has been rather marginal. As previously noted, average
WALIs have improved between 2005 and 2011, but only slightly so.62
‘Open skies’ agreements have remained a rather marginal phenomenon, in
spite of the fact that they imply only partial liberalization. More than ‘open
skies’ agreements, with an ALI higher than 30, and common aviation mar-
ket models along the EU or trans-Tasman lines (with ALIs higher than 40),
have not gained any traction, even for more trade-integrated groupings
such as MERCOSUR or ASEAN.
However, there is one element of liberalization that QUASAR, as it cur-
rently stands, is not able to capture, namely the routes that may be served
and the capacity that may be deployed on the permitted routes. Clearly,
this element is only relevant for agreements that are ‘less than open-skies’,
and that do not provide for total freedom with regards to the choice of
routes to serve and the capacity to deploy, but these are precisely the agree-
ments that account for the largest share of the total (see Chart 1).
Therefore, it is possible that liberalization may be happening ‘under the
QUASAR radar screen’, via the addition of new city-pairs, and the addi-
tion of new capacity on the city-pairs already granted. And media accounts
would seem to suggest that this is indeed the case, although, yet again, the
pace of change appears to be very slow.63 While there are cases of ‘big
bang’ capacity liberalization, with sudden trebling or quadrupling of capa-
city,64 most openings seem of an incremental nature. Anecdotal evidence
and discussions with industry and government representatives would sug-
gest that airlines feel the need for additional capacity only when their load
factors surpass 80 per cent. At that point, they seek to obtain the one or
two additional weekly flights they need to cater for the increased demand,
but do not appear to ask for more.
This leaves one unanswered question. Even in a purely duopolistic con-
text, why do fixed limits on routes and capacity seem to be preferred to
their free determination? Given that international commercial opportunities
in air transport are created exclusively through the channel of intergovern-
mental agreements, and that aviation negotiators essentially defend the
interests of their national airline(s), such limits on capacity should normally
reflect an express desire of the airlines. Why do airlines appear to prefer
capacity to be constrained even in a context where the market is essentially
reserved to two competitors only?
Skies Wide Shut 103
They would appear to believe that their revenues in a capacity-

constrained context will always be higher than revenues in a context where
they would be free to deploy as much capacity as they wished on the routes
that they chose, but might need to charge lower fares. However, this cannot
always be true,65 and may depend of the type of traffic involved (long, med-
ium or short haul, point-to-point, multisegment, low-cost, legacy, business,
economy, etc.), the local propensity to travel, the degree of substitution
between economy and business-class travel depending of the economic con-
juncture, and the like. Apart from anything else, moreover, when capacity
and routes can be self-determined, transaction costs may be expected to be
lower. Unfortunately, there is no viable way of testing these scenarios
empirically, as the profitability of individual routes, which is a closely
guarded business secret, would need to be known. Although price data
does exist,66 airlines’ operating costs on different routes are necessarily
confidential.
Multilateral Liberalization of Air Transport Services?
It is generally accepted that liberalizing air transport services would bring

significant economic benefits. It is not difficult to believe that when eco-
nomic actors are not totally disconnected from market signals, subject to
some degree of competition, and free to make their own basic production
decisions with regard to output quantities and prices, while governments
continue to intervene to address any market failures, superior economic
outcomes will realize. Indeed, although a review of the economic literature
is beyond the scope of the current article, there is abundant empirical evi-
dence pointing to the significant positive effects of air transport services lib-
eralization, in terms of additional opportunities for travel, tourism and
business and, more broadly, employment creation and economic growth
(see, for instance, Arvis & Shepherd, 2011; Fu, Oum, & Zhang, 2010;
Intervistas, 2007; Piermartini & Rousová, 2013). Against this background,
this section will first discuss the technical feasibility of progressively liberal-
izing the air transport sector on a multilateral basis, and then speculate on
how a hypothetical, totally liberalized aviation industry might look like
and, in so doing, suggest ideas for a possible research agenda.
Is It Technically Feasible?
From a technical and legal perspective, a multilateral process of progressive
liberalization of the air transport industry is not impossible to envisage.
Contrary to a commonly held belief, a multilateral organization of the

aviation sector does not necessarily imply the need for a full application
of the MFN principle. The house of multilateralism par excellence, the
WTO, for instance, had managed a bilateral textile export quotas system
(which, mutatis mutandis, resembles the air transport traffic rights sys-
tem) for more 20 years, until it evolved into a free market situation.
More recently, the WTO saw the successful extension of the Government
Procurement Agreement (GPA). The GPA is based on a conditional
application of the MFN obligation, whereby Country A opens up its gov-
ernment procurement market to countries B and C, but not countries D
and E, if it is not satisfied by the level of concessions offered the latter.
Against this background, it would be possible to imagine a similar
mechanism at play in the air transport sector, which we term ‘Optional
Dynamic Conditional MFN Mechanism’ (ODCMM). For example, if
country A had a PPoB withholding clause in its ASA with country B, and
country B had the same PPoB clause in its ASA with country C, it might
be possible, at least theoretically, for A and C to explore the possibility of
including the PPoB clause in their own ASA, thereby ‘closing the trian-
gle’.67 More broadly, if country A and country B had an agreement con-
taining the PPoB clause, and if country C and country D also had an
agreement containing the same PPoB clause, then it might be possible to
conceive of any agreements between A and C, A and D, B and C and B
and D also including this clause.68 Still, the sheer number of potential com-
binations is so high69 that it would imply the need for a structured mechan-
ism to apprehend the totality of the negotiating potential. Such a
mechanism could be based on the aeropolicy comparator concept devel-
oped in the third section, provided that the detailed information about the
ASA features was to be made publicly available.70
The ODCMM process is illustrated in Chart 2. Let us assume that A to
K are 11 different countries, and that shaded, dark grey and star-shaped
cells represent, respectively the SOEC, CoI and PPoB variants of the with-
holding clause. Starting from the status quo in Stage 1, countries are first
re-ordered on the matrix to as to bring those with the liberal PPoB clause
close together (Stage 2). This enables an immediate identification of a ‘criti-
cal mass’ of potential candidates for liberalization, i.e. the acceptance of a
PPoB withholding clause in our example (Stage 3). The process would be
dynamic: if two ‘isolated’ countries were to subsequently conclude a liberal
agreement (Stages 4 and 5), this would trigger a potential wave of added
liberalization (Stage 6).
Skies Wide Shut 105
Stage 1: Status quo Stage 2: Status quo re-ordered
A B C D E F G H I J K A C E G I J B D F H K
A A
B C
C E
D G
E I
F J
G B
H D
I F
J H
K K
Stage 3: Potential liberalization options Stage 4: New "isolated" liberal agreement
A C E G I J B D F H K A C E G I J B D F H K
A A
C C
E E
G G
I I
J J
B B
D D
F F
H H
K K
Stage 6: Potential liberalization options triggered by the

Stage 5: New "isolated" liberal agreement re-ordered new liberal agreement
A C E G I J H D F B K A C E G I J H D F B K
A A
C C
E E
G G
I I
J J
H H
D D
F F
B B
K K
Chart 2. The Functioning of the Optional Dynamic Conditional MFN Mechanism.

Withholding
face
Designation face
Tariffs face
Freedoms Capacity face

face
Cooperative
arrangements and
statistics face
Chart 3. The ODCMM Rubik’s Cube.
The matrix could clearly be enlarged to 190 ICAO members

and extended to cover all features of bilateral agreements. By selecting
the features currently captured by QUASAR, a ‘Rubik’s cube’ would
appear along the lines illustrated in Chart 3, which would provide a useful
tool to identify untapped scope for liberalization. The cube would trans-
form into a prism if other ASA aspects, such as the capacity granted, were
added.
A further step would see potential liberalization options progressively
becoming less ‘optional’. One could envisage, for instance, a transitional
opt-out that would need to be justified and explicitly renewed, a de minimis
threshold, transitional periods varying with countries’ level of development,
provisional safeguards, or asymmetric concessions.
To a certain extent, the Air Services Negotiation Conference (ICAN)
mechanism established by ICAO could be used as the embryo of such a lib-
eralization mechanism. ICANs are annual meetings where air transport
negotiators negotiate, or re-negotiate, ASAs with each other.71 This process
bears some similarity with the periodic gathering of WTO trade negotiators
when bilaterally exchanging requests and offers on the elimination of trade
barriers. Clearly, there is a key difference between the two processes,
namely the MFN applicability of those eventual concessions in a WTO
context, and not so in an aviation one. Still, the simple procedural idea is a
significant operational contribution to liberalization. Through ICAN,
ICAO has created a framework for the collective exchange of rights which
entails much lower transaction costs, in terms of time and resources, than
in a purely bilateral setting.
Skies Wide Shut 107
What Might It Look Like?

Let us speculate now and, based on the scattered information we have been
able to reorganize and analyse with QUASAR and the liberalization experi-
ences we are aware of, try to imagine what a hypothetical, totally liberal-
ized aviation industry might look like.
For short- and medium-haul traffic, the situation is likely to be similar
to that of deregulated domestic markets, such as the United States, and
common aviation markets, such as the Trans-Tasman or the intra-EU mar-
kets. The low-cost, multibase model would flourish, new city-pairs would
be served, even though some routes might eventually be eliminated,
seventh, eighth and ninth freedom rights would be effectively used and new
carriers would emerge, even if others would disappear and the competitive
landscape would probably look more concentrated in the longer term.
However, it is difficult to venture much further, also because, differently
from the effects of the US deregulation, which have been extensively ana-
lysed, very little is known about the impact of the Trans-Tasman or intra-
EU liberalization.
In this regard, useful pointers could come from assessing the market
developments in countries that have recently acceded to the European
Union. The EU accession process has, in fact, implied a very sudden
change from a bilateral, restrictive regime with their EU partners, to a fully
liberalized one through the immediate implementation of the Single
Aviation Market rules. A contrario, interesting lessons might also be drawn
from examining the situation within ASEAN, a short- to medium-haul
market where, in spite of an overall moderately restrictive aviation regime,
successful low-cost carriers have emerged.
For long-haul traffic, we have even fewer reference points. Firstly, no
long-haul country-pair has been fully liberalized. Even if a country were to
conclude a fully liberal agreement with a long-haul partner, any associated
effects could be nullified by third parties vetoing any fifth or seventh free-
dom traffic rights or imposing an SOEC withholding clause. As a result,
the effects of long-haul liberalization could be properly realized only if a
sort of global ‘big bang’ were to take place, with all major partners opening
up their markets simultaneously, an occurrence that is very difficult to con-
ceive in spite of traffic being very concentrated.72
Secondly, conflicting signals seemingly come from the two main aircraft
manufacturers about how they expect traffic to evolve in the future. While
Boeing’s 787 aircraft reflects more of a point-to-point prediction, Airbus’s
A380 points to expectations of growing hub-and-spoke, or even hub-to-
hub, traffic. All that can be said about long-haul traffic with some degree
of certainty is that it requires density of demand and, hence, in most

instances, at least a proportion of ‘feedering’73; as such, it may be difficult
to imagine situations with self-standing, point to point, seventh freedom
flights.
Should hubbing remain a key feature of the long-haul market, it may be
possible to imagine full liberalization translating into a double hub struc-
ture, ‘bone-shaped’, with airlines setting up in two distant hubs, for
instance Berlin and Abu Dhabi.74 Airlines would then radiate to spokes
from both hubs, and possibly combine longer-haul six freedom flights
between them, so as to create the greatest possible synergies from the com-
bination of two networks. Should more than two hubs be involved, the
structure would resemble the multi-base model of the short haul, low-cost
carriers. Alliances’ networks may also not change at all, however, given
that they already maximize hubbing possibilities (fifth freedom flights are
often replaced by code-share flight operated by the local alliance partner).
These unchanged networks would be simply operated under as a single
brand, or even under several local brands, by a single owner.
Such liberalization scenarios are highly unlikely, however. In the mari-
time transport sector, it took exactly two centuries until Disraeli abolished
the very restrictive Navigation Acts. Unless the bilateral system in avia-
tion75 implodes because of the competitive inroads of long-haul low-cost
operators, the demise of purely national, cash-strapped carriers or consu-
mer demand, we believe that its abolition may take just as long.
CONCLUDING OBSERVATIONS
QUASAR represents a bona fide attempt to close a gap in the literature on

the regulation of air transport services. For the first time, it has enabled a
systematic assessment of the degree of liberalization granted through bilat-
eral and plurilateral ASAs and the passenger traffic they cover. It has also
permitted a critical evaluation of five commonly held beliefs about air
transport services liberalization. Its updating has further allowed to shed
some light on the way in which such market openings are evolving.
What QUASAR has been able to quantify is that skies are not really
open. Additionally, although the sector has experienced some liberalization
over the past few years, this has been rather marginal. Indeed, the way in
which the sector is regulated internationally is unique among all productive
activities,76 with heavy government involvement and considerable
Skies Wide Shut 109
disconnect from market signals. The fact is that, in the 21st century, jet air
transport is still regulated in the same restrictive way in which sailboat
transport was in the 17th century, with the same bilateral mercantilist
approach.
At the same time, however, the results obtained through the application
of QUASAR have also exposed some of the methodology’s limitations.
These include, first and foremost, the fact that any information on routes
and capacity granted is not factored into the measure of ASAs’ degree of
openness, thereby overlooking a relevant source of liberalization. Second,
the lack of a sufficient historical dimension: the current availability of data
for only two years, additionally separated by a six-year interval, may dis-
tort the picture of how the sector is truly evolving. Third, the fact that
QUASAR covers only one segment of the aviation industry. It would be
especially valuable to extend QUASAR by applying it to cargo transporta-
tion, where seventh freedom appears prevalent and effectively used, and
price controls seem inexistent, and to domestic passenger transport, where
fully liberalized markets have, in some countries, been in existence for a
few decades. These extensions could provide useful information on the
basis of which to test many of the hypotheses we have put forward and
examine the possible effects of liberalization.
Indeed, QUASAR’s shortcomings are not insurmountable. We have pre-
sented a number of possible avenues through which they could be
addressed, such as the ODCMM or the Rubick’s cube. They would enable
QUASAR to be both refined and extended, including with the aim of quan-
tifying the welfare impact of air transport liberalization. We conceived
QUASAR not as an end in itself, but as a means to an end; not as a set of
definitive conclusions, but rather as a list of open questions underpinned by
a toolbox that would enable their consideration beyond the analytical work
it has already generated.77
NOTES
1. ICAOs contracting parties do not always conscientiously abide by their regis-

tration obligation.
2. The results of this analysis are contained in the WTO document S/C/W/270/
Add.1, dated 30 November 2006.
3. Throughout this document, the terms ASAP and QUASAR will be used
interchangeably.
4. Complemented by ASA information received from Australia, Guatemala and
New Zealand in 2007.
5. The application can be consulted at http:\\www.wto.org/asap.

6. We consulted the 2011 edition, and qualify this as the ‘latest edition’,
although we understand that the WASA database is being continuously updated.
However, as the database does not include a function permitting the identification
of the most recently added agreements, we consider the latest edition as the 2011
one, last consulted on 31 August 2012.
7. Information was extracted from the following Trade Policy Review Reports:
China (WT/TPR/S/264/Rev.1, dated 20 July 2012), Colombia (WT/TPR/S/265/
Rev.2, dated 1 August 2012), Guyana (WT/TPR/S/218/Rev.1, dated 10 August
2009), Japan (WT/TPR/S/276, dated 15 January 2013) and Norway (WT/TPR/S/
205/Rev.1, dated 16 January 2009).
8. The term ‘withholding’ clause is used by ICAO, but this clause is often also
referred to as ‘designation clause’ or, more frequently, ‘ownership clause’, given
that the ‘standard’ withholding clause requires that airlines be ‘substantially owned
and effectively controlled’ by nationals of the country in question.
9. For instance, with respect to the tariff clause, ‘dual approval of tariffs’, a
very restrictive provision, was attributed zero points, whereas ‘free pricing’, the
most liberal of the tariff provisions, is given eight points.
10. As an alternative, weights may be assigned through the use of a purely statis-
tical technique. Piermartini and Rousová (2013) in particular, constructed an alter-
native air liberalization index by means of factor analysis. Comparing it to the ALI,
they found that the two indices were highly correlated, with a correlation coefficient
of 0.97.
11. At http:\\www.icao.int/sustainability/Documents/RegionalAgreements.pdf
12. Available at http://clacsec.lima.icao.int
13. The revision has led to the exclusion of four agreements (BIMP-EAGA,
IMT-GT, ASEAN Roadmap and WAEMU) for which missing information made it
impossible to compute the ALI both in 2005 and in 2011. Another three agreements
(ECAA, EURO-MED and US-EU) were excluded from the 2005 dataset because
they were actually signed post-2005, but were included in the 2011 dataset.
14. It is often difficult to get information about the effective entry into force of
plurilateral ASAs, and even for those that have entered into force, the degree of
effective implementation varies considerably, according to both operators and the
parties concerned.
15. WALI information for individual countries and groups of countries may be
obtained from the ASAP application.
16. Indeed, Piermartini and Rousová (2013) control for the possible problem cre-
ated by data mismatch and find that it does not invalidate the results of their
analysis.
17. On-Flight Origin Destination (OFOD) ICAO data show the origin and desti-
nation of a passenger on the basis of the flight number, and as such would match
the regulatory regime under which passengers travel. However, only around a quar-
ter of the OFOD data ICAO have at its disposal are released externally to the
public.
18. Corrections and exclusions were possible, inter alia, due to the fact that the
texts of the agreements, and not only the codified summaries, were available for
analysis in 2012.
Skies Wide Shut 111
19. This is justified also in light of the fact that the four weighing methodologies
have been found by Piermartini and Rousová (2013) to be highly correlated (over
90 per cent).
20. These are 169 more agreements (representing an 8.6 per cent increase)
and two more countries compared to the dataset captured in the ‘original’
QUASAR.
21. 2.3 per cent traffic than in the original QUASAR.
22. See, for instance, the agreements between China and the Republic of Korea
(ALI 4 for over 9 million passengers); Turkey and Germany (ALI of 12 for over 7
million passengers); and the United States and Mexico (ALI 17 for over 17 million
passengers).
23. The reduction in ASAs granting seventh freedom rights compared to the ori-
ginal QUASAR is due to the fact that the text of the agreements was available in
2011, but not so in 2005. This allowed us to verify that, in the majority of cases,
seventh freedom traffic rights were granted exclusively for cargo traffic and not for
passenger traffic. Thus, for the purposes of the QUASAR analysis, which focuses
on passengers, such agreements were excluded from the total number of ASAs
granting seventh freedom rights.
24. WTO document S/C/W/270/Add.2, dated 10 September 2007.
25. Consistent statistical IATA data only go as far back as 1995.
26. Air services agreements contain confidential clauses that are neither published
nor notified to ICAO.
27. As discussed, consistent IATA traffic data would be available as of 1995.
28. By way of illustration, a number of the ASAs collected through the WTO’s
TPR process, or listed in the Reform in International Transport Services (RITS)
database of the Groupe d’Economie Mondiale in Paris (http://gem.sciences-po.fr/
content/research_topics/trade/transportation_resources_EN.htm) are not captured
in ICAO’s WASA.
29. Routes cover more than just city-pairs as they may also include the setting of
intermediary, behind and beyond points than can be served under fifth freedom
rights.
30. An additional constraint came from the fact that QUASAR, which was
developed as part of a WTO-mandated exercise, had to be produced in a relatively
short period of time with limited resources.
31. Although it is not possible to give precise figures, anecdotal evidence would
suggest that this information is available in less than 20 per cent of cases, open skies
agreement being, by definition, not concerned.
32. Allowing airlines of A and B to fly between any city in A and any city in B,
via any intermediate, behind or beyond points.
33. For instance, Piermartini and Rousová (2013) find that 96 per cent of agree-
ments containing the restrictive dual approval pricing clause also require a SOEC
withholding regime.
34. At this first stage, the analysis is limited to long-haul traffic (i.e. longer than
4000 km), so as to avoid the multiplication of airports, short and medium-haul
routes being more diffuse than long-haul ones. By way of fictitious illustration,
Delhi might account for 60 per cent of India’s total long-haul traffic, Mumbai for
25 per cent, Kolkata for 10 per cent, Chennai for 3 per cent and Hyderabad for 2
per cent, while on the German side Frankfurt might make up 60 per cent of
Germany’s total long-haul traffic, Berlin 15 per cent, Munich 15 per cent, Hamburg
8 per cent and Rostock 2 per cent.
35. This might be inscribed as a ‘none’ when the city-pair has not been
granted by the ASA (and therefore no capacity will be allowed), ‘open ended’ when
the city-pair has been granted without any quantitative capacity restrictions, or in
numerical terms when a limit on capacity has been set (e.g. number of passengers
per week).
36. Relevant information may be accessed at http:\\www.cepii.fr/CEPII/fr/bdd_
modele/presentation.asp?id=8.
37. It must be underlined that QUASAR has been conceived as an open source
methodology, with the aim of facilitating its extension.
38. Incidentally, this explains the disproportionate importance in the sector of
carriers emanating from small trading nations, such as Luxemburg’s Cargolux.
39. In trade circles, this kind of non-discrimination is also referred to as the
granting of Most Favoured Nation (or MFN) treatment.
40. The expression ‘open skies’ has no single definition. The United States uses
the term to designate ASAs with a substantial ownership and effective control
clause, fifth freedom rights but no seventh freedom rights, no limitation on routes,
tariffs or capacity. Other countries (e.g. Australia and New Zealand) use the term
to refer to ASAs that include the principal place of business clause and grant
seventh freedom and, in certain instances, cabotage rights.
41. Yet again, it is puzzling that public complaints of ‘dumped’ capacity are
never heard.
42. Examples are the EU and neighbouring countries, the Trans-Tasman region
and ASEAN.
43. We are not aware of any such studies, but our documentation may be
incomplete.
44. For more information on the interaction between those three concepts, see
WTO document S/C/W/270/Add.2, pp. 405421.
45. For developments until 2006, see WTO document S/C/W/270/Add.2,
pp. 414418 and 420. To our knowledge, these inventories have not been updated.
The newly available WTO I-TIP services database (http://i-tip.wto.org/services/)
contains recent relevant raw data that would allow such an update.
46. This openness is often bound by relevant GATS or regional trade agreements
commitments.
47. The fears raised in the United States in the late thirties by the control of a
Colombian airline by German interests, which originated from the need for German
military pilots (whose number was limited by the Versailles treaty) to be trained
clandestinely somewhere. The name of the charter subsidiary of Lufthansa,
Kondor, perpetuates the memory of it. The military provisions of the Versailles
treaty created the conditions that led to the creation of the most restrictive, durable
and pervasive provision in ASAs. Incidentally, long-distance passenger vessels,
whose services were replaced by commercial aviation in the fifties and sixties, oper-
ated already in the 20th century in a liberal investment and traffic regime, with, for
instance, unlimited fifth freedom rights. One has to go back to the 18th century to
find comparable restrictive provisions in the maritime passenger transport sector.
Skies Wide Shut 113
48. ‘Community of Interest’ allows for ownership to be split amongst a group of

countries, whereas ‘Principal Place of Business’ refers to ‘establishment’, in EU
terms, or ‘substantive business operations’, in trade terms.
49. Maritime examples are Sealand (US), P&o (UK)Nedlloyd (NL) with
Maersk (DK), Australia National lines (AUS), Wan-Hai shipping (TW) with
CMACGM (F). For an overview of consolidation in the international maritime
transport see WTO document S/C/W/315, pp. 1416.
50. To our knowledge, there is no study comparing the degree and scope of co-
operation within internationally merged airlines and within alliances. There also
appears to be no study analysing the economic and regulatory reasons that have
apparently prevented full integration.
51. Australia, Brazil, Canada, European Union, India, Mauritius, Morocco,
Panama, Singapore, Switzerland, Turkey, United Arab Emirates, United States and
Vietnam, plus, later on, New Zealand.
52. Save possibly for KLM and Austrian Airlines, for whom the inclusion of CoI
provided definitie legal certainty, instead of the fragile waiver regime to the SOEC
clause they had relied on beforehand.
53. In this respect, it can be compared to the maritime ‘Brussels Package’
(Regulation 954/79). The Regulation ‘communitarized’ and virtually redistributed
the shares of traffic that belonged to the various EU Member States in their relation
with third parties under the 40/40/20 traffic allocation system of the United Nations
Liner Code of Conduct for Maritime Conferences. The ‘communitarization’ and
redistribution of the shares was never intended to become a practical reality but the
mere legal existence of the regulation ensured conformity with EU law.
54. The clause has been introduced in 979 agreements as of July 2011.
55. The only information available concerns the number of agreements that have
been adapted. See http://ec.europa.eu/transport/modes/air/international_aviation/
external_aviation_policy/doc/table_-_asa_brought_into_legal_conformity_since_
ecj_judgments-_january_2013.pdf.
56. Switzerland is one exception, as the inclusion of a PPoB clause in Swiss
ASAs is essential to legally guarantee the continued operations of Swiss, which is
German-owned. Regardless, the Swiss authorities do not always succeed in nego-
tiating the inclusion of such a clause, and there are cases where Swiss’ operations
are covered only by a waiver to the SOEC clause.
57. The most notable exception is the Fortaleza agreement, which has an ALI of
13 both in 2005 and 2011.
58. Less than 1 per cent of countries parties to plurilateral agreements are not
ICAO Contracting Parties. They are Aruba, Dominica, Niue, Tuvalu and United
Nations Mission in Kosovo (UNMIK).
59. Although no such agreement is captured by ASAP, it might nonetheless exist.
60. It took an entire PhD thesis to establish the state of play for the
Yamoussoukro decision. See Schlumberger (2009), available at: http://digitool.
library.mcgill.ca/webclient/StreamGate?folder_id=0&dvs=1376991205575∼994.
61. Expression used by the German chancellor Bethmann Hollweg to qualify the
1831 treaty guaranteeing the neutrality of Belgium in 1914.
62. Although in certain instances changes have been more significant, such as in
the case of Japan, whose WALI increased from 15 in 2005 to over 23 in 2011.
63. In this regard, it should be noted that QUASAR, as it currently stands, is not
equipped to measure such liberalization precisely, but that, as discussed in the third
section, a refined and extended version of the methodology could help quantify it
more precisely.
64. A case in point is the Memorandum of Understanding signed by Germany
and India in 2005, which increased capacity for the 2005 winter season from 16
flights a week, accounting for 6,800 passengers, to 42 flights a week, representing
16,800 passengers, and to 50 flights a week, representing 20,000 passengers, as of
the 2006 summer season.
65. Airlines serving ‘open skies’ destination where capacity and routes are freely
determined may indeed be seen as proving the point.
66. As might be available, at a significant cost, from IATA and computer reser-
vations systems via Marketing Information Data Tapes (MIDTs), for instance.
67. Clearly, things may be very different in practice. For instance,
despite Australia being a staunch supporter in principle of fifth freedom rights, it
has not granted them to Singapore Airlines between Australia and the United
States.
68. In practice, out of all possible agreements many may never be concluded,
either for geographical or commercial and economic reasons.
69. There are 190 × 190 States, amounting to over 36,100 potential bilateral rela-
tions. Of these, a good 5,000 would probably make commercial and geographic
sense.
70. Presently, the individual ASA features on the basis of which the ALIs are cal-
culated are not publicly accessible via ASAP, as the application relies in part on
ICAOs WASA proprietary ASA information.
71. The December 2012 event resulted in 130 new or amended agreements.
See http://www.icao.int/Newsroom/Pages/centralized-ICAO-negotiation-event-pro-
duces-130-new-air-services-agreements.aspx
72. Traffic was very concentrated in 2005 and only slightly less so in 2011, when
116 agreements out of 2,224 covered over 1 million passengers and involved 56
parties.
73. This essentially maritime concept refers to the feeding of traffic to the hub.
74. Etihad, for instance, has acquired a minority stake in Air Berlin. Full control
would not have been possible.
75. It should be noted that, contrary to a very common misperception,
the Chicago Convention is silent on bilateral ASAs, and was in fact
conceived in the perspective of a future liberal, multilateral regime that never
materialised.
76. Save possibly for international road transport, which is also governed by a
bilateral, quota-based State-administered system but where national ownership
requirements are nonetheless much less prevalent.
77. For a list of those works, see http://www.wto.org/english/tratop_e/serv_e/
transport_e/transport_air_e.htm#asap.
78. For more on flags of convenience, on the demise of the air-like mercantilist
model in maritime and on maritime generally see our work for WTO in this regard:
WTO documents S/C/W/62 dated 1998, S/CSS/W/106 dated 2001 and S/C/W/315
dated 2010.
Skies Wide Shut 115
REFERENCES
Arvis, J. F., & Shepherd, B. (2011). The air connectivity index: Measuring integration in the
global air transport network. The World Bank, June. Retrieved from http://elibrary.
worldbank.org/content/workingpaper/10.1596/1813-9450-5722
Aviationpros. (2007). Like republicans, resistant to rules change on U.S. airline ownership.
January 29. Retrieved from http://www.aviationpros.com/news/10392997/democrats-
like-republicans-resistant-to-rules-change-on-us-airline-ownership
Fu, X., Oum, T. H., & Zhang, A. (2010). Air transport liberalization and its impacts on airline
competition and air passenger traffic. Transportation Journal, 49(4), 2441.
International Civil Aviation Organization. (2006). Regulatory and industry overview.
Information paper presented by the ICAO Secretariat at the 2006 Global Symposium
on Air Transport Liberalisation held in Dubai, August 15. Retrieved from http://www.
icao.int/cgi/goto_m_atb.pl?icao/en/atb/ecp/dubai2006/index.html
Intervistas. (2007). The economic impact of air service liberalization. Retrieved from http://
www.intervistas.com/downloads/Economic_Impact_of_Air_Service_Liberalization_
Final_Report.pdf
Jomini, P., Achard, P., & Rupp, J. (2009). The Istanbul declaration and the agenda for freedom
summit An opportunity for broad liberalisation. GEM Working paper, 2009. Retrieved
from http://www.gem.sciences-po.fr/content/publications/pdf/Jomini-Achard-Rupp_
Istanbul05022009.pdf
Piermartini, R., & Rousová, L. (2013). The sky is not flat: How discriminatory is the access to
international air services? American Economic Journal: Economic Policy, 5(3), 287319.
Schlumberger, C. (2009). The implementation of the Yamoussoukro decision. Mac Gill
University, February. Retrieved from http://digitool.library.mcgill.ca/webclient/
StreamGate?folder_id=0&dvs=1380188319465~904
World Trade Organization. (1998). Maritime transport services, note by the Secretariat, S/C/
W/62, November 16. Retrieved from http://www.wto.org
World Trade Organization. (2001). Maritime transport services, note by the Secretariat, S/
CSS/W/106, October 4. Retrieved from http://www.wto.org
World Trade Organization. (2006). Second review of the air transport annex, developments in
the air transport sector (part two), Quantitative Air Services Agreements Review
(QUASAR), note by the Secretariat, S/C/W/270/Add.1, November 30. Retrieved from
http://www.wto.org
World Trade Organization. (2007). Second review of the air transport annex, developments in
the air transport sector (part three), note by the Secretariat, S/C/W/270/Add.2,
September 10. Retrieved from http://www.wto.org
World Trade Organization. (2009a). Trade policy review, Guyana, note by the Secretariat,
WT/TPR/S/218/Rev.1, August 10. Retrieved from http://www.wto.org
World Trade Organization. (2009b). Trade policy review, Norway, note by the Secretariat,
WT/TPR/S/205/Rev.1, January 16. Retrieved from http://www.wto.org
World Trade Organization. (2010). Maritime transport services, note by the Secretariat, S/C/
W/315, June 7. Retrieved from http://www.wto.org
World Trade Organization. (2012a). Air Services Agreements Projector (ASAP), 2011 Edition.
Retrieved from http://www.wto.org/asap
World Trade Organization. (2012b). Trade policy review, China, note by the Secretariat, WT/
TPR/S/264/Rev.1, July 20. Retrieved from http://www.wto.org
World Trade Organization. (2012c). Trade policy review, Colombia, note by the Secretariat,
WT/TPR/S/265/Rev.2, August 1. Retrieved from http://www.wto.org
World Trade Organization. (2013a). I-TIP database, services module. Retrieved from http://
i-tip.wto.org/services/
World Trade Organization. (2013b). Trade policy review, Japan, note by the Secretariat, WT/
TPR/S/276, January 15. Retrieved from http://www.wto.org
WEBSITES
Centre de recherche français dans le domaine de l’économie internationale

(CEPII), Gravity dataset. Retrieved from http://www.cepii.fr/CEPII/fr/
bdd_modele/presentation.asp?id=8
European Commission. Retrieved from http://ec.europa.eu/transport/
modes/air/international_aviation/external_aviation_policy/doc/table_-_asa_
brought_into_legal_conformity_since_ecj_judgments-_january_2013.pdf
Groupe d’Economie Mondiale, Reform in International Transport Services
(RITS) database, Paris. Retrieved from http://gem.sciences-po.fr/content/
research_topics/trade/transportation_resources_EN.htm
International Civil Aviation Organization. Retrieved from http://www.
icao.int/Newsroom/Pages/centralized-ICAO-negotiation-event-produces-
130-new-air-services-agreements.aspx
International Civil Aviation Organization. Retrieved from http://www.
icao.int/sustainability/Documents/RegionalAgreements.pdf
Latin American Civil Aviation Commission, Multilateral Open Skies
Agreement. Retrieved from http://clacsec.lima.icao.in
Skies Wide Shut 117
APPENDIX A
Table A.1. ALI Features and Variants.

ALI Features and Definition Degree of
Variants Openness
Grant of rights
Fifth freedom The right to carry freight and passengers between two Liberal
countries by an airline of a third country on route
with origin/destination in its home country.
Seventh Freedom The right to carry freight and passengers between two Liberal
countries by an airline of a third country on a route
with no connection with its home country.
Cabotage The right to carry freight and passengers within a Liberal
country by an airline of another country on a route
with origin/destination in its home country or with no
connection with the home country.
Capacity
Free determination Capacity is to be decided by designated airlines free Liberal
of regulatory control.
Bermuda I The governments set out the capacity principles for Semi-
the designated airlines to follow but allow each airline liberal
the ab initio freedom to determine its own capacity,
subject only to ex post fact review by the governments
through their consultation procedure.
Pre-determination Capacity is to be agreed to prior to the Restrictive
commencement of the operation, either by
governments or their aeronautical authorities or
between their designated airlines subject to
government approval.
Other liberal Capacity provisions cannot be classified as any one of Fairly
the above three types but provide for the exclusion of liberal
unilateral capacity controls.
Other restrictive Capacity provisions cannot be classified as any one of Fairly
the above three types but provide for the setting of restrictive
some general capacity principles, a formula to control
capacity or a requirement to file capacity.
Pricing
Dual approval It requires the tacit or express approval by both Restrictive
parties of tariffs or agreements on tariffs before those
tariffs can take effect.
Dual disapproval Tariffs become effective unless both aeronautical Semi-
authorities disapprove them. liberal
Country of origin A party may disapprove tariffs only for originations Semi-
in its own territory. restrictive
Free pricing Tariffs shall not be subject to the approval of any Liberal
party.
Table A.1. (Continued )

ALI Features and Definition Degree of
Variants Openness
Zone pricing It involves a reference point or points around which Fairly

various types of tariff control are agreed. The parties liberal
agree to approve tariffs falling within a specified
range of prices and meeting corresponding
conditions, though tariff filing may still be necessary.
Outside of the zone, one or a combination of the
other regimes may apply.
Withholding
Substantial ownership A party would accept a foreign designated airline to Restrictive
and effective control operate the agreed services under the condition that
substantial ownership and effective control is vested
in the designating party or its nationals.
Community of interest A party would accept a foreign designated airline to Semi-
operate the agreed services under the condition that liberal
substantial ownership and effective control is vested:
(a) in one or more countries that are parties to the
agreement or by any one or more of the parties
themselves, that is a joint operating organization or a
multinational carrier created by intergovernmental
agreement; or (b) in one or more countries that are
not necessarily party to the agreement but are within
a predefined group with a ‘CoI’.
Principal place of A party’s acceptance of a foreign airline if the carrier Liberal
business is incorporated in the designating party and its PPoB
or permanent residence is also in the designating
party, including one incorporated and having its
PPoB in, and effectively controlled by the designating
party, which removes the substantial ownership
requirement.
Designation
Single designation Each party may designate one airline to exercise the Restrictive
rights to operate the agreed air services.
Multiple designation Each party may designate more than one airline to Liberal
exercise the rights to operate the agreed air services.
Exchange of statistics
No exchange of statistics The Contracting States or their airlines are not Liberal
required to exchange statistics.
Exchange of statistics The Contracting States or their airlines are required Restrictive
to exchange statistics.
Not allowed The designated airlines cannot enter into co-operative Restrictive
marketing agreements.
Allowed The designated airlines may enter into co-operative Liberal
marketing agreements.
Skies Wide Shut 119
Table A.2. ALI Features: Share of Agreements and Traffic, 2005

and 2011.
ALI Feature Percentage of Percentage of
ASAs Traffic
2005 2011 2005 2011
Grant of rights
Only fifth freedom 65.5 65.7 78.7 76.4
Fifth and seventh 0.2 0.1 0.1 0.1
Fifth and cabotage 0.2 0.2 1.2 1.1
Fifth, seventh and cabotage 0.0 0.1 0.0 0.2
Seventh only 0.0 0.0 0.0 0.0
Seventh and cabotage 0.0 0.0 0.0 0.0
Cabotage only 0.0 0.0 0.0 0.0
No fifth, no seventh and no cabotage 34.0 33.8 20.0 22.3
Capacity
Free determination 5.9 7.7 21.5 25.6
Bermuda I 16.0 14.9 24.3 16.1
Pre-determination 68.7 68.3 43.5 48.8
Not available 2.5 2.4 0.6 0.8
Other liberal 0.8 0.9 0.3 0.7
Other restrictive 5.9 5.6 9.3 8.0
Other (unspecified) 0.1 0.0 0.4 0.0
Pricing
Dual approval 81.4 78.5 66.8 61.3
Dual disapproval 9.0 9.6 20.9 19.4
Country of origin 2.9 4.0 2.5 3.0
Free pricing 1.7 2.7 4.2 5.4
Zone pricing 0.3 0.8 2.1 7.3
Withholding
Substantial ownership and effective control 88.5 86.6 88.2 87.7
Community of interest 1.1 1.4 0.2 0.2
Principal place of business 7.3 8.2 9.4 8.8
Designation
Single designation 42.8 40.7 16.5 17.9
Multiple designation 57.2 59.3 83.5 82.1
Exchange of statistics
No exchange of statistics 23.5 23.7 38.3 37.1
Exchange of statistics 76.5 76.3 61.7 62.9
Not allowed 87.1 83.8 63.8 60.9
Allowed 12.9 16.2 36.2 39.1
APPENDIX B: ‘FLAGS OF CONVENIENCE’ IN AIR

TRANSPORT A FIRST APPRAISAL
Flags of convenience originated in the Black sea in the 19th century with
Greek owners under Turkish domination using the Russian flag. They
developed with the First World War so as to maintain some safe maritime
traffic by belligerent ship-owners under the flags of neutral countries (at the
time Holland, the Scandinavian countries, Spain and the United States
until 1917). The US prohibition was the next stage of their development so
as to avoid too tight controls by the coast guards patrols. That is when the
Panama registry started to flourish. The Second World War had the same
effects as the first one and it is only after it, notably during the shipping
boom provoked by the Korea war where Greek ship-owners using
Panamanian and Liberian flag to carry oil made fortune, that flags of con-
venience became a way to escape company taxes, high labour costs and
safety regulations and controls.
Very quickly flags of convenience became dominant in the segments of
dry bulk (iron ore, bauxite coal and grains) and wet bulk (oil) where com-
petition was more cut throat than in the liner sector. The deployment of
this fleet was allowed because there was no such thing as a 50/50 bilateral
cargo sharing for bulk maritime trade as it exists for aviation passengers.
As of then, cross trade, i.e. seventh freedom in aviation, started reigning
sovereign: a Panamanian or a Liberian ship never served Panama or
Liberia, nor did they, by the same token, had Panamanian or Liberian sai-
lors. Countries or territories such as Singapore, the Isle of Man, Gibraltar,
Cyprus or Malta also opened their registries/flags to foreign ownership (the
shipping professionals prefer the euphemism ‘open registries’ to ‘flag of
convenience’).
From the 1950s to the 1980s, if not even later in certain regions, ships of
this kind were effectively sub-standards in all respects. Then, copying para-
doxically aviation, in the developed world, the origin and destination states
and all coastal states (the so-called ‘port states’), started implementing
strong safety control on all ships calling in their waters with arrest, seizure
and ban as sanctions. The 1980 Paris memorandum of understanding, for
Europe and the black list of the US coast guards, were the first examples of
this ‘port state control’ principle and were soon followed by a flurry of
similar provisions spanning on all oceans.
Panama and Liberia in particular were then obliged to expel unsafe ships
towards less demanding registries but remained social and tax heavens for
Skies Wide Shut 121
ship-owners. In the meantime, trade union of sailors and dockers that have
very early organized themselves at a global scale (the International
Transport Federation) began negotiating and imposing, if need be through
‘waterfront action’ (i.e. the ship’s boycott by the dockers), a minimum
wage for sailors, decent by third world standards (about 600 $ a month for
now) but which remains very low by developed countries standards. This
movement culminated at the beginning of the millennium by the adoption
of a worldwide maritime labour convention at the ILO setting minimum
social standards which will be implemented through port state control
mechanisms.
In the meantime, the flag of convenience phenomenon extended itself to
the more protected and less cut throat sector of liner/container shipping. It
has also extended to developing countries: Nowadays even India and
China have a considerable part of their controlled/beneficially owned fleet
under flags of conveniences. Developed countries nationally flagged fleets
(and the developed countries sailors that go with it the national flag
implying generally an obligation of nationality for the sailors) have largely
disappeared from international traffic and operate essentially on cabotage
traffics (eighth and ninth freedom in aviation parlance) which are, in most
instances, reserved to the national flags.
Indeed the international division of labour has basically chased devel-
oped countries workers from this activity to the benefit of developing
countries workers. While that is clearly the case for relatively low quali-
fied jobs (sailors/rating) that is not the case for officers. Liner shipping
companies in particular tend to keep national on board as ships have
become very large, very expensive and very difficult to manoeuver. For
these jumbos (18,000 containers) going for cheap labour at any rate is
not an option. In that respect the situation of aircraft pilots and co-pilots
is much closer from that of shipping officers than it is from that of rank
and file sailors, although that basic fact is never stated in an aviation
context.
Similarly, ships under flags of convenience, in particular in the liner
segment, have become safe ones. With tight and coordinated (via the
Equasys database) port state controls, that one could compare to the
EU aviation black list or to the CEAC controls, no substandard ship
can venture for long in the waters of countries deploying such controls
effectively.
The tax heaven effect remains, although its relative impact has dimin-
ished since many countries have largely exempted the shipping sector
from company tax, through lump sum ‘tonnage tax’ systems. The tax
heaven effect is also coming under attack notably in OECD, as shipping
heavens are also often tax heavens since they offer opacity for the same
reasons (bulk ship-owners used to eliminate risks and liabilities by the
creation of opaque structures of cascade holdings with cross-ownership
managing ‘single ship societies’, as shown by the Prestige oil spill
disaster).78
Trying to transpose this concept to aviation probably means the simulta-
neous acceptance by party A of an open foreign investment regime in avia-
tion, seventh freedom traffic rights and a PPoB withholding clause for the
airline of party B, in a context where party A has high labour costs and
party B low labour costs and where the combination of Agreements
between B and its bilateral partners other than A (C to Z) allow the airline
(s) of B to practice seventh freedom, rights between a third-party state (C
to Z) and A.
The occurrence of such a combination has been possible in the largely
unregulated environment of bulk shipping and later on liner shipping
because of the “freedom of the seas” principle. The same conjunction is
highly unlikely to happen in the regulated context of aviation: ‘freedoms of
the air’ being a joke or dream, depending how you see it and at any rate an
Orwellian oxymoron except if, as with the EU Single Aviation Market, a
free for all, abolishing all intra-restrictions, is suddenly decided.
In the EU Single Aviation Market nothing prevents Tarom, the
Romanian carrier which has comparatively low labour costs, to operate
between France and Germany and for that matter even within France on a
consecutive (eighth freedom) or stand-alone (ninth freedom) basis. Nothing
prevents either Air France or Lufthansa to take over Tarom to operate
with the Romanian flag and Romanian crews at Romanian costs between
and within France and Germany. While this is allowed and makes eco-
nomic sense, it has not happened.
Another way to test the flag of convenience concept in the aviation
realm is to play the devil’s advocate and to put oneself in the shoes of a
country willing to become an air flag of convenience and to examine what
kind of national and international regulations this would require. Imagine
therefore, for the sake of the argument, that Panama or Liberia want to
transpose their maritime transport experience to air transport. Panama is
completely free to adopt a ‘substantive business operations’ principle in its
national investment rules and allow a 100 per cent foreign ownership, as it
the case by the way in most services sector. Panama is even free, as it does
in maritime, not to impose national crews under its flag and to allow the
Skies Wide Shut 123
international recruitment of even more competitive manpower/cheaper

workforce.
But then the foreign-owned carrier flying the Panamanian flag will face
problems as soon as it will venture internationally, beyond Panama’s bor-
der. Panama will have to negotiate with each country the carrier is willing
to serve a PPoB withholding clause in the corresponding bilateral agree-
ment, say for instance Spain, France, United Kingdom, Germany or the
United States to take a few countries over which Panama may have a
labour cost advantage. If Panama does not succeed in negotiating such a
clause, it will have at least to negotiate a less legally secure waiver to the
classical substantial ownership and effective control clause.
That is a relatively weak regulatory hurdle to jump but, the cases where
it historically occurred all concerned airlines with no clear labour costs
advantages over the countries they were willing to serve: for example
Luxemburg’s owned Monarch airlines in the United Kingdom before the
EU single aviation market, Swiss since its acquisition by Lufthansa, Hong
Kong airlines due to mainland investors, Australia’s and New Zealand’s
airlines, seemingly more as a matter of principle than out of commercial
interests.
Still this purely bilateral setting does not produce in itself in a full-
fledged flag of convenience situation. To reach that degree of ‘free for all’
competition, full seventh and possibly eighth and ninth freedom would be
needed. In other terms the freedom, being from country A to serve any
other country-pairs than those involving your own country, the way
Danish-owned Liberian-flagged ships serve the traffic between China and
the United States without ever calling in Liberia. This is a huge and com-
plex negotiating challenge because the bilateral structure of air traffic gives
a veto to third parties on such effective granting of seventh freedom via the
mechanisms described above in the section devoted to the myth of liberali-
zation through seventh freedom.
To continue to play the devil’s advocate, even if such an all over seventh
freedom possibility was granted, pilots would not be in the position of sai-
lors but in the position of officers who have much less suffered from globa-
lization and have managed to maintain decent wages. On its side, aviation,
in its present regulatory setting, is not exempt from instances of labour cost
competition. The employment as early as of the eighties of Romanian pilots
by Crossair at Romanian conditions, the use of non ALPA or non
IFALPA affiliated pilots a by Southwest or Ryanair or the Irish labour
contracts of employees of Ryan air and Easyjet deployed in continental
Europe can be viewed as examples of labour cost competition.
One may note also that States have other means than a simple national-
ity requirement to protect staff. Minimum local conditions based on the
duration of stay of the personnel on the territory can be imposed without
breaching EU laws and the single aviation market, as proven by the condi-
tions imposed a few years ago to Easy Jet by France for its France-based
employees. Furthermore, to be willingly used by the developed countries
consumer the ‘flag of convenience’ carrier would need to build brand recog-
nition, trust and confidence in its safety records, all things that do not sit
well with extremely low wages and that, besides, entail many other costs
(marketing, distribution, safety, on-board services) that, for a new entrant,
will to a large extent diminish the competitive advantage of lower labour
costs.
Due to the prevalence of the hub-and-spoke structure for long-haul traf-
fic, it is not even sure that a long-haul flag of convenience carrier can profit-
ably operate in developed countries markets without itself reconstituting a
hub there with the local cost it entails. The hub structure may even destroy
the competitive labour cost advantage.
Unlike sailors, air crews, including flight attendants, are not confined to
their aircraft. They have to be based on the ground, in a territory, in most
instances the territory from which they take off. So the flag of convenience
carrier would have only two options. Either ferrying its crews from its
own territory to start the seventh freedom flight or to base its crew in the
country served. Both options entail costs. There may not be everyday a
commercially profitable third freedom flight from the flag of convenience
state to the hub allowing the carriage at marginal cost of the crew to con-
tinue a seventh freedom flight. One can also hardly imagine quasi empty
dedicated flights to ferry crews. On the other hand, one cannot imagine
the staff of a flag of convenience carrier based in the developed country
hub being paid with the very low wages of their country of origin: local
unions and authorities would inevitably intervene to end abuses and to
enforce also existing fiscal and social rules based on the duration of stay
and territoriality. Sixth freedom airlines which play to a certain extent the
‘interlope’ role of maritime flag of convenience and constitute the only
effective third-party competition on long-haul flights are, reportedly, far
from being bad and low-paying employers including for their flight
attendants.
Alternatively, if the point-to-point trend really develops and if the multi-
base model of short- to medium-haul low-cost carriers (e.g. Ryanair,
Easyjet, Air Asia or Southwest) is transposable to long-haul, then there
Skies Wide Shut 125
may be a niche for flag of convenience carrier. But the bases will be still
located in the developed country market and will entail significantly higher
costs, including labour ones. That makes a lot of ‘ifs’. Based on our analy-
sis, we would argue that flag of convenience in aviation will for long, if not
forever, remain a distant perspective.
CHAPTER 5
INTERNATIONAL MERGERS AND

ACQUISITIONS IN THE AIRLINE
INDUSTRY
James Nolan, Pamela Ritchie and John Rowcroft
ABSTRACT
Mergers and acquisitions in the transportation sector are typically

explained as attempts to capture economies of scale and scope through
shared infrastructure and related cost-saving measures. In the airline
industry, the past 15 years have seen an increasing number of interna-
tional mergers and acquisitions that would have been blocked under prior
regulatory regimes. This activity suggests that there are indeed gains
from increasing airline size.
Such gains may be largely financial in nature. One benefit to a merged
airline could be greater market power over particular routes and hubs
after merger, as well as improved contract structure and bargaining
power in operations, although greater de-regulation and more competi-
tion internationally makes these arguments less compelling. In many
cases, gains may be unique to specific airlines or operational situations.
Thus, the issue addressed in this chapter is whether, in general, increas-
ing the size or scope of airline operations enables them to function more
efficiently and whether this effect is sustained across all sizes of airline.

ISSN: 2212-1609/doi:10.1108/S2212-160920140000004004
127
128 JAMES NOLAN ET AL.
More pointedly, the chapter examines whether there exist measurable

efficiency gains that can help explain the variety of mergers and acquisi-
tions in the industry.
Keywords: Airlines; efficiency; mergers and acquisitions; data envelop-
ment analysis
JEL classifications: L93; L41; C81
INTRODUCTION
Mergers and Acquisitions (or M&As) in the transportation sector have

been typically explained by industry analysts as attempts to capture econo-
mies of scale and scope through, for example, shared infrastructure and
firm related cost-saving measures (Brooks & Ritchie, 2003; Scherer & Ross,
1990). The past 15 years have seen an increasing number of international
mergers and acquisitions among airlines, mergers that would have been
blocked under prior regulatory regimes (Table 1). There are two points
worth noting about the table. The term “creations” refers to those subsidi-
aries formed by the parent air carrier itself. In addition, acquisitions, mer-
gers or sales generally take a long time to negotiate between the carriers,
considering the regulatory bodies governing the jurisdiction(s) in which the
carriers are incorporated. Frequently, the public announcement of these
activities is subsequent to the effective date of the agreement. During this
time, there may be a significant change in the operation of the carriers
involved since they will begin operating as a combined entity (acquisition
or merger) or reduced entity (sale/divestiture). In such instances, the dates
noted for these activities are the dates when the financial reports incorpo-
rate the impact, rather than the public announcement.
Merger and acquisition activity suggests that there are gains from
increasing airline size. However, these gains may be largely financial in nat-
ure. A key benefit for a merged airline might be greater market power over
particular routes and hubs after the merger, as well as improved contract
structure and bargaining power in operations, while greater de-regulation
and competition internationally makes these arguments in favor of a mer-
ger less compelling. In many cases, gains may be unique to specific airlines
or situations. Thus, the issue addressed in this chapter is whether, in gen-
eral, increasing the size or scope of airline operations enables them to
International Mergers and Acquisitions in the Airline Industry 129
Table 1. Acquisitions, Creations, Mergers, and Sales in the Airline

Passenger Industry, 19982010.a
Year Airline Acquired/Created/Merged/Sold
1998 SAS Air Botnia acquired

British Airways Deutche BA fully acquired; GO created
1999 British Airways City Flyer Express acquired; 9% Iberia acquired
Lufthansa Air Dolomiti 26% acquired
2000 Air Canada Canadian Airlines acquired
Air New Zealand Ansett Australia acquired
British Airways Comair 18.3% acquired; Air Liberty (86% held)
sold
2001 American Air Transworld Airlines acquired
British Airways GO sold; British Regional Airlines acquired
Qantas Qantaslink created by combining Eastern,
Southern, Airlink and Sunstake
SAS Braathens acquired
2002 Qantas Australian Airlines created as an international
LCC
2003 British Airways Deutche BA sold
Delta Song created
Lufthansa Air Dolomiti fully acquired
Qantas Jetstar created
Ryanair Buzz acquired
SAS Estonian Air (49%) sold
2004 Air France KLM merger
2005 Lufthansa Swiss International acquired
2006 British Airways BA Connect sold to Flybe; Flybe 15% acquired
Cathay Pacific Dragonair created
Delta Song ceased operations
Lufthansa Condor sold
Qantas Australian Airlines ceased operations
2007 British Airways Open Skies created
2008 Delta Northwest Air merger
Lufthansa Austrian acquired; British Midland acquired;
Brussels acquired 45%
Southwest Air ATA acquired
SAS Air Baltic sold
2009 Lufthansa German Wings acquired
2010 American Air US Air merger
Southwest Air Air Tran acquired
United Continental merger
a
Data extracted from the relevant air carrier web sites.
function more efficiently and whether this situation is sustained across all
types and size of airline. For a useful discussion of scale and scope in the
airline industry, see Jara-Dı́az, Cortés, and Morales (2013). More specifi-
cally, the chapter examines whether there exist measurable technical effi-
ciency gains for airlines, gains that might explain historical mergers and
acquisitions in the industry.
Efficiency and returns to scale estimates for airlines, particularly those
operating in the United States, have been the subject of a number of stu-
dies. Some early work (Caves, Christensen, & Tretheway, 1984) concluded
that there were few economies of scale per se but that as airlines increased
in size, they experienced economies of route density. Later studies (e.g., Liu &
Lynk, 1999) suggested that there are increasing returns to network size, a
finding that helped rationalize the industry’s tendency even at that time to
pursue expansive mergers and alliances. On the policy side, airline deregu-
lation in the United States focused attention on the gains from size but also
raised concerns about the degree of competition (or contestability) that
might be expected in the deregulated airline industry (see Cornwell,
Schmidt, & Sickles, 1990; Kumbhakar, 1990).
Until recently, most work on the airline industry has concentrated on
US carriers due to the level of deregulation compared to Europe in particu-
lar, and the direct comparability of domestic US airline data (Adler &
Golany, 2001; Schefczyk, 1993). However, recent changes in international
airline regulation, coupled with the emergence of aggressive low cost car-
riers (sometimes referred to as LCCs) such as Ryanair highlight the need
for a more internationally focused comparison of efficiency and returns to
scale in the industry.
Over the past 15 years, both the world and the airline industry in parti-
cular have been subject to a number of exogenous shocks, most notably the
events of September 2001 and the subsequent changes in travel security
requirements, as well as the 2003 outbreak and movement of the SARS
virus, and the more recent prolonged worldwide economic recession. At the
same time, the airline industry has been undergoing numerous structural
changes while the US industry in particular has seen a number of carriers
seek bankruptcy protection. These issues raise the question of the extent to
which these broader economic events affected the production functions of
the world’s airlines, as opposed to just their shorter term financial situation
(see e.g., Nolan, Ritchie, & Rowcroft, 2004).
The continuing interest in airline mergers/alliances, coupled with on-
going international regulatory developments and unforeseen economic
shocks to the air travel market suggest an updated examination of airline
efficiency and returns to scale using more recent data covering a sample of
world airlines. This chapter examines the performance of world airlines
over the period 19982010, including both legacy carriers and more recent
entrants. This period contains several industrial shocks, but includes suffi-
cient time for adjustment and understanding of the consequences. The mea-
surement issues of relative efficiency and returns to scale for the airlines are
considered directly through the use of data envelopment analysis (DEA).
This is a more extensive DEA than that provided in Scheraga (2004) which
covered only the individual years 1995 and 2000. For references to other
studies of airline efficiency, see, for example, Barros and Peypoch (2009).
To examine other operational and network effects including those poten-
tially resulting from airline mergers and acquisitions, more detailed indus-
trial data are analyzed with a stepwise linear regression model estimated on
the panel.
The data set is described in greater detail in the second section, while the
third section reports the results of the DEA efficiency analysis. Regression
procedures and results are described in the fourth section. A fifth section
offers a general discussion covering the findings of both analyses, in parti-
cular as they relate to efficiency and consolidation in the industry.
Concluding remarks are offered in the sixth section.
THE DATA
One major constraint in data selection for this analysis was access to the
required data for the desired period of study. Thus, data were collected
from a sample of 18 world airline operators, each chosen to capture
diversity in geography, size, age, home airport and alliance. Data sources
were publicly available data released either directly by the airlines or
through audited financial statements and other required operational dis-
closures provided to regulatory bodies and/or investors. Access to the
data is available through the airlines’ web sites, and these are listed in
Appendix A at the end of the chapter. Table 2 provides some additional
details about the airlines included in the study. Further details about the
airline data collected for this study are available from the authors upon
request.
Annual data on each airline were collected over the period from 1998 to
2010. This period overlaps the end of some related studies such as Scheraga
(2004) and also covers the effects of September 2001 and its aftermath, as
Table 2. Airlines in the Sample.a

Airline Founded Home Base Size Alliance
Air Canada (AC) 1937 Canada 11th World Star Alliance

Air France (AF) 1933 France 5th World SkyTeam
Air New Zealand (ANZ) 1940 New Zealand 1st New Zealand Star Alliance
American Airlines (AA) 1930 United States 1st World Oneworld
British Airways (BA) 1924 England 3rd Europe Oneworld
Cathay Pacific (CP) 1946 Hong Kong na Oneworld
Continental (CON) 1934 United States 4th US SkyTeam
Delta (DE) 1928 United States 2nd World SkyTeam
Finnair (FIN) 1923 Finland 1st Finland Oneworld
Lufthansa (LH) 1926 Germany 2nd Europe Star Alliance
Northwest Airlines (NW) 1926 United States 5th US SkyTeam
Qantas (QS) 1920 Australia 1st Australia Oneworld
Ryanair (RY) 1985 Ireland 1st LCC Europe
Scandinavian (SAS) 1946 Sweden na Star Alliance
Singapore Airlines (Sing) 1972 Singapore 15th World Star Alliance
Southwest Air (SW) 1971 United States 3rd World
United Airlines (UA) 1926 United States 4th World Star Alliance
WestJet (WJ) 1996 Canada na
a
The data for Air France combine AF and KLM since the merger in 2004. Similarly data for
Continental and United are combined since the merger in 2010 and for Delta and Northwest
since their merger in 2008. The size descriptor is derived from the airline web sites and refers
to the number of passengers transported annually.
well as SARS, and the global financial crises. The industry worldwide con-
tinues to experience considerable uncertainty and structural change, for
example, a number of prominent US-based carriers sought Chapter 11
bankruptcy protection over the sample period, while some LCCs in both
the EU and the United States continue to prosper. The set of financial,
physical and network variables used in this study are listed in Table 3, and
Table 4 shows the means of some key variables (listed by airline) in the
dataset.
Not only are there differences in the accounting disclosure requirements
for different airlines, but also reporting standards differ. Reporting stan-
dards are changing significantly around the globe with the introduction of
International Financial Reporting Standards (IFRS) and agreements across
most trading nations to comply. Such compliance is not yet universal and
differences will continue to appear through 2015 when full compliance is
expected (IASB-FASB Update Report, 2012). IFRS, in the current form,
was not applied explicitly anywhere prior to 2010, and extraction of the
data from airline financial reports requires clear recognition of these
Table 3. Financial, Physical, and Network Variables Contained in the

Data Set.
Financial Variables Physical Production Variables Network Variables
(Millions US$)
Wage expense Available seat miles (millions of Number of destinations

miles) served
Fuel expense Fuel consumption (millions of US Number of aircraft
gallons)
Fuel price Number of employees Number of aircraft types
Total assets Number of regions
served
Productive (or fixed) assets
Investment in aircraft
Table 4. Means of Some Key Sample Variables 19982010 (by Airline).

Airline Total Assets ASM (Millions of Destinations Number of
(Millions US$) Miles) Aircraft
Air Canada (AC) 7646.82 55004.00 157.15 321.62

Air France (AF) 23697.07 113697.64 218.38 461.85
Air New Zealand 2945.60 19944.07 46.69 88.08
(ANZ)
American Airlines 27131.62 169546.85 158.08 759.00
(AA)
British Airways 20348.47 95862.46 158.85 308.08
(BA)
Cathay Pacific 11297.01 50613.74 90.15 95.85
(CP)
Continental (CON) 10459.25 88402.00 236.00 489.33
Delta (DE) 27188.85 161925.00 275.69 758.23
Finnair (FIN) 2083.29 13630.79 68.92 62.31
Lufthansa (LH) 23491.98 93257.46 371.92 435.08
Northwest Airlines 13699.70 94390.40 189.80 514.00
(NW)
Qantas (QS) 11940.78 65554.08 124.92 186.38
Ryanair (RY) 4773.72 21463.35 85.00 98.08
Scandinavian 6352.52 23089.31 134.23 262.46
(SAS)
Singapore Airlines 12860.73 62543.43 64.08 91.92
(Sing)
Southwest Air 11130.29 78486.48 61.38 426.08
(SW)
United Airlines 23237.54 157036.77 166.77 567.23
(UA)
WestJet (WJ) 1564.24 9194.96 33.15 49.69
differences. Furthermore, many of the airlines report in their home cur-

rency, so recognition of the impact of foreign currency translation must
also be a consideration. This study examines efficiencies in the airline
industry. Each carrier has the ability to mix the inputs to their business
activities and these mixes are likely to be the drivers of the firm’s efficiency
scores. Accordingly, some standardization is required to examine the air-
lines in this context. Thus, the choice of financial variables was framed in
the context of appropriate information being available to allow for
accounting standard differences and for exchange rate fluctuations.
The wage expense variable is the most vulnerable in this dataset with
respect to foreign currency translation, but it is not affected by any report-
ing or disclosure differences. A current expense item, under all regimes, is
reported at the actual amount paid, including all benefits and matching
contributions by the firm (also recognizing any amounts due at the end of
an accounting period). Many of the airlines in the sample have employees
outside their home currency base, and these employees are paid in the cur-
rency of the country in which they reside. However, none of the carriers
report hedging activities undertaken explicitly for this expense. Generally
in the regimes incorporated in this study, there cannot be any adjustment
of the expense amount due to hedging activities unless the hedges were
explicitly reported as being undertaken as a hedge against that expense.
Therefore, the amount reported by the carriers in this sample, is not miti-
gated by any hedging activities and so the financial data for these airlines
already incorporates foreign currency translation issues. As employees are
paid consistently over the reporting period, the average exchange rate
across the reporting period provides a fair approximation of the expense in
US$.
Fuel expense is affected not only by accounting disclosure requirements,
but also by foreign exchange rates and hedging activities. Fuel is a global
commodity, quoted and traded in US$. Fuel is a volatile commodity and
prices fluctuate throughout each day, let alone throughout a reporting per-
iod. Each of the firms in this sample undertake explicit hedging activities to
reduce the impact of the fluctuations in fuel price and the results of these
hedging activities are treated as a modifier to the fuel expense reported by
each airline. Again, since fuel consumption is an ongoing activity in the air-
line industry, the average exchange rate across the reporting period pro-
vides a fair approximation of the expense in US$.
The fuel price reported by each airline in this chapter, as well as in their
financial disclosures, is a construct. The “average” fuel price factors in pur-
chases at various different times during the reporting period, for different
quantities purchased at different prices. Where a price fluctuates continu-

ously in the market, a single figure quote over a reporting period generally
reflects a weighted average. The majority of the firms in this sample report
their average fuel price in US$. Where that is not the case, the average
exchange rate over the period has been used.
On the surface, there are some significant financial reporting and disclo-
sure differences amongst the firms in the sample. Productive assets (aircraft,
inventory of engines and other costly spare parts, etc.) are another global
commodity, with prices set at a world level. However, this price may well
be affected by both the number purchased and the time over which they
are purchased. Typically, a contract to purchase new aircraft specifies the
total number of craft to be purchased, and the number that will be pur-
chased each year. But some of these contracts extend over a long term time
horizon. As new aircraft are brought into the fleet, older or less efficient
aircraft are sold, and in addition, there is ongoing refurbishment as well as
regulated maintenance and repair, items which, due to their significant
cost, are generally capitalized by the corporation. In this case, the account-
ing differences (either historical cost or current market value) tend to be
nullified by the ongoing activity.
As foreign exchange hedging activities are undertaken to try and mini-
mize the impact of rate fluctuations, these activities modify the purchase
prices of the assets acquired or maintenance and refurbishment done. The
choice of an appropriate exchange rate is open to question, since these
acquisitions are historical and normally occur at specific times. However,
maintenance, repair and refurbishment of these assets is an ongoing activity
(much of which is capitalized) as is any related debt repayment, so the use
of the average rate throughout the reporting period is consistent with this
activity as well as with the exchange rates used throughout the analysis.
Although there were no bankruptcies in the study set, there were a num-
ber of firms in the sample that went into bankruptcy protection during a
portion of the study period. While there are certainly numerous accounting
disclosure differences while a firm is in bankruptcy protection, fundamen-
tally they do not affect the financial variables used in this study. Debt
restructuring and disposal of assets are the most common results of a per-
iod of bankruptcy protection. Union contracts, and hence employee bene-
fits that will ultimately affect “wage expense,” may also be affected by
bankruptcy protection. While contracts to purchase aircraft in the future
are often revised during bankruptcy protection, the aircraft were not owned
prior to the bankruptcy protection and therefore were not included in the
asset mix prior to going into bankruptcy protection. The costs associated
with the dissolution or revision of these contracts is not included in the

asset base either. Assets sold will be reflected in the “total assets” and in
the specific asset category if it is used in the analysis, just as it would be if
the firm were not in bankruptcy protection but had chosen to change its
fleet composition. Finally, wage expense will reflect the changes that result
from any revision of the union contracts arising from bankruptcy protec-
tion, just as they reflect revisions carried out in normal union negotiations
when the firm is not in bankruptcy protection.
Financial and physical production variables are used to represent the
direct production process of an airline. Since the emphasis in this chapter
is on airline efficiency and productivity, available seat miles (asm) was
judged to be a more appropriate output measure than revenue passenger
miles, since the latter also includes demand changes and variations in
load factors. In this regard, the small portion of total output represented
by cargo and mail shipments in the sample suggested that the construc-
tion of a combined output index was unnecessary (Ray & Mukherjee,
1996). Other variables were constructed to be consistent with modern air-
line operations. For example, with respect to physical inventory, invest-
ment in aircraft consists of the carrying value of the aircraft as well as
significant spare parts.
Network related variables were selected to reflect the volume and variety
of airline services. Destinations served are those serviced by the airline itself,
as distinct from any code share responsibility or alliance availability. To
provide a measure of international scope for each airline, the globe was
divided into ten regions Africa, Asia, AustraliaNew Zealand,
Caribbean, Central America, Europe, Eurasia, Middle East, North
America, and South America and the number of regions served by each
airline was incorporated as a variable.
Doganis (2001) and others have suggested that low cost carriers operate
more efficiently than the older or legacy carriers essentially through simpli-
fying their operations. This simplification process includes using fewer
types of aircraft in their fleets, deploying their aircraft more efficiently
across their networks, and focussing more closely on their core business
(Schefczyk, 1993). Given this, several additional variables were constructed
in an attempt to reflect these aspects of the complexity/simplicity of modern
airline operations. Specifically, the following additional variables were
developed:
Fleet simplicity = total number of aircraft/number of types of aircraft

Fleet investment = investment in aircraft/total assets
Operational complexity 1 = number of destinations/number of aircraft

types
Operational complexity 2 = number of destinations/number of aircraft
For the first three measures, a lower value represents a greater degree of
complexity in the airline’s operations. For example, a lower fleet investment
as a proportion of total assets suggests a greater diversity in the airline’s
corporate structure. However for the last variable listed, a larger number of
destinations for a given fleet size (operational complexity 2) is indicative of
greater complexity.
ASSESSING RELATIVE AIRLINE

EFFICIENCY DEA
DEA was performed using each airline observation from each year as a
separate observation, yielding a total of 230 decision-making units (referred
to as DMUs) over the sample, allowing for mergers. The analysis is also
based on the assumption that available production technology remained
largely unaltered over the study period. A review of the reference DMUs
identified in the analysis provides some support for this assumption, since
observations from both early and later sample years appear in the reference
sets.
The input variables selected were value of productive assets, number of
aircraft (of any type), quantity of fuel used (in US gallons), and total number
of employees. Unfortunately, separate flight crew data were not available
for a sufficiently large number of airlines to permit disaggregation of
employees. Output variables used in the DEA were available seat miles
(asm) and the number of destinations served.
The software used for the DEA estimates was DEA-Solver Professional
Version 9.0 (2013). Using an unconstrained input-oriented analysis, some-
times referred to as Charnes-Cooper-Rhodes (CCR-I), 25 of the 230
DMUs were ranked as technically efficient. Alternatively, constraining the
DEA to compare DMUs of similar size, the Banker-Charnes-Cooper model
(BCC-I) produces an estimate of technical efficiency without presuming
that the technology is scalable (Lovell, 1993). In this case, 57 DMUs were
evaluated as technically efficient. Finally, both models were also run using
a form of super efficiency (Andersen & Petersen, 1993) which serves to
distinguish between the efficient DMUs.
The ratio of the CCR/BCC measures provides an indication of the scale

efficiency of each DMU (Lovell, 1993). The average efficiency scores for
each airline over the sample period are reported in Appendix B, while the
scale results for each airline in the sample are summarized in Table 5.
Fig. 1 illustrates the connection between firm size and returns to scale
(increasing, constant, or decreasing) derived from the BCC-I analysis. In
this case, the variable productive assets is used as a proxy for firm size and
is measured on the vertical axis, but very similar figures can be derived
from the other variables. Along the horizontal axis DMUs are plotted in
order of their scale efficiency score (CCR/BCC). This implies the points in
the left-hand region labelled “Increasing RTS” correspond to DMUs with
a scale efficiency score of less than one, plotted in order of increasing mag-
nitude. The points in the other two illustrated regions are derived similarly.
Thus DMUs (airlines) exhibiting increasing returns to scale have produc-
tive assets between US$817m and USD$11,508m; those showing constant
returns to scale, between US$58m and US$11,652m (with an outlier at US
$23,802m); while DMUs showing decreasing returns to scale have produc-
tive assets from US$194m to US$25,330m.
Table 5. Airline Returns to Scale, 19982010.

