You may notice some images loading slow across the Open Collections website. Thank you for your patience as we rebuild the cache to make images load faster.

Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Determining factors of air passenger's choice of transfer airports in the Southeast Asia - North America… Choi, Jong Hae 2018

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata

Download

Media
24-ubc_2018_september_choi_jonghae.pdf [ 1.55MB ]
Metadata
JSON: 24-1.0371142.json
JSON-LD: 24-1.0371142-ld.json
RDF/XML (Pretty): 24-1.0371142-rdf.xml
RDF/JSON: 24-1.0371142-rdf.json
Turtle: 24-1.0371142-turtle.txt
N-Triples: 24-1.0371142-rdf-ntriples.txt
Original Record: 24-1.0371142-source.json
Full Text
24-1.0371142-fulltext.txt
Citation
24-1.0371142.ris

Full Text

   DETERMINING FACTORS OF AIR PASSENGER’S CHOICE OF TRANSFER AIRPORTS IN THE SOUTHEAST ASIA - NORTH AMERICA MARKET  by  Jong Hae Choi    A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  MASTER OF SCIENCE in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Business Administration in Transportation and Logistics)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  August 2018   © Jong Hae Choi, 2018   ii  The following individuals certify that they have read, and recommend to the Faculty of Graduate and Postdoctoral Studies for acceptance, a thesis/dissertation entitled:  Anming Zhang and Robin Lindsey  submitted by Jong Hae Choi  in partial fulfillment of the requirements for the degree of Master of Science In Business Administration  Examining Committee: Anming Zhang, Operations and Logistics Supervisor  Robin Lindsey, Operations and Logistics Supervisory Committee Member  Sanghoon Lee, Strategy and Business Economics Additional Examiner    Additional Supervisory Committee Members:  Supervisory Committee Member  Supervisory Committee Member      iii  Abstract This study investigates the factors affecting air passengers’ choice of transfer airport in the market between Southeast Asia (SE Asia) and North America (NA). It estimates demand parameters and sheds lights on the business strategies for an airport operator including identifying effective routes to better compete for transfer passengers and allocate resources. More specifically, a discrete choice model is applied to estimate air travelers’ utility from choosing transfer airports in this market.  Our data consists of 78 routes which represent the SE Asia - NA market. We estimate demand parameters by standard instrument-variables techniques to control for the endogeneity of airfare, while treating airport characteristics including minimum connection time (MCT), service, size, and detour degree as exogenous. Main empirical findings are: (1) Transfer passengers’ airport choice is determined by an airport’s characteristics including MCT and service quality, apart from those traditional factors such as airfares and travel time. (2) Based on estimated utility function parameters, we provide a quantitative evaluation of the transfer passengers’ willingness-to-pay for several transfer airport’s characteristics. (3) With our estimation results, a case study is conducted for Seoul-Incheon International Airport.  We identify the candidate route that has the greatest potential to increase the transfer passengers through the airport. This research contributes to a better understanding of air travelers’ choice of transfer airport in the SE Asia – NA market by utilizing aggregated market-level data. We identify the crucial factors that determine transfer passengers’ airport choice and estimate passengers’ willingness-to-pay for airport services and characteristics. The findings of this research provide valuable  iv  insights for an airport operator to allocate resources effectively so as to attract airlines and increase transfer passengers.                       v  Lay Summary This study investigates air travelers’ choice of transfer airport in the SE Asia – NA market which has had strong demand. We find that transfer passengers’ decisions are related to a transfer airport’s characteristics, besides well-known factors such as airfares and travelling time. The study therefore demonstrates that there exists a significant area in transfer passengers’ market where airport operators can play an important role. Based on the data and empirical estimation, we show how more efficient and responsive routes in the SE Asia - NA market can be identified from the perspective of a specific airport operator. In particular, we conduct a case study of Seoul-Incheon Airport (ICN) in South Korea. Based on the study, the route between Ho Chi Minh (SGN) and Los Angeles (LAX) is identified as the route that has the greatest potential to increase the transfer passengers and is thus recommended for marketing organization of ICN.     vi  Preface This dissertation is original, unpublished, independent work by the author, Jong Hae Choi.  vii  Table of Contents  Abstract ......................................................................................................................................... iii  Lay Summary .................................................................................................................................v  Preface ........................................................................................................................................... vi  Table of Contents ........................................................................................................................ vii  List of Tables ................................................................................................................................ ix  List of Figures .................................................................................................................................x Acknowledgements ...................................................................................................................... xi Dedication .................................................................................................................................... xii  Chapter 1: Introduction ................................................................................................................1  1.1 Research Questions ......................................................................................................... 3 1.2 Main Results ................................................................................................................... 5 1.3 Organization of the Thesis .............................................................................................. 7 Chapter 2: Literature Review .......................................................................................................9 2.1 Decision Factors for Transfer Passengers ....................................................................... 9 2.2 Discrete Choice Model ................................................................................................. 11 2.3 Endogeneity .................................................................................................................. 15 Chapter 3: Market Analysis, Data, Econometric Model and Hypotheses ..............................17 3.1 Market Analysis ............................................................................................................ 17 3.2 Data ............................................................................................................................... 19 3.2.1 Demand Data ........................................................................................................ 19 3.2.2 Airport Characteristic Data ................................................................................... 27  viii  3.2.3 Price Data .............................................................................................................. 29 3.2.4 Summary Statistics................................................................................................ 29 3.3 Econometric Model ....................................................................................................... 30 3.3.1 Utility Function ..................................................................................................... 30 3.2.2 Model of the Choice Probability ........................................................................... 31 3.4 Hypotheses .................................................................................................................... 32 Chapter 4: Methodology..............................................................................................................35  4.1 Regression Model ......................................................................................................... 35 4.2 Endogeneity and IVs ..................................................................................................... 37 4.3 Two-Stage Least Square (2SLS) Estimation ................................................................ 38 Chapter 5: Empirical Results .....................................................................................................40  5.1 Transfer Passenger’s Airport Choice ............................................................................ 42 5.2 Transfer Passenger’s Willingness-to-Pay ..................................................................... 43 Chapter 6: Effective Route - The Case Study of Incheon International Airport...................48 Chapter 7: Conclusion .................................................................................................................53  Bibliography .................................................................................................................................56 Appendices ....................................................................................................................................60 Appendix A ............................................................................................................................... 60 Appendix B ............................................................................................................................... 62  Appendix C ............................................................................................................................... 63 Appendix D ............................................................................................................................... 64   ix  List of Tables  Table 3.1 Top 10 Gateway in SE Asia - NA Market (2011-2016) ............................................... 18 Table 3.2 Airports in Southeast Asia ............................................................................................ 19  Table 3.3 Airports in North America ............................................................................................ 20  Table 3.4 Top 120 Routes in Transfer Passengers’ Market between SEA-NA ............................ 21 Table 3.5 An Example of BKK-LAX Route as of July 2016 ....................................................... 26 Table 3.6 Summary Statistics ....................................................................................................... 29 Table 5.1 2SLS Estimation Results .............................................................................................. 40 Table 5.2 Calculation Process ....................................................................................................... 43  Table 5.3 Transfer Passenger’s Willingness-to-Pay ..................................................................... 44 Table 6.1 Transfer Passengers (ICN, NRT) .................................................................................. 48  Table 6.2 China - NA Market Share (2011-2016) ........................................................................ 49 Table 6.3 Routes Ranked by Efficiency ....................................................................................... 50  Table 6.4 Routes Ranked by Responsiveness ............................................................................... 51    x  List of Figures  Figure 1.1 Research Organization................................................................................................... 7  Figure 5.1 Transfer Route: Bangkok - Vancouver ........................................................................ 47   xi  Acknowledgements First, I would like to express my sincere gratitude to my supervisor Professor Anming Zhang whose patient guidance, support and suggestions have been an immeasurable help throughout  my master study.  I also owe thanks to Professor Robin Lindsey and Professor Sanghoon Lee, for spending  valuable time giving me advice and serving as my thesis committee members.  Special thanks are owed to my fellow students, Kun Wang and Wenyi Xia, who have always provided coherent answers to my questions and have supported me throughout my years of MSc education at University of British Columbia. Finally, I want to thank my company, Incheon International Airport Corporation, that has supported me financially throughout my years of MSc education.  