Problem statement
In 2015, the aviation industry reached a historic net profit of $33 billion, nearly doubling the $17.4 billion profit from 2014 The Asia Pacific region alone contributed over $5.8 billion to this total, accounting for 31% of global air passengers, while Europe and North America represented 30% and 26%, respectively Additionally, low-cost carriers transported more than 950 million passengers, making up approximately 28% of scheduled air travel (IATA report, 2016).
The International Air Transport Association (IATA) projects that the number of air travelers will nearly double from 3.8 billion in 2016 to 7.2 billion by 2035, highlighting the rapid growth of the aviation sector Among the fastest growing markets, Vietnam is expected to see an influx of 112 million new passengers, bringing its total to 150 million Furthermore, Vietnam is recognized as one of the top seven countries experiencing significant aviation growth The Vietnamese government is prioritizing infrastructure development, aiming to establish 26 airports by 2020, with Long Thanh International Airport set to open its doors by that time.
The Vietnam airline industry, which was administered by Ministry of Transport and Civil Aviation Authority of Vietnam, has witnessed rapid growth in 2015 compared to the figures in
In 2014, the airline market served 40.1 million passengers and transported 771,000 tons of cargo, with domestic carriers accounting for 31.1 million passengers, reflecting a 21% increase The subsequent 30% drop in crude oil prices in 2015 provided a stimulus for airlines to further reduce fares, catering to the growing demand for passenger transportation.
The airline industry in Vietnam presents significant potential due to several factors With a population exceeding 90 million, the demand for travel is substantial Additionally, rising incomes among Vietnamese citizens have further increased the need for air transit Travelers now have various transport options that are not only faster but also safer Despite some global airline disasters in 2014, air travel remains the safest mode of transportation, as highlighted by the IATA Safety Report, which recorded 12 fatal accidents out of 73 incidents, resulting in 641 fatalities worldwide.
In 2014, air travel accounted for a small percentage of the approximately 33 billion passengers worldwide (IATA Annual Review 2015) However, it significantly reduces travel time; for instance, a train journey from Ho Chi Minh City to Hanoi takes about two days, while flying takes only two hours Additionally, the rise of e-commerce, facilitated by the internet, allows consumers to purchase airline tickets from the comfort of their homes, often at discounted rates during promotional periods.
In 1956, the Vietnam Civil Aviation Department was established with just five aircraft for domestic flights The national carrier, Vietnam Airlines, was founded in 1993, and by 1995, it evolved into Vietnam Airlines Corporation, comprising 20 aviation enterprises Today, Vietnam Airlines operates a comprehensive network of domestic and international services across Southeast Asia, North Asia, Europe, and Australia In July 2016, ANA Holding Inc became a strategic shareholder by acquiring an 8.77% stake Vietnam Airlines plans to reduce state ownership to 75% as part of its restructuring efforts Additionally, Skytrax has recognized Vietnam Airlines as a 4-star airline, highlighting its quality and service.
Vietjet Air, Vietnam's first privately owned international low-cost carrier, received approval to operate in November 2007 but commenced its inaugural flight in December 2011 with just three aircraft By 2015, the airline expanded its fleet to 29 aircraft, servicing 28 domestic and 12 international routes Looking ahead to 2016, Vietjet plans to increase its fleet to 42 aircraft to accommodate growing travel demand and introduce three additional domestic and five international routes.
Jetstar Pacific Airlines JSC, originally founded as Pacific Airlines in 1990, began its operations in 1991 with charter cargo services under Vietnam Airlines Corporation The airline expanded into passenger services in 2005 and joined the Jetstar network in 2008, rebranding as Jetstar Pacific In 2012, Vietnam Airlines acquired a 70% stake in the airline, leaving Qantas with a 30% ownership.
Vietnam Air Services Company (VASCO), a subsidiary of Vietnam Airlines Corporation, has been operational since 2004, offering passenger transportation from Tan Son Nhat Airport to various southern destinations, including Ca Mau, Con Dao, Rach Gia, and Can Tho In addition to its flight services, VASCO functions as a multifunctional airline, providing maintenance services for private aircraft.
Currently, four domestic airlines operate in Vietnam: Vietnam Airlines, Vietjet, Jetstar, and VASCO Historically, two additional airlines, Indochina Airlines and Air Mekong, also served the market but ceased operations due to financial difficulties Indochina Airlines suspended all flights after just one year in 2009, while Air Mekong halted its commercial services in 2013 and was officially revoked by the Ministry of Transport in January 2015.