Airline Returns to Scale
Increasing Constant Decreasing
Air Canada ✓ ✓
Air France ✓
Air New Zealand ✓
American Airlines ✓ ✓
British Airways ✓
Cathay Pacific ✓ ✓
Continental ✓
Delta ✓
Finnair ✓ ✓ ✓
Lufthansa ✓
Northwest Airlines ✓ ✓
Qantas ✓ ✓ ✓
Ryanair ✓ ✓ ✓
Scandinavian ✓
Singapore Airlines ✓ ✓ ✓
Southwest Air ✓
United Airlines ✓ ✓
WestJet ✓ ✓ ✓
30000
Productive Assets (millions of US$)
25000
Constant returns
to scale
20000
Increasing Decreasing
15000 returns to returns to scale
scale
10000
5000
0
Decision-making units (airlines)
Fig. 1. Firm Size and Returns to Scale.
The constant returns to scale findings may represent a transition phase

in the industry over the duration of the sample. But the remaining patterns
found can be linked to conventional wisdom about an industry that air-
lines grow by increasing returns to scale, and ultimately reach a mature size
characterized by decreasing returns to scale. While this finding helps to
explain the drive to expansion among airlines at various stages in their
development, it also highlights the issue concerning those aspects of an air-
line’s operations which are non-scalable.
As a check on the data, in the efficiency literature some researchers
(e.g., Tsolas & Manoliadis, 2003) use non-parametric rank tests for correla-
tion to compare scores from different DEA models. In the present analysis,
correlation was used as a broad check on the effects of size on estimated
airline efficiency, by measuring the correlation between the estimated effi-
ciency scores with various measures of airline size. All sets of DEA scores
(CCR-I and BCC-I, both in normal and in super form) correlated nega-
tively with each of the inputs used and outputs produced, with the excep-
tion of the number of destinations. The latter was not significantly
correlated with any of the scores.
These findings are summarized in Table 6. They appear to provide sup-
port for the supposition that airline efficiency does not automatically
Table 6. Correlation of Efficiency Measures with Size.

Model CCRi BCCi Super BCCi
Size Measure
Physical assets −0.4289 −0.2888 −0.3137
Number of aircraft −0.4606 −0.2384 −0.2496
Fuel −0.3934 −0.2129 −0.2487
Employment −0.5410 −0.2464 −0.2609
Number of destinations −0.3950 0.0934 −0.0655
Available seat miles −0.2649 −0.1514 −0.1926
Available seat miles/Number of destinations −0.0086 −0.2028 −0.2616
All correlations are highly significant.
improve with the size of operations or equipment. The finding with respect
to the asm/#destinations variable is particularly interesting since if consid-
ered in conjunction with Fig. 1, it suggests that increasing route density
does not tend to improve efficiency (cf. Caves et al., 1984). Similarly, the
number of destinations provides a representation of network size and again
does not appear to make a marked contribution to the measures of airline
efficiency.
Finally, as one indicator of efficiency outcomes of specific key airline
mergers over the sample, the DEA was run for two particular merger sce-
narios as a kind of simulation. In each scenario, a pair of airlines that
merged at some point within the sample was replaced over the entire sam-
ple period by a composite DMU consisting of the measured inputs and out-
puts of the two airlines combined (e.g., see Davis and Garces (2010,
chapter 8)). In each case, once merged, the airlines concerned operated at
least ostensibly as a single entity, unlike for example, Air France and KLM
which continued to market themselves as separate units subsequent to their
merger. The mergers simulated in this manner are between Delta/
Northwest and United/Continental. The results of these BCC-I estimates
are illustrated in Figs. 2 and 3.
Summarizing the merger simulations, with respect to the Delta-
Northwest merger, the composite airline (“DENW”) outperforms the con-
stituent airlines in the early part of the study period (from 1998 to 2001).
However, the performance of the composite airline is indistinguishable
from each of the separate airlines as the two converge in performance until
the date of their actual merger in 2008.
For the case of the United and Continental merger, the constituent air-
lines outperform the composite airline (“UACON”) for every year except
1.1
0.9
BCC DEA Score
DENW
0.8 DE
NW
0.7
0.6
0.5
98 99 00 01 02 03 04 05 06 07 08 09 10
year
Fig. 2. Efficiency of Delta and Northwest Compared to the Two Airlines

Combined/Merged (DENW).
1.1
0.9
BCC DEA Score
UACON
0.8 UA
CON
0.7
0.6
0.5
98 99 00 01 02 03 04 05 06 07 08 09 10
year
Fig. 3. Efficiency of United and Continental Compared to the Two Airlines

Combined/Merged (UACON).
2000, and again the separate airlines converge in performance until the
date of the actual merger in 2010. Thus, for each case examined here, the
DEA simulated merger comparison supports the contention that increasing
airline size through merger does not necessarily generate notable improve-
ments in operating efficiency.
MEASURING THE EFFECTS OF OPERATIONAL

FACTORS ON AIRLINE PRODUCTION USING
REGRESSION ANALYSIS
Regression analysis on the airline data provides a means to investigate

overall airline production from an absolute rather than a relative perspec-
tive. To account for the fact that the data are pooled using a short panel of
data, firm specific dummy variables (i.e., fixed effects) were incorporated
for each airline. The software used for the regression analysis was Stata
(StataCorp, 2003).
Early work by Caves et al. (1984) first suggested the use of the Cobb-
Douglas form for airline production. In the first instance, the single output
production function for the airline industry was estimated using the follow-
ing function:
Q = ALα K β F γ ð1Þ
Q is available seat miles (asm), L represents labor as measured by total

number of employees, F is total fuel consumption in US gallons, and K is
capital, measured by the sum of Productive Assets (PA), Total Assets
(TA), and Investment in Aircraft (IAC). To perform the analysis, a step-
wise regression was run using this production relationship as well as the
variables noted in Table 4. It is worth noting that some econometricians
are critical of stepwise regression as a process of model selection (in parti-
cular, see StataCorp, 2003). However, without prior structural knowledge
about the chosen operational and network variables and how they affect
production in this industry, a carefully constructed stepwise analysis is jus-
tified as a data mining exercise that allows identification of interesting
structural information from the data.
Apart from L, K, and F, the remaining variables were entered additively
into the basic production relationship (see Eq. (2)). Along with the
production and network/complexity variables, airline specific dummy vari-

ables were included in the variable choice set (ci).
X
Q = ALα K β F γ expð δi ci Þ ð2Þ
Furthermore, to address possible concerns about multicollinearity in the

data, the stepwise regressions were run using appropriate ratios for L and
F relative to the capital input. Thus, the final form of the estimated model
became:
α γ X
L F
Q=A Kβ expð δi ci Þ ð3Þ
K K
As might be expected, the stepwise model resulting from the estimation

of Eq. (3) on the airline data yielded a very high R-squared in the final spe-
cification. The model selected retained eight significant airline dummies.
Each of these variables were negative in sign (and hence capture idiosyn-
cratic reductions in output) with the exception of the dummy variable for
Continental Airlines. In addition, the parameter estimates for the physical
inputs yielded a significant value for returns to scale at approximately 0.99
and this near constant-returns-to-scale parameter estimate is consistent
with a mature industry. All parameter estimates for the final stepwise
model are reported in Table 7.
To identify potential output effects of the various mergers and acquisi-
tions that occurred throughout the sample, merger/acquisition dummy
variables were developed for each airline that merged or acquired another
airline during the sample period. Given the well-known time lags associated
with observing the effects of mergers and acquisitions (Scherer & Ross,
1990), an indicator variable was constructed that accounted for both the
year following an announced merger in the data and the subsequent year.
This allows a period for the particular airline to work through the initial
strain of the merger/acquisition process.
The merger/acquisition dummies were significant (and positive) for three
airlines in the sample Ryanair, SAS, and Air Canada. For each of these
airlines, there is evidence that they had either positive merger/acquisition
experiences in that they either absorbed a similar airline to themselves while
expanding their routings or alternatively the merger/acquisition served to
help strengthen their hold over the market served by the airline. Certainly
Ryanair captures the former type of merger and acquisition, whereas Air
Table 7. Production Function Results Stepwise Regression on

Available Seat Miles.
Variable Coefficient P-Value
Int Intercept (ln A) 2.784 0.000

Airline dummies
British Airways −0.163 0.000
Continental 0.084 0.001
Cathay Pacific −0.193 0.000
Ryanair −0.257 0.000
SAS −0.553 0.000
Singapore Airlines −0.131 0.000
Lufthansa −0.475 0.000
Air France −0.208 0.000
Input variables
Ln (labor/total assets) 0.177 0.000
Ln (fuel consumption/total assets) 0.476 0.000
Ln (total Assets) 0.986 0.000
Network complexity variables
number of aircraft/number of types −0.0003 0.001
number of destinations/number of aircraft types 0.0047 0.000
Merger dummies
Ryanair 0.161 0.013
SAS 0.091 0.047
Air Canada 0.155 0.010
Adjusted R2 = 0.99
Canada, which through 19992000 had near monopoly status on interna-

tional flights originating from Canada (see Canadian Transportation
Agency, 2004), represents the latter type. The reported significance of the
SAS merger variable may be driven by its unusual multinational origins as
well as its consistent merger and acquisition activity over the sample (see
Table 1).
DISCUSSION OF RESULTS
Early related empirical research (Caves et al., 1984) suggested that the pro-
duction function for airlines exhibited constant returns to scale and that
any benefits of increased size were derived from economies of route density.
In this view, airlines could be more efficient by utilizing their fleet and other
assets more fully. This proposition was used to support efficiency claims
expected from deregulating the airlines. Later analysis, post-deregulation,

laid doubts about the claim of constant returns to scale and instead sug-
gested that attempts to increase size coupled with the evolution of the hub
and spoke system were largely driven by attempts to increase monopoly
power (Creel & Farell, 2001). In some cases, this derived from a narrow
view of route competition (Brander, Cook, & Rowcroft, 1989) and cer-
tainly the success of carriers like Southwest and Ryanair has demonstrated
that the “fortress hub” model is far from impregnable. Thus, the issue of
the connection between airline size and efficiency remains an important
one.
DEA results shown in this chapter indicate that airlines’ physical inputs
and outputs in the sample follow a now familiar pattern of increasing, fol-
lowed by decreasing returns to scale as airline size grows, and this pattern
was found across the wide variety of airlines in the sample. Considered
alone, this finding would suggest that the largest carriers have reached a
point where further expansion is unlikely to be attractive, at least from a
technical viewpoint. Nevertheless, several recent major amalgamations (Air
France and KLM, Delta and Northwest, United and Continental) suggest
that growth through merger or acquisition continues to be an attractive
strategy even for large airlines.
The regression analysis returns robust results in favor of constant
returns to scale at the industry level. This result is consistent with earlier
results covering a different operational and regulatory environment in
terms of competition, routes and security (Caves et al., 1984). Any devia-
tions from constant returns to scale at the industry level appear to stem
from two reasons specific airlines and network differences. With the
exception of Cathay Pacific (CP), the airline dummies found to be signifi-
cant and negative in the regression correspond directly to those airlines
with the lowest average efficiency scores over the 8-year period (see
Appendix B). The set of older European airlines are particularly note-
worthy in this regard, as also reported by Good, Röller, and Sickles (1995).
The regression results are applicable over a considerable time period and
across major changes in industry structure as well as several exogenous
shocks. In the model, the effects of shocks such as September 11 and its
after-effects, as well as SAARs seem to have been mitigated with respect to
the chosen input and network variables, as these variables were consistently
found to be insignificant across a variety of model specifications.
With respect to the regression results, deviations from constant returns
to scale seem to be explained by the strategic decisions made by the airlines
in terms of the diversity of destinations served and the equipment used to
service them. Specifically, the positive coefficient for the number of regions
served suggests the presence of economies of scope. In fact, scope econo-
mies may be the counterpoint in the international airline context to “econo-
mies of density” as identified in prior airline literature. Economies of
density arise from more intense use of aircraft on routes within a particular
network, whereas economies of scope result from the airline’s ability to
deploy all its resources across an expanded network, and the latter may
underlie continued industrial trends towards expansion through merger,
alliance, or acquisition. The positive coefficient found for investment in air-
craft considered with the negative coefficient for number of aircraft also
seems to emphasize the critical efficiency gains that come from being
able to use larger aircraft. This finding is consistent with the growing
worldwide use of larger, fuel efficient airliners such as the Airbus A380 and
the Boeing 787.
Somewhat unexpectedly the number of aircraft types does not appear in
the final regression model. This suggests that the LCCs may owe any effi-
ciency gains to their focus on core business (investment in aircraft per se)
rather than reducing the variety of crew certifications and the heterogeneity
of their parts inventory.
CONCLUSIONS
The results reported here suggest that from a physical input and output
perspective, a constant returns to scale Cobb-Douglas production technol-
ogy as applied to the airline industry remains remarkably robust over a
diversity of times, places, regulatory changes and exogenous shocks.
Significant deviations from this industry level model can be explained by
inefficiencies for a particular airline and in other cases by the complexities
of the network in which an airline operates. These differences are reflected
at the individual airline level by the DEA. Given the interest in the so-
called “new models” of airline operations developed by the LCCs
(Doganis, 2001), it is noteworthy that the results here are sustained across
both legacy carriers and LCCs. This suggests that much of the novelty in
these new models lies in the marketing rather than the production side of
the business.
There also appears to be a continued drive for airline growth, exempli-
fied by the recent mergers of Air FranceKLM, DeltaNorthwest, and
UnitedContinental, along with continued interest by LCCs such as
Southwest in creating new alliances among themselves. This drive for

growth is probably best explained in terms of efforts to create even greater
returns to network size and fleet investment. By way of example,
Southwest’s acquisition of AirTran in 2010 (Table 1) is particularly inter-
esting in light of the DEA results clearly showing the airline experiencing
decreasing returns to scale throughout the sample.
The robustness of the results across differing competitive and regulatory
structures around the world may also be a function in part of the increased
interaction between airlines operating within major alliances, interaction
which allows each partner to continue to gain from returns to regional
diversity. However, further research will be necessary to examine in more
detail the extent to which formal alliances serve to increase market power as
distinct from improving partner airlines efficiencies. Impending mergers
may also provide interesting opportunities to infer airlines’ motives in terms
of geographic rationalization and additional fleet simplification, among
related strategies which may be critical to greater efficiency and survival.
REFERENCES
Adler, N., & Golany, B. (2001). Evaluation of deregulated airline networks using data envelop-
ment analysis combined with principal component analysis with an application to
Western Europe. European Journal of Operational Research, 132(2), 260273.
Andersen, P., & Petersen, N. (1993). A procedure for ranking efficient units in data envelop-
ment analysis. Management Science, 39, 12611264.
Barros, C., & Peypoch, N. (2009). An evaluation of European airlines’ operational perfor-
mance. International Journal of Production Economics, 122(2), 525533.
Brander, J., Cook, B., & Rowcroft, J. (1989). Congestion, concentration and contestability:
The case of the airline industry. Transportation Research Record, 1214, 3743.
Brooks, M., & Ritchie, P. (2003). Mergers & acquisitions in the air sector 19962000. Air
Transport Research Group Annual Conference, Toulouse.
Canadian Transportation Agency. (2004). 100 years at the heart of transportation: A centen-
nial perspective of the Canadian Transportation Agency and its predecessors. Minister
of Public Works and Government Services Canada, Ottawa.
Caves, D., Christensen, L., & Tretheway, M. (1984). Economies of density versus economies
of scale: Why trunk and local service airline costs differ. Rand Journal of Economics,
15(4), 471489.
Cornwell, C., Schmidt, P., & Sickles, R. (1990). Production frontiers with cross-sectional and
time-series variation in efficiency levels. Journal of Econometrics, 46, 185200.
Creel, M., & Farell, M. (2001). Economies of scale in the US airline industry after deregula-
tion: A Fourier series approximation. Transportation Research Part E, 37(5), 321336.
Davis, P., & Garces, E. (2010). Quantitative techniques for competition and antitrust analysis.
Princeton, NJ: Princeton University Press.
DEA-Solver-PRO. (2013). Retrieved from http://www.saitech-inc.com. Accessed May 11,

2013.
Doganis, R. (2001). The airline business in the 21st century. Abingdon: Routledge.
Good, D., Röller, L., & Sickles, R. (1995). Airline efficiency differences between Europe and
the US: Implications for the pace of EC integration and domestic regulation. European
Journal of Operational Research, 80(3), 508518.
IASB-FASB Update Report to the FSB Plenary on Accounting Convergence. (2012).
Compiled by H. Hoogervorst, International Accounting Standards Board and
L. Seidman, Financial Accounting Standards Board, pp 5. Retrieved from http://www.
ifrs.org/Use-around-the-world/Global-convergence/Convergence-with-US-GAAP/
Documents/r_120420d.pdf. Accessed on July, 2012.
Jara-Dı́az, S. R., Cortés, C. E., & Morales, G. A. (2013). Explaining changes and trends in the
airline industry: Economies of density, multiproduct scale, and spatial scope.
Transportation Research Part E, 60, 1326.
Kumbhakar, S. (1990). A reexamination of returns to scale, density and technical progress in
US airlines. Southern Economic Journal, 57(2), 428442.
Liu, Z., & Lynk, E. (1999). Evidence on market structure of the deregulated US airline indus-
try. Applied Economics, 30(9), 10831092.
Lovell, K. (1993). Production frontiers and productive efficiency. In H. O. Fried, C. A. Lovell,
& S. S. Schmidt (Eds.), The measurement of productive efficiency: Techniques and appli-
cations. New York, NY: Oxford University Press. (chapter 1).
Nolan, J., Ritchie, P., & Rowcroft, J. (2004). September 11 and the world airlines financial cri-
sis. Transport Reviews, 24(2), 239255.
Ray, C., & Mukherjee, K. (1996). Decomposition of the Fisher ideal index of productivity: A
non-parametric dual analysis of US airlines data. The Economic Journal, 106(439),
16591678.
Schefczyk, M. (1993). Operational performance of airlines: An extension of traditional mea-
surement paradigms. Strategic Management Journal, 14(4), 301317.
Scheraga, C. A. (2004). Operational efficiency versus financial mobility in the global airline
industry: A data envelopment analysis and Tobit analysis. Transportation Research
Part A, 38, 383404.
Scherer, F., & Ross, D. (1990). Industrial market structure and economic performance (3rd ed).
Boston: Houghton Mifflin.
StataCorp. (2003). Stata Statistical Software: Release 8.0. Stata Corporation, College Station,
TX.
Tsolas, I., & Manoliadis, O. (2003). Sustainability indices of thermal electrical power produc-
tion in Greece. Journal of Environmental Engineering, 129(2), 179182.
APPENDIX A: AIRLINE WEB SITES
www.aa.com
www.aircanada.com
www.airfrance.com
www.airnewzealand.com
www.ba.com
www.cathaypacific.com
www.continental.com
www.delta.com
www.finnair.com
www.lufthansa.de
www.nwa.com
www.qantas.com.au
www.ryanair.com
www.scandinavian.net
www.singaporeair.com
www.united.com
www.westjet.com
APPENDIX B: AVERAGE EFFICIENCY SCORES FOR

AIRLINES IN SAMPLE, 19982010
Airline Average Efficiencies
Super BCCi BCCi
Air Canada 0.9900 0.9020

Air France 0.7491 0.7491
Air New Zealand 0.8311 0.8311
American Airlines 0.7742 0.7742
British Airways 0.7576 0.7576
Cathay Pacific 0.9368 0.9224
Continental 0.9466 0.9347
Delta 0.9040 0.8924
Finnair 0.9579 0.9313
Lufthansa 0.9921 0.9636
Northwest Airlines 0.8886 0.8857
Qantas 0.8896 0.8778
Ryanair 1.0812 0.9778
Scandinavian 0.8679 0.8443
Singapore Airlines 0.9650 0.9597
Southwest Air 0.8091 0.8091
United Airlines 0.9246 0.9170
WestJet 0.9426 0.9046
CHAPTER 6
AIRFARES AND COMPETITION

ON INTERNATIONAL ROUTES
John Bitzan, Alice Kones and James Peoples
ABSTRACT
This chapter uses airline data on fares, traffic, and flight characteristics
to estimate a series of fare equations for international flights. The results
are used to examine the role of international competition as a determi-
nant of fares along international flights originating or departing from the
United States. Findings suggest that actual and potential competition are
important determinants of international airfares. We interpret these
results as indicating that pricing behavior along USinternational routes
is consistent with the theory of imperfect contestability.
Keywords: Imperfect contestability; international air fares; Open Skies

Agreements; cabotage
JEL classifications: L93; L11
INTRODUCTION
Since 1978, the US aviation industry has experienced liberalized regula-

tions, including policy changes that promote competition among domestic

ISSN: 2212-1609/doi:10.1108/S2212-160920140000004005
151
152 JOHN BITZAN ET AL.
carriers. More recently, liberalization policies have been more expansive in

their coverage, including the promotion of international competition. For
instance, the Open Skies Agreements (OSAs) between the United States
and other countries liberalized aviation rules on international air transport
by minimizing government intervention in decisions on passenger and
cargo routes, fares, and capacity for international carriers. OSAs therefore
remove barriers to entry and allow for greater competition along interna-
tional Origin-Destination (O-D) pairs. Standard economic theory suggests
that enhanced competition should promote low fares on these routes.
In support of the competitive market hypothesis, past research examin-
ing the effect of EUUS OSAs shows that between 1996 and 1999 fares on
routes under OSAs fell by 20.1% compared to 10.3% for non-Open Skies
routes (US Department of Transportation, 2000). While past work exam-
ines fare changes following the OSAs, there is a dearth of research examin-
ing price variation along international O-D pairs. In particular, the role of
OSAs in enhancing competition along routes with varying traffic levels is
not known.
Moreover, understanding the benefits of OSAs and further revisions to
OSAs requires a better understanding of the role played by competition in
determining international air fares. Studies investigating the relationship
between competition and domestic US airfares provide a framework based
on the traditional structure-conduct-performance framework and on the
theory of perfect and imperfect contestability (Graham, Kaplan, & Sibley,
1983; Leigh, 1990; Morrison & Winston, 1987). Their findings indicate
pricing behavior by domestic carriers in the US market is consistent with
the theory of imperfect contestability. Such findings are significant, in part,
because they suggest that in addition to actual competition, the threat of
potential entry facilitates carriers charging competitive fares. While such
findings contribute substantially to our understanding of fare determina-
tion, these same empirical examinations have not been performed for inter-
national airfares. To the extent that OSAs and changes in those agreements
affect the level of actual route competition or potential route competition,
this investigation is germane to understanding the impacts of these changes
on consumer welfare.
This study uses international airfares from the US Department of
Transportation’s (DOT’s) DB1B Airline Origin and Destination Survey
along with the US DOT’s T100 segment data to examine the role of inter-
national competition in determining fares along international routes. We
estimate fare equations for all US carriers serving international O-D pairs
involving the United States as an origin or destination. Our equations
Airfares and Competition on International Routes 153
examine the role of cost factors, demand factors, and competition in

explaining differing airfares among international routes.
The next section of the study examines the history of OSAs for the
United States, along with their details and likely impact on international
aviation competition. The following section includes a discussion of the
theory of perfect and imperfect contestability in the context of aviation,
their empirical implications, and findings of previous studies examining
these concepts in a domestic context. A description of empirical models,
along with a presentation and description of results follows. Finally, a
discussion of the implications for OSAs and their future is presented.
US REGULATION OF INTERNATIONAL
AVIATION OPERATIONS
Past research identifies maintenance of sovereign rights over air transporta-

tion as a major reason for establishing restrictions limiting international
operations for foreign carriers (Button, 2009).1 As early as 1919, restric-
tions were enacted that imposed national controls on foreign competition
and investment in domestic airline companies. Most countries banned
operations of foreign air carriers servicing domestic routes (Cabotage), as
well as foreign ownership of national carriers. Bilateral agreements between
governments often limited the routes, number of scheduled flights, and
capacity for international air carriers. Under the “National Clause,” bilat-
eral agreements restricted cross-border investment in foreign carriers, and
therefore restricted control and substantial ownership of carriers to a coun-
try’s citizens (Button, 2009; Robyn, Moselle, & Reitzes, 2005). In the
United States, Congress passed the 1926 Air Commerce Act (ACA), which
imposed relatively stringent restrictions on foreign investment in US airline
companies (Hardaway, 2007). This Act mandated that US citizens own at
least 51% of any aircraft registered in the United States and that at least
two-thirds of the board of directors of any US airline be comprised of US
citizens (Westra, 2007). Further restrictions on foreign participation in US
air transportation followed with passage of the 1938 Civil Aeronautics Act
(CAA) (Hardaway, 2007). This Act required US citizens to own or control
at least 75% of voting rights in any US carrier.
Following this series of legislation limiting competition of international
air transportation services, momentum developed supporting easing of bar-
riers to entry by foreign carriers. For instance, immediately following the
end of World War II, delegations from 54 countries convened at Chicago

to form the International Air Transport Agreement (IATA). Provisions
included the right for nonscheduled flights to land for noncommercial
purposes (e.g., refuel). Scheduled flights for international air services had to
seek permission to fly over or land in the territory of a contracting state.
Engaging in cabotage, however, was still prohibited. The treaty signed at
the Chicago Convention also failed to loosen restrictions on other interna-
tional air operations, and as a result bilateral state agreements were formed
to address international regulation on airfares, freight tariffs, capacity, and
frequency of flight departures and arrivals (Doganis, 1973).
The movement toward greater liberalization of international air trans-
port operations arising in the postwar era continued to develop throughout
the remainder of the 20th century. Notable among liberalization develop-
ments at that time was the Open Skies Agreement (OSA) between the
United States and Netherlands in 1992, as it established a new a framework
governing international air services. This agreement contributed to
enhanced competition from a foreign carrier as highlighted by Button
(2009) and Pitfield (2009). They observe that negotiated provisions pro-
vided the Dutch carrier KLM, the flexibility to meet market demand any-
where in the United States, as well as exempting it from anti-trust
restrictions with its Northwest Airlines alliance.
Following this initial 1992 agreement, a growing number of US and
European Union Bilateral OSAs contributed to continued liberalization
of transatlantic aviation. In 2007, the United States and EU concluded a
comprehensive USEU OSA on air transport that included all 27 EU coun-
tries.2 The legal framework within which US and EU members now operate
is captured in a number of major provisions of the comprehensive agreement
as outlined by Alford and Champley (2007).3 First, member countries are
allowed broader entry into cooperative marketing arrangements for code
sharing, franchising, and leasing. This is facilitated by a cooperative joint
committee to further airline deregulation. Second, US investors are now per-
mitted to participate as minority shareholders in any majority-owned airline.4
Third, EU carriers are now recognized as “community air carriers” by the
United States, allowing them to schedule flights between any EU member
state and the United States without touching the home country and without
any price or capacity restrictions (European Commission, 2008). For exam-
ple, a UK British Airways flight can go from Paris to the United States, with-
out having to pass through the United Kingdom. The comprehensive
agreement also grants EU carrier companies who purchase a carrier or invest
in a controlling share in a carrier from a third party that has an OSA with the
United States to continue operating without jeopardizing the acquired
airline’s rights to operate in the United States. Under the agreement, EU

member carriers are also authorized to transport certain Fly America traffic,
except for the Department of Defense. In addition, EU carriers can now sche-
dule flights between several third-party states and the United States without
touching the home country, including between the United States and mem-
bers of the European Common Aviation Area as well as African countries.5
Easing of restrictions negotiated in the comprehensive agreement has
the potential to enhance operating efficiency and promote international
competition. Potential efficiency gains arise in part from provisions
allowing carriers to fly less circuitous routes. Promotion of international
competition arises from provisions liberalizing code sharing, franchising,
leasing, pricing, and routes between the United States and EU (Button,
2009; Pitfield, 2009).
Although OSAs liberalize international aviation, significant barriers still
remain. EU carriers are still prohibited from engaging in cabotage in the
United States, as US law does not permit EU carriers the right to fly
between two points in United States before returning to Europe (Westra,
2007). Another restrictive feature of OSAs is that there are still restrictions
on carriage between third party states and the United States. For example,
while a UK-owned airline may operate direct service from Heathrow to
Chicago or from Paris to Chicago, it may not operate from Beijing to
Chicago.6 Moreover, non-EU carriers cannot transport passengers between
third party countries and the United States (e.g., a Chinese carrier can’t
transport from the United Kingdom to the United States). Furthermore,
European and other non-US airlines, still face a 25% legislated cap on
voting equity, and 25%-minus-one-share regulatory cap on nonvoting
equity in US airlines (Alford & Champley, 2007).
These remaining restrictions on foreign ownership, cabotage, and routes
continue to create efficiency challenges for non-US carriers. Prohibiting
foreign companies from acquiring US airline companies makes it difficult
for foreign airlines to experience efficiency gains through economies of
scale, scope, and density when providing service to the US market. Hsu
and Chang (2005) explain that removing provisions on foreign ownership
makes it easier for foreign carriers to acquire domestic air companies,
allowing newly allied carriers to spread fixed costs over more passengers.
Foreign acquisition promotes economies of scope by allowing newly allied
carriers to reconfigure their networks to connect more flights to hub
airports. Acquisition promotes economies of density by allowing newly
allied carriers to combine traffic to improve load factors. Remaining
restrictions on international air transportation also create impediments to
competition by limiting the number of carriers that have the ability to serve
various routes and by limiting the scope of carrier operations. Removing

these remaining restrictions then has the potential of promoting entry from
cost competitive foreign carriers.
PERFECT CONTESTABILITY, IMPERFECT

CONTESTABILITY, AND INTERNATIONAL AIR FARES
Industry analysis using the traditional structure-conduct-performance
paradigm suggests further easing of restrictions that limit entry along inter-
national routes should place downward pressure on fares (Scherer & Ross,
1990). Within this conceptual framework, lower fares are associated with
less dominance by domestic incumbent carriers on international routes. In
contrast, the theory of contestability explains under certain circumstances
the threat of potential competitors can act as an incentive to force the
incumbent to charge competitive prices. Within this theoretical framework,
if the domestic incumbent charges noncompetitive prices, it risks entry into
the market by and loss of market share to potential foreign or domestic
entrant carriers. Such entry would undercut the noncompetitive fare set by
the incumbent carrier, while allowing entrant the opportunity to generate
economic profits. Competitive pricing, therefore, would persist with or
without entry if international routes are contestable.
The pricing behavior hypothesis that is consistent with the theory of
contestability relies on three fundamental assumptions. First, it is necessary
for firms to face the same production functions when providing air trans-
port service. To satisfy this assumption, the domestic incumbent and the
potential foreign entrants must have access to the same technology and
resources to ensure that the incumbent does not benefit from a cost advan-
tage. Second, contestability requires the potential entrant to have the oper-
ating capacity to meeting market demand at competitive prices; otherwise
offering low prices does not present a viable threat to the loss of market
share for incumbent carriers.7 Finally, firms must face zero-sunk cost. If
there are significant sunk costs associated with entering the industry, this
creates a barrier to exit for potential entrants, reducing the likelihood of hit
and run entry. It should be noted that this assumption does not require
that fixed costs be equal to zero. If fixed costs exist, then the monopolist
faces these costs also, under the first assumption, and does not have an
advantage over potential entrants. The only condition placed on fixed costs
is that exit is costless, or that fixed costs will not generate losses when exit
is desired. In the absence of this condition, entrants will have a disincentive

to compete for fear of losses, meaning the incumbent will have the ability
to take advantage of monopoly power. As long as exit is costless, the
incumbent will engage in average cost pricing in a contestable market to
avoid negative profits.
Following Tirole (1988), the pricing dynamics associated with contest-
ability can be expressed more generally by supposing that there are m
incumbents (i = 1,…, m) and a nonnegative number of potential entrants,
n-m. For the market to be feasible, the market must clear
X
m
qi = DðpÞ
i=1
and firms must make nonnegative profits
pðqi Þ × qi ≥ Cðqi Þ
where C(q) is the cost function facing all firms and C(0) = 0.
With increasing returns technology, let CðqÞ = f þ cq, where f is fixed
cost and c is marginal cost. Let
π m = maxq f½pðqÞ − c qg
As shown in Fig. 1, the contestable price and output combination occurs
at the intersection of average cost ðACðqÞ = f =q þ cÞ and the demand curve.
As a result
f = ðpc − cÞDðpc Þ
P q=D(p)
pc c+f/q
c (MC)
qc q
Fig. 1. Illustration of Contestability with Increasing Returns.

Assessing the likelihood that the three primary assumptions for contest-
ability are met is critical in considering whether fare determination for inter-
national air transport service is consistent with this theory. Bailey and
Panzar (1981) present a useful argument regarding the inherent barriers to
entry in the case of a dominant firm. Examining the airline industry, they
argue that the existence of a natural monopoly between many city-pairs for
airlines does not result in barriers to entry and consequently does not impede
market efficiency. Barriers to entry are created with sunk costs, not necessa-
rily with economies of scale. Additionally, they argue that the majority of
the initial costs in airlines can be recovered at little or no cost to the firm.
While Bailey and Panzar (1981) conclude that airlines exhibit characteris-
tics consistent with contestability using data for domestic US air transport,
others find little empirical support for the contestability hypothesis. For
instance, Graham et al. (1983) measure the effect of market concentration
on fares by specifying a fare equation to include the distance flown, market
concentration as measured by the route Herfindahl Index, traffic volume,
and travelers’ valuation of time as a measure of service quality. Controlling
for market structure endogeneity, they observe a positive estimated coeffi-
cient on the Herfindahl Index, refuting the contestability hypothesis.
In addition, they include a variable for newly certificated airlines and find it
to be negative and significant, indicating that fares decline with the entry of
new firms.
Further investigation by Leigh (1990) examines whether other hypotheses
derived from perfect contestability theory hold for the airline industry.
He notes that in a perfectly contestable market, carrier profits should not
deviate appreciably from competitive levels, which he identifies as the
average industry profit level. He estimates a parsimonious profit equation
that includes personal income for a given year, carrier market share of
passenger service, and carrier dummies. Consistent with Graham, Kaplan
and Sibley, Leigh finds market concentration is associated with higher
profits. In addition, he finds profit levels vary significantly across carriers.
Neither finding supports the perfect contestability hypothesis. As an alter-
native to the perfect contestability model, Leigh observes the possibility that
establishing a hub-and-spoke network presents incumbent carriers a cost
advantage due to their ability to enhance load factors by using feeder routes
connected to their hub airport.8 To test this “network effect hypothesis”
Leigh estimates a fare equation that includes the number of connections
providing traffic feed to the hub. In support of the network effect hypoth-
esis, he finds a positive and statistically significant association between the
feeder rout parameter and air fares.9
While past research does not support the applicability of perfect contest-
ability to fare determination in air transport services, Morrison and
Winston (1987) observe that these studies do not directly test the possibility
of imperfect contestability as a framework for analyzing fare determination
in a regulatory environment that promotes competitive entry. These
authors note that perfect contestability requires that potential competition
is sufficient to achieve optimal pricing. Yet research performed prior to
their study only includes measures of actual competition and fails to
include a direct measure of potential competition. Thus, while most pre-
vious research shows that perfect contestability doesn’t exist (due to an
effect of actual competition on fares), the specification of their fare equa-
tions does not allow for the possibility of imperfect contestability; that is,
the ability for potential competition to have an effect on fares in addition
to that of actual competition. Including an analysis of the welfare effect of
potential competition is significant, because evidence of higher consumer
welfare on routes facing the threat of potential entrants would support the
hypothesis of imperfect contestability. Morrison and Winston estimate a
consumer welfare equation that includes potential competition as a deter-
minant. They depict potential competition as the number of airlines serving
at least one airport on the route without offering direct or on-line connect-
ing service for that route. They find a statistically significant negative value
of the parameter estimate, which is consistent with the theory of imperfect
contestability.10
Although a number of studies have examined competition and fares in a
domestic context, the role of competition in determining fares in an inter-
national context has not been investigated to the same extent. Nonetheless,
such an investigation is very important for informed policymaking regard-
ing the future of OSAs. For example, as stated previously, a British carrier
can travel between the United Kingdom and the United States, but not
between China and the United States. Thus, if potential competition plays
an important role in determining international fares, then relaxing this type
of restriction may create an important benefit for consumers. In particular,
if the restriction is removed the UK carrier becomes another potential
competitor for this route.
This example also highlights the importance of distinguishing the three
models about the role of competition in fare determination. Under a tradi-
tional structure-conduct-performance model where only actual competition
matters, relaxing restrictions that only allow carriers to serve markets
where at least one end point of a trip is in their home country will not have
any effect on fares unless those carriers actually start to serve such markets.
Relaxing these restrictions would not have any effect on fares under perfect
contestability either, since adding more potential competitors has no effect
on fares as long as one potential competitor already exists. However, under
imperfect contestability, relaxing this restriction would reduce fares,
since adding more potential competitors serves as a stronger discipline on
market power.
Moreover, to the extent that an investigation of international fares
provides similar findings for the role of competition as has been found for
domestic fares in the United States, some evidence would exist to suggest
that the relationship between competition and fares may be generalizable
to the airline industry as a whole. This would suggest a benefit from pro-
competitive policies in the airline industry, regardless of the market under
consideration. For example, a finding of an important role for potential
competition would provide further support for the idea that relaxing cabo-
tage restrictions in the United States would be beneficial for US consumers.
It is important to note that there may be a reason to believe the relation-
ship between competition and fares may be different in an international
context than in a domestic US context. In particular, Leigh (1990) argues
that perfect contestability doesn’t hold for domestic US air transportation
because a hub airport gives a cost advantage to carriers on routes going to
or from that hub airport, due to higher load factors. For international
flights, almost all flights travel between hub airports, and all have very high
load factors. Thus, the cost advantage associated with hubs in domestic air
transportation is not likely to exist for international air transportation, sug-
gesting that perfect contestability may be a possibility. On the other hand,
however, the existence of government-owned foreign airlines that have
access to subsidies may give a cost advantage to foreign-owned airlines that
precludes perfect contestability.11
DATA AND EMPIRICAL APPROACH
In order to investigate the relationship between competition and airfares in

an international context, we estimate airfare equations using a sample of
one-segment international fares originating or terminating in the United
States in the fourth quarter of 2012. We use the international itineraries
from the DB1B Airline Origin and Destination Survey (US DOT), which is
a 10% sample of tickets for air travel within, to, or from the United
States.12 The DB1B dataset includes the year and quarter traveled, the
airfare paid, the origin and destination airports, and the round-trip status
for US carriers.13
We match these fares with the characteristics of those flights using the
US DOT’s T100 Segment and International Market Data. Data are
matched by carrier, origin, and destination. The T100 Segment Dataset
provides information on the number of passengers traveling on each inter-
national and domestic flight, the type of plane they are traveling in (includ-
ing seat configuration), and the number of flights made for a particular
airline between a specific origin and destination during a given time period.
The T100 International Market Data provides information on the number
of passengers traveling on a specific airline from a particular origin to a
particular destination during a given time period, regardless of the number
of stops in between the origin and destination.
Two types of fare equations are estimated in this study. The first mea-
sures the relationship between actual competition and fares to test whether
international airfares might be characterized by perfect contestability.
As noted previously, while a number of studies have shown that perfect
contestability does not hold for the US domestic aviation market, there is
reason to believe that it may hold for international aviation.
The second type of fare equation considers the critique of previous
studies by Morrison and Winston (1987) that notes that a lack of perfect
contestability does not necessarily mean a lack of imperfect contestability.
That is, while having at least one potential competitor may not force
airlines to charge a socially optimal price, potential competition may still
play a role in enhancing social welfare.
Both types of reduced-form fare equations used in this study have the
same general form. The generalized fare equation estimates fares as a func-
tion of factors influencing the costs a carrier incurs in serving a particular
route, factors influencing the strength of air passenger demand on a parti-
cular route, and factors influencing the effectiveness of competition and
potential competition. The following general relationship is estimated:
Fare per Passenger Mile = f ðcost characteristics; demand; competitionÞ
We standardize the airfare by dividing it by the number of one-way miles

traveled. By controlling for demand and competition, the variables that
influence costs should influence fares in the same way. Similarly, to the
extent that demand and cost characteristics are accounted for, the competi-
tion variables will show the influence of these variables on the airline’s
ability to price above marginal cost. Variables included to account for
differences in the cost carriers incur on different routes are total flight dis-
tance, the average plane size used, and the average number of passenger
miles per seat mile (load factor). The variable included to account for the
strength of demand is the frequency of service on the route. Competition
variables include a slot control dummy for airports that have slot controls,
HerfindahlHirschman indexes of actual and potential competition, and
the number of carriers serving the route or potentially serving the route.
Model 1 Testing for Perfect Contestability
The first model assesses the role of actual competition in influencing

airfares, and therefore, is an indirect test of perfect contestability. As noted
by several previous studies, if having at least one potential competitor
forces the incumbent to charge the socially optimal price, then having more
competition along a route would have no effect on fares. The model used
to assess the role of actual competition is specified as:14
ln FPMij = β0 þ β1 ln Distanceij þ β2 ln Loadij þ β3 ln Equipij
þ β4 ln Freqij þ β5 Herf ij þ β6 Slotij þ ɛ ij ð1Þ
where
FPMij = average (2012 Q4) airfare per passenger-mile for the carrier
between origin i and destination j (where miles are one-way distance)15
Distanceij = one-way distance between origin i and destination j
Loadij = average (2012 Q4) load factor for the carrier between origin i
and destination j (passenger-miles/seat-miles — weighted by the number
of departures performed)
Equipij = average (2012 Q4) plane size for the carrier between origin i
and destination j (number of seats — weighted by the number of depar-
tures performed)
Freqij = frequency (2012 Q4) of flights for the carrier between origin i
and destination j
Herfij = HerfindahlHirschman Index (2012 Q4) for flights between
origin i and destination j (summation of carrier shares of passengers
squared ranges between 0 and 1) for US and non-US carriers
Slotij = slot controlled airport at origin i or destination j (dummy variable)
Most airfare investigations distinguish between total flight distance and

average stage length; where stage length is the distance traveled for a parti-
cular flight segment. In our data, total flight distance and stage length are
the same thing, since all flights are one-segment trips. A priori, we expect
flight distance to have a negative impact on fares through its impact on
costs, as many airline costs are a function of the number of takeoffs and
landings varying less than proportionally with distance. Examples of these
types of costs include fuelling costs, boarding costs, luggage loading costs,
security costs, landing fees, and maintenance costs.
Other variables influencing costs, and therefore expected to influence
fares in the same way, include average load factor and average plane size.
Average load factor, or the average number of passengers per seat, is
expected to have a negative influence on fares, as many airline costs vary
less than proportionally with the number of passengers. For example, flight
crew costs, maintenance costs, fuel costs, gate fees, and terminal rental do
not vary proportionally with passengers. Similarly, as shown by Graham
et al. (1983), there are cost savings associated with operating larger aircraft.
Thus, we expect that average plane size will have a negative effect on fares,
a priori.16
The frequency of service is likely to influence fares through its impacts
on demand. Higher flight frequency is likely to mean more convenient sche-
dules for passengers, increasing the demand for those flights. This suggests
an expected positive influence of flight frequency on fares.
In order to capture the impact of varying levels of competition on
fares, we include two variables: (1) a slot control dummy variable for
flights originating or terminating at a slot-controlled airport,17 and (2) a
HerfindahlHirschman Index of market concentration for flights between
the origin and destination. Because takeoff and landing slots are limited
at the slot-controlled airports, the ability of an existing or new carrier to
handle more flights is limited. In this way, the slot control can serve as an
inhibitor to competition. Thus, the slot control dummy is expected to have
a positive sign in the fare equation, a priori.
The HerfindahlHirschman Index measures the level of actual competi-
tion existing on a route. It is defined mathematically as
X
N
H= Sl 2
l=1
where Sl is the share of passengers carried between origin i and destination

j by the lth firm. This index ranges between 0 and 1, representing perfect
competition and monopoly markets, respectively. In previous studies that

have examined domestic airfares in the United States, researchers have
found that the level of competition as measured by various indices such as
the HerfindahlHirschman Index does matter, rejecting the idea that
domestic aviation is perfectly contestable. As noted previously, the hub
advantage enjoyed by carriers in the domestic US aviation market is
unlikely to be realized in international aviation markets where most flights
travel between hub airports and have very high load factors. This suggests
that perfect contestability may be more likely to hold in an international
context. On the other hand, however, government involvement in interna-
tional aviation may suggest subsidies that give some carriers a cost advan-
tage. This would suggest that perfect contestability is unlikely. Because of
these conflicting factors, the applicability of perfect contestability to inter-
national aviation is an empirical question. Thus, the expected effect of
actual competition on fares in an international context is uncertain.
Model 2 Testing for Imperfect Contestability
Estimating Eq. (1) is expected to provide useful insight regarding the role
that actual competition plays in determining international airfares.
However, as noted by Morrison and Winston (1987), potential competition
may also play an important role. Therefore, we estimate a series of models
to capture the effects of actual and potential competition in determining
international airfares. All of these models test for a role for potential com-
petition after controlling for the effect of actual competition in determining
airfares.
Three models are used to examine the roles played by actual and poten-
tial competition in determining international airfares. The models differ in
the ways that competition is measured, essentially serving to examine how
robust the estimated relationship is between actual/potential competition
and fares.
In the first of these, we measure the effect of potential competition by
adding HerfindahlHirschman indexes of concentration at origin and
destination airports to our previous equation.
The model is specified as:
ln FPMij = β0 þ β1 ln Distanceij þ β2 ln Loadij þ β3 ln Equipij þ β4 ln Freqij

þ β5 Herf ij þ β6 OHerf i þ β7 DHerfj þ β8 Slotij þ ɛij ð2Þ
where
OHerfi = HerfindahlHirschman Index of the share of all passengers
loaded or unloaded at the origin airport (2012 Q4)
DHerfj = HerfindahlHirschman Index of the share of all passengers
loaded or unloaded at the destination airport (2012 Q4)
All other variables are defined as previously.

In this estimation, the origin and destination HerfindahlHirschman
indexes measure market concentration at the origin and destination air-
ports, respectively, for all passengers, regardless of where they are travelling
to or from. Thus, while many of the passengers represented in these con-
centration indexes travel on the route in question, many others travel to or
from other destinations or origins. As noted by Morrison and Winston
(1987), carriers must have a presence in at least one of the two airports in
order to engage in hit and run entry in an airline market. This suggests that
our origin and destination HerfindahlHirschman indexes, by including
concentration of passengers at each airport regardless of whether they tra-
vel in the particular market include a mix of actual and potential competi-
tion. That is, some of the passengers captured in the origin and destination
HerfindahlHirschman indexes travel in the particular market (measure of
actual competition), while others only travel to or from one of the two
airports (measure of potential competition). Because we also control for
route-level competition in this model (actual competition), the origin and
destination HerfindahlHirschman indexes should measure the incremental
effects of potential competition on fares.
In the second model used to assess the role of actual and potential
competition, we measure the effect of potential competition by adding the
number of carriers that serve either the origin or destination, but not the
route, to Eq. (1). The model is specified as

þ β4 ln Freqij þ β5 Herf ij þ β6 PCarrij þ β7 Slotij þ ɛij ð3Þ
where
PCarrij = Number of carriers serving the origin airport and/or the desti-
nation airport, but not the route (2012 Q4)
In this estimation, the number of carriers serving at least one of the

endpoints, but not the route is the potential competition variable used by
Morrison and Winston (1987). As noted by Morrison and Winston (1987),

it is these carriers (those that serve at least one of the endpoints) that have
the potential to engage in hit and run entry. Because the level of actual com-
petition is accounted for with the inclusion of the HerfindahlHirschman
Index of route-level competition, these variables show the incremental effect
on fares from potential competition. An advantage of measuring potential
competition in this way in comparison to using the origin and destination
HerfindahlHirschman indexes is that it results in a clearer distinction
between actual and potential competition. That is, none of the potential
competitors are actually involved in carrying passengers on the route.
The final model includes the number of carriers serving the route to
measure the level of actual competition, and the same measure of potential
competition as the previous model. The specification of the model is

þ β4 ln Freqij þ β5 Carrij þ β6 PCarrij þ β7 Slotij þ ɛij ð4Þ
where
Carrij = Number of carriers serving the route (2012 Q4)
This last estimation is included essentially as a robustness test. The
number of carriers does not provide as much information about competi-
tion as the HerfindahlHirschman Index. Not only does the Herfindahl
Hirschman Index vary with the number of carriers serving a route, it also
varies with the shares of traffic that each carrier handles. Thus, it takes into
account the fact that a market with many carriers may still not be very
competitive if one carrier dominates the route. For example, if four carriers
serve a route, with one carrier handling 85% of the passengers and the
remaining three each handling 5% of the passengers, the Herfindahl
Hirschman Index would be 0.7375. If, on the other hand, four carriers each
handle 25% of passengers on a route, the HerfindahlHirschman Index
would be 0.25. Nonetheless, since the goal of this research is to identify
whether potential and actual competition both play a role in determining
international airfares (not to identify the specific magnitude of such effects),
having several models to test these effects is desirable.
Empirical Implementation
In all of these estimations, following previous researchers (Bitzan & Chi,

2006), we eliminate fares that are outliers and we average fares by route and
carrier. This is done to eliminate fare variation where no variation in

independent variables exists. For example, there may be 200 fares from
Boston to Toronto on Pinnacle Airlines in the fourth quarter of 2012. In our
estimation, each of these fares would have the same distance, the same load
factor, the same equipment size, the same flight frequency, etc. By averaging
the fares, we have one observation for each carrier/route combination.
Descriptive statistics for all of our variables are shown in Table 1.18 As
the table shows, the average fare per one-way mile is $0.35, and the average
one-way distance is over 2,100 miles. Moreover, the average plane size is
over 155 seats, and flights are nearly 80% loaded. The table also shows
that more than 40% of the flights originate or terminate at slot-controlled
airports, and average route-level concentration is fairly high with a
HerfindahlHirschman Index of 0.75. Strikingly, the Herfindahl
Hirschman indexes of concentration at origin and destination airports are
much lower at about 0.29. This suggests that there is a lot of potential
competition on these routes, in addition to the actual competition. A simi-
lar picture is obtained when examining the number of carriers serving the
route and the number of potential carriers. As the table shows, an average
of two carriers serve each route. However, the number of potential carriers
on each route averages more than 50.19
An important empirical concern in these models relates to the potential
simultaneity bias. The load factor, service frequency, and Herfindahl
Hirschman Index all may be endogeneous. That is, not only do these
variables influence fares, but fares also probably influence them. Higher
fares may lead to lower load factors, lower service frequency, and lower
concentration as more air carriers enter markets with high fares. However,
despite these concerns, a previous study with a very similar model
Table 1. Descriptive Statistics.

Variable Mean Standard Deviation
Fare per Mile $0.3540 $0.2334

Distance 2,179.04 1,743.10
Load Factor 0.789 0.115
Seats per Plane 155.03 68.74
Frequency of Service 103.74 96.21
Herfindahl Index of Route-Level Competition 0.753 0.261
Herfindahl Index of Shares at Origin Airport 0.288 0.167
Herfindahl Index of Shares at Destination Airport 0.288 0.167
Number of Carriers Serving the Route 2.14 1.35
Potential Carriers 50.16 18.52
Slot Control 0.408 0.492
specification found that two-stage least squares results were very similar to
ordinary least squares results (Bitzan & Chi, 2006). Moreover, another
study could not reject exogeneity of the HerfindahlHirschman Index in an
airfare equation (Graham et al., 1983). In this study, we report OLS
results.20
EMPIRICAL RESULTS
The empirical results for the model used to assess the applicability of
perfect contestability to international aviation markets are presented in
Table 2. As the table shows, parameter estimates on the variables that
influence costs are statistically significant and have their expected signs.
Distance has the expected negative influence on fares due to the fact that
several airline costs do not vary proportionally with distance. Similarly, the
load factor has a strong negative effect on fares due to airline costs varying
less than proportionally with the number of passengers. Flights in larger
aircraft also realize lower fares due to economies associated with plane size.
In addition, the variable influencing the demand for service also has
its expected sign. The parameter estimate for the frequency of service is
statistically significant and has a positive sign, suggesting that travelers are
willing to pay more to have the convenience associated with more frequent
flights.
Table 2. Estimation of Airfares for One-Segment International Routes

Accounting for Route-Level Competition.
Variable Parameter Estimate Standard Error
Intercept 1.9259* 0.1380

Distance −0.2540* 0.0223
Load Factor −0.1328** 0.0643
Seats per Plane −0.4010* 0.0329
Frequency of Service 0.1146* 0.0093
Herfindahl Index of Route-Level Competition 0.1906* 0.0415
Slot Control 0.1033* 0.0248
Adjusted R2 = 0.4892.
F = 285.11.
N = 1,781.
All variables except Herfindahl Index and Slot Control are in natural logarithms.
*Significant at the 1% level.
**Significant at the 5% level.
Of particular interest in this estimation is the impact of actual competi-

tion on fares. Table 2 shows that fares are higher for flights that originate or
terminate at slot controlled airports than they are for other airports.
As mentioned previously, the reduction in airport access created by
slot controls is likely to limit competition, and therefore increase rates.
The other competition variable included in this estimation is the
HerfindahlHirschman Index of actual competition. As the table shows,
the HerfindahlHirschman Index of actual competition has a positive
sign and is significant at conventional levels.21 This suggests, consistent with
the findings of previous authors in domestic aviation, that international
aviation markets are not characterized by perfect contestability. Thus, to
the extent that further liberalization of international aviation markets intro-
duces more route-level competition, consumer welfare would be enhanced.
The empirical results for the models aimed at examining the roles of
actual and potential competition in international aviation are shown in
Tables 35. As the tables show, cost variables and the demand variable
still have their expected relationships to fares in all these equations.
Moreover, the slot control dummy and the actual competition variables
maintain their expected relationships with fares.
Table 3. Estimation of Airfares for One-Segment International Routes –

Actual Competition (Route-Level Herfindahl Index), Potential
Competition (Origin/Destination Herfindahl Indexes).
Intercept 1.9369* 0.1373

Distance −0.2722* 0.0225
Load Factor −0.1077*** 0.0643
Herfindahl Index of Route-Level Competition 0.1039** 0.0456
Herfindahl Index of Shares at Origin Airport 0.2334* 0.0654
Herfindahl Index of Shares at Destination Airport 0.1942* 0.0652
F = 218.56.
N = 1,781.
All variables except Herfindahl Index and Slot Control are in natural logarithms.
***Significant at the 10% level.

Actual Competition (Route-Level Herfindahl Index), Potential Competition
(Carriers Serving Origin and/or Destination, but not the Route).
Intercept 2.0003* 0.1423

Distance −0.2539* 0.0223
Load Factor −0.1234*** 0.0644
Herfindahl Index of Route-Level Competition 0.1728* 0.0423
Number of Potential Carriers −0.0013** 0.0006
F = 245.50.
N = 1,781.
All variables except Herfindahl Index, Potential Carriers, and Slot Control are in natural
logarithms.

Actual Competition (Number of Carriers Serving the Route), Potential
Competition (Carriers Serving Origin and/or Destination, but not the
Route).
Intercept 2.2049* 0.1437

Distance −0.2527* 0.0226
Load Factor −0.1187*** 0.0647
Number of Carriers Serving the Route −0.0203** 0.0084
Number of Potential Carriers −0.0014** 0.0006
F = 242.48.
N = 1,781.
All variables except Carriers, Potential Carriers, and Slot Control are in natural logarithms.
Of particular interest in these equations is the relationship between poten-

tial competition and fares. As the tables show, no matter how potential
competition is measured, it seems to play an important role in fare determi-
nation. In Table 3, the parameter estimates on origin airport and destination
airport HerfindahlHirschman indexes are both positive and significant at
conventional levels, suggesting that a higher concentration index of poten-
tial competition at origin and destination airports increases fares. In
Table 4, the parameter estimate on the number of carriers serving the origin
and/or destination, but not both, is negative and significant at conventional
levels, suggesting that more potential competitors reduce fares.22 Similarly,
Table 5 shows that more potential competitors reduce fares, when actual
competition is measured by the number of carriers serving the route. All of
these estimations suggest that actual and potential competition are impor-
tant in determining international airfares. This evidence is consistent with
Morrison and Winston (1987) for domestic US markets, suggesting that
airline markets are characterized by imperfect contestability.
CONCLUDING REMARKS
Based in part on declining real fares following the passage of earlier bilateral
agreements, past research predicts further declining air fares along interna-
tional routes following the 2007 comprehensive USEU OSA (Hofer &
Dresner, 2007). These authors identify enhanced competition resulting from
implementation of provisions outlined in bilateral agreements as major con-
tributors to lower fares. In addition to the benefits to consumers that have
already occurred from liberalization of international air transport regula-
tions, it is possible that further opportunities exist if remaining restrictions
are relaxed. This study, however, observes that the potential for attaining
even lower fares due to further lowering of entry barriers depends on the
nature of the relationship between fares and competition. We focus on the
models of perfect and imperfect contestability as a framework for predicting
potential fare patterns if remaining restrictions on international air trans-
port were removed. Empirically testing hypotheses derived from these mod-
els requires estimating fare equations that include measures of potential and
actual competition along international routes. Findings suggest that actual
and potential competition are important determinants of international air-
fares. These findings comport well with pricing behavior in an imperfect
contestable theoretical framework. We interpret these findings as suggesting
that easing remaining restrictions on OSAs has the potential to enhance

consumer welfare, even if those policies do not generate additional route-
level competition.
NOTES
1. Early regulation of international air transport is depicted by provisions

negotiated at the 1919 Paris international aviation convention. Recognizing the
importance of sovereignty over their airspace so as to protect their citizens, the
United States and British Empire among other nations, signed the treaty which pro-
vided member countries the right to deny entry and regulate foreign and domestic
flights into and through its airspace. It should be noted that while some countries
such as the United States signed the treaty, they did not ratify it.
2. Croatia’s entrance into the EU in 2013 means that there are now 28 EU
countries in this agreement.
3. For a more detailed presentation of the provision of the 2007 comprehensive
USEU Open Skies Act see Button (2009), Pitfield (2009), Alford and Champley
(2007), Robyn et al. (2005), Westra (2007), Prokop (2014), and Fu and Oum (2014),
among others.
4. This provision means that the minority shareholders from the United States
have voting rights as long as the EU carrier is majority owned by an EU company.
It should be noted that EU majority ownership requirement can be waived through
EU comprehensive agreements with partner countries that would allow foreign
majority ownership and control comprehensive agreements (Westra, 2007).
5. ECAA member countries include Norway, Iceland, Croatia, Macedonia,
Albania, Bosnia and Herzegovina, Serbia, Montenegro Switzerland, and Liechtenstein.
6. Robyn et al. (2005) noted that the UK carrier could not transport passengers
from Paris to Chicago in 2005. However, under the 2008 agreement, the United
States now recognizes any EU carrier as a “community air carrier.”
7. This assumption is necessary for absolute entry, which implies that if a poten-
tial entrant charges a price below the incumbent’s price, then the entrant will
displace the incumbent.
8. Due to significant declines in passenger demand for air service following the
September 11 attack, extensive hub-and-spoke networks were actually a liability to
legacy carriers.
9. Several other researchers find a positive association for fares and competition
along routes. See, for example, Call and Keeler (1985) and Moore (1986).
10. Peteraf and Reed (1994) present further evidence supporting the imperfect
contestability hypothesis.
11. Foreign governments owned shares of approximately 221 airlines by 2012.
Source: http://www.icao.int/sustainability/Documents/PrivatizedAirlines.pdf.
12. Our sample is limited to one-segment movements, as we are not able to
identify origins and destinations for multiple-segment movements. All airfares are
for round-trip movements and include taxes and fees, which vary by destination
and originating airports.
13. Although the type of ticket (e.g., unrestricted first class) is identified in the
DB1B, the US Bureau of Transportation Statistics notes that different carriers may
use different standards to classify tickets. Therefore, they recommend that this
classification not be used for analysis.
14. In all models, all variables are in natural logarithms except competition
variables. All competition variables are in level form.
15. The fare per-mile variable used in this study is often referred to in other
studies as yield.
16. It is important to note that each of the variables hypothesized to influence
the costs of a particular movement may also potentially influence demand. For
example, higher load factors may mean less comfort due to crowding, decreasing
demand. As another example, as Borenstein (1989) suggests, larger aircrafts are
often more comfortable and perceived to be safer than smaller aircrafts. This may
increase the demand for travel on larger aircrafts. However, previous researchers
have shown that the cost effects of these variables dominate (e.g., Borenstein, 1989
or Bitzan & Chi, 2006).
17. The International Air Transport Association (2013) lists 160 airports as
fully coordinated level-3 airports. They define these as airports where the gate
and runway demand exceeds capacity, necessitating slot allocation through a slot
coordination process.
18. Variable descriptions and data sources are described in Table A1.
19. It should be noted that our measure of potential competition may understate
the number of potential competitors to a certain extent, because it is based on US
DOT data. The US DOT only collects traffic data for those carriers serving routes
involving the United States. Therefore, foreign carriers that originate at foreign air-
ports, but that don’t serve the United States are not counted in the airlines that
serve as potential competitors.
20. We estimated both models using 2SLS, and the results were very similar to
OLS. All signs and significance of independent variables remained the same, though
the magnitude of the parameter estimate on load factor increased some.
21. In running this same model with the natural logarithm of Herfindahl
Hirschman index, the sign remains the same and it is also significant at conventional
levels.
22. When estimating this same model with competition variables in natural
logarithms, the number of potential competitors was only significant at the 20%
level.
REFERENCES
Alford, E., & Champley, R. (2007). The impact of the 2007 USEU Open Skies Air Transport
Agreements. International Trade Administration Occasional Paper No. 07-001. US
Department of Commerce.
Bailey, E., & Panzar, J. (1981). The contestability of airline markets during the transition to
deregulation. Law and Contemporary Problems, 44, 125145.
Bitzan, J. D., & Chi, J. (2006). Higher airfares to small and medium sized communities
Costly service or market power? Journal of Transport Economics and Policy,
40(September), 473501.
industry. RAND Journal of Economics, 20(Autumn), 344365.
Button, K. J. (2009). The impact of USEU ‘opens skies’ agreement on airline market
structures and airline networks. Journal of Air Transport Management, 15, 5971.
Call, G. D., & Keeler, T. E. (1985). Airline deregulation, fares, and market behavior: Some
empirical evidence. In Analytical studies in transport economics (pp. 221247).
Cambridge: Cambridge University Press.
Doganis, R. (1973). Air transport A case study in international regulation. Journal of
Transport Economics and Policy, 7, 109133.
European Commission. (2008). The EUUS ‘Open Skies’ air transport agreement- Q&A.
MEMO/08/185 28/03/2008.
Fu, X., & Oum, T. H. (2014). Air transport liberalization and its effects on airline competition
and traffic growth An overview. In J. Peoples (Ed.), The economics of international
airline transport (Vol. 4, pp. 1144). Advances in Airline Economics. Bingley, UK:
Emerald Group Publishing Limited.
Graham, D. R., Kaplan, D. P., & Sibley, D. (1983). Efficiency and competition in the airline
industry. Bell Journal of Economics, 14, 114138.
Hardaway, R. (2007). Of cabbages and cabotage: The case for opening up the U.S. airline
industry to international competition. Transportation Law Journal, 34(1), 2007.
Hofer, C., & Dresner, M. E. (2007). The United StatesEuropean union open aviation area:
The American perspective. Journal of the Transportation Research Forum, 46(1), 129–
143. (Spring 2007).
Hsu, C.-J., & Chang, Y.-C. (2005). The influence of airline ownership rules on aviation policies
and carriers’ strategies. Proceeding of the Eastern Asia Society for Transportation
Studies, 5, 557569.
International Air Transport Association. (2013). Worldwide airport slots. Retrieved from
https://www.iata.org/policy/slots/pages/index.aspx
Leigh, L. E. (1990). Contestability in deregulated airline markets: Some empirical tests.
Transportation Journal, 30, 4957.
Moore, T. G. (1986). U.S. airline deregulation: Its effects on passengers, capital, and labor.
Journal of Law and Economics, 29, 128.
Morrison, S., & Winston, C. (1987). Empirical implications and tests of the contestability
Peteraf, M. A., & Reed, R. (1994). Pricing and performance in monopoly airline markets.
Journal of Law and Economics, 37, 193213.
Pitfield, D. E. (2009). The assessment of the EUU.S. Open Skies agreement: The counterfac-
tual and other difficulties. Journal of Air Transport Management, 15, 308314.
Prokop, D. (2014). Government regulation of international air transportation. In J. Peoples
(Ed.), The economics of international airline transport (Vol. 4, pp. 4559). Advances in
Airline Economics. Bingley, UK: Emerald Group Publishing Limited.
Robyn, D., Moselle, B., & Reitzes, J. (2005). Beyond open skies: The economic impact of a
USEU open aviation area. In S. Daniel & J. P. Quinlan (Eds.), Deep integration: How
transatlantic markets are leading globalization (pp. 50–73). Washington, DC: Center for
Transatlantic Relations and Centre for European.
Scherer, F. M., & Ross, D. (1990). Industrial market structure and economic performance (3rd
ed.). Boston, MA: Houghton Mifflin.
Tirole, J. (1988). The theory of industrial organization. Cambridge, MA: MIT Press.
US Department of Transportation (DOT), office of the Secretary. (2000). Transatlantic
Deregulation, The Alliance Network Effect, International Development. Second Report.
Retrieved from http://www. worldbank.org/data/countryclass/countryclass.html
Westra, C. (2007). U.SEU ‘Open Skies’ agreements: A dream of liberalization deferred.
Boston College International and Comparative Law Review, 32, 161–176.
APPENDIX
Table A1. Variable Definitions and Data Sources.
Variable Source
Fareij Ticket Priceij/Distanceij average for the route and carrier

Ticket Priceij DB1B Airline Origin and Destination Survey
Distanceij DB1B Airline Origin and Destination Survey
Loadij Passengersij/Seatsij weighted average for the route — weight = flight
frequency
Passengersij T100 International Segment Data
Seatsij T100 International Segment Data
Equipij Seatsij weighted average for the route — weight = flight frequency
Freqij Flight Frequency for route T100 International Segment Data
Herfij Calculated based on the number of passengers carried on the route by each
carrier T100 International Market Data
OHerfi Calculated based on the number of passengers originating or terminating
at the airport by each carrier T100 Segment Data
DHerfj Calculated based on the number of passengers originating or terminating
at the airport by each carrier T100 Segment Data
Slotij Dummy = 1 if origin = slot controlled or dest = slot controlled IATA
OCarri T100 Segment Data
DCarrj T100 Segment Data
Carrij T100 International Market Data
Variables in italics are used in regressions.