xii  Dedication To Sul Hee, Seungyeon and my families  1  Chapter 1: Introduction The airline industry has grown at a historically fast pace, with the passenger traffic having grown by about 60% over the last ten years according to Global Market Forecast 2017-2036 (Airbus, 2017).  Between January and August 2017, destinations worldwide welcomed 901 million international tourist arrivals (overnight visitors), 56 million more than in the same period of 2016. This corresponds to a robust 7% increase, well above the growth of previous years. Thanks to the exploding growth of travel and tourism industry, it is expected that worldwide aviation will maintain its growth momentum for the next 20 years. In effect, for the next 20 years, Boeing forecasts a 4.7% global annual air traffic growth according to Current Market Outlook 2017-2036 (Boeing, 2017), very similar to 4.4% of Airbus’ forecast. Amid the aviation industry growth, the role of an airport has been extended beyond the concept of traditional transportation platform. Importantly, airports are actively engaging in strengthening airlines’ network competitiveness because such a network advantage will attract more passengers and freight, leading to more airport revenues. Since the airport industry requires large-scale investment at regular intervals, expansion of stable revenue sources is an important issue. Obviously, the role of an airport is not restricted to a transportation platform any more. Policy makers are seeing airports as strategic assets to the regional and national economy (Twomey and Tomkins, 1995). When an airport is faithfully performing its role as the growth engine of the regional economy based on network competitiveness, it can be referred to as the hub airport (Button and Lall, 1999). In the case of the Netherlands’ Schiphol airport (Hakfoorta, Pootb, and Rietveld, 2001), the impacts of airport’s growth are not only to stay within the area where the airport is located, but to expand to the surrounding regions as well. Dubai aviation sector based  2  on Dubai International Airport supports over 250,000 jobs and contributes over USD 16.5 billion to Dubai’s GDP, which, according to Oxford Economics (2014), represent approximately 13% of total employment in Dubai and 16.5% of Dubai’s GDP in 2013. In addition, Schiphol and Dubai are pioneering cases of development of an “airport city” which creates its own aviation demand. According to Halpern and Graham (2015), 54% of the surveyed airports have a dedicated route development team to attract airlines to launch new routes and increase flights of existing routes, and have a marketing team to execute marketing plans for travel agencies and passengers. This kind of marketing has been recognized as a core activity at many airports, and is considered as a vital strategy to enhance network competitiveness. Currently, airport operators are actively involved in marketing airlines and passengers. Particularly, attracting the transfer passengers is one of the important targets of the marketing strategies. This importance stems from the fact that the transfer passengers basically do not directly belong to one airport’s local catchment area. This means that transfer passengers can select their transfer airports; as a consequence, the airport operators face more competition in the transfer passenger market than in the local market. Despite of severe competition, an airport cannot abandon developing transfer passengers’ market because the transfer traffic effectively contributes to expand the airport’s demand basis by making its airlines maintain their diverse routes as well as generate additional revenues. For airports to develop optimal competition strategies for the transfer passenger market, there is a need to identify, quantitatively, the preferences of transfer passengers. For example, many leading airports highlight, among others, the better transfer service quality and reputation of their own brand as a differentiation tool in order to be competitive in attracting more transfer customers (Chung et al, 2013).  3  1.1 Research Questions Specifically, this study focuses on the market of transfer passengers travelling between Southeast Asia (SE Asia) and North America (NA). The SE Asia - NA market has traditionally been a strong market. As of 2017, the market is the second largest inter-continent market in the world, next only to the Western Europe - NA market (i.e., the trans-Atlantic market), and has grown at an average rate of 8% per annum over the past five years. The Asia-Pacific including the SE Asia - NA market is also a region with huge potential demand given its continuous economic performance that outperforms the other regions of the world. According to recent statistics from Airport Council International (ACI) and International Air Transport Association (IATA), Asia-Pacific airports accounted for 36% of total passengers in the world. The overall market share of Asia-Pacific reaches 32% of global air passenger traffic. It is also forecasted that trans-Pacific demand will lead future aviation market growth mainly because the demand between two regions (SE Asia and China) and NA is highly robust. According to Boeing Market Outlook (2017), the SE Asia-NA market and the China-NA market will grow at almost 7% annually, more than double the market growth forecasted for other regions. According to the Centre for Asia-Pacific Aviation (CAPA, 2016), trans-Atlantic has grown 7%, Europe - Asia 13% – but trans-Pacific has increased 27% since 2011. In 2005, there were 25% as many trans-Pacific flights as trans-Atlantic; in 2015 this has grown to 34%. Trans-Pacific market can be expected to grow significantly at least until 2025.  Another reason for us to focus on the SE Asia - NA market is because the market has a longer flight distance than the Northeast Asia (NE Asia) - NA market or the Trans-Atlantic market. Thus, most of the flights between SE Asia and NA have more than one stops, transferring via NE  4  Asian airports as NE Asia-based airlines and airports had been developing their North American networks early on. For example, according to CAPA (2016), the market share of SE airlines currently accounts for less than 20% in this market. Instead, NE Asian airlines take the leading position mainly by providing connecting flights through their geographically advantaged hub airports, such as Taipei, Seoul-Incheon, Hongkong, Tokyo and Beijing.  Now, the competition in the market is likely to intensify as Chinese airports and airlines are dramatically expanding. China is rapidly expanding its network in NA based on its abundant market demand. Important progress has also been made in the China - SE Asia aviation market. In 2010, the Association of Southeast Asian Nations (ASEAN) and China signed an Open Skies agreement. Law et al. (2018) showed air connectivity between ASEAN and China has increased substantially since then. The number of flights operating between the two regions was 862 a week in 2010. In 2017, however, there were more than 5,000 flights operating between ASEAN and China each week.  In the NA market, the average daily number of passengers between China and the United States (US), which stood at about 5,000 in 2010, grew by more than 60% to about 8,000 in 2015. China’s share of the market has grown from 40% to 70% thanks to a significant boost in capacities to the NA market. Such an improved competitiveness of Chinese airports and airlines will likely trigger another growth in the transfer market between SE Asia and NA. Thus, based on above discussions, we are strongly motivated to analyze the transfer-passengers market between SE Asia and NA. Accordingly, this study will investigate the main factors of passengers’ choice of transfer airports in the SE Asia - NA market, and the strategies that the airport operators need to consider in order to gain a competitive edge in this transfer-passengers market. Specifically, our study focuses on the following three research questions:   5  Q1: What factors affect a transfer passenger’s decision in the SE Asia - NA market to select a transfer airport, and how does the choice react to changes in such factors? - Are the factors related to an airport’s characteristics besides traditional factors such as airfares and traveling time? Q2: How is passengers’ willingness-to-pay for changes in airport characteristics caused by improving operation policies or the implementation of marketing strategies? - Airport operators or authorities are currently in a supportive trend to develop policies for transfer passengers and implement marketing strategies to attract them. This is done by their own route development teams or marketing organizations. -  Thus, measuring the willingness-to-pay in monetary unit may become a good criterion for the problem of resource concentration and choice that are faced by the teams or organizations.  Q3: Can we compare and measure which route of the SE Asia - NA market is more efficient and responsive in attracting transfer passengers from a specific airport’s perspective?  1.2 Main Results This study applies the demand estimation approach in differentiated market by transforming observed market shares to mean utilities and estimates air travelers’ behavior in choosing transfer airports. Our data consists of cross-section data covering 33 transfer airports and 78 routes in the SE Asia – NA market. In SE Asia, 14 airports from eight countries are included. Airports in Brunei and Laos are excluded because of their restricted market size among the ten ASEAN countries. In NA, six major airports from the US and Canada are included. We estimate  6  demand parameters by standard instrument variables (IVs) techniques to control for the endogeneity of airfare, while treating airport characteristics including MCT (“minimum connection time” to make a transfer at each airport), service quality for transfer passengers, size of international passenger traffic which represents overall operation competencies and the detour degree (how much more time is needed by selecting a transfer flight instead of a direct flight) as exogenous. By adopting the two-stage least square (2SLS) method, we have the following empirical findings. First, transfer passengers’ airport choice is determined by an airport’s characteristics including MCT, service quality, size, and the detour degree, besides such traditional factors as airfares and traveling time. Second, based on the estimated demand parameters, we can evaluate passenger’s willingness-to-pay for the changes in airport characteristics resulting from airport policies or marketing strategies. Finally, the evaluation of passenger’s willingness-to-pay can provide airport operators with useful implications to execute marketing policies and strategies to attract transfer passengers. Based on the willingness-to-pay, we can identify the most “effective route” from a perspective of any one specific airport, in that this route has the greatest potential to increase the transfer passengers through the specific airport. The effectiveness is evaluated by two criteria including efficiency and responsiveness. The case study of Incheon International Airport (ICN) in South Korea is conducted. By combining the above two criteria, the route between Ho Chi Minh (SGN) and Los Angeles (LAX) is identified as the most effective route which is recommended for ICN to concentrate limited resources for implementing marketing plans and providing incentives.    7  1.3 Organization of the Thesis The rest of the thesis is structured as follows. Section 2 reviews the related literature. Section 3 presents the details of data source, econometric model and hypotheses to be investigated. Section 4 describes the econometric methodology employed, and Section 5 discusses the empirical results. Section 6 concludes. Figure 1.1 Research Organization  1.  Introduction - Research background and questions   2.  Literature Reviews   3.  Data, Hypotheses and Econometric Model    4.  Methodology - Research methodology employed   8                        5.  Estimation Results and Research Findings   6.  Conclusion  9  Chapter 2: Literature Review This study mainly intersects with three domains of literature: (1) determinant factors for the choice of transfer airport; (2) discrete model of evaluating demand for each route based on the aggregation of individual traveler’s preferences; and (3) handling endogeneity issues of the airline price in the discrete choice model of air travel. We now provide a review of the three strands of literature.     2.1 Decision Factors for Transfer Passengers There has been an extensive literature on the decision factors of air travel. Frequency, airfare and travel time have been the traditional and essential variables to explain the choice of airports and airlines. In the UK, studies including Ashford and Bencheman (1988) and Thomson and Caves (1993) found flight frequency as a critical factor to decide competitiveness of an airport. Furuich and Koppelman (1994) captured flight frequency as the most significant factor to explain airport choices in Japan. Brueckner and Zhang (2011) showed hub network which is an important factor for a transfer passenger’s choice is closely related with flight frequencies. Bradley (1998) suggested that airfare and ground access time to an airport to be the major factors for airport selection. Mason (2000) presented that air ticket price, flight frequencies and service level are the top three factors in the airport selection of business travelers in London area. Loo et al. (2005) showed that a “generalized price” including air ticket price and transportation cost to access an airport represented passenger’s preference well for international flights. The explanatory power  10  of this price variable was lower in the case of domestic flights than in the case of international flights. At the same time, a different view was presented and studied to describe passengers’ choice of airport with other variables including airport characteristics. These attempts started from the idea that traditional variables including price, frequencies and time were not enough to explain customers’ choice. For instance, one can observe an international air traveler choosing a specific airport but cannot tell whether he/she was choosing the airport only because of the air ticket price and higher flight frequencies (Loo, 2008). Harvey (1987) suggested demographic impacts from an airport choice as airport characteristics. Ndoh et al. (1990) found that average journey time, average connection time and weekly available seats on each route could largely explain choice of a hub airport.  Basar and Bhat (2004) presented the size of an airport, number of airlines serving the airport, and simply overall airport activities as desirable characteristics of an airport selection. Adler (2005) dealt with competitions among airlines in Western Europe in the 1990s and presented a “relative” view of utilizing variables, including traditional variables. This research showed that each airline’s market share in the market can be explained based on a function of flight frequencies, air fare, value of time and the decision variables of the competing airlines. The independent variables, for example relative air fare compared with competitor airlines,  were measured on a relative basis, instead of the absolute values, to reflect competition status of hub choice. Loo (2008) suggested the number of airlines, airport shopping area and airport access time as determinant factors for air passengers’ choice, based on the case of Hong Kong International Airport. By providing an analysis of the relationship between hub networks and airport charges including passenger service charge and landing fee, which is directly linked  11  to the airport's competitiveness for attracting transfer passengers, Lin and Zhang (2016) showed that airport operators are also able to play a significant role in the transfer market through the decisions on airport charges and capacity investment. Zhang (2003) provided interesting insights in terms of the factors that determine an international air-cargo hub airport. The study showed that, in addition to cost and delivery times, geopolitical location, infrastructure, customs modernization, inter-modal transportation and international aviation policy are competitive factors for developing an air-cargo hub. Burghouwt and Veldhuis (2006) suggested a new approach to explain factors for the determination of transfer airports, including concepts of connectivity and airline alliance. This approach provided an implication that it is useful to narrow the range of airlines into the airlines belonging to the same alliance as hub carriers of a transfer airport, when measuring flight frequencies at the airport. The approach may enhance empirical meaning of our study, since market data shows that most of the transfers are made by airlines belonging to the same alliance or the airlines with cooperative relations. Park and Jung (2011) suggested an interesting point of view on how transfer passengers see services at the airport. This study utilized the service level specified on transfer passengers as an important driver to explain their experience and satisfaction for the selection of transfer airports.  2.2 Discrete Choice Model The discrete choice model is widely used in estimating consumers’ choice among products. It was pioneered by McFadden (1973) by using the micro-level product characteristics data. Anderson et al. (1989) extended the discrete choice model to allow estimating demand for differentiated products based on their attributes. Later, Berry (1994) and Berry et al. (1995)  12  invented the BLP method, a structural discrete choice model, to estimate consumer utility functions using the aggregate data of product market shares. Our estimation method used in this study is partly in the spirit of the BLP method to use the transfer air product’s market share to estimate the utility function of the transfer passengers.  Discrete choice models have also been extensively used to estimate air travel choices. For example, one of the first studies of airport choice was by Skinner (1976), who estimated airport choice in three airports of the Washington DC region. This study showed that flight frequency and ground accessibility explained passengers’ choice well, and that the passengers were putting more importance on the latter. Harvey (1987) used a multinomial logit (MNL) model for airport selection, and his study revealed that accessibility to an airport and flight frequencies are significant to leisure passengers as well as business passengers. Windle and Dresner (1995), by adopting an MNL model, found that the more opportunities to use a specific airport a passenger had, the more likely the passenger was to use the same airport in the next travel. Pels et al. (2003) estimated the passengers’ airport choice in the San Francisco Bay area. Berry and Jia (2010) used the BLP method to systematically estimate the air passengers’ demand in the US market.  The MNL model has been developed to compute the market share of the airlines, given a maximal level of demand. Ben-Akiva and Lerman (1985) defined the MNL market share model as follows. 𝑀𝑆௜௝௦௔ = 𝑒௩೔ೕೞೌ ∑ 𝑒௩೔ೕೞ೎௔∈஺⁄                                               (2-1)  13  where i = origin; j = destination; s = an individual passenger; a = his choice for an airport to transfer.  Thus, 0 ≤ 𝑀𝑆௜௝௦௔ ≤  1      ∀𝑎 ∈ 𝐴, 𝑤ℎ𝑒𝑟𝑒 𝑡ℎ𝑒 𝑣𝑒𝑐𝑡𝑜𝑟 𝐴 𝑖𝑠 𝑠𝑒𝑡 𝑜𝑓 𝑎𝑙𝑙 𝑎𝑣𝑎𝑖𝑙𝑎𝑏𝑙𝑒 𝑎𝑙𝑡𝑒𝑟𝑛𝑎𝑡𝑖𝑣𝑒𝑠 ෍ 𝑀𝑆௜௝௦௔  =  1 The market share model is based on the following utility function. The utility of a passenger s represents the passenger’s preferences on frequency, price and airport characteristics. 𝑈௜௝௦௔ = 𝑉௜௝௦௔ + 𝜀௜௝௦௔                                                              (2-2) where 𝑉௜௝௦௔ is the deterministic utility attained by a passenger s in route (i,j) for the choice of alternative a.  Berry (1994) also introduced the discrete choice model as a basic frame for this research. The study showed that the utility of consumer i for product j on market m depends on the characteristics of the product and the consumer’s taste as followings. 𝑈(𝑥௝ , 𝜉௝ , 𝑝௝, 𝜃ௗ , 𝑣௜௝௠)                                                   (2-3) where 𝑥௝ are observed product characteristics other than the ticket price; 𝜉௝ is unobserved (by the econometrician) product characteristics; 𝑝௝ is product price; 𝜃ௗ is the utility (demand) function parameters we aim to estimate; 𝑣௜௝௠ captures consumer specific terms that are not observed by the econometrician. Specifically, we can express the utility function of an air traveler who choose a transfer airport as following equation similar to Eq. (2-2).  14  𝑈௜௝௠ = 𝑥௝𝛽෨௝ − 𝛼𝑝௝ + ξ௝௠ + 𝑣௜௝௠                                             (2-4)                                                           where the consumer-specific taste parameters are 𝛽෨௝ and 𝑣௜௝௠   The term ξ௝ might be thought of as the mean of consumers' valuations of an unobserved product characteristic such as product reputation, brand value and consumer experiences, while the 𝑣௜௝௠  represents the distribution of consumer preferences about this mean. 𝑈௜௝௠is a starting point to discuss because market share of product j that we can see in real market is the maximizer over discrete set of alternative utilities. 𝑠௝௠= argmax 𝑈௜௝ ,    𝑓𝑜𝑟 𝑖 = 1,2, . . . , 𝑛                                          (2-5) where 𝑠௝௠ stands for market share of product j on market m, and n is the total number of consumers. We also assume that 𝛽෨j = β (no random coefficients) and that 𝑣௜௝௠  is identically and independently distributed (IID) across products and consumers with the extreme value distribution exponential function.  Then, the market share of product j is then given by the well-known logit form. 𝑠௝௠ = 𝑒𝑥𝑝(𝑥௝𝛽෨௝ − 𝛼𝑝௝ + ξ௝௠) ∑ 𝑒𝑥𝑝(𝑥௞𝛽෨௞ − 𝛼𝑝௞ + ξ௞௠)௃௞ୀ଴ൗ = 𝑒𝑥𝑝(𝑥௝𝛽෨௝ − 𝛼𝑝௝ + ξ௝௠) 𝑉ത௠⁄  (2-6)  where 𝑉ത௠ = ∑ 𝑒𝑥𝑝(𝑥௞𝛽෨௞ − 𝛼𝑝௞ + ξ௞௠)௃௞ୀ଴  is the mean average utility of all the alternative products on market m, and J is total number of products in the market.    15  Taking the logarithm of both sides of Eq. (2-6) leads to the following semi log-linear regression model. 𝑙𝑛𝑠௝௠ = 𝑥௝𝛽෨௝ − 𝛼𝑝௝ + ξ௝௠ −  𝑉ത௠                                           (2-7) The above semi log-linear equation can be estimated with regression method, where the term 𝑉ത௠ can be controlled by the route-level specific dummy variables, and 𝜉௝௠ is the error term of the regression.  2.3 Endogeneity  One important issue in discreate model estimation is the endogeneity of the price with the unobservable product characteristics included in the error term. Berry and Jia (2010) presented methods to handle such endogeneity issue by adopting instrumental variables (IVs) to identify the fare coefficients. The study employed the percentage of rival routes that offer direct flights, the average distance of rival routes, the number of rival routes, and the number of all carriers as instrumental variables. In their study, one common strategy is to exploit the rival product attributes and the competitiveness of the market environment. One of the standard IVs is the total number of products in the same market, because, all else being equal, a product having closer substitutes is likely to have lower prices. Evans and Kessides (1993) employed an instrumental variable procedure to provide an unbiased estimate of the true effect of airline price on air traveler’s utility in US airline market. This study focused to employ a fixed-effects procedure to estimate directly the effect of route and airport  16  market shares on prices, because they argued that price was more objective variable than other factors which are difficult to measure and wanted to maximize the effect of using price variable. IVs that can affect costs but do not affect demand at the same time can be effective to handle endogeneity issue. The study also introduced other strategy to utilize variables that can affect costs, but do not affect demand at the same time. This group of instruments included the fitted values of the 25th and the 75th quantile of fares in a given route. As documented by Borenstein and Rose (1994), there is a wide fare dispersion across passengers travelling on the same route. The 25th and the 75th fitted fare quantiles are nonlinear functions of the exogenous route characteristics and allow us to parsimoniously capture the price dispersion (Berry and Jia, 2010).    17  Chapter 3: Market Analysis, Data, Econometric Model and Hypotheses  3.1 Market Analysis SE Asian airlines have been seeking to capture a larger share of the SE Asia - NA market over the past years as they launched new flights to the continent, NA. However, such efforts have not been successful so far. In 2015, Thai Airways announced that it cancelled the service from Bangkok to Los Angeles in October, ending 35 years of direct route operation to NA, due to high operation costs and low yields. Despite having the largest population of immigrants from the ASEAN countries, Los Angeles International Airport (LAX) is experiencing failures in receiving foreign carrier service originated from SE Asia, due to the tough market dynamics that renders it impossible to compete at profitable levels. The Philippines is by far the largest SE Asian destination from NA. In the year ending June 2016 the Philippines accounted for approximately 35% of all SE Asia - NA bookings according to CAPA (2016). The Philippines accounts for more than one-third of demand between SE Asia and NA. However, Philippine Airlines, the national flag carrier of the Philippines, relies almost entirely on local traffic, particularly ethnic traffic and the visiting friends/relatives segment instead of expanding aggressively to new markets. According to CAPA (2018), the main objective of the SE airlines is still to increase its share of local traffic, and thus, approximately two thirds of traffic between SE Asia and NA are travelling on one-stop options through NE Asian airports by NE Asian airlines instead. As per Table 3.1, NE Asian airlines are dominating this market.   