CHAPTER 7
THE CHOICE OF AIRPORT,

AIRLINE, AND DEPARTURE
DATE AND TIME: ESTIMATING
THE DEMAND FOR FLIGHTS
Diego Escobari and Cristhian Mellado
ABSTRACT
This chapter estimates the demand for flights in an international air

travel market using a unique dataset with detailed information not only
on flight choices but also on contemporaneous prices and characteris-
tics of all the alternative non-booked flights. The estimation strategy
employs a simple discrete choice random utility model that we use to
analyze how choices and its response to prices depend on the departing
airport, the identity of the carrier, and the departure date and time.
The results show that a 10% increase in prices in a 100-seat aircraft
throughout a 100-period selling season decreases quantity demanded by
7.7 seats. We also find that the quantity demanded is more responsive
to prices for Delta and American, during morning and evening flights

ISSN: 2212-1609/doi:10.1108/S2212-160920140000004006
177
178 DIEGO ESCOBARI AND CRISTHIAN MELLADO
and that the response to prices changes significantly over different

departure dates.
Keywords: Airline demand; discrete choice; departure time; posted
prices
JEL classifications: R41; C25; L93
INTRODUCTION
The main goal of empirical studies of differentiated product markets is the

estimation of the demand and the modeling of choices. The common prac-
tice is to estimate random utility models of consumer demand using dis-
crete choice models. A typical problem in this type of estimation is the
difficulty in obtaining the data on choices and on all the available options.
This chapter proposes using a unique dataset on prices and transactions to
estimate discrete choice models that explain individual choices at the flight
level. In our data, we observe not only the flight choices and the corre-
sponding prices but also the prices and characteristics of all the non-
booked alternative flights. We are able to estimate how air travel demand
changes depending on the departure date, the departure time, the identity
of the carrier and the departing airport. The passenger choice model pre-
sented here can be vital to the development and the assessment of new pri-
cing strategies, capacity choices, or route entry/exit decision. Moreover, it
opens the door to a large number of other discrete choice models that can
be estimated using similar datasets. These include, for example, models to
forecast demand, estimation of market power, cross-price elasticities, single
agent dynamic models, or dynamic games.
To the best of our knowledge this is the first study that estimates the
itinerary choice (i.e., flight choice) in a revealed preference setting where
information on choices and all the alternative flights is available. The infor-
mation on all options is important because this is part of the information
set available to customers when they make their choices. This allows an
easy construction of the representative utility in a random utility model.
When not all the available alternatives are observed the estimation is com-
plicated due to the lack of information on the arrival rates. For example,
some unobserved customers can arrive and purchase from a seller that is
not in the data.1
The Choice of Airport, Airline, and Departure Date and Time 179
The airline industry has already demonstrated to be a popular place for

the estimation of discrete choice models using aggregate data from the U.S.
Department of Transportation. However, these data is too aggregate to
estimate the individual choices. Estimating discrete choice models of
demand employs aggregate market level data, consumer-level data or both.
Aggregate level data usually has aggregate quantity, prices, consumer char-
acteristics, market size, and sometimes the distribution of demographics. It
has the advantage that in most cases it is easier to get. Consumer-level data
has individual choices, prices, and characteristics of all options. The distri-
bution of demographics is optional. It has the advantage that it is more
detailed than aggregate data, but it is more difficult to obtain. We use
posted prices and inventory changes following a similar collection strategy
as Escobari (2012) and Escobari and Lee (2012).
In this chapter, we focus on the international market between the three
big airports serving New York City (Newark Liberty, John F. Kennedy,
and La Guardia) and the main airport in Toronto (Toronto Pearson
International). The data covers all the 317 flights from the six carriers that
served this city pair between December 19 and 24, 2008. We cover the
advance sales during 40 days leading to the departure dates. Counting the
choices and the available alternatives we have more than half-a-million
observations in our dataset. The results from the estimation of our random
utility framework show that a 10% increase in prices throughout a 100-
period selling season decreases quantity demanded by 7.7 seats in a 100-
seat flight. We also find that demand is greater closer to departure. When
allowing the response of quantity demanded to prices to change with the
identity of the carrier we find that Delta Airlines has the most price respon-
sive demand followed by American Airlines and Air Canada. Additional
results show that less responsive demand is associated with departure dates
that have more congestion and higher prices and that the responsiveness to
prices varies significantly across departure dates. Finally, demand is more
responsive to prices for flights that depart from John F. Kennedy and dur-
ing morning flights.
There is important related literature in airlines that uses discrete choice
estimations aimed at modeling consumer choices. Pels, Nijkamp, and
Rietveld (2003) and Hess and Polak (2005) use data from an airline passen-
ger survey to estimate various logit models of airport and airline choice for
the San Francisco Bay area. Ashiabor, Baik, and Trani (2007) use the 1995
American Travel Survey to forecast travel demand, while Proussaloglou
and Koppelman (1999) and Wen and Lai (2010) also use survey data to
estimate the choice of carrier. It is important to note that survey data
basically creates trip scenarios to simulate the booking process. These sur-
veys record stated preferences and not revealed preferences based on actual
choices that result in transactions. Our data comes from the actual beha-
vior of the interaction between sellers and buyers recorded by sales and
posted prices. Carrier (2008) estimates itinerary choice using booking data,
but does not have the non-booked travel alternatives as we do. Using
revealed preferences from a single major European airline and stated pre-
ferences, Atasoy and Bierlaire (2012) present an itinerary choice model.
Related literature on airlines that used the most common Airline Origin
and Destination Survey (DB1B) from the U.S. Department of
Transportation includes Berry (1992) who estimates a structural discrete
choice model of entry for the airline industry. Berry and Jia (2010) present
a structural model and estimate the impact of demand and supply changes
on profitability during the turmoil in the industry in the early 2000s, while
Ciliberto and Tamer (2009) used a partially identified entry model to inves-
tigate the heterogeneity in carriers profits.2
While we propose using posted prices, inventory changes and discrete
choice models to explain demand side behavior, posted prices data has
been very popular for the estimation of pricing strategies and supply side
behavior. Stavins (2001) uses posted prices from the Official Airline Guide
to find that price dispersion attributed to ticket restrictions increases with
competition. More recently McAfee and te Velde (2007) looks at price
dynamics, Van Eggermond, Schuessler, and Auxhausen (2007) at travelers
itinerary in European markets, Mantin and Koo (2009) study dynamic
price dispersion, and Alderighi (2010) explains fare dispersion. Bilotkach,
Gaggero, and Piga (2011) show empirically how yield management is effec-
tive in raising a flight’s load factor, while Bilotkach and Rupp (2012) study
the intertemporal profile and the role of low-cost carriers and differences
across online distributors. The intertemporal profile of fares in also studied
in Bergantino and Capozza (2012), who find a J-curve and in Alderighi,
Nicolini, and Piga (2012) who find a U-shape and that fares increase with
occupancy rate. Escobari (2012) estimates a dynamic demand equation and
a dynamic supply equation that jointly explain the dynamics of fares and
sales as the departure date nears, Escobari and Lee (2012) estimate price
reaction functions to capture the interaction between flights, and Escobari,
Rupp, and Meskey (2013) shows empirically how airlines dynamically price
discriminate.3
The organization of the rest of the chapter is as follows. The second sec-
tion explains the collection of the data and presents the summary statistics.
The explanation of the discrete choice empirical model is presented in third
section. The fourth section presents the results, while the fifth section dis-
cusses possible extensions. The sixth section concludes the chapter.
DATA
The data for this chapter was collected from the online travel agency
Expedia.com following a similar strategy as in Escobari (2012) and
Escobari and Lee (2012). We not only have information on all contem-
poraneous posted prices at different points in time for all available
options (i.e., flights) but also on seat inventory levels. We use the
changes in inventories to identify a sale. Hence our data replicates the
information displayed to individuals who buy tickets online and records
their choices. To control for various sources of price dispersion and
demand variation across customers we focus on the lowest available one-
way non-stop economy-class nonrefundable posted fare. Looking only at
one-way non-stop tickets helps define a single inventory at each price
and helps control for tickets sold as part of round-trips or longer
itineraries.
As in Escobari (2012), even if one-way tickets are a small fraction of
the overall tickets sold, our observed prices are relevant as long as the
carriers adjust these prices based on the current inventory levels.
Moreover, sales (obtained as inventory changes) can be the result of tick-
ets sold at prices different than the one-way prices for example, as a
round-trip ticket where one of the legs is in our sample. Then our
demand estimation is also capturing the demand of round-trip tickets if
one-way tickets are always priced half of the round-trip tickets, which is
the standard assumption in the airline pricing empirical literature.4 In
addition, to make the problem tractable we focus on a single city pair,
New York City to Toronto which already generates over half-a-million
observations. Because there are three big airports that serve New York
City, we collected the data for all three airports, Newark Liberty
International Airport (EWR), John F. Kennedy International Airport
(JFK), and La Guardia Airport (LGA). The only airport in Toronto that
we consider is the Toronto Pearson International Airport (YYZ), which
is the only big airport that serves this city.
We have sales and pricing information on all the flights that departed
between December 19 and December 24, 2008. Moreover, we keep record
of the prices and inventory changes every 3 days between 40 days to
departure and 1 day to departure for all these flights. Overall this included
317 flights from American Airlines, Air Canada, Continental, Delta, Lan
Chile, and United, with 10,708 tickets sold during our period of study. The
details of the 317 flights by carrier and by departure date are presented in
Table 1, which is reproduced from Escobari and Lee (2012).5 Each time a
ticket was sold, we recorded the corresponding price and the prices of all
competing flights for the same day of departure. This makes our data repli-
cate the information available to the buyer as well as the same structure as
required for the estimation of discrete choice models. Each consumer at the
time of arriving to buy a ticket observes all posted prices and chooses the
flight that gives him the highest utility. On average, we have that for each
recorded ticket transactions there are 52 competing flights. For example, if
a traveler buys a ticket from United to fly on December 24, we also record
the contemporaneous posted prices for the other 49 competing flights as
detailed in Table 1.
Note that sales are recorded every time there is an inventory change
within the 3-day window in which inventories are tracked. When more
than one ticket is sold we record this as separate sales made to different
travelers. This makes sure we comply with the single-unit purchase
assumption of logit models. We believe this is a reasonable assumption in
airlines because it is consistent with theoretical models that explain airline
pricing where individuals are assumed to have unit demands (e.g., Dana,
1998, 1999; Deneckere & Peck, 2012). This means that we do not con-
sider sales of a bundle of tickets made, for example, by a family flying
together. Unfortunately when the change in inventories is greater than
one our data does not allow us to identify whether sales are made by
Table 1. Flights by Carrier and Date.

Friday Saturday Sunday Monday Tuesday Wednesday Total
December December December December December December
19 20 21 22 23 24
American 12 8 11 12 13 11 67
Air Canada 19 9 13 20 20 16 97
Continental 8 5 7 8 8 6 42
Delta 4 4 4 4 4 3 23
Lan Chile 1 1 0 1 0 1 4
United 17 10 10 18 16 13 84
Total 61 37 45 63 61 50 317
separate individuals or from a single individual buying more than one

ticket.6 Moreover, during this 3-day window inventories might have chan-
ged in more than one flight. Because in discrete choice models of demand
consumers choose the single alternative that gives them the highest utility,
if more than one inventory changes then these are also considered as
separate sales made to different buyers.
In addition to sales and prices that allows estimating airline demand, the
data is interesting because allows demand comparisons between departing
airports, departure dates, carriers, and departure times. Table 2 reports the
summary statistics for these variables. The first four columns report the
typical statistics, while columns 5 through 8 report the mean, standard
deviation, minimum and maximum prices for each of the classifications dic-
tated by the dummy categories. The figures show that the dominant carrier
in this city pair is Air Canada with 30.9% of the flights, followed by
United with 25.8% and by American Airlines with 20.8%. Moreover, there
is substantial price dispersion in the sample. The lowest priced ticket at
US$65 is more than 16 times cheaper than the most expensive ticket. The
busiest airport is La Guardia, while most flights depart in the morning. It is
interesting to note that while the overall average price is US$169.80 (col-
umn 1), the average price at which a transaction occurred is US$156.93
(column 5).
There are important advantages in the structure of the collection of
the data and in focusing on this particular city pair. First, direct flights
between New York City and Toronto take only one hour and a half,
hence it is reasonable to think that combination of flights connecting this
city pair with one or more stops are not a desired alternative for trave-
lers. Moreover, focusing on non-stop flights and one-way fares is useful
to control for fare differences associated to round-trip tickets and open
jaws. For example, these tickets are usually associated with Saturday-
night-stay restrictions or minimum-, and maximum-stay restrictions. This
would involve tickets of a significantly different quality. Selecting the
least expensive available nonrefundable ticket is important to control for
the existence of more expensive refundable tickets that are also available
for purchase at different points prior to departure. We do not record
sales in flights where only refundable tickets are available. Finally, focus-
ing on economy-class tickets controls for some consumer’s heterogeneity
as some higher valuation consumers may want to buy first-class tickets.
We consider refundable and first-class tickets to be of a significantly dif-
ferent quality.
Table 2. Summary Statistics.

Variables Main Variable Price
(1) (2) (3) (4) (5) (6) (7) (8)

Mean SD Min Max Mean SD Min Max
Price 169.8 133.3 65 1,075

Advance 19.00 12.43 1 40
Bought 0.0191 0.137 0 1 156.93 111.87 65 1075
Carriers:
American 0.208 0.406 0 1 192.38 194.72 76 1075
Air Canada 0.309 0.462 0 1 144.57 69.15 81 736
Continental 0.138 0.345 0 1 155.21 114.51 77 1001
Delta 0.0732 0.261 0 1 234.36 203.77 105 953
Lan Chile 0.0129 0.113 0 1 130.45 25.84 123 220
United 0.258 0.438 0 1 173.42 105.25 65 1008
Departure Dates:
December 19 0.247 0.431 0 1 142.21 95.33 76 944
December 20 0.061 0.239 0 1 279.85 241.12 65 1075
December 21 0.108 0.311 0 1 180.72 98.35 81 738
December 22 0.231 0.422 0 1 128.38 38.72 76 472
December 23 0.214 0.410 0 1 193.73 159.52 81 1075
December 24 0.138 0.345 0 1 194.51 160.96 81 1075
Airports:
Newark Liberty 0.356 0.479 0 1 150.73 101.27 65 1001
La Guardia 0.515 0.500 0 1 171.87 133.96 76 1075
John F. Kennedy 0.129 0.336 0 1 214.46 186.27 87 1075
Departure Times:
Morning 0.447 0.497 0 1 142.06 95.94 65 1075
Afternoon 0.378 0.485 0 1 192.68 162.87 77 1075
Evening 0.175 0.380 0 1 191.55 130.96 81 953
Note: The sample size is 560,244.
EMPIRICAL MODEL
Consider the following random utility model framework in which trave-

lers are assumed to be utility maximizers. We adapt the model of chapter
2 in Train (2002) for our setting.7 Let traveler n face a choice of traveling
in any of J different flights. Note that these J flights are all the available
flight options shown by the online travel agency. The utility of individual
n obtained from alternative j ∈ J is Unj. This level of utility is known by
the traveler but not by the econometrician. We assume that the traveler
already decided to fly over other transportation alternatives (e.g., driving
or renting a car). Hence, he will choose the flight that gives him the high-
est level of utility. That is, he will choose flight i if and only if Uni > Unj
for all j ≠ i. Even though the econometrician cannot observe the utility
levels, some of the flights’ attributes including the price can be observed.
We label them as xnj and pn, respectively. We will related these observed
factors to the traveler’s utility with the function Vnj = Vðxnj ; pnj Þ, which is
called the representative utility. Some attributes in xnj can include, for
example, departure time, departure date, departing airport, and identity
of the carrier.
From the econometrician’s view point, utility levels Unj contain some
random unobserved component that make Unj ≠ Vnj . We then write
Unj = Vnj þ ɛnj ; with ɛnj being the stochastic component of utility. We write
f ðɛn Þ as the joint density of the random vector ɛn = fɛn1 ; ɛn2 ; ⋯; ɛnJ g. Then
the probability that traveler n chooses alternative i is given by,
Pni = ProbðUni > Unj ∀j ≠ iÞ

= ProbðVni þ ɛ ni > Vnj þ ɛnj ∀j ≠ iÞ ð1Þ
= Probðɛnj < ɛni þ Vni − Vnj ∀j ≠ iÞ
Assume that ɛnj is distributed iid extreme value. Hence, the distribution
of each unobserved component of utility is
− ɛnj
f ðɛ nj Þ = e − ɛnj e − e ð2Þ
− ɛnj
with the cumulative distribution being Fðɛnj Þ = e − e . Because the differ-
ence between two extreme value variables is logistic we have
eɛnj − ɛni
Fðɛ nj − ɛni Þ = ð3Þ
1 þ eɛnj − ɛni
From Eq. (1) if ɛ ni is taken as given, the cumulative probability distribution

− ðɛni þ Vni − Vnj Þ
for each ɛnj evaluated at ɛni þ Vni − Vnj based on Eq. (3) is e − e .
Following the assumption of independence, this cumulative distribution
over all j≠i is the product of the individual cumulative distributions:
− ðɛni þ Vni − Vnj Þ

Pni jɛni = ∏ e − e ð4Þ
j≠i
Because ɛni is not given, we need to take the integral of Pni jɛ ni over all
values of ɛ ni weighted by the density of ɛni :
Z !
− e − ðɛni þ Vni − Vnj Þ − ɛni
Pni = ∏e e − ɛni e − e dɛni ð5Þ
j≠i
that has the integral equal to
eVni
Pni = P Vnj ð6Þ
je
which is the equation for the logit choice probability. Because we have a
panel, the utility that the traveler n obtains from buying a ticket on flight j
at time t is given by,
Unjt = βanjt þ αj pnjt þ μj þ ɛnjt ð7Þ
In Eq. (7) we model the systematic component of utility as a linear

function of the parameters, Vnjt = βαnjt þ αj pnjt þ µj . β and αj are coeffi-
cients to be estimated and the variables anjt and pnjt are the number of
days in advance the ticket was bought and the price. Our main variable
of interest is αj which captures the marginal utility of a price increase.
Of course we expect αj to be negative, hence −αj is the disutility of a
price increase. The key element in which flights are differentiated are all
time invariant and captured by µj, the time-invariant fixed effect specific
to flight j. This one controls for observed and unobserved flight specific
characteristics such as departure time, departure date, identity of the car-
rier, distance between the cities, or the carrier’s managerial capacity.
As a first approach we fix αj = α ∀ j but then we allow αj to change with
time-invariant flight specific characteristics. Then the systematic component
of utility will be modeled as Vnjt = βanjt þ ðδ0 xnj Þ pnjt þ µj where we just allow
αj in Eq. (7) to be αj = δ0 xnj . The vector of variables xnj includes departure
time, departure date, the identity of the carrier and the departing airport.
This is alternative specification in helpful to determine how demand for a
flights changes with these characteristics. There are some assumptions for
the validity of the estimation of Eq. (7) that discuss in detail along with the
results in the next section.
RESULTS
In this section, we present the discrete choice logit estimates of the demand.
For comparison purposes we also report the linear pooled OLS as well as
the linear fixed effects estimates. Table 3 presents the base model of Eq. (7)
where anjt is relabeled as Advance, the number of days prior to departure
fares and the sale was recorded. For pnjt we use Log(Price), the natural
logarithm of Price. For simplicity we omit the subscripts njt from the
names of the variables. The negative and statistically significant coefficients
on Advance in columns 1 and 3 suggest that sales (demand) increases closer
to departure. However, this result is not robust across specifications as the
fixed effects estimate finds a non-significant coefficient. Note that the
reported logit coefficients are the marginal effects obtained when evaluating
the other regressors at their mean levels.
It is important to note that Log(Price) in the estimation is potentially
endogenous. Endogeneity arises if there is correlation between pnjt and the
unobserved µj þ ɛ njt . The most common cause of this correlation is if the
carrier sets prices knowing more about the error term than the econometri-
cian. Escobari (2012) controls for potential endogeneity in a dynamic set-
ting using internal instruments. Here we control for the potential
endogeneity that arises due to correlation between pnjt and µj using flight
fixed effects. For ɛnjt we assume that it is uncorrelated with pnjt . This
assumption is reasonable given that in Escobari (2012) the point estimates
in the within specification that consider fare as exogenous is virtually the
same as the point estimate that treats prices as endogenous.8 An alternative
approach would be to follow the methods developed in Berry, Levinsohn,
Table 3. Demand Estimates, Base Model.

Variables (1) (2) (3) (4)
Pooled Within Logit Logit
Log(Price) −0.00642*** −0.0113*** −0.00704*** −0.00771***

(0.000416) (0.000465) (0.000459) (0.000501)
Advance −0.000154*** 0.000832 −0.000151*** −0.000166***
(1.53e-05) (0.00106) (1.50e-05) (1.44e-05)
Flight FE No Yes No Yes
Observations 560,244 560,244 560,244 560,244
Log-likelihood −52,836 −52,188
Note: Numbers in parentheses are standard errors.

*** Significant at 1%.
and Pakes (1995). However, this approach still needs exogenous instru-
ments which in most cases use supply side variables.
The negative point estimate for the coefficient on Log(Price) is statis-
tically significant at a 1% and it is consistent with a downward sloping
demand. The coefficient in column 3 indicates that when the price of a
ticket increases by 10%, quantity demanded decreases by 0.070 seats in a
100-seat aircraft.9 This is a reasonable value for the estimate because this
is what quantity demanded decreases each period prior to departure. If
prices are 10% higher throughout the selling season and there are 100
periods, then this particular flight will sell 7 seats less. The point estimate
in column 4, which controls for flight fixed effects, suggests a slightly lar-
ger effect. This coefficient also captures the disutility of a price increase.
Note that while we have a number of variables reported in Table 2
besides Advance that can be part of Eq. (7) (e.g., carrier identifier, airport
identifier, departure time, and departure date), we cannot separately iden-
tify the marginal effects of those variables because they are perfectly colli-
near with the flight fixed effects µj. However we can estimate interaction
terms between those variables and Log(Price) to see how the slope of
the demand changes.
Table 4 estimates a model in which the slope α is allowed to change
with the identity of the carrier. Focusing on the last column of the
table we can observe that the carrier with the most responsive demand is
Delta, followed by American Airlines and Air Canada. The least respon-
sive demand is the one for United. Note that these differences have a
relatively big economic importance. For example, all else constant, a 10%
price increase throughout a 100-period selling season in a 100-seat aircraft
decreases the quantity demanded by 18 seats in Delta flight (the largest
response) and by slightly less than one seat in a United flight (the smal-
lest response).
More inelastic demands are usually associated with higher market power
and the ability to charge higher prices. Interestingly, Delta that is the car-
rier with the less responsive demand is also the one with the lowest presence
in this city pair. As column 1 in Table 2 shows, only 7.32% of the flights in
this city pair belong to Delta. However, Delta also charges the highest aver-
age prices ($234.36) as reported in column 5 of Table 2. Air Canada, who
is the dominant carrier in this city pair with 30.9% of the flights is not the
carrier that charges the highest average fares and does not have the most
responsive (or nonresponsive) demand. It is difficult to infer about any
causality from these results because from the view point of the econometri-
cian, fares and market presence are jointly determined.
Table 4. Demand Estimates, the Role of the Carrier Identity.

Variables (1) (2) (3) (4)
Log(Price)
American −0.00630*** −0.0116*** −0.00724*** −0.0151***
(0.000429) (0.000481) (0.000480) (0.00111)
Air Canada −0.00573*** −0.0110*** −0.00662*** −0.00866***
(0.000440) (0.000491) (0.000485) (0.00115)
Continental −0.00590*** −0.0113*** −0.00681*** −0.00301***
(0.000446) (0.000498) (0.000495) (0.00112)
Delta −0.00699*** −0.0122*** −0.00828*** −0.0180***
(0.000426) (0.000476) (0.000484) (0.00189)
Lan Chile −0.00489*** −0.0104*** −0.00593*** −0.00558***
(0.000553) (0.000598) (0.000565) (5.81e-05)
United −0.00522*** −0.0103*** −0.00612*** −0.000838
(0.000427) (0.000477) (0.000472) (0.000832)
Advance −0.000153*** 0.000845 −0.000150*** −0.000158***
(1.53e-05) (0.00106) (1.49e-05) (1.42e-05)
Observations 560,244 560,244 560,244 560,244

Table 5 shows how the effect of price on quantity demanded changes

with the departure date. Demand is less responsive the Friday (December
19) before Christmas and it is more responsive 2 days before Christmas.
With the exception on December 19, there is not a big economically signifi-
cant difference in the response across the rest of the departure dates all
coefficients lie between 0.011 and 0.015. On December 21, a 10% increase
in prices throughout a 100-period selling season in a 100-seat aircraft
decreases quantity demanded by 11.1 seats while on December 23 the effect
is 15.4 seats. Compared with the other departure dates on Tuesday
December 23 the demand is relatively more responsive. This is interesting
because that Tuesday is the day where there is a large number of scheduled
flights, 61 (see Table 1), which corresponds to 21.4% of the flights in the
sample (see column 1 in Table 2). An interesting point in Table 2 is the link
between higher fares and congestion. For example, the day in which the
least number of flights were scheduled is Saturday December 20 with only
37 flights. This is also the departure date with the highest average fares
($279.85, see column 5, Table 2). Higher fares associated with more
Table 5. Demand Estimates, the Role of the Departure Date.

Variables (1) (2) (3) (4)
Log(Price)
December 19 −0.000190*** 0.000740 −0.000180*** −0.000214***
(1.55e-05) (0.00106) (1.47e-05) (1.39e-05)
December 20 −0.0113*** −0.00324*** −0.0117*** −0.0149***
(0.000472) (0.00108) (0.000492) (0.000542)
December 21 −0.00761*** −0.0163*** −0.00862*** −0.0111***
(0.000439) (0.00123) (0.000459) (0.000504)
December 22 −0.00964*** −0.0117*** −0.0102*** −0.0128***
(0.000458) (0.00135) (0.000478) (0.000525)
December 23 −0.0117*** −0.0140*** −0.0121*** −0.0154***
(0.000476) (0.00166) (0.000494) (0.000546)
December 24 −0.0108*** −0.00915*** −0.0113*** −0.0143***
(0.000447) (0.000832) (0.000472) (0.000522)
Advance −0.000190*** 0.000740 −0.000180*** −0.000214***
(1.55e-05) (0.00106) (1.47e-05) (1.39e-05)
Observations 560,244 560,244 560,244 560,244

congestion known ex-ante are evidence of systematic peak-load pricing, as

previously documented in Escobari (2009). It would be reasonable to
observe that a more responsive demand is associated with more congestion,
but the estimates in Table 5 show that there is little evidence that this is
the case.
The regression estimates in Table 6 are presented to address the role of
the departing airport on the relationship between prices and quantity
demanded. Column 4 shows that the point estimates for Newark of for
La Guardia are nearly the same, while the response in flights that depart
from John F. Kennedy is much smaller. The null that the coefficients
between Newark and La Guardia are the same is rejected at the any rea-
sonable significance levels. Note that the last three rows in the
table report the p-values of all the null hypotheses that test for pair-wise
differences in the coefficients across airports. The last two rows show that
there response in the John F. Kennedy airport is significantly different
that in the other two airports. A 10% increase in prices throughout a
100-period selling season in a 100-seat aircraft decreases quantity
demanded by 17.8 seats in a flight departing from the John F. Kennedy
Table 6. Demand Estimates, the Role of the Departing Airport.

Variables (1) (2) (3) (4)
Log(Price)
Newark Liberty −0.00593*** −0.0112*** −0.00661*** −0.00665***
(0.000436) (0.000487) (0.000477) (0.000633)
La Guardia −0.00573*** −0.0107*** −0.00639*** −0.00579***
(0.000424) (0.000474) (0.000465) (0.000563)
John F. Kennedy −0.00679*** −0.0119*** −0.00766*** −0.0178***
(0.000420) (0.000471) (0.000466) (0.00148)
Advance −0.000152*** 0.000836 −0.000150*** −0.000160***
(1.53e-05) (0.00106) (1.49e-05) (1.43e-05)
Observations 560,244 560,244 560,244 560,244
H0: βEWR = βLGA (p-value) 0.0120 1.05e-07 0.00592 0.107
H0: βLGA = βJFK (p-value) 0 0 0 0
H0: βEWR = βJFK (p-value) 0 6.14e-11 0 0
Notes: Numbers in parentheses are standard errors. For k, m = EWR, LGA, JFK, and k ≠ m
the last three rows present the p-values of the null H0: βk=βm.
airport, while for the La Guardia and for Newark this figure is 57.9 seats
and 66.5 seats, respectively.
The final set of estimates is presented in Table 7. Here the goal is to
assess the role of the departure time on the effect that prices have on
quantity demanded. We divide flight departure times in three, morning if
the flight departs before noon, afternoon if the flight departs between
noon and 5:00 p.m., and evening if the flight departs after 5:00 p.m. The
logit estimates in the last column that control for flight specific character-
istics show that quantity demanded is less responsive for flights departing
in the afternoon and about equally responsive for flights departing either
in the morning or in the evening. The last three rows reports the p-values
for various null hypotheses that the coefficients are pair-wise equal.
While focusing on nonrefundable one-way economy-class tickets helps
control for a number of dimensions that affect price dispersion in airlines,
these are usually a small fraction of the overall number of tickets sold by
airlines. Our measure of a ticket sale responds to sales that can happen via
different channels, for example, someone buying a ticket with this leg as
part of a round-trip ticket. In this case, the estimated marginal effect will
be channeled through the prices of round-trip tickets. If the round-trip
Table 7. Demand Estimates, the Role of the Departure Time.

Variables (1) (2) (3) (4)
Log(Price)
Morning −0.00777*** −0.0140*** −0.00856*** −0.00943***
(0.000450) (0.000510) (0.000495) (0.000679)
Afternoon −0.00736*** −0.0132*** −0.00814*** −0.00656***
(0.000429) (0.000485) (0.000476) (0.000533)
Evening −0.00682*** −0.0126*** −0.00759*** −0.00864***
(0.000431) (0.000486) (0.000474) (0.000698)
Advance −0.000168*** 0.000714 −0.000165*** −0.000169***
(1.54e-05) (0.00106) (1.50e-05) (1.44e-05)
Observations 560,244 560,244 560,244 560,244
H0: βMor = βAft (p-value) 7.02e-07 0 5.11e-07 6.33e-07
H0: βAft = βEve (p-value) 1.65e-07 1.67e-10 4.00e-08 0.000688
H0: βMor = βEve (p-value) 0 0 0 0.334
Notes: Numbers in parentheses are standard errors. For k, m = Morning (Mor), Afternoon
(Aft), Evening (Eve), and k ≠ m the last three rows present the p-values of the null H0: βk=βm.
tickets prices always change proportionally with the one-way ticket prices
then our demand estimates are also capturing the round-trip ticket
demand.10 Another way sales could be channeled is, for example, via pas-
sengers booking using frequent flyer miles through the carrier’s website. In
this case, it can be argued that a buyer may not be aware (or may not care)
of the prices of all the available options. In this chapter, we do not model
any potential lack of information on the part of the buyer. Because all
prices we use are readily available online we assume buyers have perfect
information about all the available options. This should be intuitively cor-
rect even for tickets bought with frequent flier miles because accumulating
miles is costly and paying for a ticket is a substitute for buying a ticket
using frequent flier miles.11
The consumer choice behavior analyzed in this chapter is part of a very
interesting problem of consumer and producer interaction during advance
sales. This problem has generated an important theoretical and empirical
literature that has mostly focused on supply side behavior (i.e., to explain
pricing). Airlines use yield management to set prices for perishable seats
and maximize profits. Theoretical pricing models (e.g., Dana, 1998, 1999;
Talluri & van Ryzin, 2004) that consider, for example, fixed capacity,
aggregate and individual demand uncertainty, and have predictions for the
price schedule and its relationship with available capacity and time to
departure are consistent with our results. The reason is that these pric-
ing models are aimed at explaining the supply side behavior. In this
chapter, we take the supply side behavior as given and focus on modeling
the demand side. In the next section, we discuss a few structural estima-
tion approaches that are capable of modeling jointly both sides of the
market.
The logit estimator used in this chapter has some limitations. As
explained in Train (2002), this logit estimator can represent systematic taste
variation but not random taste variation. Logit can approximate average
tastes fairly well even when tastes are random, but a Probit or a mixed logit
model may be better at including random taste variation. A second poten-
tial limitation is that the logit model has the independence of irrelevant
alternatives (IIA) property, which means that the ratio of the probabilities
of two alternatives does not depend on any other alternative. Finally, the
logit cannot handle unobserved factors that are correlated over time. The
main goal in this chapter is to illustrate the use of our unique data to esti-
mate flight choice models, but other logit estimators can also be used as
extensions to this research that overcome some of the limitations outlined
above. In the next section, we provide some examples for further research
using similar datasets.
POTENTIAL EXTENSIONS
In this chapter, we estimated a simple random utility model, but our data
and the discrete choice modeling approach used in this chapter can be
extended to a number of settings. The most obvious is to use these model
estimates to forecast flight-level demand. This is or particular importance
for carriers because they use forecasted demand to schedule flights.
Additional potential extensions include measuring market power at the
route level and merger evaluation using the methods proposed in Nevo
(2001) or welfare from new flights following techniques for the introduc-
tion of the minivan in Petrin (2002). Discrete choice models can also be
used to estimate pricing strategies in monopoly routes using single agent
dynamics as in Rust (1987), Hotz and Miller (1993), or Aguirregabiria
and Mira (2002). In these models agents are forward looking and maxi-
mize intertemporal payoffs. When more than one carrier serves a route
the approach can as well follow Bajari, Benkard, and Levin (2006) or
Aguirregabiria and Mira (2007) to estimate dynamic discrete games to
model and estimate the interaction between flights. The well-known chal-
lenges in these dynamic estimations are the large number of agents,
choices and states, and the existence of multiple equilibria, which means
an important computational burden. Igami (2013) overcomes these chal-
lenges by modeling a small number of state spaces and choice sets to esti-
mates a dynamic model via maximum likelihood using the nested fixed-
point algorithm of Rust (1987). Some of these topics have already been
addresses using posted prices but without discrete choice models.
Escobari (2012) estimates dynamic demand and dynamic pricing equa-
tions in a setting where agents are forward looking. Moreover, Escobari
and Lee (2012) estimate price reaction functions to model the interaction
between agents.
CONCLUSION
The choice of transportation mode is the most widely used example to illus-
trate random utility models. This follows from the seminal work of Daniel
McFadden (1974) that estimates the trip mode choice in a study of travel
demand. On this line modeling the choice of flights has also been a popular
research topic; however, obtaining the appropriated data proved to be a
difficult task. Previous studies either worked with stated preferences based
on survey data or with revealed preference based on data from a single
seller. In this chapter, we work with revealed preferences based on data on
all flights from the international air travel market between New York City
and Toronto. Our study combines two key pieces of information. First, we
have posted prices at each time period and for all the available booked and
non-booked flights. This records the menu of options available to the
buyers. Second, we observe changes in inventory levels which allow us to
identify choices. The panel structure of the data with multiple transactions
per flight at various points prior to departure allows us to control for unob-
served flight specific characteristics.
Our demand estimates show that in a 100-seat flight a 10% increase in
prices throughout a 100-period selling season reduces quantity demanded
by 7.7 seats. We further inquire how this figure changes based on key
sources of product differentiation in airline markets the departing air-
port, the identity of the carrier and the departure date and time. We find
that quantity demanded is more responsive to prices for Delta flights, fol-
lowed by flights from American and Air Canada. Moreover, there are sig-
nificant differences in this responsiveness across departure dates and
departing airport. Demand is less responsive for departures at John F.
Kennedy than for departures at La Guardia or Newark. Finally, quantity
demanded is more responsive to prices for departures in the morning and
in the evening when compared to departures in the afternoon. We discuss
the validity of our assumptions and highlight potential areas for future
research.
NOTES
1. When there is missing data one potential solution is to use the techniques
described in Newman, Garrow, Ferguson, Jacobs, and Purnomo (2012).
2. See also Armantier and Richard (2008), and more recently Gedge, Roberts,
and Sweeting (2013).
3. On the theoretical side, Deneckere and Peck (2012) present a theory to
explain price posting in a multiple period version of the models presented in Dana
(1998) and Dana (1999).
4. See for example Borenstein and Rose (1994, p. 677), and Gerardi and
Shapiro (2009, p. 5).
5. While the data collection process in Escobari and Lee (2012) is the same as in
this chapter, here the structure and approach is very different. Escobari and Lee
(2012) have 4,398 observations on posted prices. Here we have 560,244 observations
that keep track of the posted prices of all the available flights for every time we
observe a sale.
6. If bundle sales were observed, the estimation strategy would need to follow
Hendel (1999) to estimate multiple-discrete choice models.
7. Talluri and van Ryzin (2004) study a similar buyers’ choice behavior from a
revenue management perspective. For a detailed discussion of discrete choice mod-
els of airline demand see Garrow (2010).
8. See Table IV, columns (2) and (8) in Escobari (2012).
9. This calculation follows that (−0.007/100)(%ΔPrice), which is the marginal
effect given that Log(Price) on the right-hand side. A 10% increase in Price
(%ΔPrice = 10) decreases the dependent variable by 0.0007 that in a 100-seat air-
craft is 0.07 seats.
10. One-way ticket prices changing proportionally with round-trip ticket prices is
consistent with the standard assumption in empirical papers of airline pricing where
the round-trip prices are calculated as two times the one-way prices. See for example
Borenstein and Rose (1994, p. 677) or Gerardi and Shapiro (2009, p. 5).
11. We believe our approach is reasonable, especially in light of the standard
assumption in empirical papers of airline pricing that just drop observations that
are believed to be made using frequent flier miles (see, e.g., Borenstein & Rose,
1994; or Gerardi & Shapiro, 2009).
ACKNOWLEDGMENT
We thank Stephanie Reynolds and Sang-Yeob Lee for their work with the
data. We also thank the comments from an anonymous reviewer that
helped improve the chapter.
REFERENCES
Aguirregabiria, V., & Mira, P. (2002). Swapping the nested fixed point algorithm: A class of
estimators for discrete Markov decision models. Econometrica, 70, 15191543.
Aguirregabiria, V., & Mira, P. (2007). Sequential estimation of dynamic discrete games.
Econometrica, 75, 153.
Alderighi, M. (2010). Fare dispersions in airline markets: A quantitative assessment of theore-
tical explanations. Journal of Air Transport Management, 16, 144150.
Alderighi, M., Nicolini, M., & Piga, C. A. (2012). Combined effects of load factors and booking
time on fares: Insights from the yield management of a low-cost airline. Milan, Italy:
Nota di Lavoro, Fondazione Eni Enrico Mattei.
Armantier, O., & Richard, O. (2008). Domestic airline alliances and consumer welfare. RAND
Journal of Economics, 39, 875904.
Ashiabor, S., Baik, H., & Trani, A. (2007). Logit models for forecasting nationwide intercity
travel demand in the United States. Transportation Research Record, 112.
Atasoy, B., & Bierlaire, M. (2012). An air itinerary choice model based on a mixed RP/SP data-
set. Technical Report TRANSP-OR 120426, Transport and Mobility Laboratory,
Ecole Polytechnique Fédérale de Lausanne, 2012.
Bajari, P., Benkard, L., & Levin, J. (2006). Estimating dynamic models of imperfect competi-
tion. Econometrica, 75, 13311370.
Bergantino, A. S., & Capozza, C. (2012). Airline pricing behaviour under limited intermodal
competition. Working Paper SIET.
Berry, S. (1992). Estimation of a model of entry in the airline industry. Econometrica, 60,
889917.
Berry, S., & Jia, P. (2010). Tracing the woes: An empirical analysis of the airline industry.
American Economic Journal: Microeconomics, 2, 143.
Berry, S., Levinsohn, J., & Pakes, A. (1995). Automobile prices in market equilibrium.
Bilotkach, V., Gaggero, A., & Piga, C. A. (2011). Airline pricing under different market
conditions: Evidence from European low-cost carriers. Rimini Centre for Economic
Analysis, Working Paper 11-47. Available at http://dx.doi.org/10.1108/S2212-1609
(2011)0000003006
Bilotkach, V., & Rupp, N. G. (2012). A guide to booking airline tickets online. In J. Peoples
(Ed.), Pricing behavior and non-price characteristics in the airline industry (Vol. 3,
pp. 83105). Advances in Airline Economics. Bingley, UK: Emerald Group Publishing
Limited.
Borenstein, S., & Rose, N. L. (1994). Competition and price dispersion in the U.S. airline
industry. Journal of Political Economy, 102, 653–683.
Carrier, E. (2008). Modeling the choice of an airline itinerary and fare product using booking and
seat availability data, PhD thesis, Massachusetts Institute of Technology.
Ciliberto, F., & Tamer, E. (2009). Market structure and multiple equilibria in airline markets.
Dana, J. D., Jr. (1998). Advance-purchase discounts and price discrimination in competitive
markets. Journal of Political Economy, 106, 395422.
Dana, J. D., Jr. (1999). Equilibrium price dispersion under demand uncertainty: The roles of
costly capacity and market structure. Rand Journal of Economics, 30, 632660.
Deneckere, R., & Peck, J. (2012). Dynamic competition with random demand and costless
search: A theory of price posting. Econometrica, 80, 11851247.
Escobari, D. (2009). Systematic peak-load pricing, congestion premia and demand diverting:
Empirical evidence. Economics Letters, 103, 5961.
Escobari, D. (2012). Dynamic pricing, advance sales, and aggregate demand learning in air-
lines. Journal of Industrial Economics, 60, 697724.
Escobari, D., & Lee, S.-Y. (2012). Demand shifting across flights and airports in a spatial com-
petition model. Letters in Spatial and Resource Sciences, 5, 175183.
Escobari, D., Rupp, N., & Meskey, J. (2013). Dynamic price discrimination in airlines. Mimeo.
Greenville, NC: East Carolina University.
Garrow, L. A. (2010). Discrete choice modeling and air travel demand: Theory and applications.
Burlington, VT: Ashgate Publishing Limited.
Gedge, G., Roberts, J. W., & Sweeting, A. (2013). A model of dynamic limit pricing with an
application to the airline industry. Durham, NC: Mimeo, Duke University.
Gerardi, K. S., & Shapiro, A. H. (2009). Does competition reduce price dispersion? New evi-
dence from the airline industry. Journal of Political Economy, 117, 1–37.
Hendel, I. (1999). Estimating multiple-discrete choice models: An application to computeriza-
tion returns. Review of Economic Studies, 66, 423446.
Hess, S., & Polak, J. W. (2005). Mixed logit modelling of airport choice in multi-airport
regions. Journal of Air Transport Management, 11, 5968.
Hotz, J., & Miller, R. A. (1993). Conditional choice probabilities and the estimation of
dynamic models. Review of Economic Studies, 60, 497529.
Igami, M. (2013). Estimating the innovator’s dilemma: Structural analysis of creative destruc-
tion. Mimeo. New Haven, CN: Yale University.
Mantin, B., & Koo, B. (2009). Dynamic price dispersion in airline markets. Transportation
Research Part E, 45, 10201029.
McAfee, R. P., & te Velde, V. (2007). Dynamic pricing in the airline industry. In
T. J. Hendershott (Ed.), Handbook of economics and information systems (Vol. 1). New
York, NY: Elsevier Science.
McFadden, D. (1974). Conditional logit analysis of qualitative choice behavior. In
P. Zarembka (Ed.), Frontiers in econometrics (pp. 105142). New York, NY: Academic
Press.
Nevo, A. (2001). Measuring market power in the ready-to-eat cereal industry. Econometrica,
69, 307342.
Newman, J. P., Garrow, L., Ferguson, M., Jacobs, T. L., & Purnomo, H. (2012). Estimation of
choice-based models using sales data from a single firm. Mimeo. Atlanta, GA: Georgia
Institute of Technology.
Pels, E., Nijkamp, P., & Rietveld, P. (2003). Access to and competition between airports: A
case study for the San Francisco Bay area. Transportation Research Part A, 37, 71–83.
Petrin, A. (2002). Quantifying the benefits of new products: The case of the minivan. Journal
of Political Economy, 110, 705729.
Proussaloglou, K., & Koppelman, F. S. (1999). The choice of air carrier, fight, and fare class.
Rust, J. (1987). Optimal replacement of GMC bus engines: An empirical model of Harold
Zurcher. Econometrica, 55, 9991033.
Stavins, J. (2001). Price discrimination in the airline market: The effect of market concentra-
tion. The Review of Economics and Statistics, 83, 200202.
Talluri, K., & van Ryzin, G. (2004). Revenue management under a general discrete choice
model of consumer behavior. Management Science, 50, 1533.
Train, K. (2002). Discrete choice methods with simulation. New York, NY: Cambridge
University Press.
Van Eggermond, M. A. B., Schuessler, N., & Auxhausen, K. W. (2007). Consumer choice
behavior and strategies of air transportation service providers. Proceedings 7th Swiss
Transport Research Conference (Monte Verità, Ascona, Switzerland).
Wen, C.-H., & Lai, S.-C. (2010). Latent class models of international air carrier choice.
Transportation Research Part E, 46, 211221.
CHAPTER 8
AIR CARGO SERVICES AND THE

EXPORT FLOWS OF DEVELOPING
COUNTRIES
Henry L. Vega
ABSTRACT
The use of air cargo by low-income countries and the effects of freight
charges on their export flows are described. This is accomplished by illus-
trating the difference between export flows from developing countries of
perishable products and high-tech goods. Descriptive statistics are used
to highlight the importance of trade that travels by air from these coun-
tries to the United States and the European Union. Subsequently, costs
of air freight are estimated. A gravity model of trade measures the effect
of these costs on export flows. Major institutional and regulatory con-
straints that may be halting additional trade that relies on air transporta-
tion, and the implications for economic growth, are identified.
Keywords: International trade; air cargo; transportation costs;
perishable goods; gravity model; developing countries
JEL classifications: F140; Q170; O180

ISSN: 2212-1609/doi:10.1108/S2212-160920140000004007
199
200 HENRY L. VEGA
INTRODUCTION
The sustained growth of air cargo has been driven by the internationaliza-
tion of production activities, changes in consumption patterns, and heigh-
tened economic activity in several regions around the world, including
some relatively low-income countries. This chapter addresses the use of air
cargo by low-income countries and the effects of freight charges on their
export flows. It focuses on the interdependent relationships between air
freight transportation costs and their effects on trade. The chapter accom-
plishes this by illustrating the difference between exporting perishable pro-
ducts and high-tech goods between the year 2000 and 2006, as reported by
Vega (2010). It also outlines institutional and regulatory constraints that
may be halting the expansion of industry sectors that rely on air transpor-
tation and the implications for economic growth.
A causal connection between transportation and trade in developed
countries is well documented in the academic literature. Such causality, it is
argued, arises from the effects that the costs of moving goods have on the
observed amounts of volume and value of international trade. In addition
to transportation costs, there is a series of costs incurred by buyers and sell-
ers when they trade. These costs are composed of direct elements, for exam-
ple, freight charges and insurance; and indirect elements, such as the
financial cost of the time goods are in transit and their inventory cost.
Usually included among other trade costs are search costs, legal fees, and
currency exchanges. However, the effects of transportation costs, and other
transactions costs, despite their documented relevance, have been only mar-
ginally studied in developing countries.
The existent literature stems from revived academic interest following
the end of the Uruguay Round of Multilateral Trade Negotiations in 1994.
These works encompassed the development of trade theories and the
empirical assessment of international trade models using computer systems.
These theories and models were based predominantly on David Ricardo’s
theory of comparative advantage and the HeckscherOhlin’s general equi-
librium mathematical model of international trade. The results of such
works have provided evidence that currently observed trade levels are lower
than those predicted by the models. Some economists have referred to this
as the mystery of the “missing trade.”1
A major limitation in exploring the ties between transportation and
trade has been the complexity in estimating the generalized costs of freight
transportation. This is largely due to the lack of publicly available disaggre-
gated data by type of good and mode, for example, air, ground, sea, or
Air Cargo Services and the Export Flows of Developing Countries 201
multimodal. Despite this shortcoming, it has been observed that demand

for air cargo has increased, largely in response to the internationalization
of production. This in turn has required large volumes of freight to be
transported to distant locations.
Fig. 1 offers a historic trend of the increasing importance of air cargo
markets involving developing regions from 2000 to 2007. It presents infor-
mation on the revenue ton kilometers (RTKs) carried by airlines from dif-
ferent regions of the world. Compared to the stability of the RTKs of the
United States and European airlines, Asia and Pacific airlines experienced
a significant growth in RTKs carried, from 48 billion to almost 70 billion.
Middle East airlines more than doubled the amount of cargo they carried,
from 4.6 billion RTKs in 2000 to 10.8 billion in 2007.
Increased demand has been eased by the effects of liberalization in trade
as well as transportation services all around the globe. Indeed, spatial loca-
tion and industry composition have in some cases changed in those coun-
tries and regions where the availability of air cargo services (ACS) has
made an increase in the diversity of export products possible. For example,
in the economic geography literature, Bowen (2002) describes how this has
occurred in both industrial and industrializing countries. Here, supply
chain management and logistics optimization are especially important in
the exchange and transportation of intermediate inputs and high-
technology products. This is also increasingly true in non-industrialized
countries. Furthermore, existing import data for the United States (US)
80,000
70,000
60,000 U.S. airlines

RTKs, millions
50,000 Europe (except CIS)
40,000 Western hemisphere

(except US)
30,000
Middle East
20,000 Asia and Pacific
10,000
0
2000 2001 2002 2003 2004 2005 2006 2007
Year
Fig. 1. 20002007 Air Cargo Market Share by Airline’s Region of Origin.

Source: Based on Data from the Boeing Company.
202 HENRY L. VEGA
show that imports of high-tech products from developing countries have

increased significantly between 2005 and 2008 (Table 1). Brazilian exports,
for example, increased 24 percent, while Chinese exports increased nearly
42 percent.
Exports of perishables & exotics (P&E) shipped by air have also become
increasingly important. In 2008, the US Department of Commerce regis-
tered imports of fresh-cut flowers and plants worth $516 million from
Colombia and $134 million from Ecuador; $381 million in vegetables from
Peru; and $986 million in fish and seafood from Chile. Industry calcula-
tions have estimated that 15 percent of global air cargo flows are P&E with
an annual growth of 7.1 percent (von Heereman, 2006).
This chapter illustrates the importance of air freight transportation costs
as a factor underlying trade levels in several regions in the world. It offers a
quantitative analysis of the effects of transportation costs on trade distin-
guishing between high-tech goods and P&E products. Whenever feasible, it
outlines major institutional and regulatory constraints that may be halting
the expansion of industry sectors that rely on air transportation and the
implications for economic growth.
Table 1. US Imports from 2004 to 2008 (in Thousands of Dollars).

2005 2006 2007 2008
US imports from Brazil

Computers 9,026 10,132 14,337 11,017
Computer accessories, peripherals and parts 12,380 20,319 13,731 15,800
Semiconductors and related devices 3,228 5,586 4,093 9,581
Telecommunications equipment 94,205 88,931 115,033 103,856
Laboratory testing and control instruments 1,737 2,691 2,833 7,367
Other scientific, medical and hospital 24,344 29,531 30,155 32,093
equipment
Total 144,920 157,190 180,182 179,714
US imports from China
Computers 14,453,320 17,371,216 23,239,995 25,039,779
Computer accessories, peripherals and parts 25,745,996 28,931,412 28,092,735 27,012,046
Semiconductors and related devices 1,785,173 2,181,376 2,338,684 2,132,169
Telecommunications equipment 7,021,769 8,659,269 12,569,256 14,497,175
Laboratory testing and control instruments 204,460 302,199 572,495 571,512
Other scientific, medical and hospital 1,550,599 1,863,379 2,130,001 2,620,457
equipment
Total 50,761,317 59,308,851 68,943,166 71,873,138
Source: Vega (2010).

AIR CARGO FLOWS FROM DEVELOPING COUNTRIES

TO THE UNITED STATES AND EUROPE BETWEEN
2000 AND 2006
In the last two decades, the choice of air transport has become increasingly
more common in the transport of high-value goods such as diamonds,
high-tech manufacturing goods, machinery spare parts, and perishable
goods that are time sensitive. From a research perspective, however, there
has only been limited progress made toward understanding how air cargo
services (ACS) affect economic activity. In practice, there are up to four
major parties that may be involved in making transportation decisions in
the supply chain: shippers, airlines, freight forwarders, and integrators. The
shipper is the person or company that has issued the contract for carriage
(the “air waybill” in air cargo) of the goods and in whose name the carriage
is performed (Fennes, 1997). Freight forwarders’ main function is to collect
and unitize several customers’ shipments before tendering them to carriers.
While airlines move freight from airport to airport, freight forwarders
move freight from the shipper’s premises to the origin airport and from the
destination airport to the consignee’s premises (Bowen & Leinbach, 2003).2
The creation of integrators, cargo-dedicated airlines, occurred as airliners
were able to accept palletized or containerized freight. After US deregula-
tion in November 1977 provided for specialized integrated air freight ser-
vices,3 airlines started addressing the air cargo market more aggressively.
This resulted in the emergence of vertically integrated cargo-dedicated air-
lines capable of offering shippers door-to-door service at a premium
charge. Examples of integrators include FedEx, UPS, DHL, and TNT.
Air Cargo Services in Developed and Developing Countries
In the air transport sector, several studies have assessed the comparative
advantages that regions with enhanced air transport infrastructure enjoy. It
is known, for instance, that in developed economies ACS are very competi-
tive. Air cargo is the predominant mode of international transport for these
economies’ high-technology and perishable goods industries, whose supply
chains rely strongly on economies of speed. High-tech manufacturing firms
use air freight not only to deliver final products but also to receive input
components. In developing countries, existing research suggests that air
freight should play a role in allowing trade among countries with different
204 HENRY L. VEGA
capacities for production but with similar tastes. These assumptions are
based on the fact that air transport has allowed developing countries to
engage in the trade of exotics and other perishable products, and increasingly
high-tech goods, in which they have a comparative advantage. Particularly
large, island and landlocked developing countries seem to have benefited
from air transport as they have seen their production feasibility frontiers
expand by being able to access remote domestic and international suppliers
and customers. Moreover, in these countries air freight may be the only
alternative to ship goods (Radelet & Sachs, 1998). Brazil, for example, has
relied heavily on air transportation to link sparsely inhabited areas with the
country’s major economic centers (US Department of Commerce, 2006).
In recent years, the differences between the uses of air transportation
in developed versus developing countries seem to blur as more developing
countries’ economies industrialize. Industrializing economies, for instance,
increasingly supply more high-tech products. At the same time more fre-
quent air services are available to them as new hub and distribution cen-
ters open. FedEx, for example, established its principal Asian hub at
Subic Bay in the Philippines. This investment subsequently transformed
the stature of this peripheral region within an international economy in
which air cargo performs a critical integrative role (Bowen, Leinbach, &
Mabazza, 2002).
In broad terms, the contributions of air transportation to increasing eco-
nomic activity in a country derive directly from the generation of jobs and
enhanced social welfare, at times in remote regions. Often mentioned
among indirect effects are the facilitation of trade and ultimately the provi-
sion of a source of comparative advantage. Industry sources, for instance,
claim that air cargo’s main economic benefit is its spin-off effect on interna-
tional trade as a trade facilitator (Air Transport Action Group, 2005).
From a macroeconomics perspective, empirical evidence so far is limited
and mainly suggests that ACS and economic growth are positively corre-
lated.4 Cline, Ruhl, Gosling, and Gillen (1998), for example, suggest that a
one percentage increase in GDP raises the demand for air cargo by slightly
more than one percent. Some evidence from developed countries further
suggests that air cargo tends to lead trade and GDP growth, especially dur-
ing economic upswings preceding increases in total trade and GDP values
(Kasarda & Green, 2005).5 Two research pieces have looked at the link and
addressed the issue of a causal relationship between air cargo and economic
growth in OECD countries (Ying, Chang, & Hsieh, 2008) and in Taiwan
(Chang & Chang, 2009). Table 2 summarizes what has been published on
the relationship between air cargo and growth.
Air Cargo Services and the Export Flows of Developing Countries
Table 2. Empirical Evidence on Air Cargo Services and Growth.
Study Focus Data Findings Limitations
Cline et al. Aviation activity’s effect ICAO’ statistics, OECD’s and A one percent rise in GDP Inconsistent and non-
(1998) on economic activity. World Bank’s economic would raise freight traffic significant estimates.
data. by slightly over one percent.
Kasarda and Air liberalization, ICAO Agreements Database, Aviation liberalization, Mutual dependency of the
Green customs reform, 2000 World Business quality of customs, and relationships. No direction
(2005) corruption, air freight, Environment Survey, lower corruption contribute of causation established.
trade, GDP, FDI. Transparency to higher GDP and FDI.
International’s Corruption
Perception Index.
InterVISTAS- Economic impact of air Bidirectional flows, GDP, GDP largely determines Model assumes air freight’s
ga2 (2006) service liberalization. distance to markets. demand. Closer market growth as consequence of an
opportunities negatively increase in the growth of
affect demand. Air freight passenger traffic.
does not drive the links
between liberalization and
economic development.
Ying et al. Relationships between GDP per capita, trade/GDP Cointegrated relationships Value of airborne trade not
(2008) trade openness, air ratio, air cargo volumes for among variables. Positive assessed. Use of time series
freight, and GDP in 19702002. contribution of air freight analyses to a relatively short
OECD countries. to GDP after performing a data series. Regression
Seemingly Unrelated analysis cannot be used to
Regression. establish causality.
Chang and Causality between air Large dataset consisting of Bidirectional relationship Possible absence of other
Chang cargo expansion and quarterly data on real GDP between air cargo relevant variables.
205
(2009) Economic Growth in and air cargo volumes. expansion and growth.
Taiwan.
206
Study Focus Data Findings Limitations
Bowen and Use of highly specialized Survey data of 126 electronics Variation in advanced ACS Results not robust.
Leinbach advanced air cargo manufacturing firms located remains unexplained.
(2003) services as a function in Penang, Singapore, Individual managers may
of firm, product, and Manila, and Kuala influence the choice of
place characteristics. Lumpur. services.
Leinbach and Use of air cargo services Survey data of 126 electronics Use of air cargo services is No distinction made between
Bowen by the electronics manufacturing firms located driven not only by the advanced air cargo services
(2004) industry in South East in Penang, Singapore, value-to-weight ratio but and general air cargo
Asia. Manila, and Kuala also by rapid product services.
Lumpur. cycles, risk of damage,
internationalization of
material procurement sites,
knowledge-intensive
logistics and demand for
specialized services.
HENRY L. VEGA
The seven studies can be grouped in two themes. As a first theme, Cline
et al. (1998), Kasarda and Green (2005), InterVISTAS-ga2 (2006), Ying
et al. (2008), and Chang and Chang (2009) look at the relationship between
air cargo demand and liberalization and income from a macro perspective.
As a second theme, the last two studies complement each other and look at
the microeconomic impact of using ACS by electronics manufacturers. The
implications for a region’s economy derived from the provision of these ser-
vices are also explored. The major difference between the two studies is the
distinction the authors make between a firm’s use of ACS and express
(advanced) integrated ACS. Bowen and Leinbach (2003) report the impor-
tance of knowledge-rich goods, employment size, and scope of internatio-
nalization as powerful determinants of integrated ACS usage. Leinbach
and Bowen (2004) show that much of the variation in ACS usage is related
to product characteristics that go beyond simply the value-to-weight ratio.
Exports by Air of High-Tech and Perishable & Exotics to the

United States and the European Union
Import statistics for 20002006 from the US Census Bureau, Foreign

Trade Statistics and the EU Statistical System’s Eurostat easily illustrate
the importance of trade that travels by air in the Global Economy. Since
2000 the quantities of exports from developing countries that travel by air
and are destined to the US and EU markets have continued to increase.
Exports from developing countries to the US expanded from 1,442,300
metric tons (MT) in the year 2000 to 2,130,000 MT in 2006. Exports from
developing countries to the EU went from 1,300,000 MT to 2,350,000 MT
in the same period. This results in an annual growth rate of 7.9 percent to
the United States, and 13.5 percent to the EU.
Tables 3 and 4 make it easier to visualize these trends. EU members cor-
respond to those 27 countries as of the 2007 enlargement. Developing
countries are those included by the World Bank in the low-income and
middle-income groups according to 2006 gross national income (GNI) per
capita, calculated using the World Bank Atlas method.6
Among the three income groups comprising developing countries, those
considered lower middle-income countries are the ones that experienced the
highest yearly growth rate, 12.3 percent to the United States and 24.0 per-
cent to the EU.
From 2000 to 2006, the growth in the quantities of high-tech products des-
tined to the United States and EU was considerable. Exports from developing
Table 3. 20002006 Exports (MT) by Air to the United States by Country Income Level.
208
Year 2000 2001 2002 2003 2004 2005 2006 Yearly
Growth %
All developing countries

Total Exports 1,442,300 1,282,500 1,608,000 1,637,000 1,903,000 2,019,000 2,130,000 7.9
Perishables 372,000 382,000 403,000 421,000 443,000 447,000 447,000 3.4
Percentage 25.8 29.8 25.1 25.7 23.3 22.1 21.0
High-tech products 95,300 95,500 125,000 166,000 200,000 232,000 273,000 31.1
Percentage 6.6 7.4 7.8 10.1 10.5 11.5 12.8
Other products 975,000 805,000 1,080,000 1,050,000 1,260,000 1,340,000 1,410,000 7.4
Percentage 67.6 62.8 67.2 64.1 66.2 66.4 66.2
Low-income countries
Total exports 139,119 111,725 164,796 169,506 176,954 189,953 214,632 9.0
Perishables 7,374 8,963 8,881 8,540 8,562 9,314 8,420 2.4
Percentage 5.3 8.0 5.4 5.0 4.8 4.9 3.9
High-tech products 745 762 915 966 1,392 1,638 2,211 32.8
Percentage 0.5 0.7 0.6 0.6 0.8 0.9 1.0
Other products 131,000 102,000 155,000 160,000 167,000 179,000 204,000 9.3
Percentage 94.2 91.3 94.1 94.4 94.4 94.2 95.0
Lower middle income countries
Total exports 821,400 725,700 938,700 978,000 1,208,000 1,311,000 1,427,000 12.3
Perishables 208,000 202,000 211,000 233,000 265,000 262,000 277,000 5.5
Percentage 25.3 27.8 22.5 23.8 21.9 20.0 19.4
Percentage 5.4 6.4 7.9 10.8 11.6 13.0 14.1
Other products 569,000 477,000 654,000 639,000 803,000 878,000 949,000 11.1
Percentage 69.3 65.7 69.7 65.3 66.5 67.0 66.5
HENRY L. VEGA
Upper middle income countries
Total exports 483,100 445,000 502,600 491,500 515,200 519,500 489,500 0.2
Perishables 157,000 171,000 182,000 180,000 170,000 175,000 161,000 0.4
Percentage 32.5 38.4 36.2 36.6 33.0 33.7 32.9
Percentage 10.4 10.8 10.1 11.9 11.3 11.5 14.2
Other products 276,000 226,000 270,000 253,000 287,000 285,000 259,000 −1.0
Percentage 57.1 50.8 53.7 51.5 55.7 54.9 52.9

Table 4. 20002006 Exports (MT) by Air to the EU by Country Income Level.

Year 2000 2001 2002 2003 2004 2005 2006 Yearly Growth %
All developing countries

Total exports 1,300,000 1,200,000 1,570,000 1,320,000 2,600,000 2,040,000 2,350,000 13.5
Perishables 258,990 301,814 384,722 346,028 350,352 368,779 365,657 6.9
Percentage 19.9 25.2 24.5 26.2 13.5 18.1 15.6
Percentage 7.1 6.9 5.9 9.3 6.9 11.3 12.4
Other products 949,243 815,051 1,091,966 851,789 2,069,173 1,441,709 1,692,076 13.0
Percentage 73.0 67.9 69.6 64.5 79.6 70.7 72.0
Low-income countries
Total exports 301,276 307,704 334,485 355,655 549,415 424,311 485,547 10.2
Perishables 118,728 124,786 150,907 143,581 171,779 152,599 138,792 2.8
Percentage 39.4 40.6 45.1 40.4 31.3 36.0 28.6
Percentage 0.8 0.8 0.8 1.0 0.8 1.0 1.2
Other products 179,995 180,437 180,829 208,593 373,380 267,287 340,953 14.9
Percentage 59.7 58.6 54.1 58.7 68.0 63.0 70.2
Lower middle income countries
Total exports 532,886 573,362 621,067 646,045 825,473 1,060,000 1,330,000 24.9
Perishables 67,679 83,947 119,115 111,635 84,779 130,776 141,899 18.3
Percentage 12.7 14.6 19.2 17.3 10.3 12.3 10.7
Percentage 12.1 10.7 11.7 15.1 18.0 18.5 19.1
Other products 400,561 428,256 429,445 436,741 591,950 732,801 934,096 22.2
Percentage 75.2 74.7 69.1 67.6 71.7 69.1 70.2
Upper middle income countries
Total exports 468,134 314,645 612,797 322,115 1,220,000 562,361 532,638 2.3
209
210
Year 2000 2001 2002 2003 2004 2005 2006 Yearly Growth %
Perishables 72,584 93,080 114,699 90,812 93,794 85,404 84,965 2.8

Percentage 15.5 29.6 18.7 28.2 7.7 15.2 16.0
Percentage 5.2 6.2 2.9 6.5 2.3 5.1 6.1
Other products 370,984 202,069 480,041 210,270 1,098,732 448,293 415,211 2.0
Percentage 79.2 64.2 78.3 65.3 90.1 79.7 78.0
HENRY L. VEGA
countries to the United States grew from 95,300 MT to 273,000, while to the
EU they grew from 91,767 MT to 292,268 MT. Developing countries as a
group observed their export volumes of high-tech products increase by 31.1
percent to the United States and 36.4 percent to the EU. Export growth was
the highest for lower middle-income countries, 58.8 percent to the United
States and 48.8 percent to the EU. Exports of P&E products experienced a
3.4 percent yearly growth to the United States and 6.9 percent to the EU.
Exports from developing countries to the United States grew from 372,990 to
447,000 MT, and to the EU from 258,990 to 365,657 MT in the same period.
Export growth was the highest for lower middle-income countries, 5.5 percent
to the United States and 18.3 percent to the EU.
Exports of high-tech products constituted between 6.6 and 12.8 percent
of total exports by air, with an upward trend in both the United States and
European markets during the period studied. Exports of P&E products
constituted between 13.5 and 29.8 percent of total exports by air, with a
downward trend in both these markets.
Trade statistics per region are presented next in Tables 5 and 6, with the
purpose of bringing in additional insights on the importance of trade in
high-tech and P&E goods for developing regions.
In the case of exports to the United States, all regions experienced
increases in trade volumes with the exception of Europe & Central Asia,
and Central America & the Caribbean. It should be noted though, that in
Europe & Central Asia exports of P&E products did increase by 28.2 per-
cent a year, while in Central America & the Caribbean exports of high-tech
and P&E increased 11.4 percent and 2.1 percent respectively. In total
exports to the EU, all regions experienced increases in trade volumes, with
the exception of Central America & the Caribbean, and South America.
As for its significance in economies of developing countries, between
2000 and 2006 the value of P&E exported by air from developing countries
to the US market (f.o.b. values are available only for the United States)
grew 8.4 percent yearly, from $1.46 billion to $2.2 billion. Exports of P&E
from South America were the largest throughout this period ($1.56 billion
in 2006), followed by Central America & Caribbean ($370 million in 2006),
and East Asia & Pacific ($186 million in 2006).
Between 2000 and 2006 as well, the value of high-tech goods exported
by air from developing countries to the US market (f.o.b. values are avail-
able only for the United States) grew 51.7 percent yearly, from $11.5 billion
to $47.2 billion. Exports of high-tech goods from East Asia & Pacific were
the largest ($44.4 billion in 2006), followed by Central America &
Caribbean ($1.452 billion in 2006), and Europe & Central Asia ($919 mil-
lion in 2006). Table 7 provides statistics for additional world regions.
Table 5. 20002006 Quantity of Exports (MT) by Air to the United States by Geographic Region.
212
Year 2000 2001 2002 2003 2004 2005 2006 Yearly
Growth %
East Asia & Pacific countries

Total exports 610,400 532,300 760,000 783,800 950,600 1,116,700 1,213,200 16.5
Perishables 15,900 16,600 17,000 16,800 18,600 19,700 21,200 5.6
Percentage 2.6 3.1 2.2 2.1 2.0 1.8 1.7
Percentage 12.4 14.4 14.5 19.0 18.9 18.9 20.7
Other products 519,000 439,000 633,000 618,000 752,000 886,000 941,000 13.6
Percentage 85.0 82.5 83.3 78.8 79.1 79.3 77.6
Europe & Central Asia countries
Total exports 59,288 55,388 63,408 60,489 63,714 53,837 56,874 −0.7
Perishables 159 621 501 873 329 316 429 28.2
Percentage 0.3 1.1 0.8 1.4 0.5 0.6 0.8
High-tech products 9,129 6,367 4,207 3,815 4,085 5,521 4,745 −8.0
Percentage 15.4 11.5 6.6 6.3 6.4 10.3 8.3
Other products 50,000 48,400 58,700 55,800 59,300 48,000 51,700 0.6
Percentage 84.3 87.4 92.6 92.2 93.1 89.2 90.9
Central America & Caribbean countries
Total exports 192,428 154,738 164,673 156,240 177,442 168,917 178,184 −1.2
Perishables 64,000 61,000 63,000 59,000 62,000 67,000 72,000 2.1
Percentage 33.3 39.4 38.3 37.8 34.9 39.7 40.4
HENRY L. VEGA
Percentage 4.4 6.3 5.3 7.2 7.6 7.6 8.0
Other products 120,000 84,000 93,000 86,000 102,000 89,000 92,000 −3.9
Percentage 62.4 54.3 56.5 55.0 57.5 52.7 51.6
South America countries
Total exports 414,048 405,562 435,622 460,660 517,658 491,783 467,016 2.1
Perishables 288,000 298,000 317,000 339,000 358,000 355,000 347,000 3.4
Percentage 69.6 73.5 72.8 73.6 69.2 72.2 74.3
High-tech products 1,048 1,562 622 660 658 783 1,016 -0.5
Percentage 0.3 0.4 0.1 0.1 0.1 0.2 0.2
Other products 125,000 106,000 118,000 121,000 159,000 136,000 119,000 -0.8
Percentage 30.2 26.1 27.1 26.3 30.7 27.7 25.5
Middle East & North Africa
Total exports 16,417 12,983 19,586 19,176 24,072 17,684 25,549 9.3
Perishables 237 277 133 177 191 306 256 1.3
Percentage 1.4 2.1 0.7 0.9 0.8 1.7 1.0
High-tech products 380 207 353 399 481 378 393 0.6
Percentage 2.3 1.6 1.8 2.1 2.0 2.1 1.5
Other products 15,800 12,500 19,100 18,600 23,400 17,000 24,900 9.6
Percentage 96.2 96.3 97.5 97.0 97.2 96.1 97.5
South Asia
Total exports 134,313 102,327 136,626 129,908 140,443 154,687 171,417 4.6
Perishables 605 601 749 972 1,078 1,235 1,746 31.4
Percentage 0.5 0.6 0.5 0.7 0.8 0.8 1.0
High-tech products 708 726 877 936 1,365 1,452 1,671 22.7
Percentage 0.5 0.7 0.6 0.7 1.0 0.9 1.0
Other products 133,000 101,000 135,000 128,000 138,000 152,000 168,000 4.4
Percentage 99.0 98.7 98.8 98.5 98.3 98.3 98.0
Sub-Saharan Africa
Total exports 17,345 18,878 27,560 27,910 26,680 17,495 20,299 2.8
Perishables 3,845 4,708 4,213 4,302 3,188 3,838 4,394 2.4
Percentage 22.2 24.9 15.3 15.4 12.0 21.9 21.6
High-tech products 100 170 147 108 92 157 206 17.5
Percentage 0.6 0.9 0.5 0.4 0.3 0.9 1.0
Other products 13,400 14,000 23,200 23,500 23,400 13,500 15,700 2.9
Percentage 77.3 74.2 84.2 84.2 87.7 77.2 77.3
213
Table 6. 20002006 Quantity of Exports (MT) by Air to the EU by Geographic Region.
214
Year 2000 2001 2002 2003 2004 2005 2006 Yearly
Growth %
East Asia & Pacific countries

Total exports 441,880 502,786 552,670 560,685 729,233 943,964 1,100,000 24.8
Perishables 8,305 20,382 24,150 19,562 21,504 25,573 27,556 38.6
Percentage 1.9 4.1 4.4 3.5 2.9 2.7 2.5
Percentage 17.3 13.9 14.6 19.1 21.9 22.2 24.6
Other products 357,039 412,542 447,902 433,848 548,326 709,070 801,871 20.8
Percentage 80.8 82.1 81.0 77.4 75.2 75.1 72.9
Europe & Central Asia countries
Total exports 110,357 75,935 102,731 88,616 412,201 317,967 372,638 39.6
Perishables 4,005 2,928 3,792 3,231 3,037 3,925 1,111 −12.0
Percentage 3.6 3.9 3.7 3.6 0.7 1.2 0.3
Percentage 1.8 2.6 2.0 3.2 0.7 0.9 0.7
Other products 104,359 71,005 96,837 82,590 406,446 311,059 368,870 42.2
Percentage 94.6 93.5 94.3 93.2 98.6 97.8 99.0
Central America & Caribbean countries
Total exports 257,698 101,031 344,930 112,706 678,247 126,224 187,439 −4.5
Perishables 69,870 93,990 128,228 98,297 101,657 101,842 95,076 6.0
Percentage 27.1 93.0 37.2 87.2 15.0 80.7 50.7
HENRY L. VEGA
Percentage 3.4 7.1 1.8 6.2 1.8 8.8 6.4
Other products 235,212 74,980 318,015 79,644 651,133 99,468 147,240 −6.2
Percentage 91.3 74.2 92.2 70.7 96.0 78.8 78.6
South America countries
Total exports 242,275 134,533 404,144 135,521 710,350 151,336 190,746 −3.5
Perishables 56,328 75,256 107,642 72,331 86,712 86,335 66,951 3.1
Percentage 23.2 55.9 26.6 53.4 12.2 57.0 35.1
Percentage 0.5 0.9 0.3 1.5 0.8 3.4 2.2
Other products 184,671 58,020 295,098 61,133 617,894 59,843 119,655 −5.9
Percentage 76.2 43.1 73.0 45.1 87.0 39.5 62.7
Middle East & North Africa
Total exports 64,820 65,409 61,230 72,153 82,101 87,111 81,099 4.2
Perishables 14,790 20,094 33,548 39,976 11,087 51,886 46,834 36.1
Percentage 22.8 30.7 54.8 55.4 13.5 59.6 57.7
High-tech products 1,023 799 838 936 1,217 864 713 −5.1
Percentage 1.6 1.2 1.4 1.3 1.5 1.0 0.9
Other products 49,006 44,516 26,845 31,241 69,797 34,362 33,552 −5.3
Percentage 75.6 68.1 43.8 43.3 85.0 39.4 41.4
South Asia
Total exports 158,296 152,895 172,012 199,633 379,734 245,352 280,363 12.9
Perishables 18,680 22,009 23,776 27,960 32,116 29,854 38,566 17.7
Percentage 11.8 14.4 13.8 14.0 8.5 12.2 13.8
Percentage 1.3 1.4 1.4 1.5 1.1 1.6 1.8
Other products 137,592 128,702 145,779 168,684 343,600 211,504 236,702 12.0
Percentage 86.9 84.2 84.7 84.5 90.5 86.2 84.4
Sub-Saharan Africa
Total exports 212,918 222,399 227,135 217,690 231,153 234,819 262,623 3.9
Perishables 143,317 142,354 171,202 156,949 180,940 155,643 156,479 1.5
Percentage 67.3 64.0 75.4 72.1 78.3 66.3 59.6
High-tech products 1,346 1,070 1,068 1,189 1,049 1,197 1,252 −1.2
Percentage 0.6 0.5 0.5 0.5 0.5 0.5 0.5
Other products 68,256 78,975 54,864 59,552 49,164 77,979 104,892 8.9
Percentage 32.1 35.5 24.2 27.4 21.3 33.2 39.9
215
216
Table 7. Export Values of P&E and High-Tech Products that Travel by Air to the United States, Million US$.
Year 2000 2001 2002 2003 2004 2005 2006 Yearly
growth %
Exports of perishable goods

Developing countries 1,460 1,460 1,520 1,780 1,960 1,980 2,200 8.4
South America 1,040 1,010 994 1,180 1,300 1,400 1,560 8.3
Central America & Caribbean 250 240 266 270 290 340 370 8.0
East Asia & Pacific 121 132 152 182 180 172 186 9.0
Sub-Saharan Africa 28 54 85 116 159 43 49 12.4
South Asia 10 8 9 13 19 13 19 14.8
Europe & Central Asia 4 12 11 18 9 13 16 59.8

Middle East & North Africa 1 2 1 2 2 3 3 30
Exports of high-tech goods
Developing countries 11,500 12,000 19,700 25,300 31,900 41,600 47,200 51.7
East Asia & Pacific 8,710 8,700 17,200 22,900 29,200 38,600 44,400 68.3
Central America & Caribbean 1,362 2,076 1,626 1,519 1,470 1,475 1,452 1.1
Europe & Central Asia 1,250 882 621 705 891 1,200 919 −4.4
South Asia 63 76 82 101 153 155 173 29.0
South America 108 214 74 71 80 65 108 0.0
HENRY L. VEGA
Middle East & North Africa 47 21 51 53 61 58 55 2.6
Sub-Saharan Africa 8 12 10 17 18 17 19 25

In summary, trade statistics clearly show the success of developing

countries at exporting both high-tech and P&E goods to distant markets,
such as the United States and the EU. These numbers also show that
developing regions across the globe experienced increasingly higher air-
borne trade with the United States and the EU between 2000 and 2006.
However, these flows do not always seem to have a clear relationship
with the distance to the market. What is clearer is that developing
regions, including Sub-Saharan Africa, are benefiting from the advantages
of air transportation by engaging not only in the trade of P&E goods but
also high-tech goods. To better understand the factors influencing trade
flows by air of these goods, the following section assesses the importance
of some variables behind increased flows of airborne exports from devel-
oping countries.
AIR TRANSPORTATION COSTS OF PRODUCTS

ENTERING THE US MARKET
Among the competitive export alternatives that producers in developing

countries have in international markets are time-sensitive spare parts,
some high-tech products, and perishable and exotic (P&E) products.
While export performance is generally linked to producers’ ability to take
advantage of their geographic location and access to low labor costs,
other concerns related to the sustainability of these economic activities
include complex supply chains and high transportation costs. From a
research perspective, there has been an increasing interest in understand-
ing the reliability of such supply chains and the implications of its high
transportation costs, particularly air freight costs, for the movement of
high-tech goods, but very little work has been done on the movement of
P&E.
Using exports of products from throughout the world to the United
States, an empirical investigation of trade and air transportation costs of
high-tech and P&E products between 2000 and 2006 was undertaken by
Vega (2008). The dataset used offered the value of the freight rate and f.
o.b. (also referred to as free-alongside-shipping or FAS) value of ship-
ments for all traded goods, including high-tech and P&E products
between the United States and countries around the world. This provided
a way to isolate outlier high freight rates from occasional or isolated
218 HENRY L. VEGA
shipments. Products considered high-tech in this analysis were all those

cataloged as Advanced Technology Products according to the 2006 defini-
tions of the Foreign Trade Division, US Census Bureau.
The quantitative assessment of the size of this trade and of the mag-
nitude of its transportation costs across world regions revealed that
overall, in the high-tech trade, lower middle income countries paid the
highest dollar amount per kilogram of freight shipped ($2.96 versus
$2.80 for low-income countries, and $2.55 for upper middle income
countries in 2006). Upper middle countries as a group paid the lowest
amount in dollar terms per kilogram. In P&E products, low-income
countries as a group paid the highest shipping cost per kilogram when
compared to lower middle income and upper middle income countries
($2.21 versus $1.10 for lower middle-income, and $1.25 for upper mid-
dle income countries in 2006). In addition, when comparing the air
freight rates as a percentage of cargo value, low-income countries paid
the highest shipping charges, 2.6 percent in the case of high-tech and
31.3 percent in the case of perishables. These findings would suggest
that air freight costs decline as countries increase their level of eco-
nomic development.
Geographically, the world regions with the highest freight cost for high-
tech products in 2006 were East Asia and Pacific ($2.97/kg) and South Asia
($2.70/kg). The regions with the lowest air freight costs were Central
America & Caribbean ($1.11/kg) and Middle East & North Africa ($1.98/
kg). For P&E products, Europe & Central Asia, and East Asia & Pacific
faced the highest air freight costs in 2006 ($3.36/kg and $2.88/kg, respec-
tively) while Central America & Caribbean and South America ($0.73/kg
and $1.14/kg, respectively) faced the lowest costs. Table 8 summarizes these
findings. Table 9 shows the outbound and inbound flows and calculates
corresponding ratios.
The results of the analysis using descriptive statistics showed that
transportation costs are obviously related to distance. Also, when they
are assessed in dollars per kilogram, they tend to be lower for lower-
cost products, such as P&E, than for high-tech products that travel by
air. However, when transportation costs are measured as a percentage
of cargo value, they tend to be higher for lower-cost products, such as
P&E, than for high-tech products. Among the inconsistencies of the
results were those regions experiencing higher transportation costs than
far more distant regions. South America was an example of such a
region.
Table 8. Air Transportation Costs of High-Tech and P&E Products Entering the United States in Years
2000/2003/2006.
Region 2000 2003 2006
Shipment Freight Rate Shipment Freight Rate % Shipment Freight Rate %

pricea rateb % cargo pricea rateb cargo pricea rateb cargo
($/kg) ($/kg) valuec ($/kg) ($/kg) valuec ($/kg) ($/kg) valuec
High-tech products
Developing countries 121.22 2.32 1.9 152.71 2.63 1.7 172.69 2.85 1.7
Low income 90.16 2.98 3.3 110.59 2.58 2.3 109.19 2.80 2.6
Lower middle income 104.52 2.57 2.5 126.49 2.77 2.2 153.50 2.96 1.9
Upper middle income 136.46 2.08 1.5 201.02 2.36 1.2 230.30 2.55 1.1
East Asia & Pacific 115.40 2.65 2.3 153.74 2.74 1.8 177.09 2.97 1.7
Europe & Central Asia 136.56 0.91 0.7 184.47 1.76 1.0 193.31 2.26 1.2
Central America & 162.13 0.93 0.6 135.53 1.51 1.1 102.80 1.11 1.1
Caribbean
South America 102.61 1.13 1.1 107.22 1.79 1.7 106.37 2.15 2.0
Middle East & North 124.47 2.97 2.4 132.25 2.32 1.8 138.59 1.98 1.4
Africa
South Asia 89.54 3.02 3.4 106.99 2.57 2.4 103.76 2.70 2.6
Sub-Saharan Africa 73.58 1.34 1.8 135.40 3.59 2.7 89.30 3.45 3.9
Perishable products
Developing countries 3.92 0.89 22.7 4.22 1.00 23.7 4.93 1.17 23.8
Low income 6.36 1.50 23.6 16.71 2.41 14.4 7.06 2.21 31.3
Lower middle income 3.55 0.92 25.8 3.74 0.92 24.5 4.07 1.10 26.9
Upper middle income 4.29 0.83 19.2 4.24 1.04 24.7 6.29 1.25 19.9
East Asia & Pacific 7.65 2.55 33.3 10.82 2.88 26.6 8.76 2.88 32.9
219
Europe & Central Asia 24.21 2.96 12.2 20.03 2.40 12.0 38.06 3.36 8.8
3.97 0.52 13.1 4.54 0.60 13.1 5.12 0.73 14.2
220
Region 2000 2003 2006
Shipment Freight Rate Shipment Freight Rate % Shipment Freight Rate %

pricea rateb % cargo pricea rateb cargo pricea rateb cargo
($/kg) ($/kg) valuec ($/kg) ($/kg) valuec ($/kg) ($/kg) valuec
Central America &

Caribbean
South America 3.62 0.86 23.9 3.48 0.96 27.5 4.50 1.14 25.3
Middle East & North 4.94 1.55 31.4 10.48 1.90 18.1 12.47 2.58 20.7
Africa
South Asia 16.29 2.46 15.1 13.59 2.48 18.3 10.68 2.37 22.2
Sub-Saharan Africa 7.30 1.75 23.9 27.11 2.18 8.0 11.16 2.03 18.2

a
Shipment prices equal to cargo FAS value divided by quantity. Data obtained from the US Foreign Trade Statistics.
b
Freight rates calculated based on the formula as shown in Eq. (1).
c
Calculated based on the previous two columns.
HENRY L. VEGA
Table 9. US Trade Flows by Air in Metric Tons and Import/Export Ratio (Years 20002006).
Trade Flows Year 2000 2001 2002 2003 2004 2005 2006
from
Developing US exports 670,000 602,000 571,000 595,000 696,000 727,000 836,000

countries US imports 1,440,000 1,280,000 1,610,000 1,640,000 1,900,000 2,020,000 2,130,000
US import/export ratio 2.1 2.1 2.8 2.8 2.7 2.8 2.5
East Asia & US exports 156,000 153,000 160,000 184,000 209,000 219,000 262,000
Pacific US imports 610,000 533,000 760,000 784,000 951,000 1,120,000 1,210,000
Europe & US exports 37,300 35,700 33,600 37,800 46,100 53,600 66,000
Central Asia US imports 59,200 55,400 63,400 60,500 63,700 53,900 56,800
Middle East & US exports 13,600 13,900 12,700 15,900 24,100 24,700 25,600
North Africa US imports 16,400 13,000 19,600 19,200 24,000 17,700 25,500
South Asia US exports 26,800 25,600 29,100 31,500 39,300 42,300 51,800
US imports 134,000 102,000 137,000 130,000 140,000 154,000 171,000
Sub-Saharan US exports 23,900 24,700 26,300 28,200 33,500 39,400 46,400
Africa US imports 17,400 18,800 27,600 27,900 26,700 17,500 20,300
Central America US exports 160,000 127,000 129,000 114,000 121,000 112,000 119,000
& Caribbean US imports 192,000 155,000 164,000 157,000 178,000 169,000 178,000
South America US exports 252,000 221,000 180,000 183,000 223,000 236,000 265,000
US imports 414,000 405,000 436,000 461,000 517,000 492,000 467,000
221
222 HENRY L. VEGA
WHAT A GRAVITY MODEL TELLS US ABOUT AIR-

SHIPPED EXPORT FLOWS FROM DEVELOPING
COUNTRIES TO THE UNITED STATES
Gravity models allow researchers to quantify how distance, country charac-

teristics, policies such as an Open Skies agreement with the United States,
and air freight charges relate to exports of high-tech and perishable and
exotic (P&E) products from developing countries. Indeed, this research
approach can effectively isolate the effect of the air freight rate and other
independent variables on the exports of developing countries.
As mentioned earlier, one of the limitations to perform additional
research on the importance of air cargo is the availability of data.
However, one useful tool is import data recorded from the US Census
Bureau, Foreign Trade Statistics. This dataset provides the value of the
freight rate and f.o.b. values of imports into the United States. Because of
the level of detail of this dataset, its contents have been used in numerous
academic studies that assess the impacts of transportation costs on interna-
tional trade.7 Vega (2010) used the information provided by this dataset to
calculate two freight rates measured in US dollars per kilogram (AFR1 and
AFR2) for developing countries’ export of P&E and high- tech goods. Data
on the existence of an Open Skies agreement with the United States are
available on the US Department of Transportation’s website. Data on dis-
tances were calculated using the Federal Aviation Administration tool to
calculate nautical miles separating airports. This study then applied a grav-
ity model to exports which were shipped by air from developing countries
to the United States from 2000 to 2006.8 The monetary values of the data
used in the analysis are nominal. The gravity model equations in the nat-
ural logarithm form were as follows:
ln XVijtp = β0 þ β1 ln GDPit þ β2 ln DISi þ β3 OSKijt þ β4 LCKi þ β5 ISLi

ð1Þ
þ β6 ln POPit þ β7 BALQijt þ β12 ln AFRitp þ β10 YEARt þ β11 REGitp þ ɛ
where
XVijtp is the value of exports from developing country i of P&E goods to

the United States (j) in year t.
GDPit is the per capita GDP of country i in year t, expressed in logarith-
mic form.
DISi is the geographical distance between country i and the United

States measured in nautical miles from the largest airport in developing
country i to the closest major cargo airport in the United States (Los
Angeles LAX, Miami MIA or New York JFK), expressed in logarithmic
form.
OSKijt is a dummy variable taking a value of 1 in year t and the subse-
quent years when an Open Skies agreement was concluded between
country i and country j in year t − 1 or preceding years.
LCKi is a dummy variable used to identify if country i is landlocked.
ISLi is a dummy variable used to identify if country i is an island.
POPit is the size of the population of country i in year t, expressed in
logarithmic form.
BALQijt is the trade imbalance between country i and country j in
year t.
AFRitp is the air freight rate measured in dollar per kilogram, charged
to cargoes of P&E goods from country i in year t, expressed in logarith-
mic form.
YEAR is a dummy variable used to identify the year in which the trade
flow was recorded.
REG is a dummy variable used to identify the geographic region to
which country i, exporter of P&E goods, belongs.
ɛ is an error term.
And:
ln XVijth = β0 þ β1 ln GDPit þ β2 ln DISi þ β3 OSKijt þ β4 LCKi þ β5 ISLi

ð2Þ
þ β6 ln POPit þ β7 BALQijt þ β12 ln AFRith þ β10 YEARt þ β11 REGith þ ɛ
where
XVijth is the value of exports from developing country i of high-tech

goods to the United States (j) in year t.
GDPit is the per capita GDP of country i in year t, expressed in logarith-
mic form.
224 HENRY L. VEGA
DISi is the geographical distance between country i and the United

States measured in nautical miles from the largest airport in developing
country i to the closest major cargo airport in the United States (Los
Angeles LAX, Miami MIA, or New York JFK), expressed in logarith-
mic form.
OSKijt is a dummy variable taking a value of 1 in year t and the subse-
quent years when an Open Skies agreement was concluded between
country i and country j in year t − 1 or preceding years.
LCKi is a dummy variable used to identify if country i is landlocked.
ISLi is a dummy variable used to identify if country i is an island.
POPit is the size of the population of country i in year t, expressed in
logarithmic form.
BALQijt is the trade imbalance between country i and country j in
year t.
AFRitp is the air freight rate measured in dollar per kilogram, charged
to cargoes of high-tech goods from country i in year t, expressed in loga-
rithmic form.
YEAR is a dummy variable used to identify the year in which the trade
flow was recorded.
REG is a dummy variable used to identify the geographic region to
which country i, exporter of high-tech goods, belongs.
ɛ is an error term.
Countries with a higher per capita GDP were anticipated to have a lar-
ger value of goods transported by air. The distance between two countries
was expected to affect the trade negatively. The existence of an Open Skies
agreement between trading partners was expected to have a positive effect.
That a country is landlocked or an island should also have had a positive
influence on the flow of trade from country i to the United States as air
cargo is the fastest and arguably the only alternative to transport P&E and
high-tech goods. The size of the population in country i was expected to
have a positive effect on trade as the manufacturing and production of
these goods tend to be labor intensive. The trade balance variable was
expected to have a negative effect on the volume and value of exports from
country i to the United States as larger unidirectional flows are not usually
appealing for air cargo companies. Higher air freight rates were expected to
have a negative effect on trade.
Tables 10 and 11 show the parameter estimates for the equations and
the different estimations of the gravity model using OLS calculations, as
reported by Vega (2010). Dummy variables for Open Skies, landlocked
country, island, year, and region are also included.
Table 10. Estimation Results for Gravity Model of US Air-Shipped

Imports of P&E Goods from Developing Countries.
Independent Variables Eq. (1): Value of Exports
Corrected by economic Weighted

cluster regression
Per capita GDP 0.781*** 0.781** 0.820***

Distance −1.134*** −1.134 −1.121***
Open Skies 0.469*** 0.469** 0.389**
Landlocked −0.826*** −0.826 −0.840***
Island 0.651*** 0.651 0.561***
Population 0.637*** 0.637** 0.624***
Trade balance 2.135*** 2.135** 2.263***
Air freight rate 0.161** 0.161** 0.156**
Year 2000 Dropped Dropped Dropped
Year 2001 0.361 0.361* 0.294
Year 2002 0.0971 0.0971 0.0666
Year 2003 0.261 0.261 0.210
Year 2004 0.410* 0.410 0.356*
Year 2005 0.427* 0.427 0.358*
Year 2006 0.458** 0.458 0.393*
South America 1.471*** 1.471 1.455***
Central America & Dropped Dropped Dropped
Caribbean
East Asia & Pacific 0.488 0.488 0.353
Europe & Central Asia −2.030*** −2.030 −2.149***
Middle East & North Africa −2.404*** −2.404 −2.560***
South Asia −2.037*** −2.037 −2.072***
Sub-Saharan Africa 0.554 0.554 0.611
Constant 6.074*** 6.074 6.112***
Observations 643 643 643
R-squared 0.697 0.697 0.739
*significance at p ≤ 0.05.
**significance at p ≤ 0.01.
***significance at p ≤ 0.001.
226 HENRY L. VEGA
Table 11. Estimation Results for Gravity Model of US Air-Shipped

Imports of High-Tech Goods from Developing Countries.
Independent Variables Eq. (12): Value of Exports
Corrected by economic Weighted

cluster regression
Per capita GDP 2.926*** 2.926*** 2.920***

Distance −0.103 −0.103 −0.0999
Open Skies −0.0947 -0.0947 −0.0976
Landlocked 1.391*** 1.391** 1.379***
Island −0.130 −0.130 −0.147
Population 1.241*** 1.241** 1.241***
Trade Balance 0.303 0.303 0.303
Air freight rate −0.400** −0.400 −0.399***
Year 2000 Dropped Dropped Dropped
Year 2001 0.0191 0.0191 0.0180
Year 2002 0.278 0.278 0.280
Year 2003 0.384 0.384 0.384
Year 2004 0.434 0.434* 0.437
Year 2005 0.432 0.432 0.435
Year 2006 0.621* 0.621*** 0.628*
South America −2.011*** −2.011* −3.117***
Central America & 1.109 1.109 Dropped
Caribbean
East Asia & Pacific 4.002*** 4.002 2.871***
Europe & Central Asia −1.051 −1.051 −2.157***
Middle East & North Africa 0.730 0.730 -0.391
South Asia Dropped Dropped −1.119
Sub-Saharan Africa 1.798*** 1.798 0.689
Constant −30.40*** −30.40*** −29.26***
Observations 356 356 356
R-squared 0.747 0.747 0.746
*significance at p ≤ 0.05
**significance at p ≤ 0.01
***significance at p ≤ 0.001
What is readily apparent is that the year dummies are not statistically
significant. Also, the regional dummies become statistically non-significant
in the estimation of the clustered model that corrects for a country’s
income level. This finding suggests that countries sharing similar levels of
economic development have characteristics that are more important in
affecting air-shipped exports than merely the fact that these countries
belong to a certain geographical region. The coefficients of per capita GDP
were positive and statistically significant for both P&E as well has high-
tech goods. The size of the coefficients varied between P&E and high-tech
goods. Thus, while a one percent increase in per capita GDP would
increase exports of P&E goods by 0.78 percent, a similar increase in income
would increase exports of high-tech goods by 2.93 percent.
The distance variable had the expected negative coefficients in all estima-
tions. The coefficient associated with distance was not significant in the model
of trade of high-tech goods. The coefficients of the Open Skies dummy vari-
able differed in size and statistical significance between P&E goods and high-
tech goods. The coefficients in the case of high-tech goods were negative
although very small and not significant. However, the coefficients were posi-
tive and statistically significant in the case of P&E goods and in agreement
with existing research. When an Open Skies agreement existed between a
developing country and the United States, increased the developing country’s
value of exports of perishables by 46.9 percent. This result would seem to sup-
port the argument of US policymakers that developing countries would bene-
fit from liberalizing their air transport services sector.
The coefficients of the dummy variables for a landlocked country or
island, both of which are usually foreseen to have negative trade implica-
tions, presented inverse results depending on the type of product for which
trade is being estimated.
In the trade of high-tech goods, that a country is landlocked had a posi-
tive and statistically significant effect on the value of exports from these
countries to the United States. If a country was landlocked, the value of its
exports of high-tech goods increased by 139.1 percent. On the other hand,
a country being an island had a negative effect, although not statistically
significant, on the value of its exports of high-tech goods.
The coefficients of population were positive and highly statistically sig-
nificant for both exports of P&E and high-tech goods. In the case of P&E
goods, a one percent increase in the size of the population of a developing
country increased its value of exports by 0.64 percent. The effect of popula-
tion is larger in the case of high-tech goods. In this case, a one percent
increase in the size of the population of a developing country increased its
value of exports of high-tech goods by 1.24 percent
The study also found that the coefficient of the air freight rate had the
expected negative effect on the value of exports of high-tech goods. But the
effect was positive for the value of exports of P&E products. The value of
P&E exports appeared to be insensitive to freight charges, and instead, a
one percent increase in freight charges would have resulted in a 0.61 per-
cent increase in the value of exports. This could respond to seasonal higher
prices for P&E goods which are inelastic to freight charges; for example,
228 HENRY L. VEGA
during the Valentine’s Day holiday when demand for red roses is inelastic
to price (the Valentine’s effect). Alternatively, the positive sign of the coeffi-
cient could be the result of constrained uplift capacity. That is, higher
freight rates for refrigerated cargo might initially create an incentive for
additional cargo capacity. Once excess capacity is in place and freight rates
are low there will be an incentive for more production of P&E products
and eventually more exports. Moreover, it is necessary to be aware that the
coefficient is not capturing the effect of higher freight rates on the “missing
trade;” that is, additional exports that did not take place due to prohibitive
freight rates. Further studying this finding through simulation studies
would be recommended, adding additional variables to try to distinguish
the effect on airlines’ revenues of carrying cargo by integrators under refri-
gerated conditions versus cargo transported in the belly of passenger
planes. In the case of high-tech goods, the effect of the coefficient was more
straightforward: A one percent increase in air freight costs resulted in a
decrease of country i’s value of exports of 0.40 percent.
These results allow drawing some conclusions regarding the importance
of air freight rates. The profit margins that producers get for their produc-
tion of either P&E goods or high-tech goods are not high. The fact that the
trade flows are sensitive to the freight rate provides support to the claims
and requests from developing country governments that overall transporta-
tion costs should be reduced to encourage growth.
Overall, the study found that an exporting country’s per capita income
and population have a positive effect on the level of exports from develop-
ing countries that travel by air. The liberalization of air cargo services, fol-
lowing the conclusion of Open Skies negotiations with the United States,
had a positive effect on the flow of exports of P&E goods, although its
effect on the flow of exports of high-tech goods was negligible. This last
finding suggests that contrary to developing country policymaker claims
that Open Skies agreements can be detrimental to their economies, this
type of agreement can actually enhance trade.
As suspected, although distance has a negative effect on the levels of
trade, its effect is small and possibly statistically non-significant when
goods are shipped by air. Trade flows are highly sensitive to the magnitude
of the air freight rate. This indicates that, from a policy perspective, much
remains to be done to address the issue of higher transportation costs that
developing countries face.
The size of the population of a country has been observed to positively
influence the levels of trade. Hence, the larger the size of the population in
a developing country, the higher the export flows of P&E goods as well as
high-tech goods.
The results also provide some support to the claim of development econ-
omists that international transportation costs can be a possible explanation
for lower trade levels from developing countries. However, in the case of
goods shipped by air, this study shows evidence of the smaller effect of dis-
tance on trade flows. The non-significance of the distance variable seems to
support claims that distance is not a deterrent to trade in the case of goods
that are air-shipped.
CONCLUSIONS
Development economists often argue in favor of facilitating trade and low-

ering trade costs between countries on grounds of economic efficiency and
welfare enhancement. Transportation costs are usually included as one type
of those costs with the ability to reduce the levels of trade and negatively
affect the trade potential of a developing country. In the area of air trans-
portation of goods, existing academic literature on the importance of air
cargo for developing countries is rather limited. However, the effects of
transportation costs on trade of high-tech goods may be greater for any
individual country if there is competition in the supply of those goods. The
results of augmenting these analyses could have extraordinary implications
for economic growth and public policies in the area of transportation infra-
structure. Existing studies suggest that lower transportation costs may actu-
ally perpetuate or extend the life cycles of low-value and low-tech exports
from developing countries to wealthier markets. Conversely, distance,
availability of air transportation, and high transportation costs may collec-
tively encourage the specialization in production of high-tech and other
high-value-to-weight goods for which transportation costs are not so rele-
vant, but timely delivery and availability during high-demand seasons are.9
After all, air transportation enables factor mobility, not only of capital but
also labor (Button & Vega, 2008), to the regions producing these goods.10
As a consequence, the effects of factor mobility could potentially affect the
traditional comparative advantages enjoyed by some regions, and the ana-
lysis of development policies need to take into consideration intended as
well as possible unintended consequences of such interventions.
A word of caution needs to be introduced before encouraging policy-
makers to engage in investments in air transportation infrastructure, how-
ever. New investments in airports may not translate into higher economic
growth, since infrastructure alone does not affect mobility of people and
goods; airplanes and airlines operating them, qualified pilots, freight for-
warders, among other elements, are also needed.11 In addition,
230 HENRY L. VEGA
transportation services interact as well with other networks such as com-

munications, information, and bureaucracies. Investments in moderniza-
tion of customs procedures (e.g., standardization of regulations), and
liberalization of insurance markets may have a greater effect at facilitat-
ing the flow of trade than building a second airport warehousing facility.
The research work by Vega (2008) has provided evidence that although
higher air transportation costs to move P&E negatively affect the levels of
trade between developing producing countries and the United States, high
air transportation costs do not seem to have a significant effect on the
movement of high-value goods such as high-tech products. The findings
suggest that the terms of trade of goods shipped by air might be the conse-
quence of the composition of trade, rather than their transporting costs.
Thus, while nothing can be done regarding the distance from developing
countries to the international markets, a lot could be achieved by enacting
public policy to eliminate institutional constraints such as restrictive regula-
tions of air navigation markets. Despite arguments against limiting access
to air space, in reality countries treat air access between them as a tradable
service based on the premise that governments, not the consumers, have a
say on the conditions of the trade. These conditions are specified under one
country’s bilateral air services agreement (ASA) with another. These agree-
ments can be burdensome in some cases and their compliance usually repre-
sents a challenge for developing countries. They can be very restrictive since
they regulate market access for air carriers, pricing, scheduling, and access
for a third country’s air carriers. The alternative, full liberalization of avia-
tion markets, however, has often been preceded in developed countries by a
backdrop of successful liberalization of many other markets, strong eco-
nomic growth, and well defined national objectives. With the exception of
China and some other Asian countries, perhaps these conditions are not
always met by developing countries. Some optimism remains as developing
countries and international financial and development institutions increas-
ingly recognize the role of air transportation in a country’s long-term eco-
nomic growth. After all, air transportation does not require excessive
investments in inflexible, fixed infrastructure, as is the case with rail.
Distinguishing airborne export flows between perishables & exotics and
high-tech products makes it possible to distinguish the effects of air freight
charges on exports of two different economic sectors, namely agricultural
goods and knowledge-intensive high-tech products. This provides a contri-
bution toward further exploring the role of air transportation in trade
levels and the possible influences of air transportation in a country’s com-
position of trade. Taking into account such distinction between exports of
P&E and high-tech, the statistics are noteworthy. Developing countries as

a group observed their exports of high-tech products increase by 31 per-
cent to the United States annually and 36.4 percent to the EU between
2000 and 2006.
The results of the empirical analyses performed using descriptive trade
statistics have shown that transportation costs tend to be related to dis-
tance and they tend to be lower for lower-cost products, such as P&E, than
for high-tech products that travel by air when they are assessed in dollars
per kilogram. However, when transportation costs are measured as a per-
centage of cargo value, they tend to be higher for lower-cost products, such
as P&E, than for high-tech products. That is, distance is only one factor
negatively influencing trade levels and its importance can be further
debated on the grounds of technical aspects such as statistical significance.
Distance, for example, does not appear to significantly affect exports of
high-tech goods from developing countries. On the other hand, policy
issues such as the existence of an Open Skies agreement between trading
partners seem to influence exports of P&E products positively.
In the “missing trade” debate among development economists, the
relevance or irrelevance of a country being either landlocked or an island
generally supports the claim that these countries are at a competitive disad-
vantage in the world market. However, such quick conclusions, usually
based on aggregate data analyses, do not seem to always hold. That a coun-
try is landlocked seems to have a negative effect on its level of exports,
although not significantly after correcting for economic cluster. On the other
hand, when a country is an island it seems to be in a better position to com-
pete in the world market of P&E goods. Perhaps this is due to the fact that
in many cases island economies have developed highly efficient air transport
infrastructure in order to improve their links to the rest of the world.
Another frequently and highly debated issue on international trade
research in the last two decades has been the importance of transportation
or freight costs. Many development economists have claimed that trade
levels involving developing countries are suboptimal and they attribute this
to high transportation costs. The results presented in this chapter provide
some support for these claims but also undercut the case for such broad
generalizations. When distinguishing between high-tech goods and P&E
products, for example, the coefficient of the air freight rate has the expected
negative effect on the value of exports of both sets of goods. However, the
magnitude of the air freight rate’s effect is larger for the trade flow of high-
tech goods than for that of P&E goods. Thus, exports of P&E seem to ben-
efit from competitive air freight charges (although not the norm, in some
232 HENRY L. VEGA
cases perhaps due to excess supply of ACS or as the consequence of

increased scheduled passenger) that have allowed products with relatively
lower value-to-weight ratios to be exported to international markets.
Policymakers need to be aware of this before establishing their goals to
reduce transportation costs by some measure. It needs to be noted that sin-
gling out a country’s geography, marked trade imbalance, or high air
freight rates, as the causes for low air freight-based trade may not be
appropriate in future research. Often, in developing countries, issues affect-
ing the air transportation sector are just another reflection of the overall
political, regulatory, and institutional environment that affects most sectors
of developing economies. This includes an absence of antitrust laws and
regulations addressing issues of competition.
NOTES
1. Anderson and van Wincoop (2004) provide a summary of research on trade

costs, while Hummels (2007) provides insights on how improved transportation sys-
tems can affect the growth and the quality of the trade between countries.
2. A shipment can be handled for multiple airlines, two or more forwarders,
customs brokers, warehouse operators and trucking firms.
3. Under regulation, for instance, Federal Express was forced to fly very small
aircraft such as the Falcon 20. Deregulation allowed Federal Express to fly larger
aircraft such as the Boeing 727 (Kiesling, 1995).
4. Kasarda and Green (2005) found that a country’s air cargo volume can pre-
dict either GDP or GDP per capita with over 90 percent accuracy given mutual
causality.
5. Zhang and Zhang (2002) provide a discussion of some of these issues and
state that freight grows slightly faster than the growth in international trade volume
and between 1.5 and 2 times the rate of global GDP growth.
6. The groups are: low income, $905 or less; lower middle income, $9063,595;
and upper middle income, $3,59611,115. The use of the term developing is conve-
nient; it is not intended to imply that all economies in the group are experiencing
similar development or that other economies have reached a preferred or final stage
of development. Products considered P&E are all those catalogued under the Trade
Harmonized System’s chapters 01 (live animals), 02 (meat and edible meat offal), 03
(fish and crustaceans, mollusks and other aquatic invertebrates), 06 (live trees and
other plants; bulbs, roots and the like; cut flowers and ornamental foliage), 07
(edible vegetables and certain roots and tubers), 08 (edible fruit and nuts; peel of
citrus fruit or melons), 09 (coffee, tea, maté and spices), and 12 (oil seeds and oleagi-
nous fruits; miscellaneous grains, seeds and fruits; industrial or medicinal plants;
straw and fodder). Products considered high-tech in this analysis are all those cata-
logued as Advanced Technology Products according to the 2006 definitions of the
Foreign Trade Division, US Census Bureau. They include the following sectors:
(01) Biotechnology; (02) Life Science; (03) Opto-Electronics; (04) Information &
Communications; (05) Electronics; (06) Flexible Manufacturing; (07) Advanced
Materials; and (08) Aerospace.
7. Hummels (2007) provides an overview of the advantages of this dataset as
well as the limitations of other series data on international shipping charges.
8. Several model specifications were run in a pre-analysis which included differ-
ent sets of variables to assess their effects and their interaction effects. Because of
this, a dummy variable for the existence of a bilateral Open Skies agreement (OSK)
was included, although air cargo is often liberalized prior to such agreements taking
place. Also, the level of airport infrastructure is assumed to be captured by the
GDP per capita variable. Unfortunately, the macroeconomic data used for the ana-
lysis did not allow distinguishing between shipments transported by integrators ver-
sus dedicated cargo versus passenger airlines.
9. Development of high-tech and knowledge industries is currently perceived as
more conducive to sustainable and long-term economic growth than the traditional
agriculture/manufacturing/services model.
10. Factor mobility relates to the viability of moving factors of production
(labor, capital, or land) from one industry sector to another. It entails factor mobi-
lity not only between firms but also within an industry sector, across sectors within
a country, between countries within sectors, and also across sectors. Academic
research on factor mobility has grown in the last few decades; as a possible result of
an increased interest in understanding globalization’s causes and effects, including
increasing returns to scale, imperfect markets, multinational corporations, labor
migration, international development, and poverty reduction.
11. Policymakers are not always aware of the adverse effects of opening up trade
when a very poor area has no comparative advantage. In these cases, commonly
referred to as the Appalachians effect, even though neoclassical economic theory
predicts that factor mobility will lead to regional convergence of per capita income
in the long run; because of increasing returns to scale, factor mobility, in conjunc-
tion with differences in the production systems between two regions, leads to a pro-
cess of agglomeration in the wealthier region. The Appalachian region is often used
as an example of such a region. Appalachia, which was largely dependent on a sin-
gle sector, coal-mining, is located close to several growing urban areas such as
Atlanta, Nashville, and Washington, D.C. This proximity resulted in the region
being impoverished as the result of the outmigration of its mobile skilled and
higher-earning population. Immobile individuals that stayed in the region subse-
quently experienced a depression in their standards of living as a result of the insig-
nificance of the Appalachian market following the outmigration (see for example:
Buchanan & Yoon, 1994 & Bhagwati, 2001).
REFERENCES
Anderson, J., & van Wincoop, E. (2004). Trade costs. Journal of Economic Literature, 3(42),
691751.
ATAG. (2005). The economic & social benefits of air transport. Geneva: Air Transport Action
Group.
234 HENRY L. VEGA
Bhagwati, J. N. (Ed.). (2001). Selected readings international trade (6th ed.). Cambridge, MA:
Massachusetts Institute of Technology.
Bowen, J. T. (2002). Network change, deregulation, and access in the global airline industry.
Economic Geography, 78(4), 425439.
Bowen, J. T., & Leinbach, T. R. (2003). Air cargo services in Asian industrializing economies:
Electronics manufacturers and the strategic use of advanced producer services. Papers
in Regional Science, 82(3), 309332.
Bowen, J. T., Leinbach, T. R., & Mabazza, D. (2002). Air cargo services, the state and indus-
trialization strategies in the Philippines: The redevelopment of Subic Bay. Regional
Studies, 36(5), 451467.
Buchanan, J. M., & Yoon Y. J. (Eds.). (1994). The return of increasing returns. Ann Arbor,
MI: University of Michigan Press.
Button, K., & Vega, H. (2008). The effects of air transportation on the movement of labor.
GeoJournal, 71(1), 6781.
Chang, Y. H., & Chang, Y. W. (2009). Air cargo expansion and economic growth: Finding the
empirical link. Journal of Air Transport Management, 15(5), 264265.
Cline, R. C., Ruhl, T. A., Gosling, G. D., & Gillen, D. W. (1998). Air transportation demand
forecasts in emerging market economies: A case study of the Kyrgyz Republic in the
former Soviet Union. Journal of Air Transport Management, 4(1), 1123.
Fennes, R. J. (1997). International air cargo transport services: Economic regulation and policy.
Ph.D. Dissertation. Leiden University, Leiden.
Hummels, D. (2007). Transportation costs and international trade in the second era of globali-
zation. Journal of Economic Perspectives, 21(3), 131154.
InterVISTAS-ga2. (2006). The economic impact of air service liberalization. Washington, DC.
Kasarda, J. D., & Green, J. D. (2005). Air cargo as an economic development engine: A note on
opportunities and constraints. Journal of Air Transport Management, 11(6), 459462.
Kiesling, M. K. (1995). A comparison of freight distribution costs for combination and dedicated
carriers in the air express industry. Ph.D. Dissertation. University of California,
Berkeley.
Leinbach, T. R., & Bowen, J. T., Jr. (2004). Air cargo services and the electronics industry in
Southeast Asia. Journal of Economic Geography, 4(3), 229231.
Radelet, S., & Sachs, J. (1998). Shipping costs, manufactured exports and economic growth.
Paper presented at the American Economic Association Meetings, Chicago.
US Department of Commerce. (2006). “Brazil,” Washington, DC: Western Hemisphere diversifi-
cation and defense market guide. Retrieved from http://www.bis.doc.gov/Defen
seIndustrialBasePrograms/OSIES/ExportMarketGuides. Accessed on July 22, 2006.
Vega, H. L. (2008). Air cargo, trade and transportation costs of perishables and exotics from
South America. Journal of Air Transport Management, 14(5), 324328.
Vega, H. L. (2010). Developing countries and their airborne export flows of perishable and high-
tech goods. Doctoral Dissertation. George Mason University.
von Heereman, R. (2006). Air transport of perishables Economic challenges in a highly compe-
titive market. Montreal: ICAO/ATAG/WB Development Forum Maximizing Civil
Aviation’s Economic Contribution.
Ying, Y. H., Chang, C. P., & Hsieh, M. C. (2008). Air cargo as an impetus for economic
growth through the channel of openness: The case of OECD Countries. International
Journal of Transport Economics, 35(1), 3144.
Zhang, A., & Zhang, Y. (2002). Issues on liberalization of air cargo services in international
aviation. Journal of Air Transport Management, 8(5), 275287.
CHAPTER 9
AN ASSESSMENT OF THE CAUSAL

RELATIONSHIP BETWEEN AIR
PASSENGER TRAFFIC AND TRADE
IN ASIA-PACIFIC
Elien Van De Vijver, Ben Derudder and

Frank Witlox
ABSTRACT
During the last few decades, rising intra-regional volumes of trade as
well as air passenger traffic have been key characteristics of Asia-
Pacific’s economic development. Although conceptual and empirical lin-
kages between rising levels of trade and air passenger flows are often
assumed, relatively little is known about the potential causality in these
parallels. In this chapter, we seek to empirically uncover this causality
through the application of heterogeneous Time Series Cross
Section Granger causality analysis for the period 19802010. Four
scenarios are found amongst the different country-pairs: (1) there is no
co-evolution, implying that both patterns develop independently (e.g.
JapanAustralia); (2) there is ‘real’ co-evolution in that both patterns
influence each other through feedback loops (e.g. South
KoreaPhilippines); (3) air passenger traffic is facilitated by trade

ISSN: 2212-1609/doi:10.1108/S2212-160920140000004008
235
236 ELIEN VAN DE VIJVER ET AL.
(e.g., South KoreaPhilippines); or (4) trade is facilitated by air pas-

senger traffic (e.g. AustraliaMalaysia). Some possible interpretations
of this heterogeneity are discussed.
Keywords: Granger causality; Asia-Pacific; trade; air passenger travel
JEL classifications: C330; F140; O530; O560; R490
INTRODUCTION
Asia-Pacific has long been perceived as a region that was in the ‘initial’
stages of economic development compared to Western Europe and North
America. Since the 1980s, however, the region has witnessed strong eco-
nomic growth figures, which dwindled only temporarily after the Asian crisis
of 1997 and the global financial crisis of 2008. For instance, between 1980
and 2006, the region’s GDP grew more than five times, and its share in
world trade increased from 15.0% to 23.5% (Athukorala, 2010; Ozeki,
2008). Both figures are thereby closely linked, as economic development in
the Asia-Pacific region relies heavily on export-based industrialization
(O’Connor, 1995). These growth figures have led to the labeling of the 21st
century as the ‘Asian Century’ (Athukorala, 2010). It is clear that Asia-
Pacific’s global economic integration has been facilitated by, and is reflected
in profound intra-regional economic integration. The ongoing liberalization
of trade and investment regimes in Asia-Pacific countries has been the main
trigger for this intra-regional economic integration (Soesastro, 2006). For
instance, average tariff rates in the region have diminished during the last
few decades in the Philippines, Thailand, Japan and South Korea, rates
have more than halved over the last decades (Athukorala, 2010). Non-tariff-
barriers have equally diminished significantly, and some countries are taking
economic liberalization yet a step further by establishing regional free trade
agreements in order to further strengthen the economic links with favored
regional partners. ASEAN, for example, aims at realizing completion of the
ASEAN Free Trade Area (AFTA) by 2015.
The ongoing regional economic integration in Asia-Pacific has been par-
alleled by rising levels of intra-regional air transport connections (Rimmer,
2000). Especially since the mid-1980s, Asia-Pacific has been the world’s
fastest growing air passenger and freight market (Bowen, 2000; O’Connor,
1995; Shin & Timberlake, 2000). In the years preceding the Asian financial
Causal Relationship Between Air Traffic and Trade in Asia-Pacific 237
crisis of 1997, IATA estimated that 30% of global international scheduled

air passenger traffic was to, from or within the Asia-Pacific (Graham,
1995). At that time, it was forecasted that by 2010, almost 43% of the
world’s air traffic would be situated in the region (Rimmer, 2000).
However, the crisis, followed by 9/11, and the SARS epidemic in 2003, tem-
porarily restrained growth. Nevertheless, in 2010, it was announced that
for the first time in history, the volume of air passenger travel within Asia-
Pacific surpassed that in North America (IATA, 2010). As with economic
integration proper, this strong expansion of air transport in the region goes
hand in hand with more liberal policies, with countries making access to
their airports progressively easier from the 1980s onwards (Doganis, 2010).
The degree of ‘openness’ of a country’s aviation market in past and present
is of course prone to inter-country variability, depending upon the national
governments’ development goals. According to Bowen (2000), the extent to
which the Asia-Pacific states attempted orchestration of the airline industry
has indeed been explicitly guided by the purpose of accelerating economic
development. Singapore, for instance, has followed ‘Open Skies’ policies
successfully since the 1960s in order to develop as an international hub for
commercial activities.
Although the parallel developments of economic integration and grow-
ing air traffic connectivity in the Asia-Pacific region have been widely
acknowledged, relatively little is known about the actual interaction
between them. As a consequence, to date the interrelations between the two
have been asserted rather than shown. The purpose of this chapter is to
address this lacuna by shedding some light on the empirical contours of
these interrelations. Figs. 1 and 2 show two very different examples. In
Fig. 1, which shows the evolution of trade and air transport connections
between South Korea and the Philippines, we see that a 19971999 slump
in seats seems to precede a leveling out of trade, just as the subsequent rise
of seats (from 2000 onwards) takes a while to be translated into extra trade.
For this specific country-pair, this would point to a pattern where changes
in air transport often precede, and therefore may ‘cause’ changes in trade.
In Fig. 2, which shows the evolution of trade and air transport connections
between Australia and Malaysia, we can observe the opposite effect: we see
a relative surge in trade between 1993 and 1997, which is followed by a
similar surge in scheduled seats some years later (19982002). Similarly,
there was a very steep increase in trade in 2006, after which there was a
sharp surge in seats in 20082009. For this specific country-pair, then, this
would point to a pattern where changes in trade precede, and thus ‘cause’
changes in air transport connections.
South Korea - Philippines

2500
2000
1500
%
Trade
1000
Seats
500
0
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
Fig. 1. An Example of Changes in Number of Scheduled Seats Facilitating
Changes in Trade Volumes between Two Countries (South Korea and Philippines).
In Figs. 1 and 2, the presence of a temporal lag in the levels of change

intuitively suggests the presence of specific causal relationships between
growth in trade and growth in air transport connectivity. In this chapter,
we apply a methodology that provides statistical evidence of the presence
of this intuitive ‘causal’ relationship: we scrutinize the alleged co-evolution
of trade and air passenger traffic for Asia-Pacific as a whole and for
country-pairs within this region by performing Granger causality tests on
time series cross-section (TSCS) datasets running from 1980 to 2010. In
principle, there are four possible scenarios: (1) there is no co-evolution so
that both patterns develop independently; (2) there is ‘real’ co-evolution in
that both patterns influence each other through feedback loops; (3) intra-
regional trade is facilitated by air passenger traffic (e.g. Fig. 1); or (4) intra-
regional air passenger traffic is facilitated by intra-regional trade
(e.g. Fig. 2). Although one of these scenarios may occur for the region as a
whole, intra-regional variation is expected, because of the varying liberali-
zation policies in terms of both trade and air transport across Asia-
Pacific. Standard TSCS-Granger testing fails to predict this regional hetero-
geneity, due to its associated homogeneous causality assumptions.
Consequently, we use an extension of the general method, which allows
heterogeneous variability to occur in the relationship between air traffic
and total trade across the region.
Australia - Malaysia
1600
1400
1200
1000
%
800
Trade
600 Seats
400
200
0
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
Fig. 2. An Example of Changes in Trade Volumes Facilitating Changes in
Number of Scheduled Seats between Two Countries (Australia and Malaysia).
The remainder of this chapter is organized as follows. First, we discuss

the evolution of total trade and air passenger traffic in Asia-Pacific, and
evaluate their possible relationships. Second, we introduce our data and
discuss our research methodology. We then discuss our results in a third
section, and end with a summary of the main conclusions, indicating some
possible avenues for future research.
LITERATURE REVIEW
Trade and Air Passenger Travel in Asia-Pacific
The Asia-Pacific region has been characterized by an expansion of intra-

regional trade, especially from the 1970s onwards. This growth in trade
flows is to a large extent triggered by the emergence of regional produc-
tion networks. These are based on differences in amongst other things
production and labour costs (Ozeki, 2008), inducing a regional division of
labour. The differences in costs stem from the large differences in develop-
ment levels between the Asia-Pacific countries. This causes multinational
corporations to shift labour-intensive parts of the production process to
countries with lower wages and more knowledge intensive components to
more developed countries such as Singapore (Athukorala, 2010; Yamazawa,

1992). This ‘flying geese’-process first began in the 1970s with Japan as core
country and the NICs of South Korea, Taiwan, Singapore, and Hong Kong
as receiving countries. However, as the latter countries rose on the develop-
ment ladder, the pattern gradually shifted to the NICs as core countries and
ASEAN countries as receivers. Later on, the same patterns occurred among
the ASEAN countries and, more recently, also between ASEAN countries
and China and the South Asian countries (Yamazawa, 1992). The accompa-
nying heightened levels of intra-regional trade went hand in hand with gov-
ernments’ policies of increasing trade and investment liberalization within
the region, but also outside the region (Soesastro, 2006; Yamazawa, 1992).
Hence, the intra-regional integration was and still is continually accompa-
nied by global economic integration (Ng & Yeats, 2003; Soesastro, 2006),
which is characterized by large trade flows with countries outside the region
(Ozeki, 2008). Although intensive de facto intra-regional economic integra-
tion has been going on for the past couple of decades, institutional integra-
tion has been lagging far behind. The formation of ASEAN in 1967 was the
first attempt at institutionalization in the region, but ASEAN economic
cooperation progressed rather slowly. This was mainly due to the differences
in development levels of the ASEAN countries, which prevented them from
cooperating because of conflicting interests. In 1977, the Agreement on
ASEAN Preferential Trading Agreements (PTA) was signed, which was a
first joint effort to liberalize intra-regional trade. However, this agreement
has had limited impact (Imada, 1993). Another milestone was the formation
of the ASEAN Free Trade Agreement (AFTA) in 1992. Again, progress has
been rather slow, with zero tariff rates across the whole ASEAN only to be
achieved in 2015. It also took time for other free trade agreements (FTAs)
to be concluded in the Asia-Pacific region. These were, and still are, often
bilateral. The first one was signed between Thailand and Laos in 1991. A
decade later, a proliferation of bilateral FTAs has proceeded in Southeast
Asia, starting with Singapore and New Zealand in 2000 (Chandra, 2005).
These developments mainly occurred because the Asian crisis of 1997 incited
a wish for more regional economic cooperation (Lee & Park, 2005).
However, critics state that the FTAs in the region are rather ‘trade-light’, in
that they are weak and partial agreements (Sally, 2010). They are generally
limited to preferential tariff cuts on a limited range of goods, hampering
true regional economic integration.
The rising trade flows in the region have been paralleled by a strong
growth in air passenger travel. Several factors contribute to this expansion:
(1) the region’s geographical fragmentation and extent; (2) its large
population numbers; and (3) its fast economic development, which implies
rising income levels and heightened intra-regional demand (Chin, Hooper, &
Oum, 1999; Doganis, 2010; Graham, 1995). Although business travel predo-
minated in the beginning, tourism is now accounting for the bulk of air pas-
senger travel in the region (Findlay & Forsyth, 1992; Rimmer, 2000). The
strong growth in air traffic during the past three decades is facilitated by the
liberalization of the air travel market through privatization and deregula-
tion. Until the early 1990s, the Asia-Pacific market was tightly regulated by
restrictive bilateral agreements (Rimmer, 2000). These were accompanied by
every country designating one, mostly government owned, airline as their
international airline (Findlay & Forsyth, 1992). Progressively, this pattern
was interrupted by countries such as Hong Kong, Indonesia and South
Korea, which opened up competition on their international routes.
Furthermore, several countries partly privatized their former national air-
line, the first being Singapore and Malaysia in 1985 (Bowen, 2000). Others
such as Japan, Thailand and Philippines soon followed (Findlay & Forsyth,
1992). However, up till now, governments still own the lion’s share of the
property rights in most countries. It must be noted that deregulation of the
market has paved the way for the birth of LCCs, which are also flying on a
couple of intra-regional routes, contributing positively to the expansion of
air travel in the region (Forsyth, King, & Rudolfo, 2006). In general, the
regional liberalization has nonetheless proceeded slowly, because the
attempts have been often localized and uncoordinated, for example between
neighbouring countries (Doganis, 2010). This resulted in a rather regulated
Asia-Pacific airline market in the 21st century, with albeit more open
bilateral agreements as core components. A few milestones were attained,
such as the AustraliaNew Zealand Single Aviation Market Arrangements
established in 1996 (Findlay & Goldstein, 2004). Also regional ‘Open Skies’
agreements were signed in the 21st century, such as the MALIAT agreement
between a few Pacific states in 2001 and the ASEAN Multilateral
Agreement on Air Services in 2009, which entailed a gradual implementation
of more ‘freedoms of the air’ in the participating countries and an ASEAN
‘Open Sky’ by 2015. (Derudder, 2012) However, implementation of the lat-
ter has its problems, again because of the differences in development levels
of the constituting countries (Grosso, 2012). There are also some subregio-
nal agreements in the region that have very limited Open Skies characteris-
tics, such as the IndonesiaMalaysiaThailand Growth Triangle (IMT-
GT) (Forsyth et al., 2006). In spite of these attempts, bilateral agreements
continue to dominate. This results in a region that is characterized by a
wide variation in policy regimes, as was indicated by Findlay and Goldstein
(2004), ranging from the highly liberalized Open Skies policy in Singapore
to the rather regulated policies of Japan.
The Relationship between Trade and Air Passenger Travel
Several authors confirm parallel developments between economic integra-

tion and growing air traffic connectivity. Graham (1995) states that the
growth in air transport is both a cause and effect of the economic develop-
ment in the Asia-Pacific region. Shin and Timberlake (2000) and Doganis
(2010) also recognize that the rapid Asian economic growth, based on
export-oriented industries, is closely related to the development of the
Asian airline industry. After all, high quality air transport facilitates trade
between regions or countries, and the movement of factors of production
such as labour and raw materials (Button & Yuan, 2013). Bowen (2000)
equally sees air transportation as an enabling mechanism, an exogenous
factor that has permitted countries to achieve a high level of access to glo-
bal markets and foreign investments. Hence, states exert a great influence
on the relationship between air transportation and economic development
in Asia-Pacific through policies and strategies, which vary considerably
(Bowen, 2000).
Despite all these assumptions, the link between air transport and eco-
nomic integration, and its accompanying trade and investment flows, has
barely been studied. Only recently, a number of researchers studied the cor-
relation between travel and trade through gravity equation models. Poole
(2010) uses country-level gravity model regressions to estimate the impact
of bilateral international business travel on bilateral international trade
between the United States and other countries. The author concludes that
business travel acts as an input to international trade. Cristea (2011) com-
plements this work, by proposing an input demand estimation approach, in
which business-class air travel is a direct function of the bilateral export
flows between the United States and several of its trade partners. Although
these kinds of investigations indicate a strong relation between air passen-
ger travel and international trade, the used methodology does not permit
to discuss causal processes,1 due to its non-temporal character. Granger
causality analysis offers a solution to this shortcoming.
To the best of our knowledge, the paper by Kulendran and Wilson
(2000) is the only thorough analysis of the causal relationship between inter-
national trade and international travel for Australia and four of its largest
travel and trading partners (The United States, The United Kingdom, Japan
and New Zealand). The authors employ time series Granger causality tech-
niques. No clear-cut conclusion could be drawn from their analyses: the
results range from a reciprocal relationship between total travel and total
trade for the AustraliaUSA link to rather unilateral relationships for the
other countries. For example, travel precedes trade in the JapanAustralia
link, but trade influences travel in the UKAustralia link. It can be noted
that the use of traditional time series Granger causality techniques limits
their analysis to pairwise comparisons: the relationship between Australia
and each of its trading partners is investigated separately. However, in order
to study larger samples of cross-sections simultaneously over a given time
period, time series cross-section (TSCS) Granger analysis offers more relia-
bility and more flexibility (Hood III, Kidd, & Morris, 2008). In this chapter,
we use an extension of TSCS-Granger causality analysis (Hurlin & Venet,
2001) that allows for cross-sectional variability, to decipher the causal rela-
tionship between total trade and air travel for the Asia-Pacific region and
several constituting country-pairs.
DATA AND METHODOLOGY
Data
We confine our empirical analysis to 9 countries in the Asia-Pacific region,

including the five founding members of ASEAN (Malaysia, Indonesia,
Thailand, Philippines and Singapore). In contrast with the other ASEAN
members, these countries have adopted liberal and well-developed policies
towards trade and the airline industry (Forsyth et al., 2006; Hiratsuka,
2006). Japan and South Korea are key players in the region. The former
has undeniably played a key role in the Asia-Pacific region’s economic inte-
gration, because of its history of outsourcing of industrial activities to other
countries in the region, creating large regional trade flows. The latter, being
one of the early NICs, more recently copied Japan’s role as core country in
this ‘flying geese’-process. New Zealand and Australia are other significant
economies in the region. They have always been actively involved in East
Asian regional trade agreements, in part out of fear to be isolated from the
wider region’s economy (Lee & Park, 2005). In addition, they generate
significant air passenger flows due to their geographically isolated location.
In principle, including 9 countries would theoretically result in n(n − 1)/2 =
36 country-pairs as cross-sections, but for three cross-sections
(ThailandNew Zealand, IndonesiaNew Zealand, and PhilippinesNew