18  Table 3.1 Top 10 Gateway Airports in SE Asia - NA Market (2011-2016) Rank based on 2016 Airport Code City Market Share Region 2016 2011 1 TPE Taipei 22.2% 18.0% NE Asia 2 ICN Seoul-Incheon 16.6% 23.2% NE Asia 3 HKG Hong Kong 15.1% 15.2% NE Asia 4 NRT Tokyo 13.5% 17.9% NE Asia 5 PEK Beijing 4.0% 1.3% NE Asia 6 PVG Shanghai 3.5% 1.0% NE Asia 7 MNL Manila 3.2% 3.6% SE Asia 8 CAN Guangzhou 3.0% 0.5% NE Asia 9 DXB Dubai 2.7% 0.8% Middle East 10 DOH Doha 2.3% 0.6% Middle East * Source: OAG Analyzer These days, the competition among NE Asian airlines and airports over the market has become fierce. Japan, where the regional hub of the US airline is located, is strategically nurturing Haneda airport with few problems such as a curfew and expansion limit. It was announced that Delta Airlines and Korean Air, the biggest hub airline of South Korea, reached an agreement to create a leading trans-Pacific joint venture in both scope and service in June 2017. As it is the cooperation between Delta Airlines, which has overwhelming connectivity to US domestic market and Korean Air, that is offering excellent network to SE Asia, it is predicted that there will be a great impact of influence in the SE Asia-NA transfer passengers’ market. Chinese airports and airlines are dramatically expanding internationally. Capacity and number of seats to NA in Chinese main airports are growing at a very high rate. The open skies agreement between  19  China and SE Asia has steadily contributed to enhance the competitiveness of Chinese airlines and airports in this market. Therefore, the incentive to better understand the market and customer characteristics is necessary in order to win the competition in this market.  3.2 Data 3.2.1 Demand Data The most basic data is the number of transfer passengers for each airport on 78 routes. We calculated market share of each transfer airport on each route, and the market share is a starting point of this research. Before we gathered transfer passengers’ data, which route to choose was an initial question. Totally, 78 routes between 14 airports in SE Asia and 6 airports in NA were selected. The SE Asian airports in this research are given in Table 3.2. Table 3.2 Sample Airports in Southeast Asia   Suvarnabhumi Airport in Bangkok, Thailand (IATA airport code: BKK)  Mactan–Cebu International Airport in Cebu, the Philippines (CEB)  Soekarno–Hatta International Airport in Jakarta, Indonesia (CGK)  Chiang Mai International Airport in Chiang Mai, Thailand (CNX)  Da Nang International Airport in Danang, Vietnam (DAD)  Ngurah Rai International Airport in Bali, Indonesia (DPS)  Nội Bài International Airport in Hanoi, Vietnam (HAN)   Phuket International Airport in Phuket, Thailand (HKT)   20   Kuala Lumpur International Airport in Kuala Lumpur, Malaysia (KUL)   Ninoy Aquino International Airport in Manila, the Philippines (MNL)   Phnom Penh International Airport in Phnom Penh, Cambodia (PNH)   Yangon International Airport in Yangon, Myanmar (RGN)   Tân Sơn Nhất International Airport in Ho Chi Minh, Vietnam (SGN)   Singapore Changi Airport in Singapore (SIN)  These 14 airports include eight capital airports of the major members of ASEAN (Association of Southeast Asian Nations) countries: Thailand, the Philippines, Indonesia, Vietnam, Malaysia, Cambodia, Myanmar and Singapore. Airports in Brunei and Laos are excluded because of their restricted market size among the ten ASEAN member countries. Selected airports are all primary and busiest international airports for each country. Besides the capital airports, additional six airports for leisure travelers serving famous tourism cities in the region are included.  They are: Cebu, Chiang Mai, Danang, Bali, Ho Chi Minh, and Phuket. These 14 airports have the most abundant international passenger demand in SE Asia. North American airports in this research are given in Table 3.3.  Table 3.3 Sample Airports in North America   John F. Kennedy International Airport in New York, USA (IATA airport code: JFK)  Los Angeles International Airport in Los Angeles, USA (LAX)  Chicago O'Hare International Airport in Chicago, USA (ORD)  San Francisco International Airport in San Francisco, USA (SFO)  21   Vancouver International Airport in Vancouver, Canada (YVR)  Toronto Pearson International Airport in Toronto, Canada (YYZ)  These six airports are selected because they are representative airports of the most popular destinations from SE Asia. In addition, they are evenly distributed in the eastern, central and western parts of NA. The arithmetic combinations of 14 airports in SE Asia and 6 airports in NA are 84, whereas the number of actual routes which transport transfer passengers was 78, as of July 2016. As our research study's target market is transfer passengers’ market, the routes that most of the transfer requests were made in 2016 are provided in Table 3.4. As 66 out of the 78 routes make up the top-120 routes in the transfer market, the routes in this study are representative in the SE Asia - NA market. Table 3.4 Top-120 Routes in Transfer Passengers’ Market between SE Asia and NA Rank Origin Destination Annual Booking Number Included or not 1 MNL LAX       386,156   2 SGN LAX       366,152   3 BKK LAX       266,332   4 MNL JFK       223,804   5 SGN SFO       204,580   6 MNL SFO       167,980   7 MNL YYZ       165,856   8 BKK JFK       159,416   9 BKK SFO       130,252   10 SGN IAH       123,200    22  Rank Origin Destination Annual Booking Number Included or not 11 MNL ORD       123,032   12 MNL SEA       102,256   13 MNL YVR         99,424   14 SGN JFK         95,900   15 SGN YYZ         88,300   16 SGN SEA         79,460   17 BKK YVR         69,588   18 SIN JFK         65,488   19 SIN LAX         64,340   20 MNL IAH         61,992   21 BKK ORD         60,796   22 SGN IAD         59,944   23 CGK LAX         56,812   24 MNL IAD         55,916   25 DPS LAX         55,768   26 PNH LAX         54,568   27 BKK YYZ         54,384   28 SGN DFW         54,116   29 BKK IAD         52,172   30 SGN YVR         49,936   31 LAO HNL         47,596   32 SIN SFO         46,204   33 BKK SEA         45,640   34 SGN ORD         45,260   35 MNL EWR         42,556   36 KUL LAX         41,492   37 MNL HNL         39,288   38 SGN ATL         36,816   39 KUL JFK         36,780    23  Rank Origin Destination Annual Booking Number Included or not 40 SIN YYZ         35,668   41 CEB LAX         34,132   42 SGN BOS         33,880   43 SIN YVR         33,716   44 BKK IAH         33,328   45 MNL YYC         31,808   46 HAN LAX         31,556   47 MNL LAS         31,260   48 HKT LAX         30,980   49 HAN JFK         30,908   50 MNL YEG         30,484   51 DPS SFO         29,724   52 CGK JFK         29,508   53 BKK HNL         29,068   54 CEB SFO         28,828   55 SIN IAH         28,000   56 SIN ORD         27,936   57 SIN SEA         26,184   58 CGK SFO         24,932   59 BKK BOS         24,756   60 DPS JFK         24,712   61 MNL DFW         23,688   62 CEB JFK         23,504   63 DPS YVR         23,076   64 BKK YUL         22,144   65 SIN BOS         22,048   66 SIN IAD         21,568   67 SGN HNL         21,108   68 KUL SFO         20,996    24  Rank Origin Destination Annual Booking Number Included or not 69 HAN YYZ         20,852   70 SIN EWR         20,692   71 MNL MIA         20,520   72 PNH SFO         20,104   73 PNH JFK         19,860   74 HKT JFK         19,780   75 BKK EWR         19,344   76 RGN JFK         17,184   77 CGK YVR         16,852   78 MNL YWG         16,652   79 BKK DFW         16,444   80 RGN LAX         16,252   81 HAN SFO         15,628   82 KUL YVR         15,512   83 PNH SEA         15,296   84 HAN YVR         15,252   85 MNL MCO         14,920   86 DAD LAX         14,912   87 BKK ATL         14,660   88 KUL YYZ         14,644   89 CEB ORD         14,416   90 RGN SFO         13,872   91 CGK MIA         13,804   92 SIN HNL         13,160   93 MNL BOS         13,096   94 PNH BOS         13,000   95 CGK SEA         12,996   96 CEB YYZ         12,944   97 CRK SFO         12,876    25   : included in this research, OAG Analyzer The data collection period is July 2016, and so the data covers traffic of one month. We used data from 2016 to take advantage of calibrated, reliable data. July, the peak season of a year, is opportune time to analyze transfer passengers’ market due to the tourism or overseas demand of Rank Origin Destination Annual Booking Number Included or not 98 SIN DFW         12,572   99 DAD SFO         12,428   100 HKT SFO         12,300   101 MNL SAN         12,084   102 CRK LAX         11,940   103 MNL YUL         11,392   104 PNH YVR         11,240   105 CGK IAD         11,144   106 SGN EWR         10,956   107 KUL IAH         10,812   108 PNH YYZ         10,780   109 DPS MIA         10,592   110 MNL OGG         10,300   111 KUL ORD         10,156   112 SGN YUL           9,788   113 SIN ATL           9,660   114 CEB YVR           9,540   115 CEB IAH           9,488   116 DVO LAX           9,448   117 CEB SEA           9,444   118 MNL ATL           9,272   119 CNX LAX           8,896   120 CGK YYZ           8,704    26  students who are sensitive to fare differences and are willing to take transfer flights. We could find demand data from the database of OAG Analyzer. OAG is an air travel intelligence company which provides digital information and applications to the world's airlines, airports, government agencies and travel-related service companies. OAG is best known for its airline schedules database which holds future and historical flight details for more than 900 airlines and over 4,000 airports. OAG Analyzer is an online accessible platform launched in 2012 to deliver airline schedule analysis and traffic analysis in partnership with Travelport, a leading distribution services and e-commerce provider for the global travel industry. As a result, the layout of demand data used in this study is shown in an example of Bangkok - Los Angeles route (Table 3.5). As shown in Table 3.5, the market share for each transfer airport is calculated for each route of this study. Table 3.5 An Example of BKK-LAX Route as of July 2016 Origin Destination Transfer Airport Booking Number Market Share BKK LAX AUH          60  0.2% BKK LAX CAN      1,000  3.8% BKK LAX CTU        260  1.0% BKK LAX DXB        836  3.2% BKK LAX HKG      2,412  9.1% BKK LAX ICN      2,496  9.5% BKK LAX KIX        148  0.6% BKK LAX NRT      3,524  13.4% BKK LAX PVG      3,388  12.8% BKK LAX SIN        168  0.6% BKK LAX TPE    12,088  45.8%   27  3.2.2 Airport Characteristic Data We have five types of airport characteristic data: flight frequencies, international passenger volume, service quality for transfer passengers, minimum connection time (MCT) and detour degree. Flight frequencies were measured for each transfer airport. The data also came from OAG Analyzer. When flight frequencies were measured, the concept of alliance was employed as mentioned in the previous section, because it is reasonable to assume that most transfers are happening through the transfer flights provided by airlines belonging to the same alliance. It means that flight frequencies of an airport used in this study are not the number of total frequencies, but the number of frequencies of the same alliance as the hub carrier of the transfer airport. We classified alliance into four groups: Sky Team, Star Alliance, One World and non-alliance. But, we adopted a different rule for Dubai International Airport (DXB) and Abu Dhabi International Airport (AUH). It is because Emirates Airline of DXB and Etihad Airlines of AUH have extensive coalition, interline and cooperation relationships with other airlines although they do not belong to any specific alliance. The detailed classification results can be found in appendix A. International passenger volume for each transfer airport was collected from the statistics of ACI and Airport Economics Report which is published by ACI every year. Since the values are annual numbers, monthly estimates divided by twelve were used in this study. The assumption that an airport which handles bigger number of international passengers has more efficient and streamlined operation competencies lies underneath this variable. The detailed number for each airport can be found in appendix B.  28  Airport Service Quality(ASQ) survey by ACI was used to estimate service quality of each transfer airport. ASQ is the most famous international survey measuring passengers’ satisfaction while they are at the airport. ASQ targets approximately 1,700 airports around the world and its results are announced every year. ASQ survey has 36 questions in total, and the score for each question is on a one to five scale, with three decimal points. One of survey questions is an index to evaluate quality of making connections. The index is as follows: Ease of making connections with other flights. Service quality used in this study came from the score for this question. The detailed score for each airport can be found in appendix C. MCT, meaning minimum connection time, is the absolute least amount of time that an able-bodied person needs to make a connection to a continuing flight in a transfer airport.  