Zealand), there were insufficient observations. For the 33 remaining
country-pairs, we calculated the number of scheduled seats and the total
trade volume.
The transport data were derived from the Official Airline Guides (OAG)
database, which contains the number of scheduled seats on direct flights
between airports. Information on the connections of low-cost carriers, a
sector that continues to expand in Asia-Pacific, is included. In order to
obtain passenger flows between the different countries from our analysis,
we aggregated the available seats from/to all of the airports in a given
country (e.g. aggregating OsakaSingapore and TokyoSingapore seats).
The trade data were collected through the United Nations Commodity
trade (Comtrade) statistics database (http://comtrade.un.org). This data-
base contains detailed import and export statistics, reported by the statisti-
cal authorities of about 200 countries worldwide. We aggregated the value
of exports and imports of both finished goods, and parts and components
for each of the country-pairs to estimate the total volume of trade.
Granger-Methodology
To decipher the ‘causal’ relationship between total trade and air passenger
travel in the Asia-Pacific region between 1980 and 2010, the method of
time series cross-section (TSCS) Granger causality testing is employed
(Granger, 1969). Causality in this sense refers to a precedence of one vari-
able to another. A variable X is said to ‘Granger cause’ a variable Y, if tak-
ing into account past values of X enables better predictions of Y than
based exclusively on past values of Y. The variable X does not literally
‘cause’ Y, but it helps to forecast it.
TSCS-Granger causality testing allows for scrutinizing multiple cross-
sectional units, for example country-pairs, simultaneously over a given time
period. It offers several advantages in comparison with the conventional
Granger time series model: more observations are incorporated and more
meaningful and efficient results are obtained (Hood III et al., 2008; Hurlin
& Venet, 2001). We use an extension of the standard TSCS-Granger model,
which allows for heterogeneity among the different cross-sections (Hurlin
& Venet, 2001). The large diversity in causal relations between total trade
and air passenger travel was already indicated in the study of Kulendran
and Wilson (2000). This heterogeneity emerges from the large differences in
economic integration and transport connectivity the Asia-Pacific region
exhibits, and additionally from the different stances national governments

have towards deregulation of trade and air travel.
This extension of the standard TSCS-Granger model can be expressed
as:
X
p X
p
yi;t = ai þ γ k yi;t − k þ βi;k xi; t − k þ ui; t ð1Þ
k=1 k=1
In which ai are the fixed effects, γk and βki represent the autoregressive
and regression coefficients respectively, yi,t-k and xi,t-k the lagged values of
the dependent and independent variables respectively, ui,t the error term,
and p the number of time lags.
The assumption underlying this extension is that the autoregressive coef-
ficient is constant for all cross-sections, while the regression coefficient is
constant for all time lags but can vary across the cross-sections. This
addresses the problem of causal heterogeneity at the level of cross-sections
because it allows for different relationships to occur among in this
case total trade and travel across the cross-sections. The Granger causal-
ity procedure of Hurlin and Venet (2001) consists of three consecutive steps
(Hood III et al., 2008), which will be elaborated in the results section.
Prior to the analysis of the trade and air travel data, two preliminary
steps are needed. First, Granger analysis requires time series that are
stationary, implying that they have a constant mean and variance
(Lütkepohl & Krätzig, 2004). Data series that are non-stationary contain a
so-called ‘unit root’, a trend that causes a spurious regression and generates
unreliable results. Additionally, the time series are required to have the
same order of integration. Two different unit root tests, specifically
designed for TSCS-data, are implemented: the Levin, Lin and Chu test
(Levin, Lin, & Chu, 2002), and the Im, Pesaran and Shin procedure (Im,
Pesaran, & Shin, 2003). Both tests suggest that the original data for total
trade and number of seats are indeed non-stationary. This is due to the
often-sharp changes in the volume of trade and the number of scheduled
seats between the different countries throughout the 19802010 period.
The standard way of dealing with this problem is using first differences to
make the time series stationary, representing yearly growth in total trade
and yearly increase in scheduled number of seats between two Asia-Pacific
countries in the period 19812010.
Second, an appropriate time lag needs to be chosen from the yearly
data. This time lag is specified as the time difference with the maximum
level of ‘causality’. This can be calculated through the Schwarz-information
criterion, which indicates that a time lag of 4 years in both directions yields
the best results. All analyses are performed in the EViews-software.
RESULTS AND DISCUSSION
We already indicated in the introduction that there are four possible causal-
ity scenarios: (1) no causality; (2) bidirectional causality; (3) an increase in
number of scheduled seats influences growth in total trade; or (4) growth in
trade facilitates an increase in air passenger traffic. Hence, these possibili-
ties require two separate equations, using the first-order differences, and a
4-year time lag:
Δseatsit = αi þ γ i;t-1 Δseatsi;t-1 þ γ i;t-2 Δseatsi;t-2 þ γ i;t-3 Δseatsi;t-3

þ γ i;t-4 Δseatsi;t-4 þ βi;t-1 FEi Δtradei;t-1 þ βi;t-2 FEi Δtradei;t-2 ð2aÞ
þ βi;t-3 FEi Δtradei;t-3 þ βi;t-4 FEi Δtradei;t-4 þ ui;t-1 þ ui;t-2 þ ui;t-3 þ ui;t-4
Δtradeit = αi þ γ i;t-1 Δtradei;t-1 þ γ i;t-2 Δtradei;t-2 þ γ i;t-3 Δtradei;t-3

þ γ i;t-4 Δtradei;t-4 þ βi;t-1 FEi Δseatsi;t-1 þ βi;t-2 FEi Δseatsi;t-2 ð2bÞ
þ βi;t-3 FEi Δseatsi;t-3 þ βi;t-4 FEi Δseatsi;t-4 þ ui;t-1 þ ui;t-2 þ ui;t-3 þ ui;t-4
where FEi is an array of dummy variables (fixed effects) for each cross-
section. The first step of the heterogeneous Granger TSCS-analysis tests for
the ‘homogeneous non-causality’ across Asia-Pacific: the very presence of
causality is tested across an aggregation of all cross-sections. This implies
formulating a null hypothesis for both equations, stating that there is no
causal relationship between the evolution in trade (seats) and the evolution
in seats (trade) across Asia-Pacific.
H1a. For all country-pairs, Δtrade does not Granger cause Δseats.
H1b. For all country-pairs, Δseats does not Granger cause Δtrade.
The null hypotheses are each tested with an F-test statistic:
ðRSS2 − RSS1 Þ=ðNpÞ

F:
RSS1 =½NT − N ð1 þ pÞ − p
With N representing number of cross-sections, p the number of time

lags, and t the number of time periods.
Interpretation of the test relies on cross-checking this value with an F-

distribution with Np and NT − N(1 + p) − p degrees of freedom for the
numerator and the denominator respectively. The test compares the sum
of squared residuals of a restricted model (=RSS2, i.e. without taking
change in X into account by assuming that the regression coefficients
βi,t-k = 0) with the sum of squared residuals of an unrestricted model
(=RSS1, i.e. taking change in X into account) presented in Eqs. (2a) and
(2b). The results of the first F-test indicate that both null hypotheses can
be rejected at the 1% (p = 0.00) significance level, implying bidirectional
causality. For the Asia-Pacific region, there is evidence that growth in
total trade precedes growth in number of seats and vice versa: there seems
to be symbiotic effect where trade fuels growth in air transport and the
other way round.
However, this does not mean there is a bidirectional relationship
between growth in air passenger travel and total trade for every pair of
countries in Asia-Pacific. This can be established in a second step, in which
the homogeneity of the causality relationship amongst the different
country-pairs is tested. The null hypothesis states that causality can be
found in each of the 33 country-pairs.
H2a. For all country-pairs, Δtrade causes Δseats.
H2b. For all country-pairs, Δseats causes Δtrade.

The F-test now contains a new restricted model (RSS3) that does not set
the regression coefficients to zero, but equal to each other (βi,t-1 = βi,t-k) for
all country-pairs. If the null hypothesis can be rejected, there is causal het-
erogeneity, implying a causal relationship is only present for a subsection
(1 to n − 1) of country-pairs. The test results indeed indicate causal hetero-
geneity for both directions, respectively at the 1% (p = 0.00) and 10%
(p = 0.07) significance level. This implies that, although the region as a
whole shows signs of a bidirectional relationship between total trade and
air passenger travel, this is not applicable to each country-pair.
Hence, additional tests are needed to discover for which country-pairs a
Granger causal relationship exists, and in which direction(s) this relation-
ship runs. This implies testing each of the cross-sections separately, which
is done in the third and final step.
H3a. For country-pair i, Δtrade does not cause Δseats.
H3b. For country-pair i, Δseats does not cause Δtrade.

Table 1 (Panels A and B) gives the p-values of the F-tests for all the
separate country-pairs. For those country-pairs where the p-value is smaller
than 0.10, we assume that a significant causal relationship is present. These
values are printed in bold.
The results indicate a diverse pattern of causal relationships across the
region, and below we will discuss some notable overall patterns. On a gen-
eral level, it is clear that total trade and air travel between Asia-Pacific
countries do influence each other, albeit in different ways and to varying
extents. A total of 19 out of 33 country-pairs show statistically significant
signs of causality, of which two (PhilippinesSingapore and South
KoreaMalaysia) exhibit bidirectional influence2.
First, growth in trade and growth in seats are not interdependent
between economically ‘developed’ countries such as Australia, Japan,
New Zealand and South Korea. One explanation could be their similar
high development levels, which prevent companies based in these coun-
tries from engaging in the shared production networks that have been
mentioned in the literature review. These networks are primarily based
on differences in labour costs, and have been the main triggers of the
dramatic expansion of trade flows in the Asia-Pacific region during the
last three decades (Athukorala, 2010; Ozeki, 2008). This is in sharp con-
trast with the causality linkages between developed and less-developed
economies. Australia and South Korea, for instance, have a series of sta-
tistically significant causality linkages (mostly running from seats to
trade) with some of the less-developed economies in our sample, such as
Thailand and the Philippines. Our analysis suggests that the creation of
new air passenger connections between, say, Australia and Thailand, has
facilitated access to and knowledge about markets that translated into
the growth of trade in the subsequent years. A complementary explana-
tion is the fact that any relation between change in trade and change in
air passenger connections will be complicated because the motivation for
air travel is multifaceted. Although there will be much trade-related travel
between Australia and New Zealand, it seems unlikely that this is the key
explanation to today’s strong air transport connections between cities in
Australia and New Zealand (Van De Vijver, Derudder, & Witlox, 2014).
For instance, it is estimated that today more than 650,000 New
Zealanders (or about 15% of the New Zealand population) currently live
in Australia, making it the second-largest group of foreign-born migrants
after the United Kingdom. Moreover, these numbers have quasi-
continuously grown over the past decades. This integration has been
facilitated by policies such as the 1973 Trans-Tasman Travel
Causal Relationship Between Air Traffic and Trade in Asia-Pacific
Table 1. P-values from the F-test for the Separate Cross-Sections (βi,t-k ≠ 0).
AU ID JP KR MY NZ PH SG TH
Panel A
AU − 0.25 0.54 0.15 0.01 0.88 0.84 0.24 0.68
ID − 0.22 0.77 0.00 − 0.99 0.09 0.38
JP − 0.43 0.93 0.86 0.31 0.01 0.67
KR − 0.02 0.72 0.84 0.05 0.64
MY − 0.00 0.05 0.05 0.07
NZ − NA 0.14 NA
PH − 0.01 0.12
SG − 0.27
TH −
Panel B
AU −
ID 0.00 −
JP 0.99 0.05 −
KR 0.41 0.10 0.73 −
MY 0.30 0.55 0.61 0.06 −
NZ 0.41 NA 0.18 0.78 0.92 −
PH 0.01 0.92 0.21 0.01 0.60 NA −
SG 0.78 0.48 0.85 0.89 0.49 0.13 0.03 −
TH 0.00 0.16 0.15 0.60 0.41 NA 0.05 0.04 −
Panel A: Null hypothesis: For country-pair i, Δtrade does not cause Δseats.
Panel B: Null hypothesis: For country-pair i, Δseats does not cause Δtrade.
AU = Australia, ID = Indonesia, JP = Japan, KR = South Korea, MY = Malaysia, NZ = New Zealand, PH = Philippines, SG =
Singapore, TH = Thailand.
Bolded values are used significant p-values (p < 0.10).
249
Arrangement, which has allowed Australian and New Zealand citizens to

enter each other’s country to visit, live and work, without the need to
apply for authority to enter the other country before travelling. Given
the relative vicinity of both countries and the rising importance of cross-
migration (albeit especially from New Zealand to Australia) in the face
of rising levels of disposable income, the main reason for air travel
between both countries alongside tourism is visiting family and friends.
Therefore, it should not be a surprise that there is no statistical causality
in the evolution of trade and seats. The main message here is that
although a multifaceted and complex linkage between motivations for air
travel and trade is probably present for all country-pairs, the linkage is
much more outspoken for some inter-country connections (Van De
Vijver et al., 2014).
Second, strong causality patterns emerge for Singapore and Malaysia,
and mostly run from change in trade to change in seats. Singapore
gained high importance as a regional ‘entrepôt’ economy, and still
remains a center for the export industry (Siddiqui, 2010). This can be
partly explained through the city-state’s central position in the Asia-
Pacific region, and its open trade and investment regimes. Additionally,
since the 1980s, Singapore has increasingly become an important (finan-
cial) service center. It serves as the regional headquarters of many multi-
national companies with subsidiaries in neighbouring Asian countries,
where the more labour-intensive parts of the production process are
located (Athukorala & Hill, 2010). Table 1 suggests that the enormous
trade flows to and from Singapore have been translated into air passen-
ger connections. This is undoubtedly related to Singapore’s early ado-
ption of very liberal approaches towards international air travel
connections (Bowen & Leinbach, 1996), which has allowed the quick
translation of demand into supply. To a lesser extent, also centrally
situated Malaysia constitutes one of the region’s forerunners in liberal-
izing its trade policies (Athukorala, 2010), while over the last decade it
has also become an important supplier of parts to other countries
(Kimura, 2008). Additionally, from 1993 onwards, the country started
with an aggressive deregulation of its international air travel connec-
tions. Overall, this provides the basic lens through which we can frame
the trade-to-seats causality for Singapore and Malaysia: massive trade
flows that created additional demand for air travel, which could be pro-
vided because of the adoption of liberal, open skies-type of policies (as
well as infrastructure provision to enable this) (Van De Vijver et al.,
2014).3
CONCLUSION
The purpose of this chapter was to investigate the causal relationship

between trade and air passenger flows in the Asia-Pacific region. Our
results indicate that there are signs of cumulative causation in the region,
but at the same time there are rather diverse patterns amongst countries.
There is no significant causality for links between the region’s most devel-
oped economies (e.g. AustraliaJapan), but there is often significant caus-
ality running from air passenger travel to trade between more developed
and less-developed economies (e.g. AustraliaThailand). Additionally, sig-
nificant causality running from trade to air passenger connections occurs
for countries that have adopted very liberal approaches towards the air
transport industry (e.g. Singapore).
Heterogeneity in the causality relations reflects the fact that the de facto
integration of Asia-Pacific is unfolding rapidly, but in a highly uneven
manner (see Ando & Kimura, 2010). Moreover, this process occurs in the
absence of an efficient and homogeneous de jure integration. Kulendran
and Wilson (2000) already hinted at the often contrasting causal relation-
ships between trade and air passenger transport, and our heterogeneous
TSCS framework allows assessing such questions for a much larger number
of cases in a single analysis.
Future research on the trade/air passenger relations in Asia-Pacific will
probably need to break our studied time period (19802010) down in time
intervals that frame key policy shifts as regards trade and air transport lib-
eralization. Some countries in the region only adopted more liberal policies
towards trade and air transport from the 1990s onwards (e.g. Philippines),
whilst other countries (e.g. Singapore) already adopted liberal policies from
the onset of our period. Working with time intervals can capture this frag-
mented nature of liberalization tendencies.
NOTES
1. Poole (2010) uses an additional two stage least square analysis to identify cau-
sal links between business travel and international trade.
2. It should be noted that the relative lack of statistically significant relations
compared to the very clear bidirectional causality at the level of the region may in
part be the result of having smaller samples (and therefore degrees of freedom for
assessing the F-statistic).
3. In the case of Singapore (and to a lesser degree Malaysia), the absence of
causal links with Australia and New Zealand may be in part due to the use of
OAG-data, which include scheduled flights rather than actual routes flown by pas-
sengers. This can bias the results for Singapore in particular in our analysis because
of the city-state’s role as an important hub in the international air transport net-
work. Because of Singapore’s 6th freedom rights, many of the scheduled seats
between Singapore and especially Australia and New Zealand are in fact used by
passengers traveling to or from Europe. Hence, changes in scheduled seats in these
cases are less influenced by/influencing changes in trade between Singapore and
Australia/New Zealand, but rather a result of demand and supply for air travel
partly outside the Asia-Pacific region.
REFERENCES
Ando, M., & Kimura, F. (2010). The spatial patterns of production and distribution networks
in East Asia. In P. Athukorala (Ed.), The rise of Asia. Trade and investment in global
perspective. London: Routledge.
Athukorala, P. (Ed.). (2010). The rise of Asia. Trade and investment in global perspective.
London: Routledge.
Athukorala, P., & Hill, H. (2010). Asian trade and investment: Patterns and trends. In
P. Athukorala (Ed.), The rise of Asia. Trade and investment in global perspective.
London: Routledge.
Bowen, J. (2000). Airline hubs in Southeast Asia: National economic development and nodal
accessibility, 19791997. Journal of Transport Geography, 8(1), 2541.
Bowen, J., & Leinbach, T. R. (1996). Development and liberalization: The airline industry in
ASEAN. In G. Hufbauer & C. Findlay (Eds.), Flying high: Liberalizing civil aviation in
the Asia Pacific. Washington, DC: Institute of International Economics.
Button, K., & Lall, S. (1999). The economics of being an airport hub city. Research in
Transportation Economics, 5, 75105.
Button, K., & Yuan, J. (2013). Airfreight transport and economic development: An examina-
tion of causality. Urban Studies, 50(2), 329340.
Chandra, A. C. (2005). Indonesia and bilateral trade agreements (BTAs). The Pacific Review,
18(4), 541565.
Chin, A., Hooper, P., & Oum, T. H. (1999). The impacts of the Asian economic crises on
Asian Airlines: Short-run responses and long-run effects. Journal of Air Transport
Management, 5, 8796.
Cristea, A. D. (2011). Buyerseller relationships in international trade: Evidence from U.S.
States’ exports and business-class travel. Journal of International Economics, 84(2),
207220.
Derudder, B. (2012). Scanning the skies: Asian cities in airline networks. Paper presented at
the Intercity networks and urban governance in Asia Conference 89 March. Asian
Urbanisms Cluster in collaboration with Metacluster Historical Sociology of Asian
Connections of the Asia Research Institute, National University of Singapore.
Doganis, R. (2010). Flying off course. Airline economics and marketing. (4th ed.). London:
Routledge.
Findlay, C., & Forsyth, P. (1992). Air transport in the Asian-Pacific region. Asian-Pacific
Economic Literature, 6(2), 110.
Findlay, C., & Goldstein, A. (2004). Liberalization and foreign direct investment in Asian
transport systems: The case of aviation. Asian Development Review, 21(1), 3765.
Forsyth, P. J., King, J., & Rudolfo, C. L. (2006). Open skies in ASEAN. Journal of Air
Transport Management, 12(3), 143152.
Graham, B. (1995). Geography and air transport. Chichester: Wiley.
Granger, C. W. J. (1969). Investigating causal relations by econometric models and cross-
spectral methods. Econometrica, 37(3), 424438.
Grosso, M. G. (2012). Air passenger transport in the APEC: Regulatory impacts
and prospects for Asia Pacific integration. Journal of Economic Integration, 27(2),
312327.
Hiratsuka, D. (2006). Outward FDI from and intraregional FDI in ASEAN: Trends and drivers.
IDE Discussion Paper No. 77. Retrieved from http://ir.ide.go.jp/dspace/bitstream/2344/
233/3/ARRIDE_Discussion_No.77_hiratsuka.pdf. Accessed on 11, 2006.
Hood III, M. V., Kidd, Q., & Morris, I. L. (2008). Two sides of the same coin? Employing
Granger causality tests in a time series cross-section framework. Political Analysis, 16,
324344.
Hurlin, C., & Venet, B. (2001). Granger causality tests in panel data models with fixed coeffi-
cients. Working paper Eurisco 200109. University of Paris Dauphine.
IATA. (2010). Asia Pacific: challenges and opportunities Intra-Asia market eclipses North
America as world’s largest. Press release No. 4. Retrieved from http://www.iata.org/
pressroom/pr/pages/2010-02-01-01.aspx
Im, K. S., Pesaran, M. H., & Shin, Y. (2003). Testing for unit roots in heterogeneous panels.
Journal of Econometrics, 115, 5374.
Imada, P. (1993). Production and trade effects of an ASEAN free trade area. The Developing
Economies, 31(1), 323.
Kimura, F. (2008). The modality of East Asia’s economic integration. In D. Hiratsuka &
F. Kimura (Eds.), East Asia’s economic integration. Progress and benefit. Hampshire:
Palgrave Macmillan.
Kulendran, N., & Wilson, K. (2000). Is there a relationship between international trade and
international travel? Applied Economics, 32,10011009.
Lee, J., & Park, I. (2005). Free trade areas in East Asia: Discriminatory or non-discriminatory?
World Economy, 28(1), 2148.
Levin, A., Lin, C. F., & Chu, C. S. J. (2002). Unit root tests in panel data asymptotic and
finite-sample properties. Journal of Econometrics, 108, 124.
Lütkepohl, H., & Krätzig M. (2004). Applied time series econometrics. Cambridge: Cambridge
University Press.
Ng, F., & Yeats, A. (2003). Major trade trends in East Asia: What are their implications for
regional cooperation and growth. Policy Research Working Paper Series 3084. The
World Bank.
O’Connor, K. (1995). Airport development in Southeast Asia. Journal of Transport
Geography, 3(4), 269280.
Ozeki, H. (2008). Development of de facto economic integration in East Asian trade. In:
Soesastro, H. (Ed.), Deepening economic integration in East-Asia: The ASEAN
Economic Community and beyond, ERIA Research Project Report 2007, No 12.
Chiba: IDE-JETRO.
Poole, J. P. (2010). Business travel as an input to international trade. C Santa Cruz Working
Paper. Available at http://www.scu.edu/business/economics/upload/Poole.pdf
Rimmer, P. (2000). Effects of the Asian crisis on the geography of Southeast Asia’s air traffic.
Journal of Transport Geography, 8, 8397.
Sally, R. (2010). Regional economic integration in Asia: The track record and prospects. ECIPE
Occasional paper, 2:10.
Shin, K. H., & Timberlake, M. (2000). World cities in Asia: Cliques, centrality and connected-
ness. Urban Studies, 37, 22572285.
Siddiqui, K. (2010). The political economy of development in Singapore. Research in Applied
Economics, 2(2), 131.
Soesastro, H. (2006). Regional Integration in East Asia: Achievements and future prospects.
Asian Economic Policy Review, 1(2), 215234.
Van De Vijver, E., Derudder, B., Witlox, F. (2014). Causal relationships in the evolution of air
passenger and trade geographies: The case of Asia-Pacific, 19802010. Journal of
Transport Geography, 34, 142150.
Yamazawa, I. (1992). On Pacific economic integration. The Economic Journal, 102,
15191529.
CHAPTER 10
INTERNATIONAL AND NATIONAL

POLITICAL REGULATIONS OF
AVIATION’S CLIMATE IMPACT
AND COST IMPACTS ON AIR
FREIGHT
Janina D. Scheelhaase
ABSTRACT
This chapter provides an overview of the current political regulations on

aviation’s climate relevant emissions in Europe, Australia, and New
Zealand and of the planned regulations in other parts of the world. In a
next step, the cost impacts of most of these regulations on air freight will
be quantified. This way, the economic impacts of environmental regula-
tions on air freight can be estimated.
The main results indicate that cost impacts on air freight services induced
by political measures for the reduction of aviation’s climate relevant
emissions turn out to be small. This is true for both local emission
charges on nitrous oxide (NOX) and hydrocarbon (HC) emissions
which are in force at a number of European airports and the European
emissions trading scheme for the limitation of CO2 emissions.

ISSN: 2212-1609/doi:10.1108/S2212-160920140000004009
255
256 JANINA D. SCHEELHAASE
Keywords: Environment; aircraft emissions; air transport policy;

climate change; environmental economics; emissions trading
JEL classifications: L93; K32; D62; Q52
INTRODUCTION
Air transport is a substantial emitter of CO2 and, additionally contributes to
climate change by short-lived non-CO2 effects, such as emission of nitrous
oxide (NOX) or triggering contrails and contrail cirrus (e.g., Sausen et al.,
2005). In 2005, aircraft-induced CO2 contributed 1.6 percent to the total
anthropogenic radiative forcing (RF). If non-CO2 climate effects are also
considered, the contribution of aviation to total anthropogenic RF is about
three times as large, that is, 4.9 percent (Lee et al., 2009). In addition, avia-
tion is one of a few sectors expected to grow significantly in the medium
and long term. In the timeframe 20112031, world annual traffic growth
rates of 4.7 percent (passenger and freight) can be expected (Airbus, 2012).
Then, worldwide air freight traffic is expected to grow slightly higher than
passenger traffic at 4.9 percent p.a. Whilst international aviation’s carbon
dioxide emissions have been regulated in a number of countries by market-
based measures in the past, this is not the case for most of aviation’s
non-CO2 climate impacts. This chapter investigates the current political
regulations on aviation’s climate relevant emissions in the world and their
impacts on air freight costs.
This chapter is organized as follows: the second section provides an over-
view of the current political regulations on aviation’s climate relevant emis-
sions in Europe, Australia, and New Zealand and of the planned regulations
in other parts of the world. In the third section, the cost impacts of most of
these regulations on air freight will be quantified. This way, the economic
impacts of environmental regulations on air freight can be estimated.
CURRENT AND UPCOMING POLITICAL

REGULATIONS TACKLING AVIATION’S CLIMATE
RELEVANT EMISSIONS
As of October 2013, market-based measures for the limitation of aviation’s

climate relevant emissions are in force in Europe, Australia, and New Zealand.
Furthermore, the implementation of emissions trading schemes is planned
Regulations of Aviation’s Climate Impact and Cost Impacts on Air Freight 257
in China (Government of China, 2011) and in South Korea (National

Assembly of Korea, 2012), both in 2015.
In Europe, aircraft emission charges on local NOX and hydrocarbon
(HC) emissions are in force for a number of years now. NOX and HC being
emitted during the landing- and take-off-cycle are subject to this charge.
NOX and HC are the main contributors to combustion-related local air
pollution and precursors of ground level ozone. A positive more wide
spread side-effect of the charge on local NOX emissions is that it will also
reduce greenhouse gas effects to some extent: Because more NOX friendly
engines are used, the amount of the gas emitted will be reduced at cruise
level as well as below 3,000 feet during the LTO cycle. On the other hand,
a trade-off exists between the reduction of NOX and CO2 because aircraft
engines can technologically be optimized either to minimize fuel burn, and
thus CO2 emissions, or to minimize NOX emissions (Scheelhaase, 2010).
Aircraft emission charges on local NOX and HC emissions following the
so-called ERLIG guidelines (European Civil Aviation Conference (ECAC),
2003), a Europe-wide harmonized approach for calculating NOX emissions,
were introduced in Sweden and at London Heathrow Airport in 2004,
London Gatwick Airport followed in 2005. In 2008, an emission charge
based on ERLIG guidelines has been introduced in Germany at Frankfurt,
Munich, and Cologne Bonn Airport (Fraport, 2007). Hamburg Airport
and Dusseldorf International acted accordingly in 2010 and 2011. In
Germany, the introduction of local emission charges is understood as a
pilot phase with airports participating on a voluntary basis. After this
phase, the environmental and economic impacts of the charge will be inves-
tigated and the design of the charge may be subject to modifications.
Switzerland modified its system of local emission charges which had been
in force since 1997 and moved towards the Europe-wide harmonized
approach in 2010. Also in 2010, Copenhagen Airport introduced a local
charge on NOX emissions based on ERLIG guidelines (Copenhagen
Airport, 2010).
The European emission charges are designed to be revenue neutral in
the sense that they do not increase the airport’s overall revenue from air
traffic. While the goal of establishing economic incentives and the principle
of revenue neutrality are practically identical in Switzerland, Sweden,
Denmark, United Kingdom, and Germany, the methods of calculating dif-
fer with regard to the amount of the charge, determining thresholds and
the method of achieving revenue neutrality.
At the European Union (EU) level, the European Commission has
been analyzing, since 2008, whether charges on NOX emissions at
European airports can be a viable approach to reduce international
aviation’s non-CO2 climate impact. Measures under consideration of the

European Commission include a local NOX charge modified by a distance
factor, an en-route charge on NOX emissions, an increased NOX strin-
gency for LTO emissions standard and a multiplier on CO2 emissions
(CE Delft, 2008).
Lately, a number of emissions trading schemes tackling climate change
both on a national and on a supranational level affecting aviation have
been introduced. However, these trading schemes are designed rather dif-
ferently (Scheelhaase, 2011).
In order to reduce international aviation’s CO2 emissions, air transport
has been fully integrated into the EU emissions trading scheme (EU ETS)
in 2012. This trading scheme covers all flights departing from or arriving at
airports in the European Union, Norway, and Iceland. By this way, both
European and non-European airlines are participating in the EU emissions
trading scheme. Concerning the EC Directives for the inclusion of air trans-
port into the EU ETS, strongly diverting views of non-EU countries were
expressed at the International Civil Aviation Organization (ICAO)
Assemblies in 2007 and 2010. Unlike the EU Member States, most other
ICAO contracting states believe that an inclusion of non-EU carriers
is only possible on the basis of mutual agreements which do not exist
to date.
In November 2012, the EU Commission proposed to “defer the require-
ment for airlines to surrender emission allowances for flights into and out
of Europe until after the ICAO General Assembly” in autumn 2013. The
EU Commission justified this proposal with “the very positive discussions
that took place” lately “in the ICAO Council on a global market-based
approach to regulating greenhouse gas emissions from aviation”
(Commission of the European Union, 2012). This proposal has been
adopted by the EU Parliament in April 2013.
New Zealand has introduced an emissions trading system for the lim-
itation of greenhouse gas emissions in 2008 (New Zealand Government,
2012). Until 2015, several sectors will have been gradually phased in the
trading scheme. The forestry sector started in 2008. The liquid fossil fuels
sector as well as the stationary energy and industrial processes sectors
have become mandatory participants by 2010. The waste sector as well as
the importers of “synthetic” greenhouse gases such as HFCs, PFCs, and
SF6 followed. In 2015, agriculture will be included, finally. Transport
including domestic aviation has been covered indirectly by a so-called
upstream approach: The liquid fossil fuels sector is expected to pass
through the costs of compliance to the aircraft operators in the form of
increased kerosene prices. Fuel used for international aviation (and mar-
ine transport) are exempt from the scheme, consistent with the Kyoto
Protocol.
In Australia, a mandatory national emissions trading scheme has been
introduced in 2012. This scheme is covering the stationary energy sources,
parts of the transport sector, industrial processes, non-legacy waste, and
fugitive emissions (Australian Government, 2011). Domestic aviation,
domestic shipping, rail transport, and fuel used for non-transport applica-
tions are included. The scheme does not apply to fuels used for private
transport, light vehicle business transport and off-road fuel use by agricul-
ture, forestry and the fishing industry. The trading scheme started with a
fixed carbon price of AUSD23 per tonne which will be raised by 2.5 per-
cent in real terms in the years 2013 and 2014. In 2015, the carbon price
will transition to a fully flexible market price (Australian Government,
2011).
A step-wise linking of the Australian emissions trading scheme with the
European emissions trading scheme has been agreed in 2012. Until 2018,
both trading schemes shall be fully linked (Australian Government/
Commission of the European Union, 2012). By then, it will be possible to
use carbon credits from the Australian scheme or allowances from the
European Union Emissions Trading System for compliance under either
system.
China is planning to introduce a nationwide carbon trading system by
2015, according to the 12th Five Year Plan of 2011 (Government of China,
2011). Here, the main emitters of CO2 shall be included. Amongst other
sectors, domestic aviation shall be participating. Following this strategy,
China intends to reduce its growing demand for fossil fuels and to limit the
impacts of climate change since a number of Chinese provinces are highly
vulnerable to these effects. China will be experimenting with seven regional
carbon trading pilot systems. This way, various existing trading models
shall be tested with the goal of finding suitable solutions for a Chinese
national carbon trading system (Stockholm Environment Institute/
FORES, 2012). In September 2012, it was agreed that the EU will provide
expertise in setting up China’s emissions trading systems (European Voice,
2012). Australia agreed to provide expertise for building up the Chinese
trading system in July 2013 (Aston, 2013).
In May 2012, South Korea’s National Assembly has passed legislation
to introduce a national emissions trading scheme to tackle carbon emis-
sions by the year 2015 (National Assembly of Korea, 2012). This way,
South Korea’s greenhouse gas emissions shall be reduced by 30 percent
until 2020. The cap and trade system will cover about 70 percent of the
national greenhouse gas emissions (Bloomberg, 2013). Installations emit-
ting more than 25,000 tonne CO2-equivalent p.a. and entities emitting over
125,000 tonne CO2-equivalent p.a. shall be participating on a mandatory
basis. In addition, it will be possibly to opt-in on a voluntary basis (Yong-
Gun, 2012). Linking arrangements with the European, Australian, and
New Zealand trading schemes are envisaged by the Korean Government
(Europolitics, 2012).
All in all, the global framework for the limitation of aviation’s climate
relevant emissions is heterogeneous. This will have impacts on competition
within the aviation sector. Against this background several ICAO high-
level groups are working on global market-based measures for the limita-
tion of aviation’s CO2 emissions. In addition, the ICAO Council has been
tasked to establish a process to develop a framework for market-based
measures in international aviation. These goals were agreed upon at the
37th ICAO Assembly in October 2010.
In 2012, a new ICAO Council High-Level-Group was formed with the
goal to focus on environmental policy challenges. The main objective of
the group was to provide recommendations on the feasibility of a global
market-based measure scheme appropriate to international aviation, as
well as its development of a policy framework (International Civil
Aviation Organisation, 2012). Hereby guidance to the general application
of any proposed market-based measure to international air transport
activity shall be provided. This task has been completed by autumn
2013.
At the 38th ICAO Assembly in September/October 2013, it has been
agreed to introduce a global scheme for the regulation of international
aviation’s CO2 emissions by the year 2020. (International Civil Aviation
Organisation, 2013). The CO2-regulation scheme is not defined yet, most
likely it will be an emissions trading or an offsetting scheme.
ECONOMIC IMPACTS OF CLIMATE

RELEVANT POLITICAL REGULATIONS ON
AIR FREIGHT
Both local charges applied at a number of European airports and emissions

trading schemes which are in force in Europe, Australia, and New Zealand
have impacts on the costs of air freight.
Economic Impacts of Local Charges on NOX and

HC Emissions in Europe
Which costs are associated with local charges for the limitation of NOX
and HC emissions at European Airports? In principle, local charges offer
an economic incentive to reduce these emissions by allowing airlines
employing relatively environmental friendly engines to cut their landing
fees. In contrast, airlines using engines with relatively high NOX and HC
emissions have to pay higher landing fees (Scheelhaase, 2010). While the
method of calculating LTO-NOX and HC emissions is identical throughout
Europe, due to the application of the ERLIG-formula, the fee charged per
unit emission value NOX/HC differs: In Sweden SEK50 (about h6) is
charged per emission value, in United Kingdom the charge is about h7.4
(Heathrow in 2012), respectively h5.5 (Gatwick in 2012) per unit and in
Switzerland it lies between CHF1.4 (Geneva Airport) and CHF3.4
(Lugano Airport). On average, CHF 2.65 (about h2.17) is charged per unit
emission value at Swiss airports. At Copenhagen Airport, 16.55 DKK
(about h2.2) is charged per unit emission value. In Germany, the charge is
h3 per unit emission value.
Revenue neutrality is achieved on the level of the individual airports.
The methods to achieve this goal differ throughout Europe. In Germany,
for instance, this method consists of three consecutive steps: First, the local
emission charges of all airlines and all air transport movements for a cer-
tain period, say a year, are summed. Second, the same is done for the
weight-related landing fees for the same period. Third, the ratio of emission
charges and weight-related fees is calculated. This ratio determines the per-
centage by which the landing fees are decreased for each airline. This
approach can be illustrated taking a hypothetical example. If the local
emission charge at an airport adds up to h7 million a year and the weight-
related landing fees add up to h70 million, to achieve revenue neutrality,
the weight-related fee is decreased by h7 million. Therefore each airline
now pays a weight-related landing fee that is 10 percent lower. At the end
of the day the carriers altogether pay h63 million in weight-related landing
fees and h7 million for local emissions. From the airport’s point of view the
emission charge just compensates for the reduction of the weight-related
landing fee and the overall revenues from air traffic remain unchanged
(Scheelhaase, 2010).
Looking at it from an airline’s angle, the landing charge expenditure is
supposed to change. The net effect of the reduced weight-based charge on
the one hand and the additional emission-related charge on the other hand
is decisive. If the net effect is positive, the airline has to pay a higher
amount in landing fees as compared to the situation prior to the local
charge. If the net effect is negative, the carrier can cut its expenditure on
landing fees (Scheelhaase, 2010).
In principle, the amount of the local charges an airline has to pay per
aircraft landing differs depending on the engines employed, the number of
engines on an aircraft and on the European State applying the charge. For
details on the calculation of the local charges, see Scheelhaase, Hepting,
Pabst, and Wilken (2005). Table 1 gives an overview on the amount of the
charge per landing for selected freight and passenger aircraft and European
countries.
Table 2 shows the influence of the engine employed on the amount of
the local charge. Here, three aircraft types and different engine options but
with comparable maximum take-off mass (MTOM) and rated output per
engine are presented. Obviously, the choice of engines has a great influence
on the amount of the local charge. However, the airline’s decision on the
engines employed depends on a bundle of managerial factors, not just on
the total local emission charge. And the number of engine options is rather
small for some freight aircraft. This limits the possibilities for lowering the
emission charge.
Table 1. Local Emission Charges for Selected Aircraft and European

Countries.
Aircraft Type Engine Emission Charge Per Emission Charge Per
Landing in Germany Landing in Sweden
in h (h3 Per Unit (h6 Per Unit NOX
NOX and HC and HC Emission
Emission Value) Value)
Boeing 747-400F CF6-80C2B1F 133.35 266.70

Boeing 747-800F GEnx-2B67 131.74 263.48
Boeing 777-200F PW4084 139.37 278.74
McDonnell-Douglas-11 PW4460 126.87 253.74
Airbus 300F4-600 PW4158 77.56 155.12
Airbus A 340-300 CFM56-5C4 104.43 208.86
Boeing 767-300 PW4060 84.58 169.16
Boeing 737-700 CFM56-7B22 27.36 54.72
Airbus 320-200 CFM56-5A1 27.03 54.06
ATR 42-300 PW121 3.83 7.66
ATR 72-500 PW 127 5.58 11.16
DHC-8-300 PW123B 5.01 10.02
Source: DLR.
Table 2. Influence of Engine Employed on Local Emission Charge for

Selected Airframe/Engine Combinations.
Aircraft Type MTOM Engine Rated Output Emission Charge
(kg) Per Engine (kN) Per Landing in
Germany in h
(h3 Per Unit
NOX and HC
Emission Value)
McDonnell-Douglas-11 280,000 CF6-80C2D1F 273.6 114.51

McDonnell-Douglas-11 280,000 PW4460 266.9 126.87
McDonnell-Douglas-11 280,000 PW4462 276.0 142.77
Boeing 747-400 396,900 CF6-80C2B1F 254.2 133.36
Boeing 747-400 396,900 PW4056 252.4 156.52
Boeing 747-400 396,900 CF6-80C2B5F 272.5 157.70
Boeing 747-400 396,900 RB211-524G/H-T 253.0 167.94
Boeing 747-400 396,900 RB211-524H2-T 264.7 294.01
Airbus 320-200 73,500 CFM56-5B4/2P 120.1 23.08
Airbus 320-200 73,500 CFM56-5A1 111.2 27.03
Airbus 320-200 73,500 CFM56-5A3 117.8 30.66
Airbus 320-200 73,500 V2527-A5 111.2 32.29
Airbus 320-200 73,500 CFM56-5B4/P 120.1 33.85
Airbus 320-200 73,500 V2500-A1 111.2 46.30
Source: DLR.
Looking at the European local emissions charges from a total costs

perspective, it shows that their impact is limited. This illustrates the fol-
lowing table. In Table 3, total airport charges and their composition at a
number of European and two non-European Airports are presented
exemplarily for an A330-200. These figures have been calculated on the
basis of the airports’ charges manuals. Reductions of the emission charges
due to the principle of revenue neutrality could not been taken into
account here.
As shown in Table 3, the relative share of the local emission charge
amounts to from 0.73 percent (Heathrow) until 3.97 percent (Arlanda
Airport) of the total airport charges. Even for aircraft emitting a relatively
high amount of NOX the impact of the emission charge is small. For
instance for a Boeing 747-200 with RB211-524D4 engines, the emission
charge amounts to 6.4 percent of the total airport charges at Frankfurt air-
port (Scheelhaase et al., 2011).
The net cost impact of the emission charges is even smaller for some air-
lines than shown in the Tables 13 due to the principle of revenue
Table 3. Composition of Airport Charges of Selected European and Non-European Airports in 2010.
264
Airport Paris-CDG Hamburg Frankfurt Madrid Stockholm Rom London-LHR New York JFK Dubai
Total charges (EUR) 12.443,9 5.052,9 6.938,3 3.727,4 5.459,7 3.603,7 16.943,5 5.294,1 1.787,79
Landing/Take-off 1.372,7 1.224,3 727,4 1.872,7 1.771,2 1.091,0 1.406,8 1.814,8 587,0
Environmental related 1.101,2 974,1 282,1 0,0 258,7 36,1 10.011,8 0,0 0,0
Passenger related 9.970,1 2.854,6 5.928,9 1.854,7 3.429,8 2.476,6 5.524,8 3.479,4 1.200,8
Landing and take-off
Landing charges 1.372,7 612,2 363,7 1.872,7 545,5 1.146,2 1.814,8 587,0
Take-off charges 0,0 612,2 363,7 0,00 1.771,2 545,5 260,7
Environmental charges
Noise charge Take-off 0,00 421,00 75,00 0,00 0,00 18,03 0,00 0,00 0,00
Noise charge Landing 222,18 421,00 75,00 0,00 42,07 18,03 0,00 0,00 0,00
Night surcharge Landing 0,00 0,00 0,00 0,00 0,00 0,00 0,00 0,00 0,00
Night surcharge Take-off 0,00 0,00 0,00 0,00 0,00 0,00 0,00 0,00 0,00
Emission charge Landing 0,00 66,04 66,04 0,00 0,00 0,00 123,10 0,00 0,00
Emission charge Take-off 0,00 66,04 66,04 0,00 216,66 0,00 0,00 0,00 0,00
Solidarity Tax, etc. 879,00 0,00 0,00 0,00 0,00 0,00 9.888,75 0,00 0,00
Passenger charges
JANINA D. SCHEELHAASE
Passenger charge 4.219,20 1.911,83 4.131,30 1.549,24 2.184,52 1.628,35 5.524,80 2.930,00 1.200,82
Airport/infrastructure taxes 2.116,19 0,00 0,00 0,00 434,24 0,00 0,00 0,00 0,00
Civil Aviation tax 1.547,04 0,00 0,00 0,00 162,22 0,00 0,00 0,00 0,00
Security fee, etc. 2.037,63 942,73 1.797,56 305,45 648,87 848,24 0,00 549,38 0,00
Share emission charge 0,00% 2,61% 1,90% 0,00% 3,97% 0,00% 0,73% 0,00% 0,00%
Source: Scheelhaase, Grimme, and Maertens (2011).

Assumptions: A330-200 with 75 percent load factor, 293 seats, Next Destination: outside of EU, Arrival at 3 p.m., Departure at 4 p.m., Pier-Side
Position, MTOM 231 Tons, Emission Value 44.02 kg NOX.
neutrality explained above. Because revenue neutrality is achieved on the

level of the individual airports, estimations of the net impact of the local
charges depend on the availability of sufficient airport data. Scheelhaase
et al. (2005) analyzed the net cost impact for selected German airports.
This study can be summarized as follows. Data basis for the calculations
was empirical flight movement data from eight German airports jointly
provided by the German Airports Association (ADV). This data was air-
line and, mostly, engine-specific allowing NOX and HC emissions per flight
movement and airline as well as the emission charges for individual airlines
for a given period to be calculated. The same was done for airline-specific,
weight-related landing charges. Applying these findings to a revenue neu-
trality regime, the effect of reduced weight-related charges against the local
emission charges could be calculated for airlines. This is done in terms of
the changes in the airlines’ expenditure on landing charges at an airport.
Fig. 1 offers the results for an emission charge of h1.50 and h3 per unit of
emission value.
As seen in the figure, the economic impact of the local emission charge
differs depending on the particular airline. While Lufthansa can expect a
refund of at least h500,000 (assuming h3 per unit of emission value) p.a.
and will therefore clearly gain an advantage through the implementation of
the emission charge, other carriers such as Condor or British Airways will
400,000.00 €
300,000.00 €
200,000.00 €
100,000.00 €
0.00 €
Lu Co Un Br Ai Si Au Am De O
–100,000.00 € fth nd ite itis rC ng st er lta th
an d h an ap ria ica er
or Ai o n Ai s
sa Ai ad r e n rli
–200,000.00 € rli rw a Ai ne
ne ay A i rli s
s s rli ne
ne
–300,000.00 € s s
–400,000.00 €
€ 1.5 per unit emission value
–500,000.00 € € 3.0 per unit emission value
–600,000.00 €
Fig. 1. Change in Airlines’ Expenditure for Landing Charges at Frankfurt

Airport. Source: Scheelhaase (2010). Data basis is 2003.
have to raise their expenditure for landing fees. For these airlines, the net
effect of the decreased weight-related landing fee and the additional local
emission charge is positive. Condor would have to pay almost h53,400
(assuming h3 per unit of emission value) p.a. in addition. British Airways
would have to pay almost h20,500 (assuming h3 per unit of emission value)
p.a. more due to the implementation of the new charge. These differences
between airlines can be explained by aircraft sizes, engine specifications,
and numbers of air traffic movements.
The economic impacts are relatively small in relation to the total expen-
diture of an airline on landing fees. Compared to the expenditure on
weight-related landing fees per year, the saving for Lufthansa of at least
h500,000 amounts to just 1.08 percent of its expenditure at Frankfurt
Airport. From a British Airways point of view, the additional expenditure
of just under h20,500 p.a. is less than 1.26 percent the company’s expendi-
ture for landing at Frankfurt in the same year, and Condor would have to
pay an additional 2.28 percent (Scheelhaase et al., 2005).
The costs of freight operations are affected by the European local emis-
sion charges. But the impact of the local charges is limited: As shown in
Table 3, the total airport charges for an A330-200 were raised by 3.97 per-
cent, respectively h216.66 in absolute terms per landing at the most (at
Arlanda airport) in 2010. This was the maximum cost increase for freight
operations of an A330-200 in Europe for two reasons: Firstly, the net effect
of the additional emission charge and of the reduced weight-related charge
is decisive for the actual costs increase. The net effect for freight operations
at the European airports can’t be calculated here due to a lack of airport
specific data. In principle, the net effect of the emission charge is much
lower or turns even into a bonus as illustrated above for a number of
airlines.
Secondly, if freight is transported in the form of belly hold cargo, the
costs for the emission charge refer to both passengers and freight carried.
Therefore the cost increase for belly hold cargo is lower than the percentage
given. This may be illustrated by the following hypothetical example: If
20,000 kg belly hold cargo is assumed, and a 50:50 cost allocation between
passengers belly hold cargo, the costs for belly hold cargo increase about
2 percent, respectively h108.33 in absolute terms at the very most for an
A330-200. Per kilogram belly hold cargo, the emission charge raises the
costs of cargo by h0.0054. In case freight is being carried by an all freight
aircraft, the cost situation is slightly different. Then costs can’t be shared
between passengers and freight. In that case a maximum cost increase of
about 4 percent of total airport charges is possible for an A330-200 in

Europe as shown in Table 3.
The highest possible cost increase for an all freight aircraft was calcu-
lated for a Boeing 747-200 with very NOX-intense RB211-524D4 engines.
For this airframe/engine combination, the emission charge amounts up to
6.4 percent of total airport charges or h984.66 in absolute terms per land-
ing at Frankfurt airport as mentioned above (Scheelhaase et al., 2011).
Under the assumption of 95,000 kg freight carried, the costs for each
kilogram will be increased by h0.0103. Then the emission charge increases
the average total operating costs of freight by about 0.8 percent. For this
calculation, total operating costs for a B747-400 F in the amount of 1.48
USD per kg freight in the year 2007 are assumed, as data for a B747-200
is not publically available. For comparison: large freighter total operating
costs varied between 1.48 USD for a B747-400F and a B777F, 1.52 USD
for a MD-11F and 1.69 USD for an A380F per kg freight in the year
2007 (Morell, 2011).
What are the impacts of the local emission charges on air freight ship-
ping companies? Unfortunately, only minimal peer-reviewed literature on
the impact of climate protecting regulations on air freight airlines and
none on air freight shippers exists to date. Against this background, it is
only possible to provide some broad thoughts on this issue. As presented
above, the cost impact of the European emission charges is expected to
be negligible for airlines offering combined services and small for all
freight airlines. This will lead to a small cost increase for those freight
companies cooperating with all freight airlines or using all freight aircraft
in their own possession. If the additional costs can’t be passed on to the
consumers by the freight shippers, they may choose to switch from coop-
erating with all freight airlines to airlines offering belly hold capacity.
Another option would to be to optimize the routing in that way that the
European airports charging for emissions will be avoided. There are
numerous possibilities for avoiding these airports in Europe. For instance
could Cologne/Bonn Airport be substituted by Leipzig/Halle Airport or
Brussels Airport, both of which are not charging for NOX and HC emis-
sions and are geographically close to Cologne/Bonn Airport. London
Heathrow and London Gatwick could be replaced by London Stansted
or London Luton Airport, for example. Companies offering multi-modal
freight transport services (as it is the case with many air freight shippers)
could also react to the emission charges by switching to other modes
such as road, rail or sea transport.
Economic Impacts of the Emissions Trading Scheme in Europe
Emissions trading schemes for the limitation of aviation’s CO2 emissions in

Europe, Australia and New Zealand have an impact on the costs of freight
operations. Due to a lack of data on the economic impact of the trading
schemes on aviation in Australia and New Zealand, the following analysis
focusses on the economic impact of the European emissions trading scheme
on aviation.
In the European Union, Norway, Iceland, and Liechtenstein, air trans-
port has been included into the European emissions trading scheme (EU
ETS) in the year 2012. The EU ETS aims at limiting international avia-
tion’s CO2 emissions by providing an economic incentive to reduce the
demand for kerosene and thus CO2 emissions from aviation. In this
scheme, aircraft operators are obliged to hold and surrender allowances for
CO2 emissions. In case an aircraft operator reduces CO2 emissions, allow-
ances are no longer needed for the emissions saved. These additional allow-
ances can be “banked” for a possible later need or sold to another aircraft
operator. The total amount of CO2 emissions under the trading scheme is
capped. In 2012, the CO2 emission target for the aviation sector was 97 per-
cent of the so-called historical emissions of the years 20042006 (Council
of the European Union, 2009). This reduction target has been lowered by
another 2 percent in 2013, according to Directive 2008/101/EC.
Allowances allocated to aircraft operators are valid within the aviation
sector only. However, it is possible to purchase additional allowances from
participants of the emissions trading scheme for stationary sources in
Europe and to a certain extent from other non-European sources.
Exemptions from the EU ETS have been granted for flights performed
within the framework of public service obligations (PSO) on routes within
outermost regions or on PSO routes with an annual capacity of fewer than
30,000 seats. Also excluded from the EU ETS have been flights performed
by commercial air transport operators operating either fewer than 243
flights per four-month period for three consecutive four-month periods (so-
called de minimis clause) or flights with total CO2 emissions of less than
10,000 tonnes per year. The “de minimis” clause was introduced with the
goal of reducing the administrative costs for operators with a low number
of flights to and from Europe (Scheelhaase, Schaefer, Grimme, &
Maertens, 2012). Flights performed under visual flight rules have been
exempted, too, amongst some other exemptions.
Which costs can be associated with the introduction of the European
emissions trading scheme for aviation? In the recent years, a number of
studies have been conducted on the economic impact of integrating air

transport into the European emission trading scheme. Anger and Köhler
(2010) summarized the most important parameters and assumptions as well
as the results of many important studies. Tables 4 (parameters and assump-
tions), 5 and 6 (results) provide an overview. In these tables, selected stu-
dies of 20112013 have been added by the author. It should be noted that
only Derigs and Illing (2013) analyze the impact of the EU ETS on air
cargo while the other studies concentrate on passenger or combined air
services.
According to Derigs and Illing (2013), costs for allowances lie between
h380,000 and h248,000 per airline in the year 2020. Therefore, the cost
impact of the EU ETS on cargo operations is expected to be small. By opti-
mizing the routing, these costs could be reduced by up to 9.6 percent. Here,
3 generic airlines have been analyzed: a typical all-cargo airline such as
Cargolux or Lufthansa Cargo; a typical mixed carrier such as Lufthansa or
Singapore Airline; and a representative for an express carrier such as DHL,
UPS or Federal Express. Only for the “AggressiveETS” scenario with an
assumed allowance price of h70 per tonne CO2 and 100 percent auctioning
of allowances both of which not very realistic from todays’ point of
view profits of the generic airlines decreased by 1530 percent (Derigs
and Illing, 2013).
All in all, the results presented in the following tables depend to a great
extent on the assumptions on the particular regulations of the EU ETS, the
allowances prices, the assumed elasticities of demand for air services and of
the model chosen for the analyses. Most of the studies published until 2010
were conducted before the EC Directive for the inclusion of aviation into
the EU ETS had been published. For this reason, the authors had to make
assumptions on particular regulations, for instance on the level of auction-
ing. Many of these assumptions differ from the final legislation (see for
instance Boon, Davidson, Faber, & van Velzen, 2007; Ernst & Young,
2007). Therefore, most results are not directly comparable with each other.
In contrast, many of the studies published in the years 20112013
assumed the EU legislation as it went into force in 2009. Here the research
question investigated, the model used, specific assumptions on the cost-
pass-through rate and the allowance prices differ. While Derigs and Illing
(2013) focus on the impact of the EU ETS on cargo airlines under 5 differ-
ent scenarios for the EU ETS in the year 2020, Malina et al. (2012) investi-
gate the cost and CO2 impact of the EU ETS on the US aviation industry
in the timeframe 20122020. Scheelhaase et al. (2012) analyze the cost and
environmental impact of the EU ETS on all airlines under the trading
Table 4. Main Parameters and Assumptions Used in the Reviewed Studies on the Economic Impact of the EU
Emissions Trading Scheme on Aviation.
Study Trading Allowance Business as Cost-Pass-Through Fuel Efficiency Income and Kerosene Price
Period Price Per Asual Annual Rate Improvements Icome
Considered in Tonne CO2 Growth Rate of Assumed Elasticities
the Study CO2 Emissions
from Aviation
Derigs and 20122020 h15 in 4 out of Not discussed 0% Not discussed Not discussed Not discussed
Illing 5 scenarios;
(2013) h70 in
scenario
“Aggressive
ETS”
Malina et al. 20122020 $20 (2012) 1.72% for US Three scenarios: Annual increase Not discussed $2.29 (2012);
(2012) $27.45 airlines in fuel $2.77 (2020)
(2020) 1. Full pass-through of efficiency for per gallon fuel
allowances costs and
new aircraft of
opportunity costs. 1.4 %
2. Pass-through of
allowances costs only.
3. Neither allowances
costs nor opportunity
costs are passed
through
Scheelhaase 20122020 h20, h40 About 4% Discussed, but no about 1% Not discussed Not discussed
et al. quantitative
(2012) assumption
Vespermann 20122020 h25 3.4% 80% 1% Not discussed Not discussed
and Wald
(2011)
Boon et al. 20122020 h15h45 About 4% 47.3100% 1% Not clear, h1.49 (2007) per
(2007) short gallon, a short
discussion comparison
on incomes with carbon
in the price, but not
paper included to the
study
Ernst and 20112022 h6, h15, h30, 4% 2935% 1% Not discussed Not discussed
Young h60
(2007)
Frontier 2030 h27 and h40 3.55% Unclear but not 100% 1% Not discussed A short
Economics comparison
(2006) with carbon
price, but not
included to the
study
ICF (2006) 20082012 h5, h11, h21 4% Not discussed 1% Not discussed Not discussed
Mendes and 20132017 h7, h15, h30 4% 100% Not discussed Not discussed Not discussed
Santos
(2008)
Morrell 20052006 h28 ($40 430% 0% for free allocation, Not discussed Not discussed Not discussed
(2007) converted to depending on 100% for purchased
Euros using airline and auctioned
the exchange allowances
rate on
5 August
2009)
Study Trading Allowance Business as Cost-Pass-Through Fuel Efficiency Income and Kerosene Price
Period Price Per Asual Annual Rate Improvements Icome
Considered in Tonne CO2 Growth Rate of Assumed Elasticities
the Study CO2 Emissions
from Aviation
Scheelhaase 20082012 h15, h20, h30 0.54% 0% for free allocation, 11.5% Not discussed Not discussed
and depending on 100% for purchased
Grimme airline allowances
(2007)
SEC (2006) 20102030 h6h30 or 2% and 4% 100% for free 1% Not discussed Not discussed
estimated allocation, 100%
from the
models
Wit et al. 2012 h10 , h30 4% 0% or 100% for free 1% Not discussed Not discussed
(2005) allocation, 100% for
purchased and
auctioned allowances
Source: Anger and Köhler (2010) and Own Amendments.