These were also collected from OAG Analyzer. MCT used in this study is the International - International Connection Criteria because most transfers in SE Asia - NA market happen in international routes. The detailed results can be found in appendix D. Our last characteristic data is detour degree of each route. The detour degree is obtained by dividing the total distance flown by transfer flights by the distance used by non-stop flight. Some transfer flights do not have direct flights at all, so the latitude and longitude location information should be used to calculate the distances instead of using actual travelling time. The calculation is based on the following form to calculate distance from point X to point Y.  acos{cos(latitude of point X)ⅹcos(latitude of point Y)ⅹcos(longitude of point Y-      longitude of point X)+sin(latitude of point X)ⅹsin(latitude of point Y)}ⅹ6371           (3-1) where the radius of the earth is approximately 6,371 kilometers.  29  3.2.3 Price Data Price data came from OAG Analyzer. OAG Analyzer provides average fare in USD for a specific route based on the booking records. Price may have a bias depending on how or when it is measured; in this study, the bias is handled by using instrumental variables (IVs) to deal with endogeneity issue from price variable. One of IVs is a relative price variable that indicates the level of the price of a route via a transfer airport compared to a route via another transfer airport in the same market. By using the relative price variable as IV, we can reduce the bias that would occur if we used the price as the absolute price. The number of competitors in the route is also used as IV, because competition is one of crucial factors to decide a price.  3.2.4 Summary Statistics Summary statistics are contained in the following Table 3.6. Table 3.6 Summary Statistics Variable Unit No. of observations Max Min Mean Stdev Frequency weekly 459 502 18 160 93 Size thousand 33 6,971 109 2,596 1,922 Service score 33 4.805 3.067 4.012 0.312 MCT* min 33 180 35 79 36 Detour degree** min 459 1.728 1.000 1.066 0.102 Fare (one-way) USD 459 1,878 251 541 189 * Minimum connection time.  30  ** Ratio of transfer flight distance over direct flight distance.  3.3 Econometric Model 3.3.1 Utility Function To study the determining factors of air traveler’s transfer airport choices in the SE Asia - NA market, we specify a utility function of a representative passenger as follows: 𝑈௥௧௜ =  𝛼𝑋௥௧ +  𝛽𝑃௥௧ + 𝜉௥௧ + 𝜀௥௧௜                                           (3-2) where the subscript r represents the route, and t represents the transfer airport, and i indicates the individual. X is a vector representing the airport characteristics that we can observe such as the flight frequencies, international passenger volume to measure the overall operation competency, the service quality for transfer passengers, airport MCT (minimum connection time) as an efficiency index for a transfer, and increased travelling distance from the selection of transfer airport t. Prt is the air ticket price. 𝜉௥௧ represents the unobserved airline product characteristics to the researcher but known by the passenger when making choice. 𝜀௥௧௜ represents a purely random utility error term. We assume that 𝜀௥௧௜  is independently and identically distributed (IID) and follows a logistic distribution.   31  3.3.2 Model of the Choice Probability We assume that there are totally n transfer airports. In a specific route r, an individual i selects transfer airport t when utility to select t (i.e., Urti) is larger than utility to select any other airport k except t (i.e., Urki). The behavior of i follows utility-maximizing principle: i chooses the best alternative that brings the biggest utility. The choice of i is designated by dummy variables 𝑦௥௧௜ as followings. 𝑦௥௧௜ =  1    (𝑈௥௧௜ > 𝑈௥௞௜)         ∀𝑘 ≠ 𝑡                                            (2)                                                                          𝑦௥௧௜ =  0    (𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒)                                                             (3-3) The choice probability is then 𝑃𝑟௥௧௜ = 𝑃𝑟(𝑦௥௧௜ = 1) = 𝑃𝑟(𝑈௥௧௜ > 𝑈௥௞௜) = 𝑃𝑟(𝛼𝑋௥௧ +  𝛽𝑃௥௧ + 𝜀௥௧௜ > 𝛼𝑋௥௞ +  𝛽𝑃௥௞ + 𝜀௥௞௜) =𝑃𝑟(𝜀௥௞௜ − 𝜀௥௧௜ < 𝛼(𝑋௥௧ − 𝑋௥௞) + 𝛽(𝑃௥௧ − 𝑃௥௞)             (3-4) Given 𝛼 and β, the choice probability is the probability that the unobserved terms, 𝜀௥௞௜ − 𝜀௥௧௜, are below 𝛼(𝑋௥௧ − 𝑋௥௞) + 𝛽(𝑃௥௧ − 𝑃௥௞), ∀𝑘 ≠ 𝑡. Different choice models arise from different assumed distributions of 𝜀௥௧௜ for a transfer airport t and different values of 𝛼 and β. To sum up, market share of a transfer airport t for a route r can be illustrated by the probability below. 𝑃𝑟 (𝑈௥௧ > 𝑈௥௞)         ∀𝑘 ≠ 𝑡                                                  (3-5) Then the choice probability takes the form of Eq. (3-6) by MNL,  32       𝑃𝑟௥௧ =௘௫௣ (ఈ௑ೝ೟ ା ఉ௉ೝ೟ାకೝ೟)∑ ௘௫௣ (ఈ௑ೝೖ ା ఉ௉ೝೖାకೝೖ)೙ೖసభ         ∀𝑘 ≠ 𝑡                                 (3-6) where Prrt is the market share of a transfer airport t in a route r; and n is the number of alternative transfer airports. Define that 𝛬(𝑘) = ∑ 𝑒𝑥𝑝 (𝛼𝑋௥௞ +  𝛽𝑃௥௞ + 𝜉௥௞)௡௞ୀଵ  which can be regarded as the mean utility of the travel in the route. Then,  𝑃𝑟(𝑦 = 𝑡) = ௘௫ ௣(ఈ௑ೝ೟ ା ఉ௉ೝ೟ାకೝ೟)௸(௞)                                                   (3-7) Then, taking logs of both sides, 𝑙𝑛 𝑃𝑟(𝑦 = 𝑡|𝑥, 𝑝) = 𝛼𝑋𝑟𝑡 +  𝛽𝑃𝑟𝑡 + 𝜉𝑟𝑡 −  𝑙𝑛 𝛬(𝑘)                             (3-8) The above semi log-linear equation can be estimated with regression method by using route level dummy variables to control for the average utility.  3.4 Hypotheses Our basic interest lies in the initial question of what factors determine transfer passengers to select their transfer airports. The choices of transfer passengers eventually result in the market share of each transfer airport that we can observe. Thus, if we can find out the factors to influence transfer passengers’ decisions, we may draw a critical clue as to the direction that a specific airport will take to increase its market share for the transfer passengers’ market. Moreover, these implications would be even more useful for the operator of airports if they were about the characteristics of airports that are able to make policy decisions. In fact, flight frequencies or prices are easily predictable drivers of transfer decisions. However, it is debatable  33  whether airport characteristics, such as services for transfer passengers, the overall proficiency of airport operators and MCT, are influencing transfer passengers’ decision making. It is interesting to see whether such an airport's characteristics are influential, especially in the SE Asia - NA transfer market. Thus,  Hypothesis 1 (H1): Airport characteristics have significant relations to transfer passengers’ decision. The first hypothesis in this study is what factors will actually affect the market share of each transfer airport and how big the impact is in SE Asia - NA market. The main question from the first hypothesis is to verify that airport characteristics have significant relations to transfer passengers’ decision. Hypothesis 2 (H2): We can measure willingness-to-pay of a transfer passenger to consume the changes in airport characteristics. The second hypothesis is that we can measure the impact on a monetary basis. The model of this study is based on the utility felt by a transfer passenger. Therefore, the second hypothesis starts from the ability to measure how much the price must change to leave the utility felt by the transfer passenger when characteristics of the transfer airport change. The main question of the second hypothesis is how much more or less the transfer passenger is willing to pay when the transfer airport where the transfer passenger is selecting to make a transit changes its characteristics.  If something can be measured, it has the same meaning that the thing can be compared with other ones. Thus, the second hypothesis implies that an airport operator or an airport authority can choose one of several policy alternatives in terms of maximizing market share and increasing the  34  number of transfer passengers in the SE Asia - NA market. The last hypothesis is closely related with the implication. Hypothesis 3 (H3): We can identify which route is more effective in terms of efficiency and responsiveness from a perspective of a specific airport. The last hypothesis is that a specific airport operator can determine which route should be given priority. Whether to operate on a specific route or increase flights is entirely up to the airline. However, the airport operator has a marketing organization to attract airlines by practicing incentive policies and executing marketing plans. Thus, knowing the priority would mean that limited resources could be utilized effectively for the interest of the airport operator itself. The main question of the last hypothesis is that which route should be preferred from a perspective of a specific airport, and it implies that the 78 routes used in this study could be ordered by how efficient and responsive they are under the investment of the same amount of resources.   35  Chapter 4: Methodology  4.1 Regression Model We assume the deterministic part of the utility function in the following form, with several characteristics of the transfer airline services: 𝑈ഥ = 𝛼𝑋௥௧ + 𝛽𝑃௥௧ = 𝛼ଵ𝐹𝑟𝑒𝑞 + 𝛼ଶ𝑆𝑖𝑧𝑒 + 𝛼ଷ𝑆𝑒𝑟𝑣 + 𝛼ସ𝑀𝐶𝑇 + 𝛼ହ𝐷𝑒𝑡 + 𝛽𝐹𝑎𝑟𝑒      (4-1) We define the above variables in Eq. (4-1) in detail as follows:    Freq: the flight frequencies for each transfer airport are measured by the number of monthly frequencies provided by the airlines in the same alliance as the hub carrier of the airport. The data comes from OAG Analyzer. Since most transfers are handled by airlines belonging to the same alliance, frequencies provided by the same alliance could be a more reasonable measure than the total number of frequencies provided at the transfer airport.   Size: the airport size variable is measured by the average number of international passengers in each month. This variable is exogenous because our observation is on route level, while size used in this study is the total number of international passengers at the airport level. The data comes from statistics of ACI. This variable assumes that an airport which handles bigger number of international passengers tends to keep airport operation more efficient and service oriented.   Serv: the airport service quality variable is retrieved from Airport Service Quality (ASQ) survey, where the index of “Ease of making connections with other flights” is used in this study. ASQ  36  is the most famous international survey measuring passengers’ satisfaction by ACI and is the most common benchmark for evaluating service level of an airport as well. ASQ has a total of 36 indexes, and the score for each question is on a one to five scale, with three decimal points.   MCT: the MCT variable, standing for minimum connection time, is the absolute least amount of time that an able-bodied person needs to make a connection to a continuing flight in a transfer airport. The data is collected from OAG Analyzer and is measured in minutes. MCT used in this study is based on the international-to-international connection criteria.   Det: the detour degree is a ratio of a total sum of distances of transfer flights to direct flights. The distances are calculated by using Eq. (3-1) for the distance between two spots. This variable measures the efficiency of a transfer route compared to a direct route. We can also interpret this variable as increased travelling time caused by taking a transfer flight instead of a direct flight.   Fare: Fare data is retrieved from OAG Analyzer. The value is an average discounted fare for economy seats provided in airport t for route r. Pricing data for premium seats is excluded because it could cause bias due to lack of data volume. Following Eq. (3-8), we can use the observed market share of each airline transfer service to replace the predicted probability of passengers’ choices. Thus, a semi-log linear regression model can be specified as Eq. (4-2). The demand parameters 𝛼 and 𝛽 can be estimated through regression on the logarithm of the market share as a dependent variable, and employing airport characteristics and fare data as independent factors.  𝑙𝑛(𝑀𝑆௥௧) = 𝛼ଵ𝐹𝑟𝑒𝑞௥௧ + 𝛼ଶ𝑆𝑖𝑧𝑒௧ + 𝛼ଷ𝑆𝑒𝑟𝑣௧ + 𝛼ସ𝑀𝐶𝑇௧ + 𝛼ହ𝐷𝑒𝑡௥௧ + 𝛽𝐹𝑎𝑟𝑒௥௧ + 𝜌𝐷௥+𝜉௥௧ (9) (4-2)  37  The route dummy variables 𝐷௥ are included to control for the route level average utility, which is the term  𝛬(𝑘) in Eq. (3-7). Route specific dummy variables are used in the regression model, meaning each dummy variable represents each route. By using dummy variables, we can remove bias from the fact that each route has different market participants and various market environment.  4.2 Endogeneity and IVs Price is an endogenous variable.  In a simple supply and demand model, when predicting the quantity demanded in equilibrium, the price is endogenous because producers change their price in response to demand and consumers change their demand in response to price. On the other hand, a change in consumer tastes or preferences would be an exogenous change of the demand curve. In this research, 𝐹𝑎𝑟𝑒௥௧ may be an endogenous variable because it is reasonably inferred to have some relations to ε୰୲ which cannot be observed in the market. The inconsistency of regression model is often due to endogeneity of independent variable x, meaning that changes in x are associated not only with changes in dependent variable y but also changes in the error ε. The endogeneity issue from 𝐹𝑎𝑟𝑒௥௧ variable can make results of regression analysis misleading about passengers’ behavior and choice in the SE Asia - NA market. To handle this endogeneity issue, two instrumental variables are employed in this research, followed by implications of Berry and Jia (2010) mentioned in the previous section. The instrumental variables used in this study represent the degree of competition in a route. The first IV, 𝐶𝑜𝑚𝑝௥௧  is the number of airports that are providing transfer flights in a route r. This variable shows how competitive the route is. Thus, 𝐶𝑜𝑚𝑝௥௧ represents price in the route well  38  because competition in a market is one of crucial factors when an airline decides its price policy. It does have significant relation to market share of an airport t in a route r and is less likely to have correlation with error term or hidden variables as well.  