Table 5. Main Results of the Studies Reviewed by Anger/Köhler on the

Economic Impact of the EU Emissions Trading Scheme on Aviation.
Flightlength Price Range Air Fare Increase Per Round Trip (h)
Per Return
Flight (h)
SEC Ernst and Young Boon et al. (2007)
(Allowance (Allowance Price (from Allowance
Price h6 to h6 to 60h) Price h15 and Free
h30) Allocation to h45
and 100%
Auctioning)
Short 50280 0.94.6 0.11.36 1.16.9

Long 6322994 7.939.6 1.0410.32 9.459.4
Source: Based on Anger and Köhler (2010), Boon et al. (2007), Ernst and Young (2007), SEC
(2006), and Wit et al. (2005).
Short-haul flight: AmsterdamParis de Gaulle, 480 km; Long-haul flight: London Gatwick
Newark, 6404 km.
scheme in the timeframe 20122020. Derigs and Illing (2013) used a self-
designed Air Cargo Scheduling Problem (ACSP) planning model in an eva-
luation study/what-if analyses to study the cost impact of different network
options. Malina et al. (2012) linked an economy-wide computable general
equilibrium (CGE) model with a partial equilibrium model that focuses on
the US aviation industry in order to analyze the impact of the EU ETS on
this sector. Scheelhaase et al. (2012) employed a chain of software tools
covering air flight simulation, air traffic simulation and forecast scenarios
and combined the results with static spread-sheet analyses for their study.
Because of these methodological and other differences, the results of these
recent studies are not entirely comparable, too.
Keeping these differences in mind, the results of most of the studies pre-
sented herein show that the cost impact of the EU ETS is expected to be
significant for some groups of airlines, specially the European airlines offer-
ing passenger or mixed services while it will be small or even negligible for
others, for instance for European and non-European cargo airlines and for
the non-European full service network carrier. In case groups of airlines
will have to bear additional costs, the impact on the growth rates depends
on the cost-pass-through rate assumed: When a 100 percent cost-pass-
through rate is assumed, the growth rates of the airlines (as well as their
CO2 emissions) remain unchanged as it is the case with most of the studies
presented. Some studies, however, suggest that only a partial pass-through
Table 6. Main Results of Recent Studies on the Economic Impact of the EU ETS on Aviation.
274
Study Cost Impact on Airlines Under the EU ETS Competitive Disadvantage for Environmental Impacts in Million
(h Billion p.a.) Airlines Under the EU ETS Tonnes CO2
Derigs and Illing In 4 out of 5 scenarios, the cost impact on Small for cargo operations in 4 out No significant CO2 savings estimated
(2013) cargo is either negligible or small of 5 scenarios. Only in scenario for three generic airlines
(scenarios: “BasicETS,” “ETS100,” “AggressiveETS,” significant
“IXHub” and “Green”). Only in the competitive disadvantages for
“Aggressive ETS” scenario, with an cargo airlines under the EU ETS
assumed allowance price of h70 per tonne can be expected
CO2 and 100% auctioning of allowances, a
decrease in profits can be expected
Malina et al. EU ETS will have a small economic impact Asymmetric effects of the EU ETS CO2 abatement will mostly take place
(2012) on US airlines and emissions. NPV of on competitiveness not in non-aviation sectors because
allowances purchased by US airlines: “Full considered in the model abatement costs will be higher for
scenario”: $1.37 billion; “Expense aviation than for other sectors. US
scenario”: $1.41 billion; “Absorb aviation under the EU ETS will
scenario”: $1.43 billion abate 3.35 million tons CO2 in
“Full Scenario” (1.6% of total CO2
emissions 20122020), respectively
JANINA D. SCHEELHAASE
1.17 million tonnes CO2 in
“Expense Scenario” (0.6% of total
CO2 emissions 20122020)
Scheelhaase et al. h1.453.04 billion p.a. (h20 per tonne CO2); Significant for European Airlines 299.13 million tonnes CO2 covered in
(2012) h2.906.08 billion p.a. (h40 per t CO2) for 2020; abatement conducted by non-
airlines under the EU ETS aviation sectors
Vespermann/ h2.25h3.67 billion p.a. (h25 per tonne CO2) Low competitive distortions 336.9 million tons CO2 covered in
Wald (2011) for airlines under the EU ETS 2020; about 26 million tonnes CO2
saved by the aviation sector in 2020.
Most of the abatement conducted
by non-aviation sectors
Source: Own compilation.

of the additional costs will be realistic, especially in the short and medium
term. Then a reduction in growth rates of these groups of airlines under the
EU ETS can be expected.
What is the impact of the EU ETS on air freight shipping companies?
As mentioned in the section “Economic Impacts of Local Charges on NOX
and HC Emissions in Europe,” peer-reviewed literature on the impact of
climate protecting regulations on air freight airlines is limited to Derigs and
Illing (2013) to date and peer-reviewed literature on the impact of climate
regulations on air freight shippers does not exist to date. Therefore, it is
only possible to provide some general insights on this issue.
As shown above, the cost impact of the European emissions trading
scheme is expected to be small for European and non-European cargo air-
lines. Only for a highly unlikely scenario from today’s point of view with
an assumed allowance price of h70 per tonne CO2 and full auctioning of
allowances, higher cost effects can be expected. This leads to the conclusion
that the impact of the EU ETS on air freight shipping companies will only
be small under current conditions. If a pass-through of the additional costs
to the consumers is not possible, some strategies will be promising for the
air freight shippers: Firstly, employing highly fuel efficient aircraft and
rerouting services in order to avoid European airports under the trading
scheme will be a favorable option. Depending on the final destination, hubs
like Frankfurt Airport, Paris Charles de Gaulle Airport, London Heathrow
or Schiphol Amsterdam Airport could be substituted by Istanbul Ataturk
Airport, Dubai International Airport or other airports. Secondly, a switch
in transport mode to road, rail or sea could be considered, too, if the final
destination of the freight is located in Europe.
FINAL REMARKS
All in all, cost impacts on air freight services induced by political measures
for the reduction of aviation’s climate relevant emissions turn out to be
small. This is true for both local emission charges on NOX and HC emis-
sions which are in force at a number of European airports and the
European emissions trading scheme for the limitation of CO2 emissions.
In principle, local emission charges aim at setting economic incentives to
foster the use of engines with low NOX and HC emissions. These charges
are designed to be revenue neutral in the sense that they must not increase
the airports’ overall revenues from air traffic. Case studies based on
empirical data indicate that airline’s costs will be affected differently by

these charges. The cost impact depends on the airframe/engine combination
and on the amount charged per unit emission value in the particular
European State applying the charge. The costs of freight operations are
affected by the local charges, but their impact is limited: For an A330-200
with an emission value of 44.02 kg NOX, the total airport charges were
raised by 3.97 percent, respectively h216.66 in absolute terms per landing at
Arlanda airport in 2010. This was the maximum costs increase for an
A330-200 in Europe for two reasons: Firstly, for many airlines, the actual
costs are much lower due to the principle of revenue neutrality. Secondly, if
freight is transported in the form of belly hold cargo, the costs for the
charge can be shared between passengers and freight. If a 50:50 cost alloca-
tion between passengers and belly hold cargo is assumed, the emission
charge increases the costs of cargo by h0.0054 per kg. For all-cargo opera-
tions, the cost situation is slightly different. Here, a small cost increase can
be expected.
The European emissions trading scheme (EU ETS) is in force since 2012
and covers all flights departing from or landing at any airport in the
European Union, Norway and Iceland. The scheme aims at limiting avia-
tion’s CO2 emissions. A number of studies has been carried out on the eco-
nomic impacts of the EU ETS on aviation lately. The results depend to a
great extent on the assumptions on the allowances prices, on the elasticities
of demand for air services and on the model chosen for the analysis. Only
Derigs and Illing (2013) investigate the impact of the EU ETS on air cargo
while the other studies concentrate on passenger or combined air services.
According to the authors, the cost impact of the EU ETS on cargo opera-
tions is expected to be limited. In the year 2020, costs for CO2 allowances
lie between h380,000 and h248,000 per airline (Derigs and Illing, 2013).
Here, three generic airlines have been analyzed: a typical all-cargo airline
such as Cargolux or Lufthansa Cargo; a typical mixed carrier such as
Lufthansa or Singapore Airline; and a representative for an express carrier
such as DHL, UPS or Federal Express. By optimizing the routing, CO2
allowances costs could be reduced by up to 9.6 percent. Much higher cost
increases can only be expected if the CO2 reduction target and the amount
of auctioned allowances will be raised drastically a scenario which does
not seem very likely from today’s point of view.
The impact of the European climate regulations on air freight shippers is
limited. The local emission charges will lead to a slight cost increase for
those freight shipping companies cooperating with all freight airlines or
using all freight aircraft in their possession. If the additional costs can’t be
passed on to the consumers by the freight shippers, they may choose to

switch from cooperating with all freight airlines to airlines offering belly
hold capacity. Another option would to be to optimize the routing in that
way that the European airports having introduced emission charges will be
avoided. There are numerous possibilities for avoiding these airports in
Europe. For instance could Cologne/Bonn Airport be substituted by
Leipzig/Halle Airport or Brussels Airport, both of which are not charging
for NOX and HC emissions and are geographically close to Cologne/Bonn
Airport. London Heathrow and London Gatwick could be replaced by
London Stansted or London Luton Airport, for example. Companies offer-
ing multi-modal freight transport services (as it is the case with many air
freight shippers) could also react to the emission charges by switching to
other modes such as road, rail or sea transport.
As stated above, the cost impact of the European emissions trading
scheme is expected to be small for European and non-European cargo air-
lines. This leads to the conclusion that the impact of the EU ETS on air
freight shipping companies will only be small under current conditions. If a
pass-through of the additional costs to the consumers is not possible, some
strategies are promising: Employing highly fuel efficient aircraft or rerout-
ing services in order to avoid European airports under the trading scheme
may be a favorable option for freight shipping companies. In case the final
destination of the freight is located in Europe, a switch in transport mode
to road, rail or sea could be considered, too.
REFERENCES
Airbus. (2012). Global market forecast 20122031. Blagnac: Cedex.
Anger, A., Köhler, J. (2010). Including aviation emissions into the EU ETS: Much ado about
nothing? A review. Transport Policy, 17, 3846.
Aston, H. (2013). China seeks Australia’s help building emissions trading scheme. The Sydney
Morning Herald, July 11, 2013, online edition. Retrieved from http://www.smh.com.au/
federal-politics/political-news/china-seeks-australias-help-building-emissions-trading-
scheme-20130711-2prjh.html. Accessed on August 14, 2013.
Australian Government. (2011). Securing a clean energy future the Australian
Government’s Climate Change Plan. Retrieved from http://www.cleanenergyfuture.gov.
au/wp-content/uploads/2011/07/Consolidated-Final.pdf. Accessed on May 24, 2012.
Australian Government/Commission of the European Union. (2012). Australia and European
Commission agree on pathway towards fully linking emissions trading systems, media
release, 28 August, 2012. Retrieved from http://www.climatechange.gov.au/minister/
greg-combet/2012/media-releases/August/JMR-20120828.aspx. Accessed on December
13, 2012.
Bloomberg. (2013). South Korea’s emissions trading scheme. Bloomberg White Paper. May
10, 2013. Retrieved from http://about.bnef.com/white-papers/south-koreas-emissions-
trading-scheme/. Accessed on August 14, 2013.
Boon, B., Davidson, M., Faber, J., & van Velzen, A. (2007). Allocation of allowances for avia-
tion in the EU ETS The impact on the profitability of the aviation sector under high
levels of auctioning. Report for WWF. UK: Delft.
CE Delft. (2008). Lower NOX at higher altitudes Policies to reduce the climate impact of avia-
tion’s NOX emissions. final report on behalf of the EU commission. Delft, NL.
Commission of the European Union. (2012). Commission proposes to ‘stop the clock’ on inter-
national aviation in the EU ETS pending 2013 ICAO General Assembly. 12 November,
2012. Retrieved from http://ec.europa.eu/clima/news/articles/news_2012111202_en.htm.
Accessed on December 12, 2012.
Copenhagen Airport. (2010). Calculation of the emission charge Retrieved fromhttp://www.
cph.dk/CPH/UK/ABOUT+CPH/Traffic+Statistics/Airport+Charges/Calculation+of
+emissions+charge.htm. Accessed on April 10, 2013.
Council of the European Union. (2009). Directive 2008/101/EC of the European Parliament
and the Council of 19 November 2008 amending Directive 2003/87/EC so as to include
aviation activities in the scheme for greenhouse gas emission allowance trading within
the Community. Official Journal of the European Union, L8/3, Brussels, 13 January
2009.
Derigs, U., & Illing, S. (2013). Does EU ETS instigate air cargo network reconfiguration?
A model-based analysis. European Journal of Operational Research, 225, 518527.
Ernst, & Young (2007). Analysis of the EC proposal to include aviation activities in the emissions
trading scheme. Report by Ernst and Young and York Aviation. Ernst & Young York
Aviation,York.
European Civil Aviation Conference (ECAC). (2003). Recommendation ECAC/27-4 NOX
emission classification scheme. Paris: ECAC.
European Voice. (2012). EU and China team up on emissions trading. Retrieved from http://
www.europeanvoice.com/article/2012/september/eu-and-china-team-up-on-emissions-
trading/75199.aspx. Accessed on December 12, 2012.
Europolitics. (2012). South Korea to start emissions trading scheme in 2015. Retrieved from
http://europolitics.info/external-policies/south-korea-to-start-emissions-trading-scheme.
Accessed on December 12, 2012.
Fraport. (2007). Einführung Emissionsabhängiger Entgelte. Hand-out for the meeting at the
German federal ministry of transport on December 6th, 2007. Fraport AG, Bonn.
Frontier Economics. (2006). Economic consideration of extending the EUETS to include avia-
tion. Report prepared for the European Low Fares Airline Association (ELFAA).
Retrieved from http://www.elfaa.com/documents/FrontierEconomicsreportforELFAA-
Economicconsideration_005.pdf
Government of China. (2011). 12th Five Year Plan. Retrieved from http://www.britishcham-
ber.cn/content/chinas-twelfth-five-year-plan-2011-2015-full-english-version. Accessed
on December 12, 2012.
ICF. (2006). Including aviation into the EUETS: Impact on EU allowance prices Final
Report. Report for Defra and Dft, ICF Consulting, London, 2006. Retrieved from
http://www.icfi.com/Publications/environment.asp#2006aviationeu.
International Civil Aviation Organisation ICAO. (2012). New ICAO Council High-level
Group to focus on environmental policy challenges, press release, 15 November 2012.
Retrieved from http://www.icao.int/Newsroom. Accessed on November 18, 2012.
International Civil Aviation Organisation ICAO. (2013). Resolutions adopted by the

Assembly 38th Session, Montreal, 24 September 4 October 2013. Provisional
Edition November 2013. Retrieved from http://www.icao.int/Meetings/a38/Pages/reso-
lutions.aspx. Accessed on January 07, 2014.
Lee, D. S., Fahey, D. W., Forster, P. M., Newton, P. J., Wit, R. C. N., Lim, L. L., Owen, B.,
& Sausen, R. (2009). Aviation and global climate change in the 21st century.
Atmospheric Environment, 43, 35203537.
Malina, R., McConnachie, D., Winchester, N., Wollersheim, Ch., Paltsev, S.,& Waitz, I. A.
(2012). The impact of the European union emissions trading scheme on US aviation.
Mendes, L. M. Z., & Santos, G. (2008). Using economic instruments to address emissions
from air transport in the European Union. Environment and Planning A, 40, 189209.
Morrell, P. (2007). An evaluation of possible EU air transport emissions trading scheme allo-
cation methods. Energy Policy, 35(11), 55625570.
Morell, P. (2011). Moving boxes by air – The economics of international air cargo. Farnham,
UK: Ashgate.
National Assembly of Korea. (2012). Decision on ETS on 3 May, 2012. Retrieved from http://
www.assembly.go.kr/main.acl. Accessed on January 18, 2013.
New Zealand Government. (2012). Climate change information. Retrieved from http://www.
climatechange.govt.nz/. Accessed on December 13, 2012.
Sausen, R., Isaksen, I., Grewe, V., Hauglustaine, D., Lee, D. S., Myhre, G., Köhler, M. O.,
Pitari, G., Schumann, U., Stordal, F., & Zerefos, C. (2005). Aviation radiative forcing
in 2000: An update on IPCC (1999). Meteorologische Zeitschrift. 14, 555561.
Scheelhaase, J. (2010). Local emission charges A new economic instrument at German air-
ports. Journal of Air Transport Management. 16, 9499.
Scheelhaase, J. (2011). Competitive distortions in the air transport market as a result of the
upcoming worldwide emissions trading systems? Airlines International magazine for
students and professionals of aviation. E-zine Edition, 49, 15.
Scheelhaase, J. D., & Grimme, W. G. (2007). Emissions trading for international aviation
an estimation of the economic impact on selected European airlines. Journal of Air
Transport Management, 13(5), 253263.
Scheelhaase, J., Grimme, W., & Maertens, S. (2011). A first assessment of the ecological
impacts of NOX based landing charges at German airports. Presentation on 7 October
2001 at Infraday Conference 78 October 2011, Berlin.
Scheelhaase, J., Hepting, M., Pabst, H., & Wilken, D. (2005). Emissionsentgelte im
Luftverehr. Final report, study on behalf of the German Federal Ministry of
Transport, German Aerospace Centre (DLR), Institute of Air Transport and Airport
Research. Cologne.
Scheelhaase, J., Schaefer, M., Grimme, W., & Maertens, S. (2012). Cost impacts of the inclu-
sion of air transport into the European emissions trading scheme in the time period
20122020. European Journal of Transport and Infrastructure Research (EJTIR), Issue
12(4), 2012, 332348.
SEC. (2006). Commission staff working document: Accompanying document to the proposal
for a directive of the European parliament and of the council amending directive 2003/
87/EC so as to include aviation activities in the scheme for greenhouse gas emission
allowance trading within the Community Impact assessment (COM(2006)
818finalSEC(2006)1685): Unofficial advance version, Brussels, 2006. Retrieved from
http://ec.europa.eu/environment/climat/pdf/aviation/sec_2006_1684_en.pdf.
Stockholm Environment Institute (SEI)/FORES. (2012). China’s carbon emission trading: An

overview of current development. Stockholm: FORES.
Vespermann, J., & Wald, A. (2011). Much Ado about nothing? An analysis of economic
impacts and ecologic effects of the EU-emission trading scheme in the aviation indus-
try. Transportation Research Part A, 45, 10661076.
Wit, R. C. N., Boon, B. H., van Velzen, A., Cames, M., Deuber, O., Lee, D. S. (2005). Giving
wings to emission trading Inclusion of aviation under the European emission trading
scheme (ETS): Design and impacts. A report for the European Commission’. Delft: DG
Environment.
Yong-Gun Kim. (2012). Emissions trading scheme for low-carbon green growth in Korea,
Korea Environment Institute, Presentation, 18 May, 2012. Retrieved from htpp://www.
kei.re.kr. Accessed on December 12, 2012.
CHAPTER 11
PLANNING A COMPETITIVE
AEROTROPOLIS
John D. Kasarda and Stephen J. Appold
ABSTRACT
Commercial aviation continues to grow but few passenger or cargo jour-
neys begin or end at airports. “Terminal” and “last” mile costs can place
considerable drag on interregional trade in goods and services, attenuat-
ing growth and prosperity. The aerotropolis model provides a holistic fra-
mework for understanding and addressing trade costs. The central
tenets of the aerotropolis model are outlined and extended by considering
the decision to establish a new business facility. Implications are drawn
for planning a competitive aerotropolis as the global economy enters a
new era.
Keywords: Airports; aerotropolis; cities; logistics; planning; trade
JEL classifications: L93; R12; R14

ISSN: 2212-1609/doi:10.1108/S2212-160920140000004010
281
282 JOHN D. KASARDA AND STEPHEN J. APPOLD
INTRODUCTION
An aerotropolis, discussed in more detail below, can be defined as a

“planned and coordinated multimodal freight and passenger transportation
complex which provides efficient, cost-effective, sustainable, and intermo-
dal connectivity to a defined region of economic significance centered
around a major airport.”1 The concept has come into prominence for four
reasons. First, passenger and cargo air transportation has grown over the
last several decades making airports increasingly important central urban
places while sometimes placing strains on the capacity of airports, particu-
larly those which serve as key hubs. Second, the ground-based portions of
air journeys are often characterized by delays and other frictions which
decrease the attractiveness of air travel, and thus some types of interregio-
nal trade. Third, as cities grow, they expand outward, increasing pressure
on suburban land, including that surrounding airports. Fourth, land pro-
motion is a workable tool for regional growth coalitions which view air-
ports as potentially viable economic development assets.2 The value
proposition of the aerotropolis concept is that it has the potential to satisfy
all four concerns. Accordingly, the aerotropolis concept is gaining promi-
nence around the world.3
Stripped to its basics, the aerotropolis model is composed of four ana-
lytically separable but interdependent elements. Its connectivity elements
are air routes along with highways, rail systems, and surface linkages to
ports that provide the aerotropolis with internal and external accessibil-
ity. Its spatial elements consist of aviation-oriented businesses and asso-
ciated residential developments that concentrate near the airport and
outward along its transport corridors generating physically observable
form. Its functional elements include the competitiveness (welfare) gains
to producers and consumers generated by the air connectivity and airport
spatial propinquity. Its planning elements include the basic principles of
aerotropolis development and the strategies intended to make the poten-
tial benefits reality.
This chapter describes the four elements of the aerotropolis model as it
now exists, focusing on its functional value proposition. Though a dynamic
model, represents a universal ideal-type end point. By elaborating on the
firm facility investment decision, we extend the model to guide the move-
ment of air-enabled manufacturing facilities to locations which are poorly
served by air service. A further extension explores the implications of access
replacing location for the spatial structure of regions with major and minor
airports. Our aim is not only to inform theory but also help aerotropolis
Planning a Competitive Aerotropolis 283
planners and developers avoid disappointing outcomes. We close with a

discussion of some implications for the future development of the aerotro-
polis model.
THE CONNECTIVITY ELEMENT OF THE

AEROTROPOLIS MODEL
Aviation and its role in the global and regional economies are central to
the aerotropolis model. Airline routes operate as a critical component of
the “physical internet” moving products and people quickly among regions
and around the world analogous to the way the digital Internet moves data
and information (Kasarda, 1998/1999, 2014; Kasarda & Lindsay, 2011).
The routers of this physical internet are airports which are the concrete
interfaces where the global meets the local in people and product move-
ments. An illustration of the aviation physical internet is shown in Fig. 1.
As can be seen in Fig. 1, this physical internet spans much of the globe
but is, until now, most dense in the northern hemisphere, particularly
Europe and North America, with emerging clusters of density in Asia. As
is true of the Internet, most of the traffic on the physical internet is concen-
trated between the places with a combination of population and wealth.
Most of the connectivity is intra-continental with an especially dense set of
Fig. 1. Aviation’s Global Physical Internet (59,036 Routes in 2012).

routes in Europe. Even if only about 10 percent of U.S. air travel is interna-
tional (in Europe, a 45-minute flight could be international), the cross-
Atlantic routes are still prominent. The somewhat thinner Pacific Rim and
Eurasian routes are also visible as are some others. The importance of loca-
tion in the development of hub airports can be inferred from the proximity
of the Gulf region to the Great Circle routes between the travel markets of
Western Europe and South Asia. A similar advantage on the “Kangaroo
route” between Europe and Australia has helped give Singapore a boost in
air traffic. Geographic location is a key factor in the location of all airline
hubs (not necessarily readily visible in Fig. 1), and the geography of the
global system of cities has determined air transportation patterns.
The critical importance of leisure travel can be inferred from the concen-
tration of routes leading from the prosperous cities of Northern Europe to
sea-side vacation destinations, a similar focus on Florida and the
Caribbean, and heavier-than-expected linkages to selected islands in the
Indian and Pacific Oceans. Given that roughly 7080 percent of passenger
air travel is for leisure purposes, depending upon route and season, the net-
work of business travel is considerably thinner than the full network illu-
strated in Fig. 1. A good portion of the business travel network would be
between the cities near the top of the global system of central places but, as
in the case of leisure, most business travel is intra-continental (Kasarda,
2013). In the United States, one-day trips are responsible for approximately
10 percent of business air travel, meaning that the air journeys are relatively
short, perhaps of no more than an hour or two in duration. New
YorkChicago and Los AngelesSan Francisco, for example, are both
busy U.S. business routes.
Cargo is responsible for somewhat over 10 percent of airline revenues.
As in the case of passenger travel, there are several, nearly separate air
cargo physical internets. The most familiar of these may be the hub-and-
spoke distribution pattern of the express companies which specialize in
overnight parcel and document delivery. Several domestic passenger air-
lines have found market niches in same-day package delivery over moder-
ate (intra-continental) distances. In both cases, the pattern of shipments
tends to follow the urban central place hierarchy. In addition, a combina-
tion of scheduled freighters and passenger aircraft cargo much of the inter-
national air cargo along well-defined trunk routes between major airports
(Sheffi, 2012). These often are part of the production and supply networks
which stock the centers of distribution. Economies of density, similar to
those which support the hub-and-spoke system of passenger transportation,
tend to concentrate air cargo along busy trunk routes with the remaining,
“spoke,” portion of the journey often undertaken by long-distance truck.

Economies of scope and scale in airport processing tend to hold much
cargo in the major passenger airports.
This combination of economies in shipping implies that several impor-
tant passenger nodes in the system are also key cargo nodes. Much of the
international air cargo tends to move within a small number of trade lanes
between large areas of production, for example, the Pearl River Delta, and
central distribution centers in the major North American and Western
European markets, which themselves may serve continent-wide market or
catchment areas. Despite the promise of speed, international air cargo is
surprisingly slow, with door-to-door delivery times for international cargo
which have averaged about a week for several decades. Additional cargo
moves via chartered aircraft which are not constrained by market size but
react to order size.
Air transportation is only a small part of the physical internet, responsi-
ble for only a small proportion of total miles traveled and ton-miles
shipped. Much larger volumes of passenger and cargo move via surface
modes. Nevertheless, air transportation dominates other modes for moder-
ate- and long-distance passenger travel. Further, while many products are
not traded and only a small volume (less than 1 percent by weight) of inter-
national trade moves by air, that small volume is responsible for approxi-
mately 35 percent of the value (International Air Transport Association
[IATA], 2013).
Air transportation continues to grow. Much of that growth will occur in
the rapidly growing, emerging economies of the world, often from small
bases. In the high-income countries of North America, Western Europe,
and elsewhere, the industry has matured and in some markets may be
reaching saturation. The management of long-distance supply chains has
also come of age. The mode share air cargo of international shipments
increased in the 1990s but reversed course the following decade until the
economic crisis made retailers and others reluctant to hold inventory,
prompting a greater use of air cargo albeit with significantly smaller
volumes of total trade (Greis & Appold, 2011). Part of the loss of mode
share may be due to the diffusion of supply chain management skill which
allows firms to more closely and less problematically match supply with
demand while using aviation more sparingly. Another part is likely due to
the significant improvements in port and hinterland processing of ocean
freight which strengthen the alternative to shipping by air. Ocean shippers
now offer time-definite delivery. Moreover, shipping economics has been
profoundly affected by increases in the real cost of fuel.
Ironically, some of the loss of mode share may also be due to the
increasing volume of trade. Air cargo is most effective when small- or
medium-sized shipments are needed. As volumes increase, the speed advan-
tage can shift from air to surface modes. Whereas in 2000, the volume of
trade from, say, a China production platform to a European central distri-
bution center may have been sufficient to justify a weekly planeload of
cargo, volume may now be large enough and the total costs low enough to
justify a weekly shipment by train (Bradsher, 2013).
THE SPATIAL ELEMENT OF THE AEROTROPOLIS

MODEL
The connectivity element of the aerotropolis model may generate a spatial
element. As noted above, the aerotropolis model is an ideal-type urban sub-
region whose infrastructure, land use, and economy are centered on an air-
port. A compressed schematic of the spatial element of the aerotropolis
model with its airport city (Conway, 1980) core is shown in Fig. 2. No real-
Fig. 2. Aerotropolis Schematic with Airport City Core.

world aerotropolis will look exactly like this illustration but many could
eventually take on similar features, led by newer “greenfield” airports less
constrained by numerous prior decades of nonaviation-oriented surround-
ing development.
The aerotropolis model contains a full set of logistics and commercial
facilities supporting aviation-enabled businesses, cargo, and millions of air
travelers who may pass through the airport annually. By “aviation
enabled” we mean firms and industries that are able to operate in their pre-
sent form primarily because of the connectivity afforded by passenger and
air cargo transport. These include, among others, freight forwarding, third-
party logistics (3PL), warehouse and distribution facilities, hotels, recrea-
tion, wellness, convention and exhibition complexes, and office buildings
along with shopping, dining, leisure, entertainment, and tourism venues.
A list of common airport, airport city, and aerotropolis commercial
facilities, adapted from Güller and Güller (2001), is provided in Table 1.
These are organized by whether the facilities are people or goods oriented
and by typical location including in or near the terminals (airside); land-
side, on or adjacent to the airport property (airport city) and further
beyond the airport boundaries (aerotropolis). As one progresses outward
from the terminals to the airport city to the aerotropolis, most facilities in
the inner zone are replicated in the broader zones, frequently at a
larger scale.
Real-world manifestations of the aerotropolis model may (and do) unfold
differently over time, depending upon their airline routes, passenger demo-
graphy and volume, cargo demand, airport and airport area land availabil-
ity, surface transportation infrastructure, regional industry structure and
economic conditions, local labor resources, real estate markets, and other
factors. Therefore, the mix and location of commercial facilities can vary
significantly as can airport city and aerotropolis physical form.
Consequently, the spatial form and the functions listed may be most
applicable to the relatively few hub airports which serve as international
gateways and, because of the size of their metropolitan markets and geo-
graphic location, are able to anchor a large choice of flights and destina-
tions, more frequent service, and more flexibility in rescheduling (i.e., they
possess the fastest and broadest physical internet). As suggested above,
even among the major hub airports, market size implies that relatively few
are able to support more than a handful of inter-continental flights. These
cities will be near the top of the global hierarchy of central places.
Therefore, comparatively few airports will be able to support the full range
of facilities by themselves.
Table 1. Common Airport City and Aerotropolis Commercial Facilities.

People Oriented Goods Oriented
Passenger and cargo • Retail (including upscale • Air express and courier
terminals (airside) boutiques) • Cold storage and cool chain
• Restaurants (higher end and • General air cargo
themed as well as fast food) • Aircraft maintenance, repair,
• Leisure (spas, fitness, recreation, and overhaul (MRO)
cinemas, etc.)
• Culture (museums, regional art,
musicians, chapels)
Airport city (landside) • Hotels and entertainment • Logistics and distribution
• Office and retail complexes • Wholesale merchandise marts
Aerotropolis (beyond • Convention and exhibition • Free trade zones and special
airport property) centers economic zones
• Business and technology parks • Logistics parks and
• Producer services (finance, distribution centers
auditing, consulting, etc.) • Precision and time-critical
• Corporate headquarters manufacturing
• Information and communication • Biomeds and pharmaceuticals
technology firms • High-tech electronics repair
• Wellness and medical facilities • High-value agricultural and
• Large mixed-use residential food products
developments • Medical instruments
• Aviation-related industries
Adapted from Güller and Güller (2001).
THE FUNCTIONAL ELEMENT OF THE

AEROTROPOLIS MODEL
If the most recognizable feature of the aerotropolis model is its spatial ele-
ment, its operational value proposition rests on the degree to which the
spatial element contributes to its functional element. The value of an aero-
tropolis is its ability to increase welfare by reducing the costs of trade.
Accordingly, the functional benefit of the aerotropolis model rests on two
pillars: the benefits of aviation for interregional transport and the benefits
of airport proximity to reduce the costs of intraregional movement.
The first is on the ability of air transport (the connectivity element) to
lower the delivered price of an array of goods and services. Air transport,
despite the cost of the ticket, economizes on traveler time allowing sales
representatives and support personnel to cover larger market areas more
effectively. Larger market areas allow for specialization and economies of

scale in production. Reduced travel and logistics cost allows production to
be more efficiently sited. These are the gains in trade due to commercial
aviation.4
Whereas air transportation is fast, it is also expensive. Speeding physical
movement, however, reduces the need for capital investment (in inventory,
storage space, etc.). Speed to market (response time) over long distances is
particularly critical to high-tech industries and high-value perishables sec-
tors (Suri, 2010).
Aviation’s value proposition applies to producer service industries,
goods-handling sectors, and leisure services and tourism, even if portions
of this chapter concentrate on one at the expense of the others. Tourism,
by some accounts the world’s largest industry and one of the fastest grow-
ing, brings significant revenue to some regions. In addition to being a major
industry itself, leisure travel contributes to a higher quality of life. Air tra-
vel to visit friends and family helps maintain social ties while facilitating
the efficient working of the labor market by reducing the reluctance to
migrate to opportunity.
The second pillar of functional benefit in the aerotropolis model rests on
the ability of proximity to an airport (the spatial element) to increase the
cost-savings of air transport. People and product air journeys neither begin
nor end at the passenger and cargo terminals. Passengers and cargo often
spend considerable time and expense in getting to and from airports and in
negotiating the airport, creating “terminal” and “last mile” costs. Terminal
costs refer to the time, money, and effort required for travelers and ship-
ments to make their respective ways through an airport. Last mile costs
refer to the time, money, and effort required for travelers and shipments to
make their respective ways to (or from) the airport. Terminal and last mile
costs are often substantial, placing a drag on response times and trade. The
airports and regions which successfully minimize these costs are often able
to enhance operational efficiencies and therefore their attraction as a loca-
tion for business investment.
Much of the time and money cost of transportation occurs at terminals.
Airports are critical but costly points of intermodal transfer. Time spent
waiting at airports, whether checking in, going through security, or just
waiting in the terminal effectively increase the costs of travel, discouraging
the trade that the travel represents. Most of the time needed for cargo ship-
ments is spent standing still. Accelerated customs clearance processes at the
airport are helpful but cargo can wait days in a terminal area (or waiting to
be brought to a terminal area) because no flight to the desired destination
is scheduled or because those which are scheduled are full. Any savings in
time and costs at these intermodal interfaces have substantial system-wide
efficiency effects in reducing the frictions (costs) of trade.
Last mile road congestion near the airport (and downtown) can create
especially adverse air-journey conditions. Combined with terminal costs,
last mile costs imply that more than half the time spent on air travelers’
journeys between Chicago’s downtown and New York City’s downtown is
often spent on the ground, stuck in freeway gridlock, terminal congestion,
or backed up on airport taxiways. Particularly for flights of moderate
length the most common trajectory for air travel such delays are espe-
cially onerous, often effectively doubling the needed travel time.
Accordingly, shippers, freight forwarders, business service providers, and
frequent air travelers make location and travel mode choices on the basis
of the time and cost of the entire (surface and air) journey.
Time and cost reductions may result from improved ground transporta-
tion (e.g., airport expressways and airport commuter rail), land use controls
(e.g., reserving land near airports for aviation-intensive businesses), state-
of-the-art ICT infrastructure (e.g., high-speed telecommunications), and
institutional reforms (e.g., accelerated customs clearance). All are instru-
mental to creating economies of speed. This means reducing time at the air-
port (or at least the disutility of time at the airport) and reducing time
getting to and from the airport. To the extent the speedy long-distance con-
nectivity afforded by air transport is complemented by efficient airport area
spatial development, the aerotropolis is a useful strategy leading to regional
economic gains from trade in goods and services beyond those that could
otherwise be realized (Appold & Kasarda, 2010).
THE PLANNING ELEMENT OF THE AEROTROPOLIS

MODEL
The fourth element of the aerotropolis model is the principles and strategies
to maximize regional benefit. In the aerotropolis model, planning spans
commercial land use and transportation planning in concert with urban
planning. Aerotropolis planning therefore includes reconciling (1) the busi-
ness site and profitability objectives of individual firms making capital
investments, (2) airport and surface transportation planning objectives of
ensuring maximal access to the airport and business sites at the lowest pos-
sible cost, and (3) the urban planning objectives of overall economic
efficiency, aesthetic appeal, and social and environmental sustainability

(Kasarda, 2010). With respect to transportation planning, aerotropolis
planning also includes designing systems for efficient, secure cargo logistics
and for efficient, safe personal mobility.
Fig. 3 illustrates the golden ring of aerotropolis planning which crosses
(and integrates) airport, urban, and business site planning domains.
Aerotropolis planning is unique in that business, urban, airport, and sur-
face transport objectives are addressed together to foster personal and
logistics mobility along with economically and socially desirable urban
development. Aerotropolis planning thus differs from conventional airport
and urban planning by considering “inside the fence” terminal, mutually
beneficial “outside-the-fence” development, and “last mile” costs holisti-
cally. Such integrated planning can serve as an antidote to the chaos, con-
gestion, and unsightliness that has resulted from organic, haphazard
development around so many major airports, detracting from the opera-
tional functionality and image of these areas.
Basic principles that apply to commercial real estate investment in gen-
eral apply to airport city and broader aerotropolis commercial development.
Fig. 3. The Golden Ring of Aerotropolis Planning.

Yet, some specific principles are especially germane to planning and devel-
oping airport cities and the aerotropolis.
1. Airports cities and their extended aerotropolises are not simply major
capital investments which must deliver positive financial return over
many decades. They are also major “public goods.” Therefore, careful
long-term planning is called for to ensure maximum value is created
for users, investors, nearby communities, the metropolitan region, and
the nation.
2. Airport city and aerotropolis development is part of a broader invest-
ment and commercial location system. Airport city and aerotropolis
planning must be cognizant of the direction of local urban develop-
ment and competing facilities in the region. Sites in the path of out-
ward urban development from the central city will typically benefit
while alternative sites in the region may compete with them for com-
mercial facilities investment. In some cases, sites external to the region
may compete, as well.
3. Aligning key stakeholders is essential for successful aerotropolis devel-
opment. Aerotropolis development is a fundamentally collaborative
venture among land owners, investors, developers, and infrastructure
and aviation service providers, including government bodies and air-
lines. Therefore, aerotropolis planning needs to understand not only
the potential costs and market considerations which may influence the
location decisions and facility investments of potential aerotropolis
businesses, but also how government and airline decisions may impact
these investment and location decisions.
4. Regional economic conditions and real estate market demands shape
the development pace and characteristics of each airport city and aero-
tropolis. Since form follows function, both airport commercial prop-
erty planning and greater aerotropolis facility planning should be
coordinated and supported by an ever-improving analysis of unmet
regional business needs and local real estate demand in order to man-
age investor risk and to better position airport city and aerotropolis
offerings. Aerotropolis planning is not only urban planning; it is also
economic planning based on business logic. It has been our experience
that airport commercial facilities and aerotropolis development pro-
jects underpinned by solid documentation of market demand usually
get funded and prosper. Careful assessments of market conditions,
investment risk, and regional competitors are necessary prerequisites
all too often overlooked in airport city and aerotropolis master plans.
5. In the aerotropolis model, the 3As (accessibility, accessibility, accessi-

bility) supersede the 3Ls (location, location, location) as the pertinent
commercial real estate development principle. Minimizing timecost
access to the airport and other critical metropolitan nodes is a primary
objective of efficient aerotropolis planning.
6. A successful airport city and aerotropolis will build on evolving econo-
mies of speed, scale, and scope in providing benefits to tenants, users,
investors, businesses, and the region. Yet, those economies usually only
fully exist at or near development maturity. Therefore, both urban and
airport planners and managers need to construct development path-
ways which will generate shorter term investment returns and contin-
ued infrastructure improvements through the earlier stages of airport
commercial property and greater aerotropolis growth.
7. Aerotropolis residential communities housing airport area workers and
frequent air travelers should be developed that are welcoming, provide
a sense of place, and offer on-site or nearby services and urban ame-
nities appealing to modern lifestyles. These communities should be
built outside flight paths but in proximity to aerotropolis job clusters
and surface transportation (including public transport) to reduce com-
mute times and costs.
8. Aerotropolis development and “smart” urban growth can and should
go hand in hand. Redensification around airports and planned cluster
development outward can be an antidote to sprawl and other hapha-
zard development that detracts from airport area functionality, sustain-
ability, and image.
9. The ultimate success of the aerotropolis rests on the aviation-enabled
advantages it provides to firms and the value it brings to regions and
their residents. These will be measured primarily in terms of business
revenues, aggregate regional wealth, and quality of life generated.
10. Getting the aerotropolis right will require integrating airport planning,
urban planning, and business site planning. In absence of such inte-
grated planning, the aerotropolis will not be as economically efficient,
attractive, or as environmentally and socially sustainable as it
might be.
Integrated land use and transportation planning is a valued ideal that

ideal is rarely achieved. While the aerotropolis represents the functional
integration of the airport, its airlines, surrounding municipalities and the
broader region, master planning all too often remains silo-ed (fragmented)
in airport planning, air route planning, surface transportation planning,
and land-use planning at various local government or territorial levels. Far

less than optimal outcomes may result for the airport, airlines, surrounding
municipalities, and the broader region.
AEROTROPOLIS FIRM SITING DECISIONS
The previous sections of the chapter outlined the aerotropolis model as it

has developed to the present, concentrating on the central causal connec-
tions of the model. That model is perhaps best suited to understanding air-
port regions after they have developed. Planning models need to
understand and help direct change. Available empirical evidence is briefly
reviewed and the model extended in this and the subsequent section.
The aerotropolis model holds that air connectivity the physical inter-
net creates traffic at key airports the routers generating a spatial
impact surrounding the connected airport. The model further holds that
improving air connectivity (measured by the number of markets served
times the frequency of service to those markets, sometimes weighted by the
size of the markets served) will increase the volume of regional high-value
exports because producers will find such airport areas attractive to invest-
ment in facilities. Critically, the model holds that proximity to an airport
(the spatial element) will decrease the costs of firm operations, leading to a
regional competitiveness boost. The multifaceted nature of contemporary
economies implies that integrated planning is necessary to optimize aero-
tropolis development.
These tenets need to be considered in the light of recent experience. For
example, as specific regions have lost their niches in the global economy,
they have frequently lost air service. The excellent connectivity formerly
held by some U.S. cities, such as Pittsburgh, Cincinnati, St. Louis,
Cleveland, Kansas City, and other now dehubbed or deemphasized air-
ports, was not sufficiently strong to grow or even preserve employ-
ment, undermining a key aerotropolis prediction. Even the direct
connections to Asia at an important node such as Detroit have not necessa-
rily translated into significant airport development or regional prosperity.
Some of the regions named have made large expenditures for passenger
terminals and other infrastructure investment which are now severely
underutilized. At the same time, other regions, once lacking in air connec-
tivity, have become attractive to specific types of business investment and
subsequently develop air service, with portions of Asia and the Middle East
offering the most spectacular evidence.
Given the predominance of leisure travel, the available empirical evi-

dence suggests that regional air service is largely a consequence, rather than
a cause, of economic development. Although there are historical excep-
tions, skilled labor tends to cumulate in urban central places and air service
tends to be more extensive where there are large numbers of well-paid peo-
ple, particularly those working in professional and business services. As air-
lines serve markets not airports, the size and commercial scope of an
airport’s catchment area has a critical impact on the development of air
service.
To be sure, economies of scale in certain aspects of airport operation,
economies of density along key trunk routes, and economies of cross-
subsidization tend to hold air services at particular airline hubs for a period
after the local need for air service has declined. These can similarly delay
service after demand has increased. Yet, overall the rise and decline of the
competitiveness of regions and their corresponding markets have driven
changes in air service. As prosperity diffuses throughout the global econ-
omy, we can expect air service to increase broadly but expand the fastest in
rapidly growing regions.
In order to understand how these shifts can take place, we focus on the
growth of new air service in new places through the lens of the business
facility investment decision. Three factors, which typically interact (Button,
2010), regularly appear with respect to aerotropolis business siting:
1. Air service
2. Labor force, and
3. Regional market size.
To the extent these factors are important location factors, investment
decisions would tend to reinforce existing spatial patterns. Large central
places might attract all aerotropolis investment. Understanding how
that inertia can break down is important to greenfield aerotropolis
development.
The business facility investment decision is a complex hierarchical deci-
sion comprised of three broad stages. Not all factors are mentioned here.
The process begins with the perceived need for capital investment. That
need is generally based on the pending introduction of a new product or
the perception that existing physical capacity is insufficient for existing or
anticipated demand. Firms are also likely to invest in new locations as
existing facilities near the end of their economic lives. The basic investment
decision often triggers a two-stage search for a new site: first, an optimal
region is chosen based on considerations of market size, labor skills, and
other firm needs along with land, labor, and transportation costs, and then
a specific site within the region is selected. In this section, we consider the
first step in that two-stage search. The second stage is considered in the
next section.
We use Foxconn’s recent decision to locate a facility in Zhengzhou,
China as a vehicle to interpret the decision to invest in a “remote” region.
Taiwan-based Hon Hai Precision Industry Co. is now the world’s largest
electronics contract manufacturer and is responsible for the assembly of
approximately 40 percent of all consumer electronics products sold.
Subsequent to its decision to expand to Zhengzhou in late 2009, the
Foxconn facility has grown to employ 240,000 by 2013, producing over 70
percent of Apple’s iPhones worldwide. 5 The Zhengzhou facility also
assembles other products. Fig. 4 illustrates the aviation-dependent global
supply chains which have resulted as a consequence of Foxconn’s location
decision. In 2008, neither the factory nor the connectivity existed, however.
Therefore, air service was not a factor in the site selection process, even if
the potential for air connectivity was.
Today, Asia is a rapidly growing consumer market in its own right but,
in the 1960s, when U.S. and European firms began using air transport to
off-shore production in a process of labor arbitrage, it was primarily a
source of inexpensive labor (Jackson, 2013). The Zhengzhou complex is
such an export-oriented facility. Despite the higher transport costs entailed
Fig. 4. Global Supply Chain Apple iPhone5 Zhengzhou, China Assembly

Complex.
by production in Asia, overall costs can decrease by tapping lower wage

labor. Electronics was one of the first products to be off-shored because
assembly was labor-intensive and total logistics costs were are just a few
percentage points of total delivered cost to the point of purchase. These
characteristics continue to apply to several contemporary microelectronics
devices such as smart phones which are typically shipped internationally
by air.
While extreme, some economists have begun to consider production
location patterns in which moving certain goods is nearly costless
(Glaeser & Kohlhase, 2004). The substantial decline in long-haul air
transportation costs experienced in the 1960s was instrumental to the
emergence of globally widening systems of production and supply.
Without low-cost long-distance transportation, global supply and distri-
bution channels would not be practical for many goods. China certainly
could not have become the “factory of the world” in absence of low-cost
long-haul transportation. In particular, without today’s wide-body jet air-
craft, China would not be the dominant location for smart phone and
other digital product assembly.
Off-shoring resulted in the growth of export-oriented production nodes.
These received a large boost when China opened selected areas for foreign
investment. Foxconn developed a large “city,” employing hundreds of thou-
sands near Shenzen which relied on labor migrants from the internal regions
of the country. By the mid-2000s, rising labor costs were eroding many of
the advantages of a Pearl River Delta location and many firms were consid-
ering relocating to cities in Western China (and were being encouraged to
do so by the Chinese government). Similar to the earlier rounds of off-shore
investment, such relocations had the potential to decrease total wage costs,
but at the cost of higher transportation costs. Firms but were hesitant to
relocate, given the perceived risks involved, however.
At the same time, Foxconn was facing a capacity shortage due to rising
demand for its products, fulfilling the first condition for a facility invest-
ment. A firm in that situation might expand an existing facility or locate a
new facility at the development frontier of the metropolitan region in which
they were located. They could also locate a facility in a different, region
with the labor supply, support firms, and transportation options required
if one could be found. Due to the economies summarized above, air
transportation costs, and possibly total logistics costs, might be minimized
by locations near busy hubs. However, high labor costs sometimes push
firms away from regions with those advantages. Consequently, the regions
where production costs were sufficiently low to compensate for increases in
transportation costs, however high they may be, become candidates for
location.
Two factors allowed Foxconn to choose Zhengzhou, a relatively remote
site, as a region for production in the first stage of the site selection process.
First, the firm received an incentive package from Zhengzhou which
included a special bonded zone for the iPhone assembly plant, full site per-
mitting at all government levels, and a factory complex including dormi-
tories and ancillary services adequate for over 100,000 workers built and
ready to operate. The package reduced the cost and risk of relocation, in
part by decreasing the opportunities for local officials to exploit the firm, a
significant risk given the perceived level of corruption in China. At the same
time, the firm’s massive size and solid record of performance reduced the
risk to the local government that it would invest in a firm which would fail.
The second factor freed the firm from the limits of existing air service.
The volume of planned production meant that the location decision was
not constrained by available air service. Economies of scale in production
(much of which is internal to establishments) imply that, as the single estab-
lishment would supply a large majority of global demand for the iPhone,
air service could be attracted to whatever location chosen solely to serve
Foxconn. Given sufficiently large shipments, producers can use dedicated
freighters which may be chartered. Note that Fig. 4 shows a very modest
number of routes. Had a larger number of routes been necessary from the
start, the new factory would have been more difficult to establish in that
location. Similarly, had the planned volume been smaller, the relocation to
a remote location would not have been feasible. Conversely, the Foxconn
facility would be difficult to establish on a true greenfield site. Its location
on the periphery of a city of 8 million implies that urban services could be
cost-effectively delivered. Further, the city can offer the amenities and
transportation options needed to attract workers.
This pattern of negotiation, incentive, and employment filtering down
the urban hierarchy has played itself out in several Western Chinese cities
recently. Over 50 cities are engaged in a hunt for new employers using land
supply policies, free trade zones, tax incentives, and accelerated permitting
as enticements (Zhou, 2013). Given that the sectors targeted are those for
which air shipments are viable, an airport capable of landing a 747 freigh-
ter is frequently a component of the negotiations, even if new rail connec-
tions are substituted for some links in the supply chain.
Two features are key in these negotiations. First, the primary focus was
on recruiting a large employer and minimizing total costs of productive
capacity, respectively. Air service was a secondary in aerotropolis
development. Second, minimizing costs and risks to both sides was critical.
An aerotropolis is not necessarily a rewarding real estate investment.
Subsidized land and other facilities for an initial tenant is an investment
which can be recouped only if follower firms find increased operational
economies worthy of compensatory prices. Returns are not assured.
SPATIAL EVOLUTION PATTERNS

After the need for a new facility is apparent, the second stage of the loca-
tion decision is the selection of a site within the chosen region. Two unfold-
ing land-use patterns offer important orientations for airport city and
aerotropolis planners. The first relates primarily to the commercial growth
of the airport, the second to the path of broader urban development.
As is illustrated in Fig. 5, airport cities grow outward from the passenger
and cargo terminals as rising levels of passenger and cargo traffic induce
greater numbers and sizes of facilities to meet expanding volumes of activ-
ity, exhausting available terminal space. With air traffic growth, increasing
numbers of commercial and logistics support activities achieve threshold
levels, and some of those activities can profitably exist further away from
passenger and cargo terminals. Rising air traffic volume acts as an attrac-
tion in itself, possibly drawing additional commercial and logistics func-
tions to the airport area.
Fig. 5. Airport Cities Grow Outward from Terminals as Activities Fill Capacity.
Note: Base loosely on Schiphol Airport development.
Space-constrained airports sometimes relocate parking and car rental

facilities off-site in order to give precedence to higher value aeronautical
activities, as the Atlanta airport has done. In other cases, private operators
do the same. Hotels frequently congregate around airports and sometimes
locate directly adjacent to terminals. These land uses are important but
aviation-oriented businesses occupy only a small portion of the available
land surrounding most airports. Just as air freight is only a small fraction
of the cargo processed in metropolitan regions, air passengers, despite the
sometimes large numbers, are only a small portion of the people moving in
or through most regions.
On the other hand, because airport cities and aerotropolises are parts of
broader regions, the path of urban development outward along radial corri-
dors, sometimes creating suburban subcenters, will have a significant effect
on the timing, nature, and process of airport city and aerotropolis develop-
ment (Appold, 2013b). In some cases, cities have expanded outward toward
their airports. In other cases, the dominant development pattern is away
(e.g., Atlanta) or in a different direction (e.g., Detroit). In some cases, urban
growth was toward the airport, before it was even planned. In other cases,
the land toward the airport is simply the last land available. Fig. 6 illustrates
Fig. 6. Cities Grow out to Airports as Air Transport Exerts Pull. Source: Based
loosely on Washington, D.C. and Dulles Airport development. Note: Numbers
indicate sequence of development.
a prototypical order of urban development from the city outward, in this

case, toward the airport.
While airports often exert a substantial gravitational pull on firms mak-
ing location decisions, so do many other factors, including a need for access
to the metropolitan area’s central business district, sector-specific concen-
trations of firms, labor pools, and amenities. Consequently, in the metropo-
litan regions which grow, commercial development proceeds from the
center out. Office buildings, business parks, and suburban edge cities may
thus develop along major highway corridors linking the metropolitan cen-
tral city to the airport, allowing tenants access to the airport as well as
needed inputs the central city offers to their business processes.
Such development may even “leapfrog” the airport to underutilized land
on its other side, accelerating the outward aerotropolis growth pattern illu-
strated in Fig. 6. This often results in a concentration of overflow urban
white-collar functions in the airport area as is the case for Amsterdam
Schiphol, Dallas-Ft. Worth, and other airports. Chicago’s O’Hare area has
the second largest concentration of Class-A office space in the entire U.S.
Midwest and the Washington Dulles region has more private sector office
space than the District of Columbia (Appold & Kasarda, 2013). The
growth of the producer services sector has overwhelmed the capacity of
CBDs to accommodate the attendant facility needs.
The importance of urban factors in bringing commercial development to
airports is highlighted by a consideration of corporate headquarters. The
gravitation of headquarters to airport areas means that more than 50 per-
cent of Fortune 500 corporate headquarters are located within 10 miles of
U.S. hub airports (Stilwall & Hansman, 2013). Revealed preferences, how-
ever, suggest that closer proximity may not be salient. FedEx, for example,
relocated its headquarters from near the Memphis airport to a more distant
suburban location. Boeing, when relocating its corporate headquarters
from Seattle to Chicago, bypassed the O’Hare area for a location near the
Loop. Airport areas often lack the prestige and amenities needed to attract
or hold high-level corporate functions.
Some airport areas do support corporate headquarters locations. For
example, Dick’s Sporting Goods, a U.S. retailer, may be the only large
firm in the United States to locate its headquarters directly on airport
property. The firm’s first-choice location was over 15 miles away. They
were unable to circumvent key site restrictions, however. The airport prop-
erty location creates a convenience for the constant flow of supplier repre-
sentatives who fly in to pitch their products. But because the facility is
located outside the fence, the headquarters is a four-mile drive from the
airport terminal. Similarly, Porsche Automobiles is in the process of relo-

cating its North American headquarters to a site adjacent to the Atlanta
airport, citing highway visibility and a substantial land price advantage
for their choice. A combination of factors is important in business loca-
tion decisions, and airport areas may offer advantages which are unrelated
to airport access.
The location of the metropolitan “favored quarter” (Leinberger, 1995)
with respect to the airport has a significant influence on the growth of
employment near airports. Favored quarters appear to repel goods-
producing sectors and attract producer services and corporate headquar-
ters. When the favored quarter is far from the airport, as in Detroit and
Atlanta, professional employment follows, shifting the mix of airport area
employment to goods-producing and lower-level administrative tasks or
suppressing employment altogether (Appold, 2013c).
Aerotropolis developers often face a daunting task if urban expansion
patterns have not favored the airport area. In those cases, the advantages
of airport access are outweighed by other considerations. In order to
attract commercial development, it may first be necessary to attract a labor
supply. Quality mixed-use housing developments near airports (but outside
high-noise contours) containing good schools, upscale shopping, fine din-
ing, vibrant night life, and cultural and leisure venues could well be a differ-
entiating competitive factor for the airport area attracting “white-collar”
and other knowledge-intensive workplaces to the airport area. Aerotropolis
planners and developers always need to be aware that competing locations
in the metropolitan region may be more appealing to younger managers
and professionals than a typical airport area.
Airport cities and aerotropolises such as Helsinki, Hong Kong, and
Incheon have been able to overcome an initial local labor supply disad-
vantage by carefully controlling commercial land prices, ensuring adequate
commuter transportation, and by providing desirable housing close to the
airport at attractive prices. Other large commercial real estate develop-
ment projects have found it necessary to subsidize initial anchor facilities
that bear a greater portion of investment risk by being pioneers.
Subsequent follow-on commercial tenants can compensate for foregone
initial revenues that were necessary to seed early-stage commercial facility
development, should those potential tenants find an airport area location
enticing. As on the interregional scale, without the demand induced by a
growing region facing constrained land options, the level of cross-subsidy
needed to attract tenants may not be financially viable on an intraregional
scale.
The commercial facilities that make up the aerotropolis cannot be

planned independently from metropolitan region land use and development
patterns. Nor can commercial facility investments by airport management
or aerotropolis site developers be planned independently of potentially rein-
forcing or competitive facilities elsewhere in the region, since these other
facilities may either support or detract from such investments. Given ade-
quate ground transportation, location in an immediate airport area may
lend firms little operational advantage. Therefore, as in the cases listed
above, airport area real estate ventures often need to compete on land
price.
SUMMARY AND CONCLUDING REMARKS
The aerotropolis model rests on a small number of interconnected theses.

Air routes operate as the physical internet for aviation-enabled trade in
goods and services with airports serving as the routers and globallocal
interfaces. These globallocal interfaces are physically manifest in aero-
tropolis spatial form, consisting of a multimodal, multifunctional airport
city core centered on passenger and cargo terminals. That combination of
air connectivity and spatial propinquity yields significant advantages to
businesses and to regions. Optimal outcomes of aerotropolis investments
depend on bringing together and aligning multiple stakeholders who cross
numerous public- and private-sector domains.
Empirical research supports aspects of the aerotropolis model for parti-
cular situations. Planning a competitive aerotropolis may require decenter-
ing aviation and airports in aerotropolis planning model, however. As the
FoxconnZhengzhou case study suggests, air transportation is catalytic,
rather than causative. As broader economic trends, notably rising demand
for labor-intensive products and increasing labor costs at the points of
existing production, dictated, cities such as Zhengzhou became attractive
sites for production for firms such as Foxconn, commercial aviation was
key to catalyzing the reaction that took place. Without those background
factors, the aerotropolis would not have been competitive.
Several factors need to be considered in assessing the catalytic power
of aviation in specific situations. For production platforms, total produc-
tion and delivery costs and total market size and growth are certainly
central. Over the past several decades, export-oriented production direc-
ted toward the large markets of Western Europe and North America has
been an effective route to increased welfare for several low-income
economies. As costs rise and the absorptive capacity of core markets is

reached, several international economists have predicted that further eco-
nomic growth will need to rely more heavily on internal consumer
markets.
The commercial aviation industry is nearly a century old and it has
been six decades since air transportation volume began to accelerate in
growth. As noted above, demand in high-income economies could be
approaching the saturation point, limiting the possibilities for additional
air-enabled development. In addition, the economics of aviation fre-
quently encourage airline and hub consolidation, further defining the pos-
sibility space for aerotropolis development. Moreover, collectively, firms
are becoming better at obtaining returns from commercial aviation. For
several decades, in the United States and elsewhere, each additional unit
of GDP produced required a rising amount of air transportation. Today,
the level of air transportation used is high but the revenue miles needed
for each dollar is stable and, for some purposes, decreasing.
A renewed wave of aircraft and air system innovation promises to
reduce the impact of fuel costs on facility location and transportation deci-
sions. This will likely slow, but not stop, the arrival of market saturation.
That doesn’t mean that commercial aviation won’t grow in absolute
volume over the near and medium term. It does make the business case for
a competitive aerotropolis more challenging.
Airports will likely shape urban economic development in the subtle
ways experienced with waterborne ports in the 18th century, railway term-
inals in the 19th, and highway exchanges did in the 20th. When seaports
and cities were founded contemporaneously, they anchored each other.
When seaports were added later, as large airports now are to existing cities
such as Zhengzhou, the seaport became a subcenter but not the center.
When towns were established as new rail lines were laid, the terminal often
became downtown. If urban growth was moderate, the rail terminal
remained the focus. Even when the establishment of rail service and the
city coincided, if urban growth was extensive, the rail terminal lost its posi-
tion as center of the city.
More importantly, in the aerotropolis model the relevant metric is not
spatial distance but time and cost of moving people and products to, from,
and through important nodes, internally and externally. In terms of the
connectivity element, users often benefit from using a distant airport, which
has the economies of the route density which reduces price and increases
service frequency, rather than using a closer facility which does not. For
inter-continental trips, a business traveler located in, say, the Düsseldorf
airport city might be better served by the direct high-speed rail connections
along the Rhine River to Frankfurt Airport, than by relying on a local
flight. Based on detailed travel models, cities and regions are often willing
to invest in efficient means of connecting airports with center cities and
other important employment and residential nodes.
Such connectivity not only enlarges the catchment areas of major air-
ports, it also reduces the need for businesses and travelers to locate near
airports. The benefit of airport proximity often resembles a step function
wherein immediate access to the terminal has a large benefit for, say, hotels,
but once an auto or public transport journey is needed, the benefit does not
fall off rapidly with distance. Similarly, once cargo is loaded onto a truck,
a five-mile journey may not be significantly more onerous than a quarter-
mile journey.
Even if businesses may not benefit from proximate access to airports,
they may benefit from the reduced cost of airport area land. Further, the
region as a whole benefits from having noncompatible uses sufficiently far
from airports. Aside from the necessity of reserving land for likely airport
expansion, obstacle avoidance and third-party risk minimization lead aero-
nautical agencies to proscribe certain land uses to ensure the safety of air
passengers and area residents. Noise disamenity, by far the most common
airport environmental concern, has a significant impact on the ability of
residents to enjoy their homes and on students to concentrate at school. To
the extent large footprint businesses, such as warehouses, offices, and some
factories and retail facilities can be encouraged to use such noise-impacted
land rather than other metropolitan land, regional welfare will be
increased, even when a direct connection to aviation is absent.6 As regions
approach buildout, efficient utilization of airport area land becomes
important.
The most important element of an aerotropolis is its cost-saving fea-
tures. The days when a traveler could arrive at an airport at the last minute
and dash to the gate are long past, diminishing the advantage of flying for
many purposes. Reducing the considerable terminal costs to passenger and
cargo transportation, will require concerted effort and creative thought.
Progress appears immanent with respect to processing international cargo.
Passenger processing may be more difficult to improve. Some of the pro-
gress entails differentiating among passengers and shippers, so that trust-
worthy parties are subject to lighter scrutiny. Additional progress stems
from moving airport functions backward in the journey away from the air-
port and away from departure time, much as electronic check-in reduces
the need to wait at the airline counter.
Charles Lindblom’s 1959 paper on the “science of muddling through”

directed attention to the process of incorporating evolving circumstances
and needs into the planning process. Airport planning is now increasingly
incorporating such flexibility in order to react to uncertain demand pat-
terns in a cost-effective manner (de Neufville & Odoni, 2003). Aerotropolis
planning needs to specify relevant stages of development and the determi-
nants of different constellations of regional needs.
Although comprehensive aerotropolis planning remains in its infant
stage, introductory textbooks now incorporate airport area real estate
development and its potential benefit to airports and regions (Ashford,
Mumayiz, & Wright, 2011). Several large consulting firms offer turn-key
airport cities and a range of consultants integrate specialized expertise
within a broad framework. The critical importance of commercial aviation
is well understood and supported by a range of international, national, and
local public agencies. Transportation planners have developed sophisti-
cated tools to help such agencies and airlines optimize airline routes and
airport location. A complementary set of tools help maximize ground
accessibility. In many countries, land use planning is increasingly integrated
with transportation planning to maximize benefit. To be sure, cross-
jurisdictional governance challenges remain. The net impact of this pro-
gress is to point toward a partial decentering of commercial aviation and
the airport in planning a competitive aerotropolis.
NOTES
1. Paraphrased slightly from [112nd] H.R.658 : FAA Air Transportation

Modernization and Safety Improvement Act, 2011.
2. A more extensive discussion of the central issues discussed in this chapter can
be found in Appold (2013a).
3. See http:\\www.aerotropolis.com for a list of existing and planned aerotropo-
lises. Several of these are described in the publication links included.
4. New trade theory which combines comparative advantage, costs of transport
and inventory, economies of scale in production and distribution, and economies of
density in transport (e.g., Krugman, 1980; Overman, Rice, & Venables, 2010;
Venables & Limão, 2002) provides an understanding of the condition under which
the value of commercial aviation to the regional and global economies can be
maximized.
5. Information on this case provided, in part, by Foxconn and the City of
Zhengzhou to John Kasarda during May 2013 site visit.
6. Although a major airport concern, aviation is not the major source of metro-
politan noise disamenity.
REFERENCES
Appold, S. J. (2013a). A short primer on aerotropolis and airport city planning, UNC Kenan-
Flagler Research Paper, most recent version. Retrieved from http://papers.ssrn.com/
sol3/papers.cfm?abstract_id=2314713. Accessed on August 23, 2013.
Appold, S. J. (2013b). Employment in large U.S. metropolitan airport areas, UNC Kenan-
sol3/papers.cfm?abstract_id=2306834. Accessed on August 07, 2013.
Appold, S. J. (2013c). The extent and reasons for U.S. airport city development, UNC Kenan-
sol3/papers.cfm?abstract_id=2333302. Accessed on October 01, 2013.
Appold, S. J., & Kasarda, J. D. (2010). Catalytic effects in the context of product cycle theory.
In U. Knippenberger & A. Wall (Eds.), Airports in cities and regions: Research and prac-
tise. Karlsruhe: Karlsruher Institut für Technologie (KIT), Scientific Publishing.
Appold, S. J., & Kasarda, J. D. (2013). The airport city phenomenon: Evidence from large
U.S. airports. Urban Studies, 50(6), 12391259.
Ashford, N. J., Mumayiz, S., & Wright, P. H. (2011). Airport engineering: Planning, design,
and development of 21st century airports. Hoboken, NJ: Wiley.
Bradsher, K. (2013). Hauling new treasure along the Silk Road. New York Times, Sunday
Business Section, B1, July 20.
Button, K. (2010). Transport economics (3rd ed.). Cheltenham, UK: Edward Elgar.
Conway, H. M. (1980). The airport city: Development concepts for the 21st century. Atlanta,
GA: Conway Publications.
de Neufville, R., & Odoni, A. (2003). Airport systems: Planning, design, and management. New
York, NY: McGraw-Hill.
Glaeser, E. L., & Kohlhase, J. E. (2004). Cities, regions, and the decline of transport costs.
Papers in Regional Science, 83, 197228.
Greis, N., & Appold, S. J. (2011). Critical issues facing the Air Cargo Market: 20102025,
report prepared for Boeing Commercial Aircraft, 15 November.
Güller, M., & Güller, M. (2001). From airport to airport city. Brussels: Airport Regions
Conference.
International Air Transport Association (IATA). (2013). The value of Air Cargo. Retrieved
from https://www.iata.org/cargo.aspx. Accessed on November 15, 2013.
Jackson, J. K. (2013). Outsourcing and insourcing jobs in the U.S. economy: An overview of evi-
dence based on foreign investment data. Washington, DC: Congressional Information
Service, Library of Congress. Congressional Research Service.
Kasarda, J. D. (1998/1999). Time-based competition and industrial location in the fast cen-
tury. Real Estate Issues, 23(4), 2429.
Kasarda, J. D. (2010). The way forward. In J. D. Kasarda (Ed.), Global airport cities (pp.
1536). London: Insight Publishing. Retrieved from www.aerotropolis.com.
Kasarda, J. D. (2013). Airport cities: The evolution. Airport World, 18(2), 2427.
Kasarda, J. D. (2014). Aerotropolis: Business mobility and urban competitiveness in the 21st cen-
tury. In K. Benesch (Ed.), Cultures of mobility. Heidelberg: Heidelberg University Press.
Kasarda, J. D., & Lindsay, G. (2011). Aerotropolis: The way we’ll live next. New York, NY:
Farrar, Straus, and Giroux.
Krugman, P. (1980). Scale economies, product differentiation, and the pattern of trade. The
American Economic Review, 79, 950959.
Leinberger, C. B. (1995). The changing location of development and investment opportunities.