The second IV, 𝑅𝑎𝑡𝑖𝑜௥௧ is the ratio of 𝐹𝑎𝑟𝑒௥௧ to the overall average fare in a route r. That is, if n airports are competing on that route, it is a comparative indicator of 𝐹𝑎𝑟𝑒௥௧ (the price of an airport t) as compared to the average price offered by the competitors. This instrument has a strength that the distortion caused by using absolute value of prices can be removed. It indirectly represents the market’s competition situation by showing relative extent of the fare. This ratio demonstrates the fare dispersion on one route, reflecting the competition intensity, as competition drives airlines to offer differentiated products via different transfer airports. Thus, this ratio reflects competition intensity to affect airline price. It also does have significant relation to market share of an airport t in a route r and is less likely to have correlation with error term or hidden variables as well.  4.3 Two-Stage Least Square (2SLS) Estimation Under the assumption that we have enough instrumental variables to handle endogeneity of price variable, the method that we use in this research is to apply two-stage least square method (2SLS). The reason why we use 2SLS instead of simple regression analysis is to pursue higher reasonability and efficiency. We can get efficient estimators to understand behaviors of transfer passengers in the SE Asia - NA market and attain insights when the given market conditions have been changed. Based on the knowledge for preferences of transfer passengers in this  39  market, we want to advance our discussion into job of an airport operator including marketing plans, incentive policy and operation practices to expand its market share in the transfer passengers’ market. Why transfer passengers’ market is important has already been mentioned.   Acknowledging this, we first need to check if our econometric model and method is statistically significant. In other words, we need to have at least one exogenous independent in this first step. It seems that we have such independent variables including 𝐹𝑟𝑒𝑞௥௧, 𝑆𝑖𝑧𝑒௧, 𝑆𝑒𝑟𝑣௧, and 𝐷𝑒𝑡௥௧. As expected, we obtained positive coefficient values for 𝐹𝑟𝑒𝑞௥௧, 𝑆𝑖𝑧𝑒௧ and 𝑆𝑒𝑟𝑣௧ variables and negative coefficient values for 𝐷𝑒𝑡௥௧. However, we find that the sign for 𝑀𝐶𝑇௧  and 𝐹𝑎𝑟𝑒௥௧ is different from our expectation. In the simple regression analysis, we obtained positive coefficient values for 𝑀𝐶𝑇௧  and 𝐹𝑎𝑟𝑒௥௧, unexpectedly. Due to the endogeneity issue, without using instrumental variables, we can’t get statistically significant coefficients for some exogenous variable including 𝑀𝐶𝑇௧  and 𝐹𝑎𝑟𝑒௥௧ in this study. After checking out the endogeneity issue, then comes the second step regression. We deal with this endogeneity issue by adopting two instrumental variables as discussed. In the next section, we show the 2SLS estimation results produced by Stata commands and analyze what the results mean. We also want to extend the scope of the discussion to how these results can be utilized in an actual field.    40  Chapter 5: Empirical Results We have the 2SLS estimation results listed in Table 5.1. We analyze and discuss the estimation results of 2SLS in Sub-sections 5.1 and 5.2 below.  Table 5.1  2SLS Estimation Results LnMS Coef. z P>|𝑍|     Fare ( 𝛽) -.0005765 -2.77 0.006     Freq (𝛼ଵ) .0049749 5.43 0.000     Size (𝛼ଶ) .0001348 3.10 0.002     Serv (𝛼ଷ) .5657545 2.31 0.021     MCT (𝛼ସ) -.0047612 -1.99 0.046     Det (𝛼ହ) -5.601131  -7.70   0.000             Market id Coef. z P>|𝑍| Market id Coef. z P>|𝑍| 2 -0.38923 0.470447 -0.83 41 1.118781 0.4662 2.4 3 0.49876 0.545183 0.91 42 1.32279 0.481069 2.75 4 -0.34768 0.526553 -0.66 43 1.829542 0.42218 4.33 5 0.178912 0.492261 0.36 44 2.006488 0.705694 2.84 6 0.402597 0.507109 0.79 45 0.449863 0.428509 1.05 7 1.187317 0.575682 2.06 46 0.490852 0.417117 1.18 8 1.133699 0.685298 1.65 47 0.943277 0.59563 1.58 9 1.520953 0.544531 2.79 48 0.212501 0.59341 0.36 10 1.246206 0.733434 1.7 49 1.063234 0.499811 2.13 11 1.436657 0.657526 2.18 50 0.859261 0.568512 1.51 12 1.445853 0.626849 2.31 51 0.340889 0.427292 0.8 13 0.904828 0.479951 1.89 52 -0.31335 0.548494 -0.57 14 0.798665 0.574499 1.39 53 0.365469 0.56076 0.65 15 0.855013 0.705954 1.21 54 0.355572 0.475193 0.75  41  Market id Coef. z P>|𝑍| Market id Coef. z P>|𝑍| 16 1.025434 0.658243 1.56 55 0.250919 0.553589 0.45 17 0.944379 0.686471 1.38 56 0.505735 0.511561 0.99 18 1.264348 0.509939 2.48 57 0.809059 0.622794 1.3 19 2.319646 0.623524 3.72 58 0.006779 0.727978 0.01 20 1.874147 0.475088 3.94 59 2.463786 0.48124 5.12 21 1.643591 0.570474 2.88 60 0.836111 0.641401 1.3 22 2.182711 0.450534 4.84 61 0.997778 0.805628 1.24 23 1.66068 0.458133 3.62 62 1.400922 0.492602 2.84 24 1.613071 0.568076 2.84 63 0.904586 0.577405 1.57 25 1.939292 0.458308 4.23 64 2.000848 0.981362 2.04 26 1.679941 0.429463 3.91 65 1.23897 0.638763 1.94 27 1.924239 0.41861 4.6 66 2.562434 0.661989 3.87 28 1.060113 0.465592 2.28 67 -0.01803 0.495523 -0.04 29 0.729514 0.634248 1.15 68 -0.56539 0.800051 -0.71 30 1.401994 0.723496 1.94 69 1.085003 0.420792 2.58 31 0.889044 0.666705 1.33 70 -0.18724 0.523661 -0.36 32 1.17567 0.647754 1.81 71 0.099901 0.532114 0.19 33 1.353434 0.489152 2.77 72 0.518649 0.590992 0.88 34 0.629928 0.652988 0.96 73 -0.50534 0.560203 -0.9 35 0.386045 0.558618 0.69 74 0.292341 0.503534 0.58 36 1.544802 0.635988 2.43 75 0.536579 0.500489 1.07 37 0.652063 0.662729 0.98 76 1.021085 0.598317 1.71 38 1.186268 0.939942 1.26 77 0.163477 0.521655 0.31 39 0.661046 0.807319 0.82 78 0.482843 0.616573 0.78 40 1.396137 0.458111 3.05      42  5.1 Transfer Passenger’s Airport Choice This research identifies that transfer passengers’ airport choice is determined by an airport’s characteristics including service quality, size and MCT, besides traditional factors such as airfares and traveling time. Service quality, size and MCT variables are statistically significant to explain the market share in SEA-NA routes as shown in Table 5.1. The coefficients of these variables have the expected sign. Specifically, the better service the airport provides, the higher market share it has. The bigger airport generally shows the higher market share in transfer passengers’ market. The coefficient for MCT has a negative sign, because the shorter MCT means that a transfer passenger has an opportunity to select more available transfer flights. Transfer passengers prefer shorter MCT because it can provide them with more opportunities to choose departure flights. Traditionally, airfares and travelling times have been crucial factors to explain transfer passengers’ choice for transfer airports. It has been recognized that transfer passengers accept longer travelling time for lower fares. As an extension of this discussion, the role of airport operators in the transfer passengers’ market becomes absent because airfares and total travelling time are mainly dedicated to the decision area of airlines. However, the recent trend is different from the traditional view. It is widely accepted that airports are more than just transportation facilities in an actual business field. They actively attract passengers in competition, cooperate with airlines to improve network by launching a new route and increasing capacities on the existing route and even have the capability to create air demand by developing adjacent areas such as airport city. We can find actual success stories related to an airport city including Dubai and Schiphol, which creates its own aviation demand.  43  Accordingly, the first research finding can provide a proof that an airport operator can play a significant role in transfer passengers’ market. Factors including service and MCT clearly belong to the decision area of operation policy and facility capabilities of an airport operator. Size is also related with competency and experience of the operator. Based on the first finding, an airport operator can carry out its marketing strategy and investment plan more strategically and more scientifically from a perspective of increasing passengers and related revenues.  5.2 Transfer Passenger’s Willingness-to-Pay In this subsection, using the transfer passengers’ utility functions, we calculate the willingness-to-pay for several airport characteristics. Because we have estimated values of the coefficients (𝛼ଵ, 𝛼ଶ, 𝛼ଷ, 𝛼ସ, 𝛼ହ and β) in Eq. (4-2), we can calculate how much fare increase a transfer passenger is willing to accept for an improvement in airport characteristics such as reduced minutes in MCT, improved quality of service or an opportunity of using a bigger size airport.  The calculation process is specified in Table 5.2.  Table 5.2  Calculation process     We have estimated that 𝛼ଷ is close to 0.5657545 and β is close to -0.0005765   When one unit (point) of service level is improved in a ‘t’ airport, it is available to raise fare 981 USD to keep the same market share according to a calculation of dividing 𝛼ଷ by β.     44   It can be said that the transfer passengers in SE Asia - NA market are willing to pay 98 USD more to make a transit in an airport providing better service by 0.1 point in transfer convenience index of ASQ survey by ACI.      It is realistic to discuss 0.1 unit(point) increase or decrease of service level, since the service level comes from the score of ACI survey, the range of survey is 0 to 5, the maximum score is 4.805, the minimum score is 3.067, and 79% of scores are located between 3.700 and 4.500.  The calculated willingness-to-pay is shown in Table 5.3. Table 5.3 Transfer Passenger’s Willingness-to-Pay  When an airport t adds one more weekly flight for a route r, transfer passengers using the route have a higher chance to transfer at the airport t. Then a hub airline operating the route r at an Variable Unit Willing-to-pay Amount (USD) Note Frequency Increase of 1 flight (weekly) 35 - Size Increase of 0.1 million int’l pax (monthly) 23  Monthly number of international passengers of 58% of transfer airports in this study is concentrated between 2mil and 6mil  Service Level Increase of 0.1 point  98  79% of scores in service index are  between 3.700 and 4.500 MCT Reduction of 10 minute 82  85% of values of MCT are  between 45min and 120min Detour Degree Decrease of 0.01  97  91% of values of detour degree are  between 1 and 1.17  45  airport t can have a room to increase the fare without losing its market share. That implies increasing fare does not cause decrease in demand, leading to a new market equilibrium.  Therefore, the airport operator has a reason to involve in a route r. The airport operator has enough incentive to encourage the airlines to increase the number of flights in the route, because it could be beneficial for both sides. The airline can charge an additional USD 35 per transfer passenger without losing current demand level as per Table 5.3. The airport can enhance the network connectivity as well because connectivity is ultimately determined by the total number of routes the airport services. If the route has an important position for overall connectivity of an airport t, it gives an additional reason for the airport to provide better incentive such as reducing airport usage fee to the airline for longer periods than in other routes. This means that differentiated incentive policies can be established. Meanwhile, it seems that transfer passengers in this market do not consider size of the transfer airports seriously. A transfer passenger has an intention to pay only USD 23 more to use a bigger size airport that handles 0.1 million more passengers monthly. The amount seems not significant, when we consider almost 60% of transfer airports in this study handles international passengers between 2 million and 6 million monthly. On the other hand, transfer passengers in this market are sensitive to the airport service levels. A transfer passenger is willing to pay an extra USD 98 to transfer at an airport where ASQ survey shows a 0.1 point higher for transfer convenience index. Given that 79% of surveyed airports have scores between 3.700 and 4.500, and 0.1 point is not a small difference, it is reasonable to interpret that the passengers are sensitive to service quality. This has an important meaning because service is an area that can have a direct impact on an airport operator’s investment and  46  decision of operational practices. It can also be expected to play a meaningful role in prioritizing the future improvement and expansion investment decisions of an airport operator. A transfer passenger is willing to pay an additional USD 82 for 10 minute reduction in MCT. 10 minute reduction in MCT is a significant number, because reducing 10 minutes entails huge facility investment and input of human resources. Difficult negotiation and cooperation with airlines are a prerequisite as well. Reducing MCT by 10 minutes means that a transfer passenger can board a flight departing 10 minutes earlier than in other airports that have 10 minutes longer MCT, because airlines sell transfer tickets considering MCT in mind. Therefore, in this case, transfer passengers have a wider choice of flights. Given that flights do not depart every minute in most airports, this can have a significant impact on transfer passenger’s utility. The difference of 0.01 in detour degree is like the difference between transferring at Beijing China and transferring at Taipei Taiwan, when passengers move from Bangkok to Vancouver by transfer flights as following Figure 5.1. It is approximately one and half an hour's difference in flight time. The number of 0.01 in detour degree is significant because 91% of values of detour degree are concentrated in the range of 1.00 and 1.17. It is calculated that the customer in this market is willing to pay USD 97 more for a smaller detour degree by 0.01. This is consistent with the general tendency of transfer passengers affected by total travelling time.    47  Figure 5.1 Transfer Route: Bangkok - Vancouver   48  Chapter 6: Effective Route - The case study of Incheon International Airport Our estimated transfer passenger utility function and the calculated willingness-to-pay can be used by an airport operator to stipulate effective marketing strategies to compete with other transfer airports. Specifically, in this section, we conduct a counter-factual analysis and discuss policy implications from the perspective of an airport. We take the Incheon International Airport (ICN) as our subject, which has become the hub airport in NE Asia. The transfer passenger business has been essential for ICN, because the airport has a small domestic OD market compared to major competitors in China and Japan due to restricted population and stagnant population growth rate. Since 2011, ICN has been positioning itself as the hub airport to handle the largest number of transfer passengers in NE Asia, aiming to replace the traditional market leader, Narita International Airport (NRT). To achieve this goal, ICN has strong motivation to attract transfer passengers in the lucrative and fast-growing SE Asia - NA market which has been traditionally strong transfer axis due to ICN’s geographic location connecting two continents, SE Asia and NA. Table 6.1 compares the total transfer passenger numbers handled by ICN and NRT in recent years. Table 6.1 Transfer Passengers (ICN, NRT) (million) 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 ICN 4.4 5.2 5.2 5.7 6.5 7.0 6.6 6.7 6.4 6.4 NRT 6.0 6.4 6.0 5.3 5.7 5.7 5.3 5.0 4.8 4.0 1                                                  1 The sharp decline in the number of transfer passengers at Narita Airport in 2017 should also be taken into account by the policy decisions of the Japanese aeronautical authorities, which have diverted the allocation of international air routes between Narita Airport and Haneda Airport.  49  * source: Incheon Airport, Narita Airport  However, the number of transfer passengers handled by ICN peaked at 7 million in 2013 and has decreased to 6.4 million in 2017. As Chinese airports and airlines increase their supply to North America, Chinese airports are undermining transfer passenger business at ICN by increasing the number of direct passengers and intra-China transfer passengers as per Table 6.2. Since the China - NA transfer market has been the largest two transfer passengers’ axes of ICN along with the SE Asia - NA market, this is a huge blow to ICN. Given that China and US are in the process of discussing open skies agreement, this trend is unlikely to be reversed. This indicates the urgency for the ICN to form better competitive strategy to maintain and develop its transfer passenger market to effectively compete with the Chinese airports. Table 6.2 China - NA Market Share (2011-2016) Year 2011 2012 2013 2014 2015 2016 Direct 41.4% 41.3% 41.7% 43.6% 44.7% 45.7% Transfer 58.6% 58.7% 58.3% 56.4% 55.3% 54.3%  PEK (Beijing) 7.5% 6.9% 6.9% 9.1% 10.8% 10.5%  PVG (Shanghai) 4.0% 4.1% 3.7% 3.8% 4.5% 6.4%  HKG (Hong Kong) 5.1% 5.6% 5.4% 5.6% 5.1% 4.1%  ICN (Seoul-Incheon) 7.7% 8.3% 9.3% 7.4% 5.0% 4.0%  ORD (Chicago) 8.8% 7.2% 6.1% 5.6% 4.9% 4.0% * source: OAG Analyzer The SE Asia - NA market that has been one of the largest connecting axes due to geographical location is an important market to protect for ICN. This is the main reason why we choose ICN  50  for a case study. In this subsection, our goal is to present which routes are the most effective and responsive (competitive) for ICN to focus on when ICN can help airlines increase the frequency. We assume that ICN can increase flight frequencies by 5% in any one route by providing favorable terms for the operating airlines (such as lower airport charge or favorable slot). We then calculate the change in market share of each route as a result of the increased frequencies, and then compare which route is more efficient and responsive than others. The first perspective is which market is more efficient. Better efficiency means higher returns when the same amount is invested. In this study, it means which route leads to the greatest increase rate in market share when ICN increases flight frequencies by 5% in that route. The results are shown in the following Table 6.3. Table 6.3 Routes Ranked by Efficiency Rank Route Current M/S (A) To-Be M/S (B) B/A #1 SIN - LAX 10.4% 11.4% ↑9.9% #2 SGN - LAX 26.1% 28.6% ↑9.2% #3 SGN - SFO 10.1% 11.0% ↑8.9% #4 BKK - LAX 19.9% 21.6% ↑8.5% #5 HAN - LAX 23.3% 25.3% ↑8.4% ….. #64 CNX - JFK  2.08% 2.14% ↑2.8% #65 KUL - YYZ 6.7% 6.9% ↑2.3% #66 CNX - YVR 10.5% 10.7% ↑2.0% We can see that the efficiency is ranked in the order of SIN-LAX route, SGN-LAX route, and SGN-SFO route. Therefore, in terms of efficiency, ICN needs to concentrate resources on these routes.   51  The second perspective is which market is more responsive. Better responsiveness means more quantitative growth in absolute number when the same amount is invested. In this study, it means which route leads to the greatest increase in the number of transfer passengers. The results are shown in the following Table 6.4. Table 6.4. Routes Ranked by Responsiveness Rank Route Current M/S (A) To-Be M/S (B) Transfer Pax Increase* #1 SGN - LAX 26.1% 28.6%    ↑12,258 #2 MNL - LAX 51.4% 54.8%      ↑11,764 #3 BKK - LAX 19.9% 21.6%        ↑5,338 #4 MNL - JFK 30.8% 32.5%        ↑5,020 #5 SGN - SFO 10.1% 11.0%        ↑2,554 ….. #64 HKT - YYZ 9.1% 9.4% ↑6 #65 CNX - JFK 2.1% 2.1% ↑5 #66 CNX - YVR 10.5% 10.7% ↑3 * Estimated increase in number of annual transfer passengers We can see that the responsiveness is ranked in the order of SGN-LAX route, MNL-LAX route, and BKK-LAX route. Therefore, in terms of responsiveness, ICN needs to concentrate resources on these routes. In fact, the responsiveness depends on size of the existing market. Accordingly, demand-rich markets have a high tendency to become a major target. The route that has both high efficiency and responsiveness is SGN-LAX. It has the best responsiveness, which is reasonable because the route already has rich existing demand reflecting the recent trend that Vietnam is emerging as an Asia’s new manufacturing hub and Ho Chi Minh  52  is an economic capital of Vietnam. Investments into Vietnam have surged to USD 8-9 billion per year over the last 5 years, and it will naturally result in the rapid growth of aviation market. The recent growth of inbound tourism in Ho Chi Minh also contributes to the expansion of demand basis.  It is also superior in efficiency perspective. Thus, SGN-LAX is recommended as one of main targets for the ICN marketing organization. Based on the recommendation, ICN can decide a priority of marketing plans and the corresponding execution of a differentiated incentive policy among the candidates.    53  Chapter 7: Conclusion This study investigates the factors affecting air passengers’ choice of transfer airport in the SE Asia – NA market. It estimates demand parameters, and sheds lights on the marketing strategies for an airport to increase the volume of transfer passengers. We find that transfer passengers’ decision is related to an airport’s characteristics including service quality for transfer, airport size and MCT, besides traditional factors such as fare and travel time.  The effects of changes in an airport’s characteristics can be measured by willingness-to-pay of a transfer passenger. This measurement is important because it gives an airport operator a quantitative analysis and economic comparison. Since a change in airport characteristic usually accompanies cost increase and investment, the measurement of willingness-to-pay can be helpful to quantify how effective it is by estimating changes in expected revenue.  Finally, we present the case of ICN to show which route is more efficient and responsive in terms of the highest increase rate of market share and the largest absolute number of increased transfer passengers when flight frequencies on the route are adjusted. The route connecting Ho Chi Minh (SGN) and Los Angeles (LAX) is shown as a recommended route which is superior in efficiency and responsiveness from a perspective of ICN. To sum up, this research contributes to the understanding of air travelers’ choice of transfer airport in the SE Asia – NA market by utilizing aggregated market-level data. We identify the crucial factors that determine transfer passengers’ airport choice and estimate passengers’ willingness-to-pay for airport characteristics. The findings of this research provide valuable insights for an airport operator to allocate resources effectively to attract airlines and increase transfer passengers. It also takes a role as a strong proof representing the recent trend that airport operators are expanding  54  their role in marketing airlines and passengers. The single biggest development for the trans-Pacific market including the SE Asia - NA market will be open skies between China and the US. Law et al. (2018) presented the impacts of deregulation on the airline industry and on the traffic and tourism flows in the Asia-Pacific region are significant and positive. The year 2019 presents a possible opening for China and the US to sign an open skies agreement (CAPA, 2017). The China-US open skies will have a profound impact on trans-Pacific and intra-Asian aviation because the China - SE Asia aviation market has already grown steadily since the two parties signed an open skies agreement in 2010. This would principally lift restrictions on flights between two major regions and result in strong attractions for transfer traffic in the trans-Pacific aviation market. The China-US open skies will likely result in other trans-Pacific markets to improve access (even if stopping short of full liberalization), while intra-Asian markets may also liberalize. Future research can be undertaken from this point of view by taking these developments into account. This study is limited in that it does not take into consideration dynamics of the supply side in which a change of one airport characteristics and policies may induce the strategic reactions of other competing airports. Although estimating behavior of transfer passengers in a specific market may be challenging,  it is certain that this attempt is informative for an airport operator to involve more strategically into attracting airlines and marketing air routes to enhance network connectivity. Airport's ultimate competitiveness is connectivity. Therefore, such an attempt is worthwhile from the viewpoint of airport management as well as government policy making. The model used in this study could be applied to markets other than the SE Asia - NA market in further studies. Research for another market will help an airport operator to understand the unique behavioral characteristics of each market, and support the operator to establish optimized  55  tailored operational policies and marketing strategies. It is our hope that this study would promote further research on transfer passengers’ market from a perspective of airport operator.    56  Bibliography ACI (2015). Key performance indicators 2015. ACI (2015). 2015 ACI airport economics report. ACI (2016). 