Urban Land, 54(May), 3136.
Overman, H. G., Rice, P., & Venables, A. J. (2010). Economic linkages across space. Regional
Studies, 44, 1733.
Sheffi, Y. (2012). Logistics clusters: Delivering value and driving growth. Cambridge, MA: MIT
Press.
Stilwall, J. D., & Hansman, J. (2013). The importance of air transportation to the U.S. economy:
Analysis of industry use and proximity to Airports. Cambridge, MA: MIT International
Center for Air Transportation. ICAT-2013-03 Vol.
Suri, R. (2010). It’s about time: The competitive advantage of quick response manufacturing.
New York, NY: Productivity Press.
Venables, A. J., & Limão, N. (2002). Geographical disadvantage: A HeckscherOhlinvon
Thünen model of international specialization. Journal of International Economics, 58,
239263.
Zhou, J. (2013). The Current situation of China’s airport economy, Civil Aviation
Management Institute of China.
CHAPTER 12
AIRPORT AND AIRLINE

SUBSTITUTION EFFECTS IN
MULTI-AIRPORT MARKETS
Dan Mahoney and Wesley W. Wilson
ABSTRACT
Airline travel is composed of business and nonbusiness travelers, each

with different preferences that give rise to differences in demand elasti-
cities and substitution not only across airlines but also airports. In this
study, we develop and estimate a model of airline wherein consumers
choose which airports and airline to use that allows for unobserved dif-
ferences between travelers (e.g., business and nonbusiness travelers).
The results point to the role that airports themselves play in the ulti-
mate selection of a flight, and that there are strong interactive effects
between the airlines’ networks and the consumers’ preferences across
airports.
Keywords: Airline demand; airline and airport substitution; choice
modeling
JEL classifications: L93; R4

ISSN: 2212-1609/doi:10.1108/S2212-160920140000004011
309
310 DAN MAHONEY AND WESLEY W. WILSON
INTRODUCTION
In 2012, there were nearly 3 billion passengers in the world airline markets,
and the average growth since 1970 is about five percent per year.1 In the
United States, nearly a billion passengers travel every year and approxi-
mately $1.3 trillion in total economic impact annually.2 Travel in a global
economy or any economy occurs for a wide variety of reasons but com-
monly travel can be grouped into business and nonbusiness travelers. For
some travelers, there may be little decision making, that is, destination air-
port and, in some cases, even the airline may be fixed, while for other trave-
lers there may be considerable discretion over the destination, the airline or
both. In this chapter, we estimate a model of consumer demand to identify
preferences for both airline and airports along with substitution patterns
both across airports and across airlines, allowing for differences among
business and nonbusiness travelers.
For consumers in some geographic locations, there is only one feasible
origindestination pair; however, many of the largest markets are served by
multiple airports. While many studies focus on demand for air travel in
aggregate or by airline, the purpose of this study is to better understand the
relative importance of the airports themselves in the consumers’ decision-
making process. There are many reasons why consumers may prefer a parti-
cular airport. It may be a feature of location, such as distance or access
infrastructure (roadways, public transportation, etc.). It may be particular
airport amenities, or it may simply be due to a consumer’s history with a
particular facility. The interaction between airports and airlines may also be
a factor. The effects of airline dominance of an airport have been well docu-
mented, going back to Borenstein (1989). Often dubbed the “hub premium,”
there is ample evidence that consumers are willing to pay a premium to fly
with the airline with a predominant market share at a given airport.
In this chapter, we adapt the model of airline demand from Berry,
Carnall, and Spiller (2006) to address the subject of consumer substitution
patterns between airports. This approach is a discrete-choice, random coef-
ficients demand model derived from market-level data, that is used to esti-
mate consumer demand parameters for airport and airline characteristics.
The estimated parameters can then be used to estimate change in consumer
behavior in response to the set of available products. In particular, it
focuses on how consumers substitute across different origin airports in a
multi-airport market when faced with hypothetical fare increases localized
to a single airport. We also examine how consumers substitute across air-
ports when faced with a fare increase from a particular airline.
Airport and Airline Substitution Effects in Multi-Airport Markets 311
Evaluating the results across different markets, substitution out of the

market tends to dominate. In response to an airport-wide price increase,
approximately 70% of those passengers that choose to abandon their origi-
nal airport will opt out of the air travel market entirely, rather than fly
from an alternative airport, though there was considerable variability
across markets, and even across airports within the same market. Among
the consumers who do switch to a different airport, again the results vary,
with no discernible patterns based upon the data available. The overall
magnitude of substitution is another feature that shows wide variation
between markets. There is relatively high substitutability in the New York
City metropolitan market (characterized by own-price airport elasticities
greater than 2), and relatively low substitutability in the Washington, DC,
metropolitan market (characterized by elasticities less than 1).
Such results may be of interest to policy-makers, who are considering
infrastructure decisions. The price changes considered in this chapter could
be driven by direct taxes or fees on the departing airports, or they could
also be thought of as being driven by ground access costs. This chapter
provides initial estimates on the extent that airport price changes drive cus-
tomers in or out of the market, and to what extent they simply cause a real-
location of customers among the existing airports in the market.
LITERATURE REVIEW
There is a rich and growing literature on the air industry. This literature
has given a plethora of knowledge that applies to the industry but also has
influenced the more general economics literature in areas such as network
analysis or consumer choice, among others. In this section, three distinct
areas are discussed. The section “Airlines and Airports” covers the relation-
ship between airlines and airports. The section “Market Definition”
addresses the question of the relevant market of an airport, while the sec-
tion “Consumer Choice” presents an overview of consumer choice model-
ing, as applied to the airline industry.
Airlines and Airports
Airlines rely on airports to conduct their operations, and the relationships

between the two can have significant effects on the outcome of the market,
particularly the demonstrated market power of firms. Since the onset of

deregulation in 1978, market power and pricing have been the focus of
much academic research. Graham, Kaplan, and Sibley (1983) test two
hypotheses of deregulation in particular: first, that air carriers were running
excess capacity prior to deregulation, and second, that potential competi-
tion would keep fares low, even in highly concentrated markets. Their
results are consistent with airline load factors increasing significantly in the
years following deregulation. They also find that broad market demand
characteristics can explain a high percentage of observed fares; however,
they reject the hypothesis that potential competition is sufficient to drive
down fares. Instead, observed airfare is highly correlated with measures of
market concentration. This result ran counter to earlier results, such as that
by Bailey and Panzar (1981) which claimed that airlines were perfectly con-
testable. Morrison and Winston (1987) also test the contestability of airline
markets, and similarly find that the markets were imperfectly contestable.
Though the Graham, Kaplan, and Sibley find market concentration was
correlated with higher observed fares; however, they stop short of identify-
ing the source of the pricing power, even in markets that appeared to be
contestable. Borenstein (1989) examines the role of airport dominance in air-
line pricing power. By estimating a pricing equation that includes both mea-
sures of concentration at the route level, as well as market concentration at
the origin and destination airports, he finds that a carrier’s share of both
route and total airport traffic have significant effects on pricing. While it is
expected that airlines with a greater share of route traffic are able to charge
higher prices as a result of their market power, it is less apparent why the
airline’s overall presence should influence pricing on a particular route. The
explanation may lie in the prevalence of consumer loyalty programs. One
such loyalty program frequent flyer miles rewards customers who do
repeated business with a particular airline. When frequent flyer programs
are present, customers may prefer an airline that offers the most flight
options from their local airport, as their airline decision depends on both
the current flight as well as expected future flights. Other potential explana-
tions include travel agent commission override bonuses, which pay travel
agents for directing a specified level of traffic to a particular airline. There
may also be common advertising costs for an airline in a local market.
Though the exact mechanisms were left unidentified, it was clear that subse-
quent studies of airline demand needed to account for carriers’ presence at
an airport, not just along a particular route.
Airline presence at an airport has a strong influence on pricing, and so it
is natural to further study the nature of the vertical relationship between
airports and airlines. As pointed out by Oum and Fu (2008), airport reven-
ues come from two primary sources. The first source is charges for aero-
nautical services. These include take-off and landing fees, terminal rental,
aircraft parking, and other such services directly related to the facilitation
of flights. The second source of airport revenue comes from nonaeronauti-
cal services, such as parking, concessions, office rental, and other commer-
cial uses of airport land. For these services, airports possess significant
market power, since price elasticity of demand is very low. Several key fac-
tors determine airport market power. The first is airport capacity relative
to demand. As noted in Oum and Fu (2008), air traffic demands have been
increasing by approximately 5% per year, and airport infrastructure has
not kept up with this growth throughout most of the United States,
Europe, and Asia. The second is regional airport competition; when multi-
ple airports serve the same metropolitan area, market power among both
airports is reduced, so long as these airports do not share common owner-
ship. The share of connecting passengers also is an important determinant
of airport market power. While local traffic is relatively inflexible, both pas-
sengers and airlines are free to choose between different hub airports.
Because of the intertwined relationship between airports and airlines, it
may often be beneficial to adopt some level of integration between the two.
These relationships may serve to guard against risk, internalize demand
externalities, or gain a competitive advantage over other airports and air-
lines. This integration may take several forms. Airlines may own shares in
the airport, or may engage in long-term contracts to guard the airport
against risk; in exchange offering the airline favorable rates.
Airportairline relationships often serve to strengthen the position of
the airport’s dominant carrier who is best able to negotiate favorable terms
with an airport. These long-term contracts can create a barrier to entry for
new firms in the market. Ciliberto and Williams (2010) investigate the role
of these arrangements in terms of the “hub premium” the difference in
between fares to or from airports where major airlines have hubs relative to
comparable trips that do not originate or terminate at a hub airport.
Estimating a log-linear pricing specification, Ciliberto and Williams find
that the hub premium is present and increasing in the fare.
Unconditionally, they find the hub premium to vary from approximately
10% at the 10th percentile of fare distribution to 20% at the 90th percentile
of fare distribution. The apparent hub premium decreases in magnitude
when controls for barriers to entry and airport congestion are added to the
model. The hub premium also decreases with the presence of low-cost
carrier Southwest Airlines, suggesting that increased competition may eat
away at the markup. Airport congestion and airport barriers may explain a
significant portion of pricing power, as represented by the hub premium;
however, they only account for approximately 50% of the observed hub
premium. They attribute the remaining 50% to the hubbing market power
factors outlined by Borenstein (1989), such as loyalty programs, travel
agent commissions, and familiarity biases.
Though airports provide a barrier to entry that can increase market
power among the airlines in the market, they also serve as a source of con-
gestion. The relationship between barriers to entry and airport congestion
is the subject of a paper by Dresner, Windle, and Yao (2002). They exam-
ine several barriers, including slot controls, gate constraints, and gate utili-
zation during peak operating periods. They estimate both a choice model
for the airline’s entry decision, as well as a standard regression on passen-
gers and yield (defined as average price per passenger mile). Their findings
indicate that all three variables have a statistically significantly positive
effect on yield. Only one barrier, gate utilization during peak operating per-
iods, had a significant effect on airline entry into a market. Their results are
indicative that although contracts between airports and dominant airlines
may correlate with greater market power, unless the airport is capacity con-
strained, these contracts will not be able to inhibit new entry.
Another concern associated with airport congestion is the costs imposed
by an airline’s flight due to congestion. Though weather is the single largest
source of delays in the U.S. airline industry, in most cases “volume” delays,
caused by traffic exceeding airport capacity, is the second largest source of
delay. Brueckner (2002) considers the effects of congestion pricing in the
airline industry and compares it to the results of the road-pricing literature.
Contrary to road pricing, in the airline industry, firms with market power
internalize some of the congestion costs of their own flights. In the case of
the monopolist, the congestion costs will be fully internalized. In the case
of an oligopoly, the firms internalize the portion of the congestion costs
imposed on themselves. Pels and Verhoef (2003) derive a similar model of
congestion costs with market power and, like Brueckner, find that a naı̈ve
congestion toll will be too large, and may actually be welfare reducing.
Their model also incorporates regulator coordination issues, particularly in
the case where origin and destination airports are located in different coun-
tries, and subject to differing regulatory agencies. Without coordination,
the incentive to reduce tolls to the optimal level is disproportionately
reduced, leading to an inefficient outcome.
Airport congestion is also affected by the size of airplanes. As the
number of runways, gates, and departure times are fixed in the short
term, larger airplanes may be the only way to increase passenger volume.
Wei and Hansen (2005) estimate a nested logit model to study the rela-
tionship between aircraft size, service frequency, seat availability, airline
fares, and market share. They find that airlines can realize higher returns
from increasing flight frequency compared to utilizing larger aircraft.
Though there may be cost savings associated with a larger aircraft, hold-
ing other factors constant, passengers do not display preference for a par-
ticularly sized aircraft. Instead, passengers display a preference for
greater choice in departure time. In this case, the airlines choose to fly
airplanes that are smaller than those that would minimize the cost per
passenger mile.
Related to airport congestion, a critical issue to understand is the opti-
mal market size of a city-pair route at an airport. As airport market size
increases, unit operating costs decrease as airlines are able to use larger
aircraft filled to greater capacity. A larger airport, however, may face
greater delays as it encounter capacity constraints. As the airport
increases its market size, the average airport access costs rise, as custo-
mers must travel from further away. Hsu and Wu (1997) attempt to
model this problem and solve for the optimal airport market size using
linear programming techniques. Using hypothetical estimates of various
parameters, they find that airports operate more efficiently in markets
with greater population density. Cities with greater per-capita income
allow an airport to serve a larger market size, along with a larger market
area. Finally, they find that stability among passenger demand allows air-
ports to operate more efficiently.
Market Definition
More generally, the question of market identification is an important one

in airline research. For demand models, identifying which airports are in
the consumers’ choice set is necessary to obtain proper estimates, and sub-
sequent models of pricing and competition also require such a market to be
properly identified.
Forsyth (2006) outlines several of the potential issues when a city’s
dominant airport faces competition from smaller, fringe airports. Most
major cities feature a single dominant airport, located either within or
near the city limits. More recently, there has been growth in secondary
airports, which has been associated with the growth of low-cost carriers
(LCCs). The secondary airports are often less convenient for consumers,
and so they compete largely on price; appealing to the more price-

sensitive consumers who are willing to sacrifice some of the benefits of
flying with the larger, full service carriers (FSCs). When the LCCs at
fringe airports enter the market, it may or may not improve overall effi-
ciency in the market. In the case when a major airport has excess capa-
city, and the markup above marginal cost is designed to cover the
airport’s substantial sunk costs, the airlines may not be able to adjust
their pricing to appropriately compete, and an inefficient allocation will
be realized. Inefficient allocations may also arise if the secondary airports
are receiving subsidies. Conversely, if the secondary airports and the
LCCs cost advantages are due to greater efficiency, competition in the
market will have a positive effect.
Morrison (2001) attempts to directly estimate some of the gains offered
by LCCs operating out of regional airports. In a study commissioned by
Southwest Airlines, he looks at the effects of Southwest’s competition on
the U.S. airline industry. When considering the effect of an LCC, such as
Southwest, competition may come by the LCC serving the same route in
question as the major carriers, or it may come by the LCC serving some
combination of the same or adjacent airports. Estimating the effects of
Southwest Airlines on fares, for a single year (1998), Morrison finds that
competition from Southwest resulted in $12.9 billion in savings, $3.4 of
which from Southwest’s own fares, while the remaining savings came from
other airline’s lower fares. The cost savings are greatest when Southwest
serves the same route in question as the FSCs; however, even when
Southwest doesn’t serve the market in question, but has a presence at either
of the end points (or their adjacent airports), the threat of entry results in a
statistically significant decline in average airfare.
Brueckner, Lee, and Singer (2010) offer a comprehensive evaluation of
competition and airline pricing. They estimate the model allowing for in
market, adjacent competition as identified by Morrison (2001). Unlike
Morrison (2001), they consider not only LCC competition from adjacent
airports, but also legacy carrier competition from adjacent airports. The
second contribution of the paper is to distinguish between competition
from nonstop flights and competition from connecting flights. Bruecker
et al. find that in-market competition from LCCs contributes to lower fares
significantly more than competition from legacy airlines. This pattern
extends to adjacent competition from LCCs. They find that in many cases,
adjacent airport competition from legacy carriers has no effect on airfare.
This result holds for competition among both nonstop flights, as well as
connecting flights.
Consumer Choice
Driving these price effects between adjacent airports is an underlying con-

sumer choice problem. Though not all consumers face a realistic choice of
airports to suit their travel needs, several of the largest airline markets,
including New York, Los Angeles, Washington, DC, San Francisco, and
Chicago all feature multiple large airports within close geographic proxi-
mity to the city. There have been a number of studies done to model the
consumer choice problem when both the flight and the airport are choice
parameters. One such study by Windle and Dresner (1995) uses survey data
for the Washington, DC, metropolitan area. They found that there were
strong proximity effects, but controlling for passengers with similar access
times to multiple airports, flight frequency appeared to be the driving deter-
minant. Not surprisingly, they also found that business travelers valued
flight frequency and airport proximity relatively more than leisure travelers,
who were more price sensitive.
Pels, Nijkamp, and Rietveld (2001) perform a similar study using survey
data from the San Francisco Bay Area. They model passengers as first
choosing their departure airport, and subsequently their particular flight,
utilizing a nested logit framework. They find that this model significantly
outperforms a direct multinomial logit model. Further extensions of an
airportairline choice model come from Basar and Bhat (2004), who
hypothesize that the airport choice set may vary between potential consu-
mers. They implement a probabilistic choice set multinomial logit model
and find that models presenting a uniform choice set across consumers pro-
duce biased estimates.
To estimate an airportairline choice model, it is ideal to have data on
individual consumers and their choices. Such data, however, is not widely
available; and consequentially, the aforementioned choice studies tend to
rely on common datasets capturing only a few markets over a relatively
short period of time. An alternative approach from Berry et al. (2006) uses
only aggregate data to estimate consumer demand. As such data are widely
available, adopting such an approach allows for greater breadth among the
estimation results. They use market shares to estimate a random-coefficient
choice model, along the lines of Berry, Levinsohn, and Pakes (1995). They
use this choice model to examine the impact of hubbing on both costs and
demand.
MODEL
I model consumer decision making with a choice model. The model used
follows those developed by Berry et al. (2006) and Jia and Berry (2010). It
is a random-coefficient, discrete-choice framework. This model assumes a
set of consumers in each market who choose from the menu of that mar-
ket’s available products, each offering some utility level (u). Specifically,
consumer utility function is assumed to take on the following form, where
the utility for consumer i, in market t, and product j is given by
uijt = xjt βi − αi pjt þ ξjt þ νit ðλÞ þ λEijt ð1Þ
where xjt is a vector of observable attributes of product j in market t, pjt is

the product’s price; νit and λ are nested logit parameters designed to pattern
those who participate in the market and those who don’t; Eijt is an i.i.d.
error term; and ξjt represents product characteristics that are unobserved to
the econometrician, but observable to the consumer, as presented in Berry
et al. (1995). Collectively, the model parameters ðα; β; λÞ will all be consid-
ered as part of a single parameter vector, θ. The consumer chooses the pro-
duct, j, whenever

u pj ; xj ; ξj ; θ ≥ uðpk ; xk ; ξk ; θÞ ∀k ð2Þ
Not all consumers may choose to purchase one of the products in the
market (in this case, airline travel). Some may choose alternatives means of
travel, such as automobile or train, while other consumers may choose not
to travel at all. The utility of those who do not participate in the market
(those who have implicitly chosen some “outside good”) is normalized to
ui0t = Ei0t ð3Þ
The random consumer taste parameters βi and αi are assumed to take

on a two-point distribution with γ and 1 − γ representing the probability
that a given consumer is of type 1 or type 2. Colloquially, the two types of

consumers are referred to as “business” and “leisure” travelers (as is consis-
tent with prior demand studies that show that those two groups tend to
vary particularly in their price sensitivity); however, in the data, the rea-
son for travel is never explicitly observed, and so the consumers are identi-
fied purely by their demand parameters.
With the consumer utility specified, the market shares can be estimated
by integrating the choice probabilities over the number of consumers in the
market. If the additive error term takes on an extreme-valued, i.i.d. distri-
bution, the choice probabilities will take on the traditional logit form.
Conditional upon purchasing some product, the probability of a consumer
of type r choosing product j is

xjt βi − αi pjt þ ξjt
exp λ
srjg = P ð4Þ
xkt βi − αi pkt þ ξkt
k ∈ J exp λ
While the probability that a type r consumer chooses any product in the
market is given by
P λ
k ∈ J exp λ
srt = P λ ð5Þ
1þ k ∈ J exp λ
The total observed market share of product j in market t is
sjt ðx; p; ξ; θÞ = γs1jg s1t þ ð1 − γ Þs2jg s2t ð6Þ
where θ is the complete set of parameters to be estimated, including

βi ; αi ; λ; and γ. The estimation procedure uses the generalized method of
moments (GMM) estimation procedure introduced in Berry et al. (1995).
The GMM estimator is based on the assumed independence of the unob-
served error component, ξ, and a set of instrumental variables, Z. These
instruments are made up of variables which are expected to be correlated
with the price, but uncorrelated with the error term, ξ. They include all
demand variables (except price), cost variables, and market-level attributes.
The procedure attempts to find a set of demand parameters, θ, that
minimize the difference between the theoretical moment condition and its
sample equivalent (in this case, the independence of ξ and the set of
instruments).
Specifically, the procedure works as follows. For a given set of para-
meters, the vector of unobserved product attributes can be solved for by
inverting the above market shares equation.
ξ = s − 1 ðx; p; s; θÞ ð7Þ
To solve for the set of parameters that satisfied the moment condition
Eðξðx; p; s; θÞjzt Þ = 0 ð8Þ
where zt is a vector of instruments. Consequentially, for any function of

instruments hðzt Þ,
Eðhðzt Þξðx; p; s; θÞÞ = 0 ð9Þ
In practice, estimating this system first requires inverting the market

shares, given by Eq. (5), to solve for the unobserved product error term, ξ.
As this equation cannot be inverted analytically, this is done by means of a
contraction mapping, as outlined in Berry et al. (1995), and modified for
this application in Berry et al. (2006). The vector ξ is found by means of
the recursive equation

ξN = ξN − 1 þ λ ln s0 − ln s x; p; ξN − 1 ; θ ð10Þ
which is iterated until the maximum difference between ξN and ξN − 1 is less

than some specified tolerance. Dubé, Fox, and Su (2012) present numerical
analysis of the convergence of this “inner loop” (the process by which the
market shares are inverted). They stress a stringent convergence tolerance,
to insure that the subsequent “outer loop” (the minimization of the
demand parameters) optimization converges appropriately. The aforemen-
tioned outer loop optimization involves the minimization of the sample
analog to Eq. (8) over the parameter vector, θ.
The final step is to estimate consumer substitution patterns between air-
ports. Using the demand parameter estimates from above, we estimate the
change in predicted market shares (Eqs. (4)(6)) in response to hypotheti-

cal price changes. We do this for two cases. In the first case, we compute
the share response to a hypothetical price increase across all flights from a
particular airport. Here, consumers may find it worthwhile to switch to a
different flight (possibly from the same airline) at a different airport. In the
second scenario, we compute the share response to a price increase only to
a particular airline (across all airports in the market, if it has a presence at
more than one). As consumers substitute flights from other airlines, some
may find it worthwhile to choose a different departing airport as well.
There are some concerns as to the applicability of this model to the
situation. By the convention established in Berry et al. (2006), products are
defined, in part, by their prices. After airlines schedule flights, they engage
in dynamic pricing behavior to maximize revenue. As airlines raise or lower
their prices in response to perceived demand and competition, the effective
consumer choice set varies. As the model assumes that all products are
available at all times, this can potentially lead to biased estimates. Ideally,
some facet of product availability is captured in the unobserved product
attribute component, ξ; however, this is an imperfect solution to the pro-
blem of product availability. To address concerns about the impact of pro-
duct availability, Jia and Berry (2010) perform Monte Carlo experiment to
estimate the extent of the bias. They conclude that the bias is small and is
unlikely to significantly alter the parameter estimates.
Using this methodology, we are able to produce consumer utility func-
tion estimates, which can be applied to hypothetical changes in the avail-
able product set to provide some insight on consumer substitution patterns
between airports; however, there are a few caveats. The aforementioned
issues concerning product availability continue to be present when evaluat-
ing substitution patterns in response to a hypothetical price increase. These
estimates assume a full complement of alternatives is available. In the short
run, airlines are capacity constrained and may not be able to support an
increase in passengers. Furthermore, if certain flightfare combinations are
offered at a fixed quota, its market share would not grow, no matter how
its rivals’ prices changed. In such cases, the “Results” section may be
upwardly biased, overestimating the substitution among consumers.
There are further concerns about the consistency of the parameter esti-
mates across markets. It is reasonable to expect the composition of consu-
mers to vary greatly by the destination, particularly between standard
“tourist” destinations, like Orlando or Las Vegas, and more “business”-
oriented destinations like Chicago. To address this concern, we estimate
both a model encompassing all U.S. airline markets, as well as a specific
model for each of the origin cities of interest. As discussed below, we find
the significance of the localized model estimation to vary based on the mar-
ket, but do not exhibit any clear pattern in their influence of the results.
DATA
The primary source of data for this study is the United Sates Department
of Transportation (DOT) Airline Origin and Destination Survey (DB1B).
These data were supplemented by the DOT’s Air Carrier Segment Data (T-
100). Population and income measures came from the Bureau of Economic
Analysis’s (BEA) Local Area Personal Income tables.3
The DB1B data are a 10% sample of airline tickets sold from reporting
carriers and collected by the Bureau of Transportation Statistics.
Consistent with Berry et al. (2006) and Jia and Berry (2010), we consider
only round trip itineraries, with at most four total flight segments. The
sample was further restricted to those in the lower 48 states, serving mar-
kets with at least 850,000 people where the market size is defined as the
geometric mean of the populations at the end point cities. Round trip fares
above $5000, and below $200, were dropped, as these may have been indi-
cative of either data processing errors or may simply represent extreme out-
liers that are not reflective of the preponderance of the data.
For this study, a market is defined as a directional city pair so, for exam-
ple, a round trip from New York to Los Angeles is distinct from a round
trip from Los Angeles to New York. Most cities are served by a single pri-
mary airport; and thus, those markets were represented by a unique airport
pair. Several large cities (or metropolitan regions) have commonly been
identified as being served by multiple airports, though the exact groupings
are not always clear.4 In all, there were six such groups of airports that
were sufficiently close to warrant grouping them.
Following Berry et al. (2006), a product is identified as a unique
origindestination flight, from a particular carrier, for a fixed number of
connections, at a particular fare. For the purposes of this study, the loca-
tion of the connection was not specified that is, it was assumed that con-
sumers cared whether or not their flight had a connection, but not where
that connection took place. This was mostly done for computational sim-
plicity, and it is not assumed to bias the results significantly. Along those
same lines, fares were clustered into $25 bins again, this was largely for
computational simplicity.
This study uses data from the first quarter of 2010. After all the restric-
tions were put in place, there remained 251,206 products, representing
2,307 different origindestination pairs. An assortment of variables
was used, intended to capture product-specific characteristics, as well as
airportairline interaction effects. The product-specific characteristics include
fare, connection, distance, and online ticket sale. Airportairline interactive
features used were a hub dummy variable, and the number of nonstop desti-
nations served by each airline at a particular airport. These, combined with
airline dummy variables, make up the bulk of the parameters.
To address the question of heterogeneity across different airline markets,
we perform the estimation routine for both the full sample and several loca-
lized markets individually. The full sample includes all flights to or from
airports serving a market of greater than 850,000 people (where, again, a
market is defined as the geometric mean of the populations of the end point
cities). Six localized markets were singled out for this study; these markets
were chosen as they were the six markets identified in Berry et al. as being
served by multiple airports. A list of the six cities and the airports they
encompass are presented in Table 1, and it is readily apparent that these
airports tend to be major international airport hubs.
Table 2 presents summary statistics for all the key variables used in this
study. However, in addition to the demand variables, there is also a need
for a number of instrumental variables. It is assumed that price is endogen-
ous, and central to the method of moments estimation procedure outlined
in the section “Model” is a vector of instruments. In addition to the set of
demand variables (excluding price), additional instruments were chosen
that would reflect cost parameters, and competition factors that would
affect price. These instruments include a hub variable, if the flight
Table 1. Cities and Airports.

City New York Washington, Chicago Dallas San Los Angeles
DC Francisco
Airports Newark Baltimore/ Chicago Dallas Love Oakland Bob Hope

Liberty Washington Midway Field (OAK) (BUR)
(EWR) (BWI) (MDW) (DAL)
John F. Reagan O’Hare Dallas Fort San Los Angeles
Kennedy (DCA) (ORD) Wort Francisco (LAX)
(JFK) (DFW) (SFO)
LaGuardia Dulles (IAD) Mineta San Long Beach
(LGA) Jose (LGB)
(SJC)
Table 2. Summary Statistics.
Sample Size Full Sample New York Washington, DC Los Angeles San Francisco Chicago Dallas
2,025,688 153,866 94,943 69,954 66,983 111,407 72,864
Mean Standard Mean Standard Mean Standard Mean Standard Mean Standard Mean Standard Mean Standard
Deviation Deviation Deviation Deviation Deviation Deviation Deviation
Products
Share 9.38E-05 1.76E-04 9.42E-05 1.20E-04 7.43E-05 9.64E-05 7.04E-05 1.21E-04 7.34E- 1.11E-04 1.27E-04 1.58E-04 1.37E-04 1.85E-04
05
Fare 433.80 219.49 469.43 289.43 453.92 239.53 440.80 271.39 439.30 256.15 413.64 188.58 478.81 266.22
Direct flight 0.65 0.48 0.86 0.34 0.72 0.45 0.75 0.43 0.72 0.45 0.91 0.29 0.87 0.33
Distance (1,000 1.23 0.64 1.42 0.73 1.27 0.73 1.62 0.80 1.56 0.89 1.11 0.46 1.02 0.33
miles)
Distance2 1.93 1.88 2.55 2.29 2.14 2.11 3.27 2.37 3.22 2.75 1.45 1.07 1.15 0.69
Nonstop destination 48.31 41.89 47.18 26.79 43.31 24.89 39.43 18.19 36.59 21.84 89.23 33.04 108.48 42.30
Online ticket sales 0.71 0.46 0.84 0.37 0.78 0.42 0.81 0.39 0.81 0.40 0.80 0.40 0.86 0.34
Hub 0.78 0.42 0.45 0.50 0.90 0.30 0.73 0.45 0.71 0.45 0.95 0.22 0.97 0.18
Slot controlled 0.06 0.24 0.58 0.49 0.32 0.47 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Airlines
Southwest 0.18 0.39 0.00 0.06 0.22 0.41 0.20 0.40 0.28 0.45 0.22 0.42 0.09 0.28
American 0.14 0.35 0.13 0.33 0.10 0.30 0.21 0.41 0.11 0.32 0.34 0.47 0.75 0.43
Delta 0.20 0.40 0.19 0.39 0.12 0.32 0.14 0.35 0.09 0.29 0.03 0.18 0.03 0.17
United 0.11 0.32 0.05 0.21 0.23 0.42 0.17 0.38 0.27 0.44 0.32 0.46 0.03 0.16
Continental 0.09 0.29 0.33 0.47 0.03 0.16 0.05 0.21 0.05 0.21 0.01 0.09 0.01 0.11
Northwest 0.03 0.18 0.01 0.12 0.01 0.11 0.01 0.11 0.01 0.09 0.00 0.06 0.00 0.05
U.S. airways 0.12 0.32 0.07 0.26 0.18 0.38 0.06 0.24 0.06 0.24 0.03 0.18 0.06 0.23
JetBlue 0.04 0.19 0.18 0.38 0.02 0.14 0.04 0.21 0.02 0.16 0.00 0.06 0.00 0.00
Airtran 0.03 0.16 0.01 0.09 0.06 0.24 0.01 0.10 0.00 0.05 0.02 0.15 0.01 0.10
Other airlines 0.06 0.24 0.04 0.19 0.04 0.18 0.11 0.31 0.10 0.30 0.02 0.13 0.02 0.14
Market
# Carriers 3.50 1.74 4.04 1.97 3.98 1.96 3.99 2.27 4.13 2.13 3.42 1.45 3.65 1.74
# Products 43.59 82.00 171.90 216.91 153.55 198.24 121.57 175.90 140.47 200.64 100.52 114.07 88.28 115.04
originates, departs, or connects through an airline’s hub, a slot controlled

dummy variable, and route-level characteristics such as the number of com-
peting airlines in a market. Further instruments were selected from rival
product attributes, such as the average rival fare on a route, and the aver-
age number of connections. Further instruments, as used in Jia and Berry
(2010) are fitted values of the twenty-fifth and seventy-fifth quantile of fares
along a given route.
RESULTS
The complete results from the estimated model are presented in Table 3.
Column 1 presents the parameter estimates when the full sample of origin
and destination airports is included in the sample. Columns 27 represent
the parameter estimates when the sample is restricted to a particular origin
city (e.g., column 2 includes all round trip destinations originating from
New York City). By restricting the sample to a single origin city, it becomes
feasible to include origin-airport dummy variables in the model. This better
captures unobservable airport effects, than simply having them collected in
the error term, as is the case with the full model. The city-specific model is
also estimated recognizing that there may be parameter heterogeneity
between different markets.
The model estimates presented in Table 3 are taken and used to con-
struct airport price elasticities these represent the percentage change in
originating airport passengers in response to an airport-wide percentage
price increase. Though not explicitly addressing the cause of such a price
increase, such price increases might arise in response to higher gate or run-
way fees implemented to combat congestion. These cross-airport elasticities
are presented in Table 4.
The elasticity estimates in Table 4 are the percentage response of quanti-
ties to a one percent change in all round trip flights originating at a specific
airport. For example, a 1% fare increase to all round trip flights originating
at John F. Kennedy International Airport would result in a 2.3% decrease
in passengers departing from that airport (corresponding to approximately
10,000 passengers), a 0.29% increase in the passengers at Newark Liberty
International Airport, and a 0.24% increase in the passengers at
LaGuardia Airport (both corresponding to approximately 2,200 and 1,800
passengers, respectively). In the Washington, DC, metropolitan market, a
1% fare increase at Reagan International Airport would result in a 0.58%
Table 3. Parameter Estimates.
326
Parameter Full Model New York Washington, DC Chicago Dallas San Francisco Los Angeles
(1) (2) (3) (4) (5) (6) (7)
Type 1
Fare −0.0032* −0.0011* 0.0001 −0.0063* −0.0010* −0.0016* −0.0011*
(0.0001) (0.0001) (0.0001) (0.0003) (0.0001) (0.0003) (0.0002)
Constant −9.1551*
(0.1891)
Connection 0.4100* −0.6970* −0.6343* −2.5416* −0.4216* −0.9527* −0.7047*
(0.0828) (0.0490) (0.0026) (0.0567) (0.0490) (0.2333) (0.0490)
Type 2
Fare −0.0024* −0.0001 −0.0009* −0.0013* −0.0005* −0.0011* −0.0014*
DAN MAHONEY AND WESLEY W. WILSON

(0.0000) (0.0001) (0.0001) (0.0003) (0.0001) (0.0001) (0.0003)
Constant −5.8173* 0.0462
(0.0297)
Connection −0.8790* −1.0498* −0.4475* 0.5976* −0.3481* −0.4870* −0.7315*
(0.0113) (0.0462) (0.0425) (0.0799) (0.0462) (0.0370) (0.0483)
Common
Nonstop 0.0025* 0.0099* 0.0048 −0.0357* 0.0281* 0.0115* 0.0013*
destinations
(0.0000) (0.0007) (0.0118) (0.0049) (0.0007) (0.0055) (0.0003)
Distance 0.5604* 1.3407* 0.8786* 2.0567* −0.5265* −0.8184* −0.0238
(0.0073) (0.0424) (0.0112) (0.0546) (0.0424) (0.0525) (0.0577)
Distance2 −0.1984* −0.4229* −0.3379* −0.6394* 0.0288* 0.1337* −0.0785*
(0.0022) (0.0110) (0.0381) (0.0182) (0.0110) (0.0134) (0.0148)
Online 0.2355* 0.4981* 0.2571* 0.4504* 0.5599* 0.3804* 0.3475*
(0.0024) (0.0108) (0.0306) (0.0133) (0.0108) (0.0135) (0.0165)
Airlines
Southwest −0.0508* −0.5091* 0.0513 2.5735* −1.3551* −0.3171* 0.0467
(0.0060) (0.0483) (0.0311) (0.3364) (0.0483) (0.0387) (0.0395)
American 0.0033 0.0494 −0.3188* 0.9394* −2.5140* 0.0003 0.1759*
Airport and Airline Substitution Effects in Multi-Airport Markets
(0.0059) (0.0300) (0.0346) (0.3421) (0.0300) (0.0251) (0.0338)
Delta −0.0051 −0.2148* −0.4034 0.7438* −0.6260* −0.0212 0.1681*
(0.0056) (0.0309) (0.0336) (0.0929) (0.0309) (0.0299) (0.0305)
United −0.0877* −0.0102 −0.3356* −1.3001* −0.2873* −0.1097* 0.0311
(0.0058) (0.0307) (0.0375) (0.4731) (0.0307) (0.0283) (0.0361)
Continental 0 0.0147 −0.3605* 0.0018 −0.2271* 0.0785 0.1702*
(0.0061) (0.0402) (0.0331) (0.0647) (0.0402) (0.0435) (0.0350)
Northwest −0.2735* −0.2763* −0.5688* 0.0524 −0.4565* −0.2048* −0.2177*
(0.0068) (0.0352) (0.0495) (0.0620) (0.0352) (0.0276) (0.0425)
U.S. airways −0.1008* −0.0037 −0.2671* 1.1421* −0.6464* −0.1262* −0.0236
(0.0057) (0.0308) (0.0362) (0.2661) (0.0308) (0.0448) (0.0291)
JetBlue −0.0674* 0.5878* 0.1570* −1.7648 −0.2064* 0.0321 0.3872*
(0.0104) (0.0351) (0.0740) (0.0982) (0.0351) (0.0681) (0.0481)
Airtran −0.2813* −0.2749* −0.1434 0.9564* −0.1187 −0.0831
(0.0081) (0.0434) (0.0791) (0.0674) (0.0711) (0.0489)
Model
Gamma 0.4987* 0.5001* 0.501 0.5005* 0.5001* 0.499 0.4997
(0.0365) (0.2159) (0.3424) (0.0792) (0.1096) (0.5418) (0.4654)
Lambda 0.7480* 0.8325* 0.8199* 0.8926* 0.7770* 0.8063* 0.7906*
(0.0083) (0.0356) (0.0468) (0.0550) (0.0484) (0.0411) (0.0375)
*significance at 5% level.
327
328
Table 4. Airport Elasticities.
Full Model Local Model
New York EWR JFK LGA Market EWR JFK LGA Market
Initial share 0.395 0.295 0.301 0.395 0.295 0.301

EWR −2.465 0.528 0.599 −0.632 −2.138 0.491 0.486 −0.549
JFK 0.302 −2.430 0.248 −0.521 0.299 −2.300 0.251 −0.483
DAN MAHONEY AND WESLEY W. WILSON

LGA 0.607 0.438 −2.393 −0.373 0.504 0.426 −2.032 −0.305
Washington, DC BWI DCA IAD Market BWI DCA IAD Market
Initial share 0.453 0.299 0.248 0.453 0.299 0.248

BWI −1.724 0.338 0.333 −0.598 −0.584 0.124 0.129 −0.195
DCA 0.365 −1.731 0.374 −0.259 0.124 −0.584 0.133 −0.085
IAD 0.293 0.305 −1.735 −0.206 0.098 0.101 −0.636 −0.083
Chicago MDW ORD Market MDW ORD Market
Initial share 0.309 0.691 0.309 0.691

MDW −2.203 0.164 −0.567 −1.682 0.111 −0.443
ORD 0.960 −1.457 −0.711 0.733 −0.990 −0.458
Dallas DAL DFW Market DAL DFW Market
Initial share 0.208 0.792 0.208 0.792

DAL −1.914 0.059 −0.351 −1.055 0.059 −0.172
DFW 0.876 −1.107 −0.696 0.468 −0.723 −0.476
San Francisco OAK SFO SJC Market OAK SFO SJC Market
Initial share 0.262 0.531 0.207 0.262 0.531 0.207

OAK −1.862 0.110 0.151 −0.398 −1.298 0.068 0.099 −0.283
SFO 0.523 −1.471 0.633 −0.514 0.408 −0.846 0.420 −0.256
SJC 0.277 0.246 −1.835 −0.177 0.193 0.136 −1.201 −0.126
Los Angeles BUR LAX LGB Market BUR LAX LGB Market
Initial share 0.161 0.758 0.081 0.161 0.758 0.081

BUR −1.879 0.091 0.258 −0.213 −2.315 0.118 0.204 −0.267
LAX 0.846 −1.063 0.905 −0.596 0.994 −1.427 1.014 −0.839
LGB 0.052 0.020 −2.350 −0.167 0.061 0.036 −2.536 −0.168
Notes: Cells refer to a hypothetical percentage increase in all fares for all itineraries originating from the row airport, and the subsequent
percentage change in passengers originating from the column airport. The Market column gives the total percentage change in passengers
across all airports in response to a price change at a single airport.
329
Table 5. Exit Shares.

Exit Rate (Full) Exit Rate (Local)
New York EWR 0.649 0.650

JFK 0.727 0.712
LGA 0.517 0.498
Washington, DC BWI 0.765 0.739
DCA 0.501 0.488
IAD 0.480 0.527
Chicago MDW 0.833 0.851
ORD 0.706 0.670
Dallas DAL 0.880 0.784
DFW 0.793 0.831
San Francisco OAK 0.815 0.833
SFO 0.658 0.569
SJC 0.465 0.506
Los Angeles BUR 0.703 0.717
LAX 0.740 0.776
LGB 0.876 0.819
Mean 0.694 0.686
Notes: This table presents the share of passengers who choose to exit the market entirely, con-
ditional on switching away from their originating airport in response to a price increase. The
mean is the unweighted mean across markets.
decline in traffic (approximately 11,000 passengers), while Baltimore/

Washington International and Dulles International airports would both see
increase of about 0.13% (corresponding to approximately 2,200 and 1,900
passengers, respectively).
Comparing the results across markets, consumers appear to be less
responsive to a hypothetical price change at the largest airport in the mar-
ket. This is consistent with the literature, as the largest airport is typically
home to the trunk carriers, often using the airport as a hub. These airlines
compete most strongly on nonprice characteristics, such as offering direct
flights. As rival prices become less competitive, it is natural to see consu-
mers flock to the dominant carriers.
Comparing the elasticity estimates of the full model to the estimates of
the localized models, they are typically quite close. The largest disparities
come from the Washington, DC, and San Francisco metropolitan areas.
The full model tends to overstate the substitution effect relatively to the
local models.
Though there is substitution across airports, this tends to be dominated
by passengers who choose to exit the market entirely. Though the market
elasticities presented in Table 4 are smaller than the proportional share of

the particular airport, they are still large. Table 5 presents the shares of pas-
sengers who, conditional on switching from their original origin airport,
choose to exit the market rather than adopt an alternate origin. On aver-
age, slightly more than 30% of passengers who have abandoned their origi-
nal airport in response to this hypothetical price change will choose to stay
in the market. Across markets, the share is highest at LaGuardia
International Airport, and Reagan International Airport, where slightly
more than 50% of the passengers will stay in the market, and lowest at
Chicago Midway, where fewer than 15% of the passengers stay in the
market.
Tables A.1A.3, in the appendix, present similar results to Table 4,
except rather than reporting the change in airports’ traffic given a change
in the prices at an airline, it presents the predicted change in traffic at an
airport if a particular airline changes its price. For example, in the New
York City market, Continental Airlines (which has presently merged with
United Airlines, but was operating independently at the time of the sample)
operated a hub out of Newark Liberty International Airport. A 1%
increase in Continental’s fares would cause nearly a 1% drop in Newark’s
traffic (approximately 7,000 passengers), while JFK and LaGuardia would
see an increase of approximately 1,000 and 500 passengers, respectively.
Though not all passengers are expected to switch airports, or even airlines
in response to a price increase, substitution to the outside good (no air tra-
vel) tends to significantly outrank substitution within the market
(Table B.1).
CONCLUSION
We estimate a model of airline demand, similar to that of Berry et al.

(2006), with particular attention focused on a set of multi-airport markets
which are major domestic and international airports. Using the estimated
demand parameters, we estimate consumers’ preferences and substitution
patterns between airports. The degree of substitutability across airports
varies based on the market, with the most cross-airport substitution occur-
ring in New York and Los Angeles, and the least in Washington, DC.
Looking at airlineairport interactions, particularly vulnerable are the air-
ports that cater to LCCs, who may not have the networks in place to
attract passengers if their prices become less attractive.
The “Results” section provide an overview of the consumers’ airport-

airline decision-making process, identifying flight-specific parameters,
airportairline interactive parameters, and purely airport characteristics.
Estimating elasticities from these parameters, substitutability between air-
ports appears to be higher among the customers of the LCCs, who may
turn to the large hub airports supported by the trunk carriers when their
low fares are no longer so attractive. In all cases, substitution to the outside
good (i.e., consumers choosing not to fly) in response to a hypothetical
price increase significantly outweighs substitution patterns within the
market.
NOTES
1. Calculated from Worldbank figures: http://data.worldbank.org/indicator/IS.

AIR.PSGR.
2. Air transportation statistics from the International Air Transport Association
annual report, available online at http://www.iata.org/pressroom/Documents/
annual-report-2011.pdf.
3. BEA Local Area Personal Income data available from their web site, http://
www.bea.gov/regional/reis/.
4. Brueckner, Lee, and Singer (2011) is devoted to the topic of which airports
should be considered clustered. Though this paper chooses to focus on the six multi-
airport cities of Berry et al. (2006), it could just as well be applied to an extended set
of multi-airport cities.
REFERENCES
Bailey, E. E., & Panzar, J. C. (1981). The contestability of airline markets during the transition
to deregulation. Law and Contemporary Problems, 44(1), 125145.
Baskar, G., & Bhat, C. (2004). A parameterized consideration set model for airport choice: An
application to the San Francisco bay area. Transportation Research: Part B, 38, 10.
Berry, S., Carnall, M., & Spiller, P. T. (2006). Airline hubs: Costs, markups and the implica-
tions of customer heterogeneity. In D. Lee (Ed.), Advances in Airline Economics (Vol.
1): Competition Policy and Antitrust. North Holland: Elsevier.
Berry, S., Levinsohn, J., & Pakes, A. (1995). Automobile prices in market equilibrium.
Econometrica: Journal of the Econometric Society, 63, 841890.
industry. The Rand Journal of Economics, 20(3), 344365.
Brueckner, J., Dyer, N., & Spiller, P. (1992). Fare determination in airline hub-and-spoke net-
works. The RAND Journal of Economics, 23(3), 309333.
Brueckner, J. K. (2002). Airport congestion when carriers have market power. The American
Economic Review, 92(5), 1357.
Brueckner, J. K., Lee, D. N., & Singer, E. S. (2011). Alliances, codesharing, antitrust immu-
nity, and international airfares: Do previous patterns persist? Journal of Competition
Law and Economics, 7(3), 573602.
Dresner, M., Windle, R., & Yao, Y. (2002). Airport barriers to entry in the US. Journal of
Transport Economics and Policy, 36, 3.
Dubé, J.-P., Fox, J. T., & Su, C.-L. (2012). Improving the numerical performance of static and
dynamic aggregate discrete choice random coefficients demand estimation.
Econometrica, 80(5), 22312267.
Forsyth, P. (2006). Airport competition an: Regulatory issues and policy implications. In D.
Lee (Ed.), Advances in airline economics: Competition policy and antitrust (pp.
347368). Amsterdam: Elsevier.
Graham, D. R., Kaplan, D. P., & Sibley, D. S. (1983). Efficiency and competition in the airline
industry. The Bell Journal of Economics, 14(1), 118138.
Hsu, C., & Wu, Y. (1997). The market size of a city-pair route at an airport. The Annals of
Regional Science, 31(4), 391.
Jia, B. P., & Berry, S. (2010). Tracing the woes: An empirical analysis of the airline industry.
American Economic Journal: Microeconomics, American Economic Association, 2(3),
143.
Morrison, S. A. (2001). Actual, adjacent, and potential competition: Estimating the full effect
of Southwest airlines. Journal of Transport Economics and Policy, 35(2), 239256.
Morrison, S. A., & Winston, C. (1987). Empirical implications and tests of the contestability
Oum, T. H., & Fu, X. (2008). Impacts of airports on airline competition: Focus on airport per-
formance and airport-airline vertical relations. Paris: Joint Transport Research Centre of
the OECD and the International Transport Forum.
Pels, E., Nijkamp, P., & Rietveld, P. (2001). Airport and airline choice in a multiple airport
region: An empirical analysis for the San Francisco bay area. Journal of Planning
Literature, 16(1), 80163.
Pels, E., & Verhoef, E. T. (2003). The economics of airport congestion pricing. Amsterdam
[u.a.].
Wei, W., & Hansen, M. (2005). Impact of aircraft size and seat availability on airlines#8217
demand and market share in duopoly markets. Transportation Research: Part E, 41(4),
315.
Windle, R., & Dresner, M. (1995). Airport choice in multiple-airport regions. Journal of
Transportation Engineering, 121, 4.
APPENDIX A: ADDITIONAL TABLES
Table A.1. AirlineAirport Elasticities (1).

Full Model Local Model Full Model Local Model
EWR JFK LGA Market EWR JFK LGA Market BWI DCA IAD Market BWI DCA IAD Market
New York Washington, DC

AA −0.021 −0.238 −0.403 −0.203 −0.005 −0.229 −0.363 −0.182 AA −0.075 −0.230 −0.059 −0.117 −0.023 −0.072 −0.024 −0.038
AS −0.012 −0.024 0.003 −0.011 −0.031 −0.026 0.007 −0.018 AS 0.004 −0.018 0.002 −0.003 0.002 −0.007 0.000 −0.001
B6 0.024 −0.437 0.033 −0.109 0.025 −0.439 0.052 −0.104 B6 0.007 0.013 −0.067 −0.010 0.002 0.006 −0.031 −0.005
CO −0.969 0.238 0.066 −0.292 −0.871 0.233 0.071 −0.253 CO −0.084 −0.096 −0.003 −0.067 −0.035 −0.031 0.000 −0.025
DL −0.130 −0.787 −0.401 −0.408 −0.081 −0.749 −0.318 −0.352 DL −0.195 −0.296 −0.113 −0.205 −0.081 −0.103 −0.036 −0.076
F9 0.010 0.012 −0.044 −0.006 0.014 0.016 −0.055 −0.007 F9 0.010 −0.044 0.011 −0.006 0.003 −0.014 0.004 −0.002
FL 0.020 0.025 −0.094 −0.014 0.012 0.013 −0.056 −0.009 FL −0.122 −0.034 −0.006 −0.067 −0.037 −0.014 −0.006 −0.022
NK 0.005 0.004 −0.020 −0.003 0.004 0.003 −0.014 −0.002 NK 0.001 −0.004 0.001 0.000 0.000 −0.001 0.000 0.000
NW −0.020 −0.021 −0.088 −0.041 −0.017 −0.018 −0.086 −0.039 NW −0.035 −0.040 −0.001 −0.028 −0.016 −0.019 −0.001 −0.013
UA −0.208 −0.059 −0.201 −0.162 −0.169 −0.053 −0.203 −0.145 UA −0.088 0.047 −0.749 −0.211 −0.017 0.016 −0.272 −0.070
US −0.292 −0.127 −0.287 −0.242 −0.246 −0.073 −0.243 −0.194 US −0.198 −0.439 0.036 −0.211 −0.057 −0.138 0.017 −0.063
VX 0.010 −0.076 0.005 −0.017 0.012 −0.085 0.006 −0.019 VX 0.008 0.006 −0.042 −0.005 0.001 0.001 −0.007 −0.001
WN 0.017 0.021 −0.074 −0.010 0.015 0.018 −0.065 −0.009 WN −0.308 0.081 −0.046 −0.127 −0.108 0.030 −0.021 −0.045
YX 0.011 0.005 −0.039 −0.006 0.006 0.006 −0.025 −0.003 YX 0.007 −0.034 0.007 −0.005 0.003 −0.014 0.003 −0.002
MDW ORD Market MDW ORD Market DAL DFW Market DAL DFW Market
Chicago Dallas
AA 0.282 −0.445 −0.221 0.219 −0.293 −0.135 AA 0.478 −0.532 −0.323 0.256 −0.370 −0.240
AS 0.013 −0.016 −0.007 0.006 −0.007 −0.003 AS 0.009 −0.018 −0.012 0.001 −0.004 −0.003
B6 0.004 −0.008 −0.004 0.002 −0.003 −0.001 CO −0.149 −0.041 −0.063 −0.046 −0.032 −0.035
CO 0.031 −0.052 −0.026 0.019 −0.031 −0.016 DL −0.018 −0.137 −0.112 −0.030 −0.085 −0.074
DL −0.310 −0.090 −0.158 −0.280 −0.066 −0.132 F9 0.034 −0.043 −0.027 0.011 −0.017 −0.011
F9 −0.172 0.013 −0.044 −0.098 0.007 −0.026 FL 0.023 −0.026 −0.016 0.021 −0.023 −0.014
FL −0.224 0.016 −0.058 −0.142 0.009 −0.037 NW −0.030 −0.008 −0.013 −0.023 −0.007 −0.010
NK 0.003 −0.003 −0.001 0.002 −0.001 0.000 SY 0.000 0.000 0.000 0.000 0.000 0.000
NW −0.051 −0.009 −0.022 −0.068 −0.008 −0.027 UA 0.025 −0.086 −0.063 0.026 −0.067 −0.048
U5 0.001 −0.001 0.000 0.001 −0.001 0.000 US 0.146 −0.191 −0.121 0.071 −0.102 −0.066
UA 0.413 −0.644 −0.318 0.317 −0.429 −0.199 WN −1.569 0.048 −0.288 −0.880 0.049 −0.144
US 0.108 −0.147 −0.068 0.089 −0.120 −0.056 YX 0.004 −0.010 −0.007 0.003 −0.006 −0.004
WN −1.350 0.099 −0.348 −1.021 0.067 −0.269
335
OAK SFO SJC Market OAK SFO SJC Market BUR LAX LGB Market BUR LAX LGB Market
San Francisco Los Angeles
AA 0.112 −0.133 −0.320 −0.107 0.112 −0.090 −0.290 −0.078 AA −0.127 −0.166 0.108 −0.138 −0.360 −0.241 0.248 −0.220
AS −0.053 −0.016 −0.037 −0.030 −0.022 −0.013 −0.006 −0.014 AS −0.068 −0.040 −0.179 −0.056 −0.026 −0.043 −0.033 −0.039
B6 −0.085 −0.005 −0.001 −0.025 −0.086 0.000 0.000 −0.022 B6 −0.021 0.007 −0.842 −0.066 −0.027 0.013 −1.227 −0.094
CO 0.051 −0.080 −0.121 −0.054 0.040 −0.063 −0.062 −0.036 CO 0.031 −0.059 0.031 −0.037 0.046 −0.085 0.043 −0.054
DL −0.110 −0.205 −0.111 −0.161 −0.112 −0.105 −0.070 −0.100 DL −0.012 −0.210 −0.489 −0.201 0.059 −0.257 −0.299 −0.210
F9 0.025 −0.021 −0.048 −0.014 0.019 −0.013 −0.039 −0.010 F9 0.014 −0.011 0.015 −0.005 0.027 −0.027 0.031 −0.014
FL 0.007 −0.014 0.006 −0.004 0.005 −0.009 0.004 −0.002 FL 0.012 −0.021 0.012 −0.013 0.023 −0.034 0.026 −0.020
G4 −0.003 0.000 0.000 −0.001 −0.004 0.000 0.000 −0.001 G4 0.000 −0.001 0.000 0.000 0.002 −0.007 0.000 −0.005
NW 0.011 −0.031 −0.020 −0.018 −0.001 −0.016 −0.010 −0.011 NK 0.000 −0.001 0.001 0.000 0.001 −0.002 0.003 −0.001
UA 0.024 −0.465 −0.025 −0.246 −0.038 −0.258 −0.011 −0.149 NW 0.010 −0.017 0.012 −0.010 0.029 −0.043 0.012 −0.027
US −0.269 −0.096 −0.107 −0.144 −0.269 −0.054 −0.057 −0.111 RJ 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000
VX 0.019 −0.044 0.021 −0.014 0.011 −0.018 0.010 −0.004 SY 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000
WN −0.795 0.004 −0.294 −0.267 −0.358 −0.001 −0.155 −0.126 UA −0.180 −0.241 0.271 −0.189 −0.034 −0.303 0.211 −0.218
YX 0.002 −0.002 0.001 −0.001 0.002 −0.002 0.001 0.000 US −0.217 −0.104 −0.335 −0.141 −0.414 −0.109 −0.571 −0.196
VX 0.030 −0.020 0.030 −0.008 0.016 −0.026 0.047 −0.013
WN −0.472 −0.057 0.158 −0.107 −0.626 −0.091 0.171 −0.156
YX 0.006 −0.004 0.007 −0.002 0.009 −0.009 0.006 −0.005
APPENDIX B
Table B.1. Airline Codes.

Carrier Code Carrier Name
AA American Airlines Inc.

AS Alaska Airlines Inc.
B6 JetBlue Airways
CO Continental Air Lines Inc.
DL Delta Air Lines Inc.
F9 Frontier Airlines Inc.
FL AirTran Airways Corporation
G4 Allegiant Air
NK Spirit Air Lines
NW Northwest Airlines Inc.
RJ Alia-(The) Royal Jordanian
SY Sun Country Airlines d/b/a MN Airlines
SY Sun Country Airlines
U5 USA 3000 Airlines
UA United Air Lines Inc.
US US Airways Inc.
VX Virgin America
WN Southwest Airlines Co.
YX Midwest Express Airlines
CHAPTER 13
BUYER SUBSIDIES IN TWO-SIDED

MARKETS: EVIDENCE FROM
ONLINE TRAVEL AGENTS
Volodymyr Bilotkach and Nicholas G. Rupp
ABSTRACT
Platforms in two-sided markets are known to provide subsidies to either

buyers or sellers, in order to take advantage of cross-group externalities
inherent in such industries. Online travel agents can be thought of as
platforms facilitating trade between passengers and travel service provi-
ders (airlines). This chapter evaluates the effects of a buyer subsidy pro-
vided by one major US online travel agent a low-price guarantee
offered by Orbitz. We find evidence consistent with increased airline par-
ticipation with this travel agent upon implementation of the low-price
guarantee policy. Our results also confirm the theoretical claims that
most-favored customer low-price guarantee policies are procompetitive.
Keywords: Two-sided markets; price matching; price guarantees;
most-favored customer; meet or release
JEL classifications: D4; L4; L93

ISSN: 2212-1609/doi:10.1108/S2212-160920140000004012
339
340 VOLODYMYR BILOTKACH AND NICHOLAS G. RUPP
Our 2008 initiatives to reignite growth succeeded and resulted in good growth in net rev-
enue … customers increasingly recognize the value of Orbitz Price Assurance and choose
to book at Orbitz.com. Steven Barnhart, CEO and president of Orbitz Worldwide
(Third Quarter, 2008 Earnings Release, November 10, 2008)
INTRODUCTION
In many industries across the world, interaction between buyers and sellers
is facilitated via one or more intermediaries. Where those markets are char-
acterized by cross-group externalities, they are termed “two-sided mar-
kets.” For instance, the value of a credit card to a buyer depends on the
number of merchants that accept the card; a game console is more valuable
to a gamer the more games he or she can play on the console; an adverti-
ser’s value of a newspaper or a web site depends on the number of readers
or visitors, etc. Intermediaries on such two-sided markets are often called
platforms. One outcome stemming from the existence of these cross-group
externalities is that platforms tend to provide subsidies to either buyers
(e.g., some newspapers are distributed free of charge) or sellers (software
producers supporting application developers). While some empirical studies
(e.g., Kaiser & Wright, 2006) showed that pricing in two-sided markets
appears to be consistent with the presence of such subsidies, previous
research has not documented how subsidies affect competition between
platforms. This is the contribution of this chapter.
The online travel services distribution industry is a typical two-sided
market. Online travel agents can be thought of as platforms facilitating
trade between passengers and travel service providers (airlines). Spread of
the Internet has led to emergence of several global ticket distribution plat-
forms, while some online travel agents remain focused on local markets.
Our study exploits a recent development in this industry, in which buyers
are effectively provided a subsidy. Among online US travel agents, Orbitz
initiated a most-favored customer (MFC) type of low-price guarantee pol-
icy beginning June 6, 2008 with the launch of “Orbitz Price Assurance.” All
consumers who purchase airline tickets on Orbitz are automatically
enrolled in this program (there are no claims to file or lower prices to
locate). Consumers receive a check for the difference (between $5 and $250)
in airfares if another Orbitz consumer subsequently purchases an airline
ticket with an identical itinerary (same carrier, dates, flight numbers, and
class of service) prior to departure. This constituted a departure from the
Buyer Subsidies in Two-Sided Markets 341
industry standard of refunding the difference should a customer locate a

lower quote on a competing platform (the so-called meet-or-release (MOR)
type of low-price guarantee). The other two platforms opted not to match
this buyer subsidy (interestingly, Cooper (1986) and Neilson and Winter
(1993) suggest models consistent with such an outcome). This move by
Orbitz was expected to attract more buyers, making this platform more
attractive for sellers (airlines), potentially inducing a higher degree of air-
line participation. Increased participation would manifest in the data
through cheaper tickets being available via Orbitz after the introduction of
the new low-price guarantee policy. In this chapter, we use a sample of US
domestic airfare quotes collected (in 2006 and 20082009) before and after
the MFC adoption by Orbitz to conduct a difference-in-differences (DID)
examination of airfare quotes to evaluate the competitive effects of this
buyer subsidy.
Results of the data analysis are consistent with increased airline
participation with the Orbitz platform following the subsidy offered by this
travel agent to the buyers. Potential changes in contracts between the plat-
form and the sellers are unfortunately unobserved due to confidential
airlineagent contracts; however, specifications that attempt to control
for such changes still point to increased participation of the airlines with
Orbitz. Since increased participation translates into the airlines’ lowest
fares being more readily available via more platforms; we can say that the
buyer subsidy offered by one of the online travel agents has brought bene-
fits to the traveling public.
Even though the focus of our investigation is on platform strategies in
two-sided markets, this study also contributes to understanding of the
effects of low-price guarantees. Previous theoretical studies (Neilson &
Winter, 1993; Schnitzer, 1994) and empirical work (Chen & Liu, 2011) sug-
gest that MFC-type low-price guarantees have a higher procompetitive
potential than MOR-type clauses. Our study provides additional empirical
evidence to support these claims.
Finally, this work adds to our understanding of the complex and intri-
guing online travel services distribution industry. In this respect, the most
similar paper to ours is Bilotkach and Pejcinovska (2012), who examine
price dispersion for the three largest online travel agencies. That investiga-
tion provided a rather static snapshot of that market, whereas in this chap-
ter we are able to observe and investigate effects of a strategic action by
one of the online travel services distribution platforms.
The rest of the chapter is organized as follows. The second section
reviews the relevant literature. The third section describes institutional
details of the US ticket distribution market. The fourth and fifth sections
describe the data and present results of our analysis, respectively. The sixth
section discusses and concludes.
LITERATURE REVIEW
Our study is related to two broad strains of literature works on two-

sided markets and studies of low-price guarantees. We contribute to both
bodies of work in the following way. First, in the two-sided markets litera-
ture, this chapter provides the first empirical analysis of effects of buyer
subsidies in such markets. Second, our contribution to the literature on
low-price guarantees is that we provide empirical support for the previously
proposed suggestion that MFC-type low-price guarantees have a higher
procompetitive potential than MOR-type clauses.
Theoretical literature on two-sided markets is very extensive. Recent
papers providing excellent overviews of these studies include Rysman
(2009) and Rochet and Tirole (2003, 2006). Generally, two-sided markets
are defined as industries where two sets of agents interact through an
intermediary or platform; and decisions of each set of agents affect the
outcomes of the other set of agents, typically through an externality
(Rysman, 2009). Due to network externalities inherent in these industries,
theoretical studies of two-sided markets have mostly focused on how
platforms set access prices for both sides of the market. One important
insight that comes from this literature is that a platform might set prices
below cost to the more price elastic side of the market, effectively subsi-
dizing participation of the more price sensitive side, hoping to recoup the
losses via increased participation on the other side, due to the network
externality.
Empirical literature on two-sided markets is understandably less
developed. Most empirical studies focus on pricing and market power in
the newspaper industry (other empirical studies of two-sided industries
include credit cards (Rysman, 2007) and video games Derdenger (2014,
forthcoming) and Kaiser and Wright (2006) analyze data on magazines in
Germany: their results are consistent with magazines subsidizing readers,
recouping the corresponding losses from higher advertising revenue.
Argentesi and Filistrucchi (2007) examine a structural model of Italian
newspaper industry with the aim of estimating the firms’ market power.
Chandra and Collard-Wexler (2009) find that mergers in the Canadian
newspaper industry do not lead to higher prices for either subscribers or

advertisers, consistent with their theoretical model.
Our contribution to the empirical analysis of two-sided markets is two-
fold. First, we provide the initial examination of travel agents as platforms.
While previous studies (Bilotkach & Pejcinovska, 2012; Chen, 2006;
Clemons, Hann, & Hitt, 2002) have compared price quotes across travel
agents, none of these prior papers examined agents as platforms in a two-
sided market, nor addressed the agents’ business strategies. Second, we
examine the effect of a buyer subsidy on competition between platforms.
In addition to examining travel agents as platforms, this chapter also
contributes to the rather extensive literature on price guarantees, which has
yet to come to a consensus on whether low-price guarantee policies are
pro- or anticompetitive. On one hand, Holt and Scheffman (1987), Deck
and Wilson (2003), Dugar and Sorensen (2006), and Chatterjee and Roy
(1997) assert that low-price guarantees support tacit collusion. Yet, Hviid
and Shaffer (1999) show that such tacit collusion can easily collapse in the
presence of hassle costs; Moorthy and Winter (2006) and Lurie and
Srivastava (2001, 2005) suggest that price matching serves as a credible sig-
nal to consumers that a firm is low priced. Empirical studies of low-price
guarantees include Arbatskaya, Hviid, and Shaffer (2004, 2006), Manez
(2006), Scott-Morton (1997a, 1997b), Crocker and Lyon (1994), and Chen
and Liu (2011).
Schnitzer (1994) finds that the “MOR” clause has more anticompetitive
potential than the “most favorable customer” clause. We effectively pro-
vide evidence to support this theoretical conjecture; this resonates with evi-
dence for the consumer electronics industry in Chen and Liu (2011).
INSTITUTIONS
Ticket Distribution Market
Recently, there has been a growing body of literature on the online US tra-
vel services industry. Bilotkach and Pejcinovska (2012) suggest that travel
agents are more than simple “technical” intermediaries, in that they appear
to strategically choose which fare quotes to present. Chen (2006) finds little
disparity in fares quoted by the major online travel agents (Travelocity and
Orbitz) and by the airlines themselves on the New YorkLos Angeles air
travel market. Clemons et al. (2002), however, observe substantial
differences in fare quotes across five unidentified online travel agents. All
the aforementioned studies treat the ticket distribution industry as a “black
box” with little attempt being made to understand the inner structure of
this market or participants’ incentive. We attempt to open up this black
box slightly by speaking directly with representatives from the online travel
agencies. These conversations were only moderately successful, since
important details of agentairline relationships remain concealed in confi-
dential contracts. The institutional details described come from our conver-
sations with various sources within the air travel services industry, all of
which requested to remain anonymous.
The visual scheme of distribution of airline travel services appears on
Fig. 1. An airline in the United States sells its tickets either directly (using
its call center and web site) or via travel agents, by posting fares on one or
more computer reservation systems (CRS), which travel agents access to
book tickets on behalf of customers. CRSs were launched by the airlines in
the 1970s. Since that time, the CRS companies have taken a life of their
own as independent entities. The four major CRS companies are Sabre
(with about 45% US market share and over 30% global market share),
Worldspan (over 25% US market share and 15% worldwide), Galileo and
Amadeus (share of these two systems on the US market keeps declining
Airline
Computer Reservation Systems Direct

(Sabre, Worldspan, Amadeus, distribution –
Galileo) call center or
web site
Brick-and- Online travel

mortar travel agents
agents
Customer
Fig. 1. Airline Ticket Distribution.

while they remain solid players on the worldwide arena, with combined
market share of over 50%).
Prior to 2001, the total price customers paid for tickets did not depend
on the ticket distribution source, since airlines paid two commissions: one
to the travel agent for selling the ticket (industry sources indicate a stan-
dard commission is 10% of the ticket price) and a second commission or
“booking fee” to the CRS. After September 11, 2001, the airlines, in an
effort to control costs, gradually stopped paying travel agent commissions.
This change prompted travel agents to begin charging customers booking
fees (ranging from $525 per ticket). Online travel agents in the United
States eliminated booking fees for travelers around March 2009. Presently,
airlines still pay commission to CRSs, as well as to the largest travel agents,
including dominant firms on the online segment of the travel services distri-
bution market.
The airlines are currently free to choose which systems to participate in
and at what level to do so.1 Most carriers actively participate in multiple
systems. An exception is Southwest Airlines, which only participates in
Sabre, and does so at a low level. Consequently, Southwest Airlines consu-
mers can only book tickets directly through the carrier’s web site, call cen-
ter, or via some brick-and-mortar agents. As for the online US travel
agents covered by this study, Travelocity is linked to Sabre, whereas
Expedia and Orbitz are both linked to Worldspan CRS.
The Internet has altered the travel services industry dramatically.
Emergence of online travel agents was a major innovation in the industry.
Airlines also saw potential in selling their tickets via their own web sites.
As late as in 1990s, the brick-and-mortar travel agents sold over three quar-
ters of all airline tickets; with the remaining 25% sold directly by the airline
operated call centers. By 2002, online travel agents captured about 15% of
the market from brick-and-mortar agents. The estimate by Citigroup
Investment Research, quoted by Forbes, states that in 2005 online travel
agents have over 25% market share in the airline ticket distribution indus-
try; the airlines still sell about a quarter of all tickets, primarily via their
web sites.2 Hence, the brick-and-mortar agents’ market share has declined
to about 50%.
On the US market, the online travel agent segment of the ticket distribu-
tion market is dominated by the three major players: Travelocity (owned
by Sabre, currently a privately held company), Expedia (founded within
Microsoft in 1995, and an independent publicly traded company since
2005), and Orbitz (started through a partnership of several major airlines
in 2001, currently a subsidiary of Travelport, owned by the Blackstone
Group a private equity company). According to the US Department of

Transportation, in 2002, 28.5% of all bookings with online travel agents
were on Travelocity; 28.7% on Expedia and 21.3% on Orbitz, for a total
of 78.5% of all US travel reservations via an online travel agent.3 While
more recent estimates of the agents’ market shares are not available it does
not appear that the market structure has changed dramatically over the last
decade.
Table 1 revenues include both domestic and international operations.
Due to Expedia’s large international presence, it has considerably larger
gross booking revenue than the other two travel agencies. If we exclude
international revenue, then the growth in booking revenues between 2006
and 2009 for Expedia (13%) would have been closer to its peers. Assuming
no drastic changes in the combined market share of the three biggest
players, we can say that approximately one in five domestic trips is booked
via one of the three largest online travel agents. Thus, our study is reflective
of the entire online travel agent industry, and the segment we are looking
at comprises a nontrivial part of the US air travel distribution market.
Online US Travel AgentAirline Relationships
According to our industry sources, airlines enter into individual agreements

with the large distributors of their services. While a small brick-and-mortar
agent typically deals directly with a CRS; large online agents enter into
direct agreements with the airlines. Those agreements are typically renego-
tiated every three to five years. Sometimes these negotiations are quite con-
tentious, as a 2011 dispute between American Airlines and two large online
travel agencies (Expedia and Orbitz), led to the delisting of American
Airlines fares on these two online travel agencies web site during the first
Table 1. Total Annual Gross Travel Booking Revenue

(in Millions US$).a
2006 2007 2008 2009 % Change 20062009
Expedia $16,882 $19,632 $21,269 $21,800 29.1

Orbitz $9,780 $10,791 $10,808 $10,140 3.7
Travelocity $10,100 $10,689 $10,567 $10,600 5.0
Total $38,768 $43,119 $44,652 $42,540 9.7
Source: Annual reports and earnings releases for Expedia, Orbitz, and Sabre Holdings.
a
Gross bookings include hotel, car, and airfare for domestic and international. operation.
quarter of 2011. Large ticket distribution platforms are quite valuable to

airlines, as a recent study (Bilotkach, Rupp, & Pai, 2013) finds that when
American Airlines fares were removed from Expedia and Orbitz,
American’s revenue dropped by $50 million. The airlineagency contracts
specify the commission paid by the airline to the travel agents, as well as
the conditions for travel agent’s access to the airline’s inventory. Compared
to hotels and car rentals, travel agents receive small commissions ranging
between 2% and 3% for selling airline tickets (Salzman, 2013). Details of
the contracts constitute confidential information. The crucial parameters
which affect the structure of commissions paid by the airlines are known in
the industry as content and participation.
Content refers to the inventory that the airline makes available to the
agent. It is our understanding that airlines prefer to offer relatively low
level of content to the agents, meaning that carriers seek to keep the lowest
fares to themselves rather than selling those tickets via the intermediaries.
The agents clearly prefer to obtain full access to the airlines’ inventory
this is known as full content. In negotiations, the agents offer the airlines
commission discounts for full content access.
Participation refers to whether the agents are able to access the airline’s
inventory in real time. When participation is asynchronous, the fare quoted
on the agent’s first screen may differ in either direction from the actual
fare at the point of booking the flight.4 Clearly, agents prefer real-time par-
ticipation, as it increases their reliability. No data exists that we are aware
of which reveals the reliability of an agent’s quotes. Our industry contacts
estimate that the agents’ “failure rate” (i.e., the likelihood that a different
fare appears between the first query and the time of booking) is about 5%.
We were unable to receive answers to all our questions from our indus-
try contacts. For example, we do not know whether the contract obliges
the agent to show the lowest available airfare offered by the airline;
whether discounts may be offered by the agents to large corporate clients;
or to what extent the large online agents are used by the corporate travel
departments (we do know that the agents take steps to target this segment
of the industry). Nevertheless, the description of institutional structure pre-
sented above allows us to develop a simple model describing the airline’s
ticket distribution strategy via a travel agent.
Suppose the airline has the predetermined set of fares Pmin < ⋯ < Pmax .
Without loss of generality, assume that each seat is offered at a different
price; further, let us assume that each offered seat sells with the same prob-
ability, holding the number of travelers visiting the travel agent constant.5
The airline’s problem then becomes that of choosing the lowest price quote
P^ ∈ ½Pmin ; Pmax to be offered via the travel agent. Given our assumptions,
we can interpret the airline’s problem as that of choosing the number of
seats to offer for sale via the travel agent. Specifically, when the lowest
price offered via the agent is P, ^ and assuming continuous demand function
QðPÞ, the airline will offer seats for sale via the travel agent. In the above
expression, N denotes the number of customers visiting the agent, and f ðNÞ
represents the probability of selling a seat (clearly, f 0 ðNÞ > 0). Denoting via
PðQÞ the inverse demand function, and using gðNÞ for the respective trans-
formation of f ðNÞ, we can write the expression for the airline’s expected
revenue from selling Q^ seats as:
Q^ = f ðNÞQðPÞ
^
Z Q^
^ = gðNÞ
ERðQÞ PðQÞdQ
0
Next, if we suppose that the airline pays the travel agent a fee of C
monetary units for every seat it sells via the agent; we can formulate the air-
line’s objective function as:
Z Q^
^ = gðNÞ
πðQÞ PðQÞdQ − C Q^
0
The airline’s problem then becomes that of choosing Q^ to maximize the

above expression. For the illustration purposes, let us assume a simple lin-
ear inverse demand function PðQÞ = a − 2Q. Substituting this into the air-
line’s objective function yields:
^ = gðNÞ½aQ^ − Q^ − CQ^
πðQÞ
2
Which easily solves for the optimal number of seats the airline will offer
via the agent as:

^ 1 C
Q = a−
2 gðNÞ
Since a higher number of potential travelers visiting the agent increases

the probability of sale; it comes trivially that:

∂Q^
∂N > 0
That is, the airline will respond to any action by the travel agent that
increases the number of potential customers visiting the agent, by increas-
ing the level of content offered through this outlet. This description of
the airline’s problem is very stylized, relies on assumptions of equal prob-
ability of sale of low- and high-priced tickets, and omits discussion of
possible strategic interaction between the airline and the agent; the issue
of competition between travel agents is also left out. This description,
however, allows us to obtain a clear and intuitive result to motivate our
empirical analysis.
Low-Price Guarantees as Subsidies
As a platform in a two-sided market, an online travel agent strives to

increase the number of customers using the platform to make their travel
arrangements. The most conspicuous strategy used for this purpose is
advertising. Clearly, the more customers that visit a travel agent, the higher
the value of the agent to an airline, and hence the more favorable terms the
agent can negotiate.
To make themselves appear more attractive to the traveling public,
online travel agents have started offering low-price guarantees. Of the
major online travel agencies, Orbitz was the first to launch an MOR policy
applicable to airline tickets in October of 2004. That policy promised a $50
travel voucher to any customer who located a lower priced ticket identical
to the one purchased via Orbitz. To qualify, the fare had to be at least $5
less than the Orbitz fare (not including the booking fee then charged by the
agent), and the customer had to notify Orbitz of the lower price by mid-
night of the day of purchase. Note that this policy did not entitle the trave-
ler to a refund of the price difference; it only provided the customer with a
$50 travel voucher for future Orbitz purchases.
Expedia (in January of 2006) and Travelocity (in August of 2006) also
introduced low-price guarantees, which provided more generous perks to
travelers who located lower fares (details of their price guarantees appear
in Table A2). These price guarantees stipulate that consumers receive a
refund of fare difference in addition to a $50 travel voucher for locating a
lower fare quote within 24 hours of purchase (as with Orbitz’s policy,
booking fees charged by the online travel agents did not count toward the
applicable fare difference). These MOR low-price policies for both Expedia
and Travelocity are still in effect as of 2012.
Thus, since August 2006 all major online travel agents have implemented
a type of MOR low-price guarantee policy for airline ticket sales. In June
2008, however, Orbitz decided to fundamentally change its price guarantee
policy for airline tickets with the launch of “Orbitz Price Assurance,” which
offered the MFC clause type of guarantee (see Appendix for details).
Under this arrangement, if another Orbitz customer subsequently books
the identical itinerary (up to the dates and flight numbers involved) at a
lower price, then Orbitz will automatically issue a refund for the price dif-
ference (from $5 to $250) to the customer who paid the higher price. The
refund is issued several weeks after travel has been completed (should a tra-
veler cancel her plans, the policy does not apply) in the form of a mailed
refund check. The customer no longer has to track airline prices or file a
claim, and there is no time limit on when the price reduction must occur (it
simply has to occur prior to departure). Neither Travelocity nor Expedia
has chosen to match the MFC clause offered by Orbitz; in the Spring of
2009, however, another player in the online travel market, Priceline,
adopted a similar MFC policy.6
Clearly, Orbitz’s price assurance policy represents an effective increase
in the subsidy offered to the passengers, with the aim of increasing custo-
mer traffic. Moreover, since Orbitz automatically issues a refund, this reduces
the transactions costs to the consumer of obtaining a refund. Such a policy
should lead to the airlines increasing the level of content offered via Orbitz.
In the data, this will manifest through lower fare quotes being offered via
Orbitz after it has introduced the MFC type of low-price assurance policy.
DATA COLLECTION AND DESCRIPTION
Collection Process
Domestic US airfare data were collected over two separate periods. First,
we use 2006 airline price quote data, when all three online travel agents
offered MOR pricing policies. Second, we gather more recent fare quotes
in 20082009, after Orbitz adopted its MFC policy. In both periods, the
same data collection techniques are employed. We track fare quotes for the
three leading online travel agents (Travelocity, Expedia, and Orbitz) on 50
airport-pair markets, randomly selected out of the top 100 US domestic

routes as measured by nonstop passenger traffic.7 Consequently, this selec-
tion criterion led to the inclusion of many markets with large hub airline
operations (e.g., Atlanta, Denver, and Dallas-Fort Worth).
The data collection operates on the notion that a traveler is making a
last-minute purchase (within 48 hours prior to departure) for a three-day
trip.8 We opt to track last-minute fares since there is some evidence (com-
ing from datasets provided by CRSs and travel agencies) that a majority of
flight bookings are made at the last minute (e.g., Puller, Sengupta, &
Wiggins, 2009). As a result, we are explicitly assuming that the traveler’s
uncertainty about whether or not she is going to fly has been realized.
Thus, our last-minute traveler is more concerned with finding a low price
than a refundable ticket.9 Hence, we collect the lowest available fare quotes
for round-trip coach-class tickets.
Our traveler is also assumed to have a strong preference for both a given
airport pair, and for nonstop flights. At the same time, she does not exhibit
brand loyalty. Airport-pair markets are directional, for example,
JFKLAX market is different from LAXJFK. Consequently, we collect
fares on both directions for the 50 airport pairs, resulting in 100 directional
routings. A complete listing of all 50 randomly selected airport pairs used
in this analysis appears in the Appendix (Table A1).
The “before” sample covers fare quotes collected between October 20
and November 17, 2006, while the “after” sample involves the period
between December 7, 2008 and January 4, 2009. Clearly, the before and
after sample periods track fares over different months. Since we are
comparing fare offerings between travel agents for identical itineraries (e.g.,
likelihood of offering the lowest fare on a route), we have no reason to
believe that a change in seasons will cause online travel agents to change
their strategy. The higher demand experienced during the Christmas season
for the postperiod affects all online travel agents equally. All fare quotes
are collected on Tuesdays (for departure on Thursday and returning on
Saturday of the same week) and Fridays (for departure on Sunday and
returning the following Tuesday). We collect the data on nine different
dates (preperiod) and eight different dates (postperiod). For each airport-
pair market on each day of data collection, we attempted to achieve near
simultaneity of obtaining fare quotes by launching three browser windows
in parallel.
The sample for the preperiod includes 823 (out of 900 possible) and 683
(out of 800 possible) date-airport-pair market combinations, respectively.
During the data collection process several computer issues hindered the
data acquisition. In addition, web site maintenance at Orbitz.com pre-

vented us from gathering Orbitz quotes for one day as well.10 For every
other date, we obtain fare quotes from each of the three online travel
agents. Finally, in both samples, we omit routes in which only Southwest
Airlines offers nonstop service (HOUDAL and OAKBUR), since this
carrier does not sell tickets via any of the three online travel agents.
For each date-airport-pair market-travel agent combination, we observe
the lowest fare quotes offered by every airline providing nonstop service on a
given airport-pair market. For example, in the LAXJFK market, nonstop
services are offered by Delta Air Lines, American Airlines, United Airlines,
JetBlue Airways, Alaska Airlines, and US Airways. Then, on each day of
data collection we record up to six different fare quotes for each of the
three travel agents (sometimes an agent would not provide a fare quote for
an airline). The result is a total of 10,026 observations, of which 5,264
observations correspond to the “after” (collected in 20082009) sample,
and the rest constitute the “before” sample.
Descriptive Statistics
Before proceeding with more sophisticated data analysis of the competitive

effects of adopting an MFC price assurance policy, we start with the raw
data. Table 2 provides a summary of the average lowest prices for the pre-
and postperiods, at the travel agent level. In the “before” sample
Travelocity offers the lowest average last-minute fare quotes ($548), fol-
lowed by Expedia ($576) and Orbitz ($618). Approximately two years later,
after Orbitz changed its low-price guarantee policy to the MFC scheme, we
find that ordering of the three travel agencies has not changed; however,
Table 2. Summary Statistics by Online Travel Agent for 50 Domestic

Routes: Lowest Last-Minute Ticket Price Quotes for Each Carrier with
Nonstop Service in the Airport-Pair Market.
OctoberNovember 2006 December 2008January 2009
Mean Standard Deviation CV Mean Standard Deviation CV
Expedia 575.99 405.41 0.704 527.96 281.52 0.533

Orbitz 618.27 403.54 0.653 547.45 312.28 0.570
Travelocity 547.68 273.66 0.500 494.40 259.15 0.524
Note: We drop cases where only one agent provides fare quotes for a route.
the difference between travel agencies has shrunk. While Travelocity still
offers the lowest average prices in the “after” sample ($494), Orbitz ($547)
is now closer to Expedia ($528). Alternatively, when comparing the price
changes between these two sample periods, we see that Orbitz has the lar-
gest price reduction ($71 or 11.5%), while Travelocity and Expedia prices
dropped by $53 (9.7%) and $48 (8.3%), respectively. We also note a reduc-
tion in the variation of lowest fares in the more recent sample since the
coefficient of variation decreases substantially for two of the three travel
agents. Unlike Table 2, which tracks the lowest fare for each nonstop car-
rier on the route, Table 3 compares the lowest available fare quotes or
“best fare” offered on the route for the two sample periods when all three
travel agents provide last-minute fare quotes. The table tracks the instances
where the last-minute fare quote is the highest and lowest among the three
online agents.
In the 2006 sample period, when all travel agents offered similar MOR
price matching policies, Orbitz’s best fare was rarely the lowest available
on the route (6.0%) and frequently it was the highest for the route
(71.5%). Two years later, after Orbitz adopts an MFC price assurance pol-
icy; we find the situation has changed considerably, with infrequent occur-
rences of Orbitz having the highest fare quotes (16.2%); and in exactly half
of the observations Orbitz’s posted price is the lowest offered on the route.
These results are consistent with the hypothesis that an agent who offers
the MFC price policy may be reluctant to report high airline fares for fear
that the airline may drop the fare at a later time, forcing the agent to
refund the difference to the customer.
Table 3. Comparison of “Best” Last-Minute Fares Offered by Online US

Travel Agents.
OctoberNovember 2006 December 2008January 2009
Cases Highest Lowest % Cases Highest Lowest %

Fare Fare Lowest Fare Fare Lowest
Expedia 520 91 279 53.7% 684 549 99 14.5%

Orbitz 520 372 31 6.0% 684 111 342 50.0%
Travelocity 520 44 457 87.9% 684 26 632 92.4%
Notes: We only include route-day combinations in which all three travel agents reported fares.
In 2006, there were 22 of 520 situations (2.5%) where all three travel agents reported identical
low fares. These situations were not included in the “highest fare” count. Similarly, in
20082009, there were 63 of 684 cases (9.2%) where all three agents offered identical low
fares.
Fig. 2 provides a histogram of the difference between travel agents’ max-

imum and minimum “lowest” or “best” nonstop fare quotes by airport-
pair market for both sample periods.11 A few observations are readily
apparent from the histogram. First, for approximately half of all last-
minute fare quotes, we find minimal differences (less than $2) between these
three travel agents. Second, differences between travel agents are becoming
only slightly less common over time. For example, in 2006 the difference
between the best fares offered by these three online travel agents exceeded
$25 for 38.1% of the sample. This proportion drops to 37.4% in
20082009. Differences of last-minute best fares that exceed $50 occur for
30.2% of the routes, with this difference shrinking to 28.2% in 20082009.
In Fig. 3 we report histograms for the differences between the individual
travel agent’s “lowest” fare and the best fare quoted on the route across all
travel agents. Comparing the distribution of differences in fares across tra-
vel agents reveals two “stylized facts.” First, Travelocity (see Fig. 3(b)) is
the undisputed low-price leader in both sample periods. Travelocity’s best
fares are rarely significantly higher than its competitors’. In fact,
Travelocity’s fares are over $50 higher than its competitors’ in only 6.7%
(in 2006) and 5.4% (in 20082009) of the cases. In comparison, large fare
0.3
0.25
0.2
Share
0.15
0.1
0.05
0
0
25
00
1+
1–
–5
–7
15
20
30
3–
–1
30
26
51
1–
1–
1–
76
10
15
20
Differences ($) 2008–2009 (blue) and 2006 (red) reflect nonstop prices for 50 domestic
airport pairs (e.g., if the lowest travel agent price on the route is $200 (Travelocity), $203
(Orbitz), and $205 (Expedia), then the maximum difference is $5).
Fig. 2. Histogram of Difference between Online US Travel Agents Maximum and

Minimum “Lowest” Nonstop Fares: 2006 versus 20082009.
differences are two to three times more likely for both Orbitz and Expedia
in 2006 and 20082009 (see Figs. 3(a) and (c)).
The occurrence of large (over $50) fare differences at Expedia has been
relatively unchanged between 2006 and 20082009 at 17.1% and 17.0%,
respectively. Somewhat surprisingly, however, is that Orbitz is more likely
to provide significantly higher fare quotes in the more recent 20082009
period (16.4%) compared to 2006 (13.0%). Hence, our second stylized fact:
in 1 of 4 last-minute fare quotes we find modest to large differences in
Orbitz and Expedia’s lowest price compared to the best fare offered on the
route across all travel agents in both sample periods. Travelocity, as noted
above, seldom posts substantially higher fares than its competitors.
Table 4 provides a comparison of the lowest price quotes by travel agent
and airline for the 50 domestic routes included in our investigation.
Comparing the differences in fares between 2006 and 20082009 samples,
we detect no distinct trend in increasing or decreasing airfares between the
sample periods. While there may be no clear trend in pricing, one result
that should not escape the reader’s attention from Table 4 is the stickiness
in which an online travel agent provides the lowest prices of a particular
airline. Of the 13 airlines listed in Table 4, just three carriers changed the
travel agent with whom they post their lowest prices between 2006 and
20082009 samples. Or stated conversely, 10 of the 13 carriers posted their
lowest prices with the same online travel agent in both samples.
Specifically, we find that Travelocity in 2006 had the lowest average prices
for the following carriers: American, Delta, Frontier, AirTran, Hawaiian,
Northwest, and US Airways. Expedia had the lowest average prices in 2006
for Alaska, Continental, Island, United, and Mesa. While Orbitz in 2006
had the lowest prices for a single airline: Spirit. In 20082009, there are
only three airlines with different lowest price online travel agencies as
Hawaiian and Island now post their lowest fares with Orbitz, while
Alaska’s lowest fares now appear in Travelocity. This “stickiness” result
that the same online travel agent presents the lowest fare quotes for the
same airlines in both sample periods suggests that the importance of indivi-
dual airlineagent agreements.
DID Estimator
We employ reduced-form price specifications (see Eqs. (1) and (2)) to inves-
tigate price changes in airport-pair markets. Specifically, we use a DID esti-
mator to determine how fare quotes posted by online travel agents have
(a) 0.8
0.7
0.6
0.5
Share
0.4
0.3
0.2
0.1
0
0
25
00
1+
1–
–5
–7
15
20
30
3–
–1
30
26
51
1–
1–
1–
76
10
15
20
Differences ($): 2008–2009 (blue) and 2006 (red) reflect nonstop travel agent fares for 50
domestic airport pairs at date t (e.g., if Expedia's lowest price on the route is $203 vs. $200
(Travelocity or Orbitz) then the difference is $3).
(b) 0.9
0.8
0.7
0.6
Share
0.5
0.4
0.3
0.2
0.1
0
0
25
00
1+
1–
–5
–7
15
20
30
3–
–1
30
26
51
1–
1–
1–
76
10
15
20
domestic airport pairs at date t (e.g., if Travelocity's lowest price on the route is $203 vs.
$200 (Expedia or Orbitz) then the difference is $3).
(c) 0.8
0.7
0.6
0.5
Share
0.4
0.3
0.2
0.1
0
0
25
00
1+
1–
–5
–7
15
20
30
3–
–1
30
26
51
1–
1–
1–
76
10
15
20
domestic airport pairs at date t (e.g., if Orbitz's lowest price on the route is $203 vs. $200
(Travelocity or Expedia) then the difference is $3).
Fig. 3. (a) Comparing Difference between Expedia’s Best Fare and “Lowest”
Nonstop Fare Available on Route 2006 versus 20082009. (b) Comparing
Difference between Travelocity’s Best Fare and “Lowest” Nonstop Fare Available
on Route 2006 versus 20082009. (c) Comparing Difference between Orbitz’s Best
Fare and “Lowest” Nonstop Fare Available on Route 2006 versus 20082009.
Table 4. Comparison of Lowest Price Quotes by Online Travel Agent and Airline for 50 US Domestic
Buyer Subsidies in Two-Sided Markets

Routes.
Travel Agent Airline OctoberNovember 2006 December 2008January 2009 Price Change
Expedia American 509.53 218.77 0.429 606.71 362.52 0.598 Increase

Orbitz American 540.68 249.34 0.461 611.49 349.27 0.571 Increase
Travelocity American 508.57 194.34 0.382 531.08 251.38 0.473 Increase
Expedia Alaska 552.94 223.68 0.405 643.13 326.78 0.508 Increase
Orbitz Alaska 690.60 299.89 0.434 620.19 259.47 0.418 Decrease
Travelocity Alaska 665.17 306.33 0.461 586.80 254.38 0.433 Decrease
Expedia Continental 522.71 214.07 0.410 630.94 315.83 0.501 Increase
Orbitz Continental 536.30 226.63 0.423 824.42 195.42 0.237 Increase
Travelocity Continental 664.59 74.79 0.113 640.90 324.19 0.506 Decrease
Expedia Delta 599.05 236.25 0.394 545.37 255.84 0.469 Decrease
Orbitz Delta 646.08 270.99 0.419 542.68 254.26 0.469 Decrease
Travelocity Delta 596.76 194.09 0.325 505.49 218.49 0.432 Decrease
Expedia Frontier 426.74 145.72 0.341 411.05 106.94 0.260 Decrease
Orbitz Frontier 448.98 138.12 0.308 432.89 113.68 0.263 Decrease
Travelocity Frontier 405.50 135.06 0.333 395.02 99.55 0.252 Decrease
Expedia AirTran N/A 459.57 135.63 0.295
Orbitz AirTran 507.77 116.95 0.230 468.74 141.67 0.302 Decrease
Travelocity AirTran 477.46 109.93 0.230 440.63 135.85 0.308 Decrease
Expedia Hawaiian 89.77 9.02 0.100 140.11 15.25 0.109 Increase
Orbitz Hawaiian 94.48 2.18 0.023 138.39 16.85 0.122 Increase
Travelocity Hawaiian 88.64 9.62 0.108 139.32 15.18 0.109 Increase
Expedia Island 118.00 0.00 0.000 154.52 10.46 0.068 Increase
Orbitz Island 118.62 2.84 0.024 150.59 10.36 0.069 Increase
Travelocity Island 118.46 8.09 0.068 158.89 10.93 0.069 Increase
Orbitz Spirit 455.00 133.54 0.293 413.13 53.88 0.130 Decrease
357
Travelocity Spirit 491.21 152.29 0.310 415.50 55.10 0.133 Decrease
358
Travel Agent Airline OctoberNovember 2006 December 2008January 2009 Price Change
Expedia Northwest 704.43 397.90 0.565 745.55 285.36 0.383 Increase

Orbitz Northwest 693.85 416.25 0.600 697.17 302.85 0.434 Increase
Travelocity Northwest 502.67 241.09 0.480 601.68 243.29 0.404 Increase
Expedia United 485.24 189.64 0.391 476.49 226.37 0.475 Decrease
VOLODYMYR BILOTKACH AND NICHOLAS G. RUPP

Orbitz United 548.76 267.02 0.487 518.52 274.69 0.530 Decrease
Travelocity United 512.31 224.28 0.438 493.33 260.18 0.527 Decrease
Expedia US Airways 1050.78 736.12 0.701 589.51 250.97 0.426 Decrease
Orbitz US Airways 1126.10 637.40 0.566 668.78 487.67 0.729 Decrease
Travelocity US Airways 802.29 383.21 0.478 562.93 402.47 0.715 Decrease
Expedia Mesa 86.92 8.64 0.099 142.82 19.95 0.140 Increase
Orbitz Mesa 89.24 8.73 0.098 147.86 18.03 0.122 Increase
Travelocity Mesa 87.14 8.90 0.102 150.14 18.63 0.124 Increase
Notes: Some carriers operated in 2006 but not in 2008 (e.g., Aloha and America West). Likewise, others operated in 2008 but not in 2006
(e.g., JetBlue and Virgin American). These carriers are not reported above.
The online travel agency which has the lowest average price appears in bold.
changed between the two sample periods (2006 vs. 20082009). We con-
duct the following OLS pricing specifications for our sample of last-minute
fare quotes:
Pij = X 0 β1 þ 08τ1 þ AGENTi γ 1 þ AGENTi 08δ1 þ AIRLINEj θ1

ð1Þ
þ AIRLINEj 08φ1 þ AGENTi AIRLINEj ζ 1 þ ɛij
Pij = X 0 β2 þ 08τ1 þ AGENTi γ 2 þ AGENTi 08δ2 þ AIRLINEj θ2

þ AIRLINEj 08φ2 þ AGENTi AIRLINEj ζ 2 þ AGENTi ð2Þ
AIRLINEj 08η þ ɛij
where Pij denotes the lowest price quoted by online travel agent i for a
round-trip coach-class ticket on airline j which provides nonstop service in
the airport-pair market; X0 is a vector of demand shifters, consisting of the
variables conventionally employed in the literature. We include controls for
income per capita and MSA population for the airport-pair markets, since
we expect demand to rise with both income and population. We also
include two tourism measures the absolute temperature difference
between the airport-pair along with an indicator variable for Florida/Las
Vegas destinations. Airports with greater absolute temperature differences
may indicate a higher proportion of leisure travelers (Brueckner, Dyer, &
Spiller, 1992). To account for cost conditions on the route we include the
scheduled nonstop travel time.12 We expect fares to rise with travel times.
To control for market conditions, we include the traditional measure of
concentration: the airport-pair market Herfindahl index. A recent study by
Hernandez and Wiggins (2014) finds that HHI has a modest impact on air-
line fares. Finally, we include airport-specific dummies for both origin and
destination airports to control for potential heterogeneities. AGENT repre-
sents the three online travel agents (Travelocity, Orbitz, and Expedia); 08 is
an indicator variable for the more recent 20082009 sample period;
AIRLINE is an indicator for each carrier that offers nonstop service in the
market.
The idea behind the DID estimator involves controlling for the agent-
specific, airline-specific, and agentairline specific time-invariant effect,
along with the time-varying effect. Airline- and agent-specific effects will
control for the respective players’ strategies. Airline08 and agent08
interactions will account for changes in the respective players’ strategies over
time. Airlineagent interactions will control for the effects of airlineagent

contracts. The difference between specifications (1) and (2) is the presence
of the airlineagent08 interactions in the latter. These interactions are
used to account for changes in the respective contracts, which have
remained unobserved to us.13 The 08 indicator variable captures the “after”
time effect, which is assumed the same for all agents. The AGENT*08
interaction term is the key variable since it captures the effect of a change
in the specific agent’s strategy that has not previously been captured by the
time trend and agent effect.
The DID estimator allows us to determine how individual travel agents
have changed their airfare presentation strategies between the 2006 and
20082009 sample periods, holding everything else constant. While the
focus of this chapter is on the competitive effects from the change in the
low-price guarantee policy implemented by Orbitz (which we interpret as a
subsidy to passengers by a platform on the two-sided market); we note that
the DID estimator measures the effect of all relevant developments in the
travel agents’ strategies over this two-year period. In particular, it is possi-
ble that the other online travel agents (Travelocity and Expedia) changed
their behavior in response to Orbitz’s new MFC policy. While neither
Travelocity’s nor Expedia’s price guarantee program changed, these online
agents may have changed their pricing behavior (e.g., lowering prices) in an
effort to compete with Orbitz. Such competitive responses would provide a
bias against finding procompetitive effects from Orbitz’s new MFC policy.
For our estimator to be valid, therefore, we need to be certain that the
MFC price assurance policy of Orbitz was the only notable agent-specific
development in the industry during our two-year period of interest. Indeed,
the online travel agents’ business models have been established for over half
a decade, and the three agents initially offered similar price matching
MOR policies; the market has a rather established structure, with the main
players’ market shares being rather symmetric and stable. One notable
industry development occurred in March 2009 (after our 20082009 sample
period ends), as the big three online travel agents began waiving booking
fees on most flights.14 This change constitutes a subsidy to travelers.
However, as such a subsidy has been implemented by all major online travel
agents nearly simultaneously; we do not expect effects of this strategic devel-
opment to bias our estimates.
A priori, hypotheses consistent with our story would be δ1 < 0 and δ2 < 0
for Orbitz*08 interaction variable. Rejection of these hypotheses can be
interpreted in two ways: (1) either Orbitz’s inability to convert its subsidy
to travelers into better access to the airlines’ seats and fares inventory or
(2) competitors renegotiating their contracts with the airlines, giving the
carriers more favorable terms in return for better access to content. This
possibility of ambiguous interpretation underscores the need to control for
possible unobserved changes to the airlineagent contracts, which is
the purpose of the triple-interaction AirlineAgent08 terms in specifica-
tion (2).
Variable Definitions
Below are the detailed definitions for each of the control variables used in
our estimations:
• Log (travel time) is the natural logarithm of the minimum nonstop travel
time between airports, and serves as a proxy for distance.
• Income per capita is the geometric average of end points’ income per
capita, at the metropolitan statistical area (MSA) level.
• Population is the geometric average of end points’ population, at the
MSA level.
• Temperature difference between the trip’s origin and destination (this
variable serves as a measure of the route’s attractiveness for vacation
travelers).
• Florida/Las Vegas destination is an indicator variable that equals 1 for
travel to Florida and Las Vegas (conventional vacation destinations).
• Airport-pair market Herfindahl index.
In addition to the above listed variables, all regressions include airline,
airlineagent, origination, and destination airport indicator variables.
Coefficients for these variables are not reported to save space.
DATA ANALYSIS
DID Estimates
We conduct our data analysis at two “levels.” First, we present DID esti-
mates for the entire sample of 10,026 observations; second, we repeat the
analysis for the subsample of lowest fare quotes for each agent on a specific
route on a given day of data collection (e.g., lowest fare quotes offered by
Expedia, Travelocity, and Orbitz on a given day of data collection by
market) this subsample includes a single data point for each possible
agentroutedate combination, and results in 4,540 observations.15 The
purpose of this second level of analysis is to determine whether the adop-
tion of a low-price assurance policy has affected the best airfares available
via a given online travel agent.
Table 5 presents DID regression estimates of specifications (1) and (2)
for the entire sample. This table provides results from four different regres-
sions. Odd-numbered regressions correspond to specification (1), and esti-
mation results for specification (2) are presented in even-numbered
regressions. Note that F-tests overwhelmingly reject the joint insignificance
of the AirlineAgent08 interaction variables, both as the broad group,
and at the individual agent level.16 Otherwise, each regression includes the
same set of control variables with different combinations of travel agent
and “after” sample interaction dummy variables. Specifically, the baseline
(omitted) agent is changed from Orbitz in regressions [1] and [2] to Expedia
in regressions [3] and [4].
Table 6 presents results from regressions on a subsample that only
includes observations for the lowest available fares offered by each travel
agent on a specific route and date. Just like in Table 5, odd-numbered
regressions correspond to specification (1), while estimation results for spe-
cification (2) are presented in even-numbered regressions. Once again, F-
tests reject the joint insignificance of the AirlineAgent08 interaction
variables, both as the broad group, and at the individual agent level.
Before we begin examining the estimation results in more detail, we note
that all specifications control for possible unobserved changes in airline-
agent contracts across the two time periods. This specification has a dra-
matic impact on the key variables of interest Agent*08 interaction terms,
while the coefficients for the remaining control variables remain stable.
Regressions [1] and [3] in Table 5 suggest that between the two time peri-
ods (and presumably following the change in Orbitz’s low-price guarantee
policy) Expedia’s lowest offered fares increased by 5.8% compared to
Orbitz’s, accounting for all of the above-mentioned trends and time-
invariant effects. Compared to Expedia, Travelocity, and Orbitz decreased
their fare quotes by equivalent amounts. In sum, we find that Orbitz’s fares
have fallen more between the two sample periods compared to one of its
competitors’ (Expedia). Orbitz fare quotes, however, have fallen by a simi-
lar proportion as its other rival (Travelocity). The latter can be interpreted
as Travelocity’s competitive reaction to Orbitz’s subsidy to travelers, which
was however not matched by Expedia.
Table 5. DID Estimates of Online US Travel Agent’s Lowest Airfare by

Carrier on 50 Domestic Routes in OctoberNovember 2006 and December
2008January 2009.
[1] [2] [3] [4]
Online travel agents

Travelocity 0.0834 0.0544 0.1815** 0.1469*
(0.0690) (0.0689)** (0.0610) (0.0628)
Travelocity*08 0.0114 −0.0123 −0.0463** 0.1620
(0.0163) (0.0914) (0.0168) (0.1652)
Expedia −0.0981 −0.0926
(0.0888) (0.0896)
Expedia*08 0.0577** −0.1743
(0.0183) (0.1652)
Orbitz 0.0981 0.0926
(0.0888) (0.0896)
Orbitz*08 −0.0577** 0.1743
(0.0183) (0.1652)
Control variables
08 −0.1176** −0.0404 −0.0599** −0.2147
(0.0127) (0.0649) (0.0135) (0.1519)
Log (travel time) 0.6748** 0.6760** 0.6748** 0.6760**
(0.0169) (0.0166) (0.0169) (0.0166)
Income per capita −0.0001** −0.0001** −0.0001** −0.0001**
(0.00001) (0.00001) (0.00001) (0.00001)
Population geometric average 4.24E-08** 4.20E-08** 4.24E-08** 4.20E-08**
(6.11E-09) (5.94E-09) (6.11E-09) (5.94E-09)
Temperature difference −0.0250** −0.0171** −0.0250** −0.0171**
(0.0043) (0.0044) (0.0043) (0.0044)
Florida/Las Vegas destination 1.2867** 1.0979** 1.2867** 1.0979**
(0.1046) (0.1062) (0.1046) (0.1062)
HHI airport market 0.4196** 0.3345** 0.4196** 0.3345**
(0.0397) (0.0392) (0.0397) (0.0392)
Agent*airline*08 dummies? No Yes No Yes
R-squared 0.656 0.678 0.656 0.678
Observations 10,026 10,026 10,026 10,026
Notes: White-robust standard errors appear in parentheses. Regressions also include controls
for both “Airline Fixed Effects” and “Airport Fixed Effects” due to the inclusion of indicator
variables for each airline and each origination and destination airports. In addition, all above
estimates include agent*airline interaction terms and a constant. * and ** indicate statistical
significance at 5% and 1%, respectively. The 50 domestic routes were randomly selected from
the Top 100 domestic routes based on 2005 passenger traffic. The selected airport-pair routes
appear in the Appendix. We include coach-class round-trip airfares whenever at least two tra-
vel agents provide fare quotes for the airport-pair market.
Table 6. DID Estimates of Lowest Available Price on 50 Domestic

Routes by on-US Travel Agents in OctoberNovember 2006 and
December 2008January 2009 Line.
[5] [6] [7] [8]
Online travel agents

Travelocity 0.0588 0.0201 −0.0467 −0.0847
(0.0523) (0.0475) (0.0549) (0.0532)
Travelocity*08 −0.0498* −0.4933** −0.0354 −0.6772**
(0.0233) (0.1500) (0.0227) (0.1534)
Expedia 0.1055 0.1048
(0.0689) (0.0666)
Expedia*08 −0.0144 0.1840**
(0.0249) (0.0559)
Orbitz −0.1055 −0.1048
(0.0689) (0.0666)
Orbitz*08 0.0144 −0.6565**
(0.0249) (0.1370)
Control variables
08 −0.0689** 0.1612** -0.0833** 0.3452**
(0.0186) (0.0335) (0.0176) (0.0445)
Log (travel time) 0.5795** 0.6009** 0.5795** 0.6009**
(0.0265) (0.0262) (0.0265) (0.0262)
Income per capita −0.0001** −0.0001** −0.0001** −0.0001**
(0.00001) (0.00001) (0.00001) (0.00001)
Population geometric average 6.21E-08** 6.14E-08** 6.21E-08** 6.14E-08**
(8.71E-09) (8.71E-09) (8.71E-09) (8.71E-09)
Temperature difference −0.0141* −0.0139* −0.0141* −0.0139*
(0.0058) (0.0059) (0.0058) (0.0059)
Florida/Las Vegas destination 0.7197** 0.7823** 0.7197** 0.7823**
(0.1342) (0.1363) (0.1342) (0.1363)
HHI airport market 0.4532** 0.3678** 0.4532** 0.3678**
(0.0558) (0.0552) (0.0558) (0.0552)
Agent*airline*08 dummies? No Yes No Yes
R-squared 0.697 0.720 0.697 0.720
Observations 4,540 4,540 4,540 4,540
Notes: White-robust standard errors appear in parentheses. Regressions also include controls
for both “Airline Fixed Effects” and “Airport Fixed Effects” due to the inclusion of indicator
variables for each airline and each origination and destination airports. In addition, all above
estimates include agent*airline interaction terms and a constant. * and ** indicate statistical
significance at 5% and 1%, respectively. The 50 domestic routes were randomly selected from
the Top 100 domestic routes based on 2005 passenger traffic. The selected airport-pair routes
appear in the Appendix. We include coach-class round-trip airfares whenever at least two tra-
vel agents provide fare quotes for the airport-pair market.
All of the effects that we have attributed to the change in Orbitz’s low-
price guarantee strategy, reported in regressions [1] and [3], however, disap-
pear completely (see regressions [2] and [4]) once we control for the possible
changes in the airlineagent contracts over time. This is done by including
the three-way AirlineAgent08 interaction variables into regressions.
Since these interaction variables are statistically significant as a group, as
evaluated by the regular F-test, this effectively confirms that any effects of
MFC low-price guarantees reported in regressions [1] and [3] are driven by
the changes in the airlineagent relationships. Taken together, these find-
ings are generally consistent with the two-sided market story: a subsidy to
travelers makes Orbitz a more attractive platform for airlines, and allows
travel agents to negotiate for more complete access to the airlines’ seat and
price inventory (content); Travelocity appears to respond by changing its
contracts with the airlines (we find evidence consistent with this); while
Expedia chooses to perhaps leave its contracts with the carriers unchanged.
Table 6 presents results from regressions on a subsample that only
includes observations for the lowest available fares offered by each travel
agent on a specific route and date. Regressions [5] and [7] present weak evi-
dence of the effect of the change in Orbitz’s low-price guarantee policy on
the lowest fares at the market level. Once we control for the possible
changes in the airlineagent contracts over time; however, the generally
procompetitive effects at the market-travel agent level become very evident
(see regressions [6] and [8]). Specifically, even though Expedia’s lowest fare
quotes at the market level have between the two sample periods been trend-
ing upward as compared to Orbitz’s; one is more likely to find lower prices
on both Orbitz and Travelocity after the former has introduced a new low-
price guarantee policy. Moreover, since effects measured by the
AGENT*08 interaction are above and beyond the time, airline, agent,
airlineagent, and market effects, and changes in the airlineagent con-
tracts have also been controlled for, the above-mentioned interaction vari-
able measures the effect of the only significant change in the ticket
distribution industry between our two samples the introduction of the
MFC-type low-price guarantee by Orbitz.
Summarizing our findings, we can say that there is evidence of procom-
petitive effects of the change in the low-price guarantee policy. The level at
which such effect is found, however, crucially depends on whether we con-
trol for the potential changes in the airlineagent contracts over time.
Specifically, when such controls are introduced, we no longer find any
effects at the airlineagent level. This means that even if Orbitz or
Travelocity might be presenting lower fare quotes by an individual carrier,
beyond what would be expected by the general trends, market and airline-
specific effects; this change is not due to the change in Orbitz’s strategy.
Rather, concurrent changes in the contracts that individual carriers have
with the carriers are to blame. At the same time, we find that the change in
the low-price guarantee policy has affected the lowest fare quote a custo-
mer can obtain at the agent-market level, above and beyond time trends,
agent, airline, market, airlinemarket, and airlineagent effects and
trends. Such effects are generally procompetitive.
Online Travel Agents and Nonstop Ticketing Options
In this subsection, we further explore whether travel agents have changed

their strategy in deciding which last-minute nonstop fares to offer to consu-
mers between 2006 and 20082009. One possibility is that Orbitz might
have be reluctant to present higher fare quotes, fearing that the airline
might drop its price later on, and hence Orbitz would have to issue refunds
to traveler who (driven by, for instance, the flight’s nonprice characteristics
or brand loyalty) purchased the higher fare ticket before the price drop.
This story is not as implausible as a skeptical reader might think. We rea-
lize that our data covers a sample of “last-minute” fare quotes; so that a
reader can rightfully question both how likely an airline will drop the price
and a whether a different customer can purchase a ticket with the same itin-
erary within a couple of days remaining between the date we observe the
fare quote and departure date. At the same time, the literature on “flight-
offer” curves (with researchers tracking fare quotes for a given flight over
time) has documented numerous instances of last-minute price drops. In
sum, if Orbitz is suppressing the number of higher fare quotes, then we
should see a reduction in the number of nonstop ticketing options for
Orbitz in the 20082009 sample compared to its peers.
Table 7 displays the number of nonstop ticketing options (i.e., the total
number of nonstop fare quote airline combinations in the data we col-
lected) by online travel agency in 2006 and 20082009 when at least two
travel agents offer fares on the route. In 2006, Travelocity provided the
most nonstop ticketing options in our sample (1,910), whereas Orbitz and
Expedia provided 1,474 and 1,409 such quotes, respectively. Since on one
of our nine data collection days in 2006 we were unable to obtain fares at
Orbitz.com, a better comparison of ticketing options may be reflected by
restricting the sample to include routes in which all three carriers report
fares. In these situations, Travelocity still provides the traveler with the
Table 7. Average Number of Unique Nonstop Carriers Flight Options

Presented by Online Travel Agent.
Agent 2006 20082009
Total Unique Average Options Total Unique Average Options

Carrier Options per route Carrier Options per route
Expedia 1409 1.74 1635 2.09

Orbitz 1474 1.82 1800 2.30
Travelocity 1910 2.36 1825 2.33
Total 4793 1.97 5260 2.24
Notes: These observations reflect situations when at least two online travel agents reported
route fares. If we require that all three travel agent report route fares, then the 2006 sample of
total unique carrier options would include: Expedia (1071), Orbitz (1425), and Travelocity
(1445); while only minor changes occur in the 2008 sample: Expedia (1612), Orbitz (1775), and
Travelocity (1789).
most carrier options in our sample routes (1,445); however, Orbitz is now a
close second with 1,425 flights, followed by a distance third: Expedia with
1,071 offerings. When all agents report fares, Travelocity and Orbitz pre-
sent 33% more nonstop options (at the airlineroute level) than Expedia
in 2006.
In 20082009, Travelocity and Orbitz are providing online ticket pur-
chasers with the most nonstop carrier flight options in our sample of
airport-pair markets: 1,825 and 1,800 observations, respectively. While
Expedia has narrowed the gap in the number of nonstop carrier options,
nonetheless, it still remains in last place with 1,635 nonstop flights in our
sample; or 10% fewer nonstop flight offerings than its peers. This differ-
ence is more meaningful if we adjust the number of observations by the
number of routes; we find that Travelocity provides an average of 2.33 car-
rier ticketing options per route followed by Orbitz (average of 2.3 different
carriers) and Expedia (2.09 carriers) in 20082009. Clearly, we find no evi-
dence that Orbitz is suppressing the number of ticketing options for its con-
sumers. Hence the above findings are consistent with the hypothesis that
Orbitz’s MFC low-price guarantees are procompetitive.
CONCLUSION
Our investigation of the online travel agency industry is cast within the
more general framework of an empirical study of buyer subsidies in two-
sided markets. Indeed, online travel agents can be considered platforms

that work to facilitate trade between buyers (travelers) and sellers (air-
lines) of travel services. Unlike previous empirical studies of two-sided
markets, we are able to cleanly identify an event, which can be considered
a buyer subsidy; and evaluate the competitive effects of such a subsidy
via a simple DID estimation strategy. Overall, we find evidence to suggest
that the buyer subsidy implemented by one of the major online travel
agents has led to contract changes between the agents and the airlines;
and the outcome of this change has been an increased availability of the
airlines’ seat and price inventory for two of the three major online travel
agents.
There are some limitations with these data. Foremost, we only observe
airline ticket prices and not quantity. The total booking revenue modestly
increased for Orbitz during our sample periods, but it’s not clear whether
the low-price guarantee led Orbitz to sell more tickets. In addition, the reli-
ance on last-minute fares is not ideal in studying the adoption of a new
price guarantee offering since there is little time for future price drops
which will trigger automatic refunds to Orbitz consumers. We rely on last-
minute fares since these data were collected prior to Orbitz’s policy change.
Nonetheless, if fare differences are present in last-minute fares, then they
are likely present in more advance notice as well.
This chapter also makes a contribution to the literature on low-price
guarantees. We find evidence in the online travel industry consistent with
the procompetitive effects after Orbitz switches its low-price guarantee pol-
icy from an MOR clause to an “MFC” clause in 2008. This change by
Orbitz in their low-price guarantee policy provides a natural experiment for
this study. We observe the ticket prices of Orbitz and its two largest compe-
titors (Expedia and Travelocity) both before and after MFC clause is
adopted; absence of other significant changes in the industry along with
rather rigid airlineagent relationships enable relatively simple DID identi-
fication strategy.
To successfully implement our DID identification strategy, we must con-
trol for time-invariant airline, agent, and airlineagent effects, as well as
for time trend and the changes in airlines’ behavior over time. Online
agents typically enter into separate confidential agreements with individual
airlines. Controlling for possible changes in individual airlineagent agree-
ments, our results suggest that MFC price policies provide procompetitive
effects above and beyond pricing trends, airline, agent, market, airline-
market, and airlineagent effects for the subset of lowest available fare
quotes on the route. Our results also suggest that while a traveler exhibiting
strong brand loyalty may not find much in terms of benefits from the
change in Orbitz’s low-price guarantee strategy (we do not find effect at the
airlineagent level); there are benefits for a customer shopping for the low-
est fare on the market. Finally, we find no evidence to suggest that
the MFC price assurance program initiated by Orbitz has suppressed the
number of nonstop fare offerings to its customers. In sum, our findings
suggest that the buyer subsidy implemented by Orbitz has benefited the tra-
veling public.
NOTES
1. Initial CRS regulation, adopted in 1984, stipulated that an airline owning or
marketing a CRS must participate in competing systems. This rule was scrapped in
2004.
2. Online Travel Gets Personal, Forbes.com, accessed 02/17/2006.
3. Travel agency market share data are from “Computer Reservations System
Regulations: Final Rule,” 14 CFR Part 255.
4. When participation is not real time, the agent first shows you the fare quote
it encountered the last time it queried the airlines’ inventory. If you want to follow
through on that quote, the agent queries the inventory again.
5. This supposition is not as unrealistic as it seems suppose lower fares are
offered well in advance of the departure date, when there are fewer potential trave-
lers visiting the agent; while higher fares are offered shortly before the departure,
when the number of potential travelers is higher.
6. Priceline is famous for its “name your own price” pricing policy, whereby the
customer places a bid for an air ticket, hotel, or car rental. Identity of the provider
of the service is not revealed until after the customer has committed to the
purchase.
7. The top 100 routes are determined from the 2006 T-100 Segment dataset.
8. While these last-minute fare data may not be best suited to examine buyer
subsidies since there is a short window in which consumers can receive refunds from
future price reductions, nonetheless, if we can detect differences in last-minute fares
following Orbitz’s new MFC policy, then these differences will also likely be present
for more advance notice fares as well.
9. One can legitimately claim that our hypothetical customer may still prefer
some flexibility, especially with respect to being able to change the time of the return
flight. Conditional on the customer traveling, however, the difference between the
refundable and the nonrefundable ticket is similar to that between a lottery and a
certain outcome (see also Escobari & Jindapon, 2014): once (and even before) the
trip has begun, a customer on a nonrefundable ticket can change her travel plans
for a fee.
10. In an effort to minimize the loss of observations, we included airport-pair
market fares whenever two or more travel agencies provided nonstop fare quotes.
11. For example, for a given airport-pair if Travelocity’s lowest price at date t on
the route is $200, Expedia’s is $203, and Orbitz’s is $205, this route would have a
maximum difference between travel agents of $5.
12. Since the contractual terms between the airline and travel agency are confi-
dential, we do not know if the terms of their agreement changed between 2006 and
2008/2009. Hence, we assume that the costs for each airline did not change over this
two-year period, and airline indicator variables capture any cost differences across
carriers.
13. Since the airlinetravel agent contracts are confidential agreements between
these two parties, we are unable to observe whether changes occurred to these con-
tracts during our sample periods, hence we include airlineagent08 interaction
terms for every airline and travel agent.
14. Expedia was the first major travel agent to waive flight booking fees on
March 11, 2009. Thirty days later, every other major online travel agent
(Travelocity, Orbitz, and Priceline) was waiving booking fees.
15. When several airlines offered identical lowest fares; we simply used the corre-
sponding airline indicators as independent variables for such observations.
16. For example, for regression [2] the F-test = 50.54 and 49.26 (both p-values <
0.0001) for the joint insignificance of all airlineagent08 and airlineExpedia08
interaction terms, respectively.
REFERENCES
Arbatskaya, M., Hviid, M., & Shaffer, G. (2004). On the incidence and variety of low-price
guarantees. Journal of Law and Economics, 48(8), 307332.
Arbatskaya, M., Hviid, M., & Shaffer, G. (2006). On the use of low-price guarantees to dis-
courage price cutting. International Journal of Industrial Organization, 24, 11391156.
Argentesi, E., & Filistrucchi, L. (2007). Estimating market power in a two-sided market: The
case of newspapers. Journal of Applied Econometrics, 22, 12471266.
Bilotkach, V., & Pejcinovska, M. (2012). Distribution of airline tickets: A tale of two market
structures. In J. Peoples (Ed.), Pricing behavior and non-price characteristics in the air-
line industry (Vol. 3, pp. 107138). Advances in Airline Economics. Bingley, UK:
Emerald Group Publishing Limited.
Bilotkach, V., Rupp, N., & Pai, V. (2013). Value of a platform to a seller: Case of American
Airlines and online travel agencies. Available at SSRN: http://ssrn.com/abstract=
2321767. Retrieved on August 6, 2013.
Brueckner, J. K., Dyer, N. J., & Spiller, P. T. (1992). Fare determination in airline hub-and-
spoke networks. RAND Journal of Economics, 23(3), 309333.
Chandra, A., & Collard-Wexler, A. (2009). Mergers in two-sided markets: An application to
the Canadian newspaper industry. Journal of Economics and Management Strategy, 18,
10451070.
Chatterjee, S., & Roy, S. (1997). Consumer reactions to price-matching signals. Advances in
Consumer Research, 24, 400404.
Chen, J. (2006). Differences in average prices on the internet: Evidence from the online market
for air travel. Economic Inquiry, 44, 656670.
Chen, J., & Liu, Q. (2011). The effect of most favored customer clauses on prices. Journal of
Industrial Economics, 59, 343371.
Clemons, E. C., Hann, I.-H., & Hitt, L. M. (2002). Price dispersion and differentiation in
online travel: An empirical investigation. Management Science, 48, 534549.
Cooper, T. (1986). Most-favored-consumer pricing and tacit collusion. RAND Journal of
Crocker, K., & Lyon, T. (1994). What do facilitating practices facilitate? An empirical investi-
gation of most-favored-nation clauses in natural gas contracts. Journal of Law and
Deck, C., & Wilson, B. (2003). Automated pricing rules in electronic posted offer markets.
Economic Inquiry, 41, 208223.
Derdenger, T. (2014, forthcoming). Technological tying and the intensity of competition:
An empirical analysis of the video game industry. Quantitative Marketing and
Economics.
Dugar, S., & Sorensen, T. (2006). Hassle costs, price-matching guarantees and price competi-
tion: An experiment. Review of Industrial Organization, 28, 359378.
Escobari, D., & Jindapon, P. (2014). Price discrimination through refund contracts in airlines.
International Journal of Industrial Organization, 34(3), 18.
Hernandez, M. A., & Wiggins, S. N. (2014). Nonlinear pricing strategies and competitive con-
ditions in the airline industry. Economic Inquiry, 52(2), 539561.
Holt, C., & Scheffman, D. (1987). Facilitating practices: The effects of advance notice and
best-price policies. RAND Journal of Economics, 18, 187197.
Hviid, M., & Shaffer, G. (1999). Hassle costs: The achilles’ heel of price-matching guarantees.
Journal of Economics and Management Strategy, 8, 489521.
Kaiser, U., & Wright, J. (2006). Price structure in two-sided markets: Evidence from the maga-
zine industry. International Journal of Industrial Organization, 24, 128.
Lurie, N., & Srivastava, J. (2001). A consumer perspective on price-matching refund policies:
Effect on price perceptions and search behavior. Journal of Consumer Research, 28,
296307.
Lurie, N., & Srivastava, J. (2005). Price-matching guarantees and consumer evaluation of price
information. Journal of Consumer Psychology, 15, 149158.
Manez, J. (2006). Unbeatable value low-price guarantee: Collusive mechanism or advertising
strategy? Journal of Economics and Management Strategy, 15, 143166.
Moorthy, S., & Winter, R. (2006). Price-matching guarantees. RAND Journal of Economics,
37, 449465.
Neilson, W., & Winter, H. (1993). Bilateral most-favored-customer pricing and collusion.
RAND Journal of Economics, 24, 147155.
Puller, S., Sengupta, A., & Wiggins, S. (2009). Testing theories of price dispersion and scarcity
pricing in the airline industry. NBER Working paper No. 15555.
Rochet, J.-C., & Tirole, J. (2003). Platform competition in two-sided markets. Journal of the
European Economic Association, 1, 9901029.
Rochet, J.-C., & Tirole, J. (2006). Two-sided markets: A progress report. RAND Journal of
Rysman, M. (2007). An empirical analysis of payment card usage. Journal of Industrial
Economics, 55, 136.
Rysman, M. (2009). The economics of two-sided markets. Journal of Economic Perspectives,
23, 125143.
Salzman, A. (2013). This means war! Barron’s, April 13, 2013.
Schnitzer, M. (1994). Dynamic duopoly with best-price clauses. RAND Journal of Economics,
25, 186196.
Scott-Morton, F. (1997a). The interaction between a most-favored-customer clause and price
dispersion: An empirical examination of the medicaid rebate rules of 1990. Journal of
Economics and Management Strategy, 6, 151174.
Scott-Morton, F. (1997b). The strategic response by pharmaceutical firms to the medicaid
most-favored-customer rules. RAND Journal of Economics, 28, 269290.
APPENDIX
Table A.1. List of US Airport-Pair Markets.

Anchorage Seattle Dallas-Fort Worth Orlando
Atlanta Boston Dallas-Fort Worth Miami
Atlanta Washington (National) Dallas-Fort Worth Chicago (O’Hare)
Atlanta Denver Dallas-Fort Worth San Diego
Atlanta Dallas-Fort Worth Dallas-Fort Worth San Antonio
Atlanta Fort Lauderdale Dallas-Fort Worth Seattle
Atlanta Jacksonville Detroit Minneapolis-St. Paul
Atlanta Las Vegas Fort Lauderdale New York (JFK)
Atlanta Los Angeles Honolulu Lihue
Atlanta Orlando Honolulu Kahului
Atlanta Miami Houston (Intercontinental) Los Angeles
Atlanta Chicago (O’Hare) New York (JFK) Los Angeles
Atlanta West Palm Beach New York (JFK) Orlando
Atlanta Philadelphia New York (JFK) San Juan, Puerto Rico
Atlanta San Francisco Las Vegas Phoenix
Atlanta Salt Lake City Las Vegas Los Angeles
Baltimore Atlanta Los Angeles Seattle
Denver Dallas-Fort Worth Los Angeles San Francisco
Denver Los Angeles New York (LaGuardia) Chicago (O’Hare)
Denver Chicago (O’Hare) Miami San Juan, Puerto Rico
Denver San Francisco Oakland San Diego
Dallas-Fort Worth Las Vegas Chicago (O’Hare) San Francisco
Dallas-Fort Worth Los Angeles Tampa Atlanta
Dallas-Fort Worth New York (LaGuardia) Seattle Minneapolis-St. Paul
Notes: Markets selected for the study are 50 airport-pair markets randomly chosen from
among the top 100 US airport-pair markets by nonstop traffic in 2006. For each of the above
airport-pair markets, we collected round-trip airfares for both directions (e.g., Anchorage to
Seattle and Seattle to Anchorage).
Table A.2. Online US Travel Agency Price Guarantees.

A. Orbitz: How Orbitz Price AssuranceSM works?
Once you book on Orbitz, we start tracking to see if another Orbitz customer subsequently
books the same flight or hotel reservation on Orbitz at a lower price.
If that happens, we’ll issue a refund for the difference. Amounts range from $5 to $250 per
airline ticket or $5 to $500 per hotel reservation.
We’ll continue tracking until the day you leave. So each time the price drops and another
customer subsequently books your same itinerary for a lower amount, your refund amount
will increase.
Expect to receive your refund check approximately 68 weeks after your trip is complete.
We’ll mail it to the billing address for the credit card used to make the booking. You’ll have
90 days from the date the check is issued to cash it.
B. Travelocity: What is the Travelocity Low-Price Guarantee?
1. The Travelocity Low-Price Guarantee (“Low-Price Guarantee”) is available to travelers
who have booked travel on Travelocity.com (“Travelocity”) on or after August 25, 2006.
2. If you find a Qualifying Lower Rate within 24 hours of your booking, we will provide you
with the following per booking:
• One $50 Promo Code for a future “Good Buy” Hotel or Flight + Hotel vacation package
booking on Travelocity and
• A refund of the difference between the price you paid through Travelocity and the
Qualifying Lower Rate.
A Qualifying Lower Rate is a lower rate found on Travelocity or another US-based
web site that satisfies the requirements of these Terms and Conditions, as determined
by Travelocity in its sole discretion.
C. Expedia: Best Price Guarantee
1. Best Price Guarantee. In the unlikely event that you find a lower rate on Expedia.com or on
another US-based web site within 24 hours of booking with Expedia.com, we will credit or
refund to you the difference. In addition, we will give you a 50 dollar ($50.00) coupon good
on a future booking of an Expedia Special Rate hotel or air plus hotel package on Expedia.
com. This Best Price Guarantee is subject to the terms and conditions listed below.
2. Notifying Expedia of a Claim. You must contact us at 1-800-EXPEDIA within 24 hours
after your Expedia.com booking to make a claim under the Best Price Guarantee. The lower
rate must be available for booking at the time you contact us, as determined by our
customer service representatives.
3. Must be “Apples to Apples” Comparison. The Best Price Guarantee is available only for
exact itinerary matches, for example, specific carrier or provider (including class of
service) … applicable refund policy, and the exact same dates and times of travel or service
as booked through Expedia.com.
Source: A: http://www.orbitz.com/pagedef/content/legal/priceAssuranceTCs.jsp?tcs=false&
popupsDisabled=false (Accessed Orbitz.com on October 16, 2009); B: http://svc.travelocity.
com/info/info_popup/0,2766,TRAVELOCITY:EN|GUARANTEE_TERMS,00.html
(Accessed Travelocity.com on October 16, 2009); C: http://www.expedia.com/daily/highlights/
best-rate-guarantee/ default.asp? mcicid= hp.why bpg#terms (Accessed Expedia.com October
16, 2009).

Economics of International Airline Transport PDF

Uploaded by

Document Information

Original Title

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

Economics of International Airline Transport PDF

Uploaded by

Copyright:

Available Formats

THE ECONOMICS OF

United Kingdom North America Japan

First edition 2014

Copyright r 2014 Emerald Group Publishing Limited

Reprints and permission service

No part of this book may be reproduced, stored in a retrieval system, transmitted in

British Library Cataloguing in Publication Data

Certificate Number 1985

LIST OF CONTRIBUTORS vii

EDITORIAL BOARD MEMBERS xi

CHAPTER 1 INTRODUCTION AND OVERVIEW

CHAPTER 2 AIR TRANSPORT LIBERALIZATION

CHAPTER 3 GOVERNMENT REGULATION OF

CHAPTER 4 SKIES WIDE SHUT AN ASSESSMENT

CHAPTER 5 INTERNATIONAL MERGERS AND

CHAPTER 6 AIRFARES AND COMPETITION ON

CHAPTER 7 THE CHOICE OF AIRPORT, AIRLINE,

CHAPTER 8 AIR CARGO SERVICES AND THE

CHAPTER 9 AN ASSESSMENT OF THE CAUSAL

CHAPTER 10 INTERNATIONAL AND NATIONAL

CHAPTER 11 PLANNING A COMPETITIVE

CHAPTER 12 AIRPORT AND AIRLINE

CHAPTER 13 BUYER SUBSIDIES IN TWO-SIDED

Stephen J. Appold Research Associate, Center for Air

Dan Mahoney Graduate Teaching Fellow, Department of

Janina D. Scheelhaase Head of Air Transport Economics,

Volodymyr Bilotkach Theodore E. Keeler

John Bitzan B. Starr McMullen

Jan K. Brueckner Steven Morrison

Kevin Cullinane Nicholas G. Rupp

Martin Dresner Ian Savage

David Gillen Wayne Talley

Timothy J. Hazledine Wesley W. Wilson

INTRODUCTION AND OVERVIEW

Globalization has placed a premium on the ability of countries to create

The Economics of International Airline Transport

Furthermore, growth in international air transportation of passengers for

LIBERALIZATION OF INTERNATIONAL AIR

The first set of chapters describes the details of international agreements

competition in the air transportation services market, increased flight fre-

restrictive regulatory environment for international air transportation ser-

AIR COMPANY PERFORMANCE AND BEHAVIOR IN A

Chapters 57 empirically examine company performance and behavior,

Consolidation of international air transportation services could raise

INTERNATIONAL AIR TRANSPORTATION AS A

on export flows from developing countries to the US and to countries in

regulations tend to be small, due in part to the airframe/engine combina-

THE ECONOMICS OF INTERNATIONAL AIRPORTS

Critical to airlines competing successfully in this era of increasingly liberal-

Economic theory suggests introduction of more liberal airline service agree-

1. Ocean liners are also major international transporters of high-volume high-

AIR TRANSPORT LIBERALIZATION

Xiaowen Fu and Tae Hoon Oum

The Economics of International Airline Transport

and computational network methods. These approaches should be selec-

Air transportation is becoming increasingly important to the world’s econ-

THE ECONOMIC EFFECTS OF AIR TRANSPORT

Under the 1944 Chicago Convention on international air transportation,

Although all of these regulations on domestic routes and international

The Effects of Major Liberalization Events

fares, flight frequencies or aircraft types. The number of Canadian cities

Dynamics of Liberalization: Increased Competition and

increased competition brought on by deregulation and liberalization, much