2016 worldwide airport traffic report. Airbus (2017), Global Market Forecast 2017-2036. Retrieved from http://www.airbus.com/company/market/forecast/?eID=maglisting_push&tx_maglisting_pi1%5BdocID%5D=86756. Adler, N. (2005). Hub-spoke network choice under competition with an application to Western Europe. Transportation Science 39(1), 58-72. Anderson, S.P., De Palma, A., Thisse, J.F. (1989). Demand for differentiated products, discrete choice models, and the characteristics approach. The Review of Economic Studies 56(1), 21-35. Ashford, N., Bencheman, M. (1988). Passengers' choice of airport: an application of the multinomial logit model. Transportation Research Record 1147, 1-5. Basar, GA., Bhat CR. (2004). A parameterized consideration set model for airport choice: an application to the San Francisco Bay Area. Transportation Research Part B 38(10), 889-904. Ben-Akiva, M.E., Lerman, S.R. (1985). Discrete choice analysis: theory and application to travel demand. MIT Press. Berry, S.T. (1994). Estimating discrete-choice models of product differentiation. The RAND Journal of Economics 25(2), 242-262. Berry, S.T., Levinsohn, J., Pakes, A. (1995). Automobile prices in market equilibrium. Econometrica 63(4): Journal of the Econometric Society, 841-890. Berry, S.T., Jia, P. (2010). Tracing the woes: an empirical analysis of the airline industry. American Economic Journal: Microeconomics 2(3), 1-43. Bilotkach, V., Clougherty, J.A., Mueller, J., Zhang, A. (2012). Regulation, privatization, and airport charges: panel data evidence from European airports. Journal of Regulatory Economics 42(1), 73-94.  57  Boing (2017), Current Market Outlook 2017-2036. Retrieved from http://www.boeing.com/resources/boeingdotcom/commercial/about-our-market/assets/ downloads/ Boeing_Current_Market_Outlook_2017.pdf. Borenstein, S., Rose, N.L. (1994). Competition and price dispersion in the U.S. airline industry. Journal of Political Economy 102(4), 653–83. Brueckner, J.K., Zhang, Y. (2001). A model of scheduling in airline networks: how a hub-and spoke system affects flight frequency, fares and welfare. Journal of Transport Economics and Policy 35(2), 195–222. Burghouwt, G., Veldhuis, J. (2006). The competitive position of hub airports in the transatlantic market. Journal of Air Transportation 11(1), 106-130. Centre for Asia-Pacific Aviation (2016). Press release database. Retrieved from https://centreforaviation.com/insights/analysis/southeast-asia-us-market-part-3-new-nonstops-need-to-overcome-stiff-one-stop-fsc--lcc-competition-303022. Centre for Asia-Pacific Aviation (2016). Press release database. Retrieved from https://centreforaviation.com/insights/airline-leader/chinas-airlines-begin-to-dominate-north-pacific-markets--and-to-leverage-transfer-options-294994. Centre for Asia-Pacific Aviation (2017). Press release database. Retrieved from https://centreforaviation.com/insights/analysis/us-china-open-skies-window-in-2019-with-alignment-of-airline-partnerships--airport-infrastructure-340603.  Centre for Asia-Pacific Aviation (2018). Press release database. Retrieved from https://centreforaviation.com/insights/analysis/north-america-asia-future-of-the-trans-pacific-airline-market-409286. Centre for Asia-Pacific Aviation (2018). Retrieved from https://centreforaviation.com/insights/analysis/southeast-asia-us-aviation-market-thai-airways-and-vietnam-airlines-to-enter-following-faa-upgrade-391751 Chung, T.W., Jang, H.M., Han, J.K. (2013). Financial-based brand value of Incheon International Airport. The Asian Journal of Shipping and Logistics 29(2), 267-286. Chung, T.W., Lee, Y.J., Jang, H.M. (2017). A comparative analysis of three major transfer airports in Northeast Asia focusing on Incheon International Airport using a conjoint analysis. The Asian Journal of Shipping and Logistics 33(4), 237-244.  58  Bradley, M.A. (1998). Behavioral models of airport choice and air route choice. Travel behavior research: Updating the State of Play, 141-145. Evans, W., Kessides, I. (1993). Localized market power in the U.S. airline industry. The Review of Economics and Statistics 75(1), 66-75. Furuich, M., Koppelman, F.S. (1994). An analysis of air travelers' departure airport and destination choice behavior. Transportation Research Part A 28(3), 187-195. 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), 24-41. Halpern, N., Graham, A. (2015). Airport route development: a survey of current practice. Tourism Management 46, 213-221. Harvey, G. (1987). Airport choice in a multiple airport region. Transportation Research Part A: Policy and Practice 21(6), 439-449. Incheon International Airport. Statistics. https://www.airport.kr/co/ko/cpr/statisticCategoryOfDay.do. InterVISTAS (2015). Asia-Pacific commercial air transport: current and future economic benefits. InterVISTAS-NA (2017). Asia-Pacific aviation: growth and challenges.  Law, C.C.H., Zhang, Y., Zhang, A. (2018). Regulatory changes in international air transport and their impact on tourism development in Asia-Pacific. In: X. Fu and J. Peoples (eds.), Advances in Airline Economics, Volume 8 (Airline Economics in Asia), Elsevier, forthcoming. Loo, B.P.Y. (2008). Passengers’ airport choice within multi-airport regions. Journal of Transport Geography 16, 117-125. Loo, B.P.Y., Ho, H.W., Wong, S.C. (2005). An application of the continuous equilibrium modelling approach in understanding the geography of air passenger flows in a multi-airport region. Applied Geography 25(2), 169-199. Lin, M.H., Zhang, A. (2016). Hub congestion pricing: discriminatory passenger charges. Economics of Transportation 5, 37-48. McFadden, D. (1973). Conditional logit analysis of qualitative choice behavior. In: P. Zarembka (ed.), Frontiers of Econometrics, New York: Academic Press.  59  Narita International Airport. Statistics. https://www.naa.jp/en/airport/traffic.html. Ndoh, N., Pitfield, D., Caves, R.E. (1990). Air transportation passenger route choice: a nested multinomial logit analysis. Spatial Choices and Processes, 349-365 OAG Analyzer. https://analytics.oag.com/analyser-client/home Oxford Economics (2014), Quantifying the economic impact of aviation in Dubai. Retrieved from http://www.dubaiairports.ae/docs/default-source/Publications/oxford_economics_quantifying_the_economic_impact_of_aviation_in_dubai_november_2014_final(1)cc4dc38b5d08685a9b2fff000058806b.pdf?sfvrsn=0. Park, J.W., Jung, S.Y. (2011). Transfer passengers’ perceptions of airport service quality: A case study of Incheon International Airport. International Business Research 4(3), 75-82. 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: Policy and Practice 37(1), 71-83. Thomson, A., Caves, R. (1993). The projected market share for a new small airport in the north of England. Journal Regional Studies 27, 137-147. Travel Codex (2015). Press release database. Retrieved from https://www.travelcodex.com/thai-airways-cancels-los-angeles-route-ends-35-years-of-serving-north-america. Tsui, K.W.H., Yuen, A.C.L., Fung, M.K.Y. (2018). Maintaining competitiveness of aviation hub: empirical evidence of visitors to China via Hong Kong by air transport. Current Issues in Tourism 21(11), 1260-1284.  Windle, R., Dresner, M. (1995). Airport choice in multiple-airport regions. Journal of Transportation Engineering 121(4), 332-337. Zhang, A. (2003). Analysis of an international air-cargo hub: the case of Hong Kong. Journal of Air Transport Management 9, 123-138.    60  Appendices  Appendix A   Hub Carriers and Alliances Airport’s Code Airport’s Name Hub Airline Star Alliance Sky Team One World AMS Amsterdam Schiphol Apt KLM -  - AUH Abu Dhabi International Apt Etihad Airways    BKK Bangkok Suvarnabhumi International Apt Thai Airways  - - BNE Brisbane Airport Qantas Air - -  CAN Guangzhou Airport China Southern Airlines -  - CDG Paris Charles de Gaulle Apt Air France -  - CPH Copenhagen Kastrup Apt Scandinavian Airline  - - CTU Chengdu Airport China Eastern/ Southern Airlines -  - DEL Delhi Airport Air India  - - DOH Doha Airport Qatar Airways - -  DTW Detroit Metropolitan Wayne County Delta Airlines -  - DXB Dubai International Emirates    FCO Rome Fiumicino Apt Alitalia -  - FRA Frankfurt International Apt Lufthansa  - - HEL Helsinki-Vantaa Finn Air - -  HKG Hong Kong International Apt Cathay Pacific - -  ICN Seoul Incheon International Airport Korean Air, Asiana Airlines   - IST Istanbul Ataturk Airport Turkish Airlines  - - KIX Osaka Kansai International Airport Japan Airlines - -  LHR London Heathrow Apt British Airways - -   61  * Airline website   MUC Munich International Airport Lufthansa  - - MXP Milan Malpensa Apt Alitalia -  - NKG Nanjing Apt Air China  - - NRT Tokyo Narita Intl Japan Airlines, ANA  -  PEK Beijing Capital Intl Apt Air China  - - PVG Shanghai Pudong International Apt China Eastern Airlines -  - SFO San Francisco Airport United Airlines  - - SIN Singapore Changi Apt Singapore Airlines  - - SYD Sydney Kingsford Smith Apt Qantas Air - -  TPE Taipei Taiwan Taoyuan International Apt China Airlines, Eva Air   - WUH Wuhan China Eastern/ Southern Airlines -  -  62  Appendix B  International Passenger Volume  * ACI, Airport Economics Report, Airport website  Airport’s Code Airport’s Name Annual Int’l Pax (2016) Airport’s Code Airport’s Name Annual Int’l Pax (2016) AMS  Amsterdam Schiphol Apt 63,533,504   ICN  Seoul Incheon International Airport 57,152,206   AUH  Abu Dhabi International Apt 25,964,178   IST  Istanbul Ataturk Airport    41,035,985   BKK  Bangkok Suvarnabhumi International Apt 45,291,073  KIX  Osaka Kansai International Airport    19,151,000   BNE  Brisbane Airport      5,589,738   LHR  London Heathrow Apt    71,030,114   CAN  Guangzhou Airport      9,108,438   MUC  Munich International Airport    32,569,420   CDG  Paris Charles de Gaulle Apt    60,384,622   MXP  Milan Malpensa Apt    18,978,773   CPH  Copenhagen Kastrup Apt    26,394,122   NKG  Nanjing Airport      1,880,984   CTU  Chengdu Airport      3,105,349   NRT  Tokyo Narita Intl    31,911,208   DEL  Delhi Airport    14,200,000   PEK  Beijing Capital Intl Apt    26,000,000   DOH  Doha Airport    37,216,179   PVG  Shanghai Pudong International Apt    24,865,920   DTW  Detroit Metropolitan Wayne County      2,868,069   SFO  San Francisco Airport    12,362,668   DXB  Dubai International    83,654,250   SIN  Singapore Changi Apt    58,158,000   FCO  Rome Fiumicino Apt    29,096,160   SYD  Sydney Kingsford Smith Apt    14,900,000   FRA  Frankfurt International Apt    53,707,953   TPE  Taipei Taiwan Taoyuan International Apt    41,876,848   HEL  Helsinki-Vantaa    14,504,796   WUH  Wuhan Airport      1,313,282   HKG  Hong Kong International Apt    70,098,216      63  Appendix C  ASQ Survey * Index: 20. Ease of making connections with other flights  Airport’s Code Airport’s Name Index Score (2016) Airport’s Code Airport’s Name Index Score (2016) AMS  Amsterdam Schiphol Apt 3.705   ICN  Seoul Incheon International Airport 4.805   AUH  Abu Dhabi International Apt       3.944   IST  Istanbul Ataturk Airport       3.742   BKK  Bangkok Suvarnabhumi International Apt       4.108   KIX  Osaka Kansai International Airport       3.915   BNE  Brisbane Airport       3.899   LHR  London Heathrow Apt       3.693   CAN  Guangzhou Airport       4.690   MUC  Munich International Airport       3.936   CDG  Paris Charles de Gaulle Apt       3.421   MXP  Milan Malpensa Apt       3.067   CPH  Copenhagen Kastrup Apt       3.980   NKG  Nanjing Airport       4.200   CTU  Chengdu Airport       4.210   NRT  Tokyo Narita Intl       4.324   DEL  Delhi Airport       4.488   PEK  Beijing Capital Intl Apt       4.583   DOH  Doha Airport       4.064   PVG  Shanghai Pudong International Apt       4.730   DTW  Detroit Metropolitan Wayne County       4.156   SFO  San Francisco Airport       4.110   DXB  Dubai International       4.015   SIN  Singapore Changi Apt       4.714   FCO  Rome Fiumicino Apt       3.776   SYD  Sydney Kingsford Smith Apt       3.887   FRA  Frankfurt International Apt       3.371   TPE  Taipei Taiwan Taoyuan International Apt       4.442   HEL  Helsinki-Vantaa       4.128   WUH  Wuhan Airport 4.467  HKG  Hong Kong International Apt       4.647      64  Appendix D  Minimum Connection Time (MCT) * OAG Analyzer, International-International connection   Airport’s Code Airport’s Name MCT (2016, min) Airport’s Code Airport’s Name MCT (2016, min) AMS  Amsterdam Schiphol Apt              65   ICN  Seoul Incheon International Airport              70   AUH  Abu Dhabi International Apt              60   IST  Istanbul Ataturk Airport              60   BKK  Bangkok Suvarnabhumi International Apt              75   KIX  Osaka Kansai International Airport              90   BNE  Brisbane Airport              50   LHR  London Heathrow Apt             105   CAN  Guangzhou Airport             150   MUC  Munich International Airport              45   CDG  Paris Charles de Gaulle Apt             150   MXP  Milan Malpensa Apt             120   CPH  Copenhagen Kastrup Apt              45   NKG  Nanjing Airport              60   CTU  Chengdu Airport              60   NRT  Tokyo Narita Intl             100   DEL  Delhi Airport              75   PEK  Beijing Capital Intl Apt              90   DOH  Doha Airport              60   PVG  Shanghai Pudong International Apt             120   DTW  Detroit Metropolitan Wayne County             150   SFO  San Francisco Airport             105   DXB  Dubai International              75   SIN  Singapore Changi Apt              60   FCO  Rome Fiumicino Apt              45   SYD  Sydney Kingsford Smith Apt              60   FRA  Frankfurt International Apt              45   TPE  Taipei Taiwan Taoyuan International Apt              75   HEL  Helsinki-Vantaa              35   WUH  Wuhan Airport              60   HKG  Hong Kong International Apt              60      65   

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            data-media="{[{embed.selectedMedia}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0371142/manifest

Comment

Related Items