Numerous studies have explored customer behavior and airline choice, with a 2015 IATA report revealing that passengers primarily select airlines for nonstop flights (15%) and lowest fares (14%), while recommendations from travel agents and in-flight services account for only 4% and 3%, respectively Despite the rapid growth of Vietnam's airline industry in recent years, research on this topic remains limited Understanding passenger preferences is essential for aviation companies and foreign investors alike.
The policies of the three carriers are tailored to meet the needs of the Vietnamese population, assisting investors in assessing the airline market to make informed investment decisions.
Research objectives
factors affecting the choice of passengers, and thus provide information for carriers in identifying their target market segments and efficiently improving their services.
Research questions
When choosing an airline, travelers consider various attributes that significantly impact their decision-making process Additionally, demographic factors such as age, income, and travel frequency also play a crucial role in influencing airline preferences among air travelers.
Scope of the thesis
This research focuses on the airline preferences among three major carriers in Vietnam: Vietnam Airlines (VNA), Vietjet (VJ), and Jetstar (BL), while excluding VASCO due to its limited operations in the Southeast, primarily servicing short flights such as those from Sai Gon to Ca Mau, Rach Gia, and Con Dao VASCO's primary business is aircraft maintenance rather than passenger transport, resulting in a minimal market share, making its exclusion from the analysis a non-critical issue.
Structure of thesis
The study comprises four chapters, beginning with Chapter 2, which explores the theory of random utility, stated preference, and revealed preference data, alongside an empirical analysis of choice models in the airline industry Chapter 3 details the research methodology, including the questionnaire design, survey process, and empirical model employed In Chapter 4, the data collected from the survey is thoroughly described, along with the results derived from the model Finally, Chapter 5 summarizes the key findings and acknowledges the study's limitations.
This chapter begins by exploring the economic literature on individual choice, which serves as the basis for empirical research into the decision-making processes of economic agents, particularly air travelers It then reviews various empirical studies that have examined passenger preferences among different airlines Building on these findings, a model is developed to analyze the choices of air travelers among three key airlines: Vietnam Airlines, Vietjet, and Jetstar.
Theoretical review
Random Utility Model is commonly used to represent individual choice behavior Thurstone
In 1927, a law of comparative judgment was introduced, laying the groundwork for the binary probit model, which assesses respondents' reactions to varying levels of psychological stimuli Marschak (1960) expanded on this concept by developing the notion of utility within the random utility model, highlighting that while decision-makers may understand the utility of each alternative, researchers often do not have complete knowledge This necessitates accounting for uncertainty, which divides the utility model into deterministic and random components The deterministic components are observable and interpretable by analysts, whereas the random components remain unknown Manski (1977) identified four primary sources of uncertainty: measurement errors, the use of proxy variables, unobserved attributes of the chooser, and unobserved attributes of the alternatives.
Discrete choice models, grounded in random utility theory, operate under the premise that decision-makers select from a finite set of mutually exclusive and collectively exhaustive alternatives to maximize their utility Each alternative's utility is influenced by deterministic factors, often represented as a linear function of various attributes The probability of an individual selecting a particular alternative is derived from the choice model, which incorporates random components as essential elements Variations in the assumptions regarding the distribution of error terms lead to different types of choice models, with key examples including logit, generalized extreme value (GEV), probit, and mixed logit models, as noted by Train (2009).
The logit model assumes that error terms are independent and identically distributed (iid), meaning that unobserved factors are uncorrelated and share the same variance across alternatives While this assumption provides a convenient form for calculating choice probabilities, it can be restrictive and may not be appropriate in all situations Additionally, the model operates under the independence assumption, indicating that each choice is made without regard to others Due to its ease of use, the logit model is widely employed by researchers to analyze various aspects of air choice behavior (Escobari & Mellado, 2014; Warburg, 2005; Yoo & Ashford, 1996).
To address the assumption of independence in logit models, generalized extreme value (GEV) models were developed, allowing for a broader distribution of extreme values (Train, 2009) This generalization enables the incorporation of unobserved factors related to alternatives, functioning as a specific case of the logit model when such correlations are absent The flexibility of these correlations varies depending on the type of GEV model used For example, a simpler GEV model organizes alternatives into groups, or nests, where unobserved factors share correlations within the same nest but not across different nests Researchers like Hess (2008) and Pels et al (2001) have applied the nested logit model to analyze air travel behavior and passenger preferences regarding airports and airlines, respectively.
The probit model addresses three key limitations of the logit model, as highlighted by Train (2009) These limitations include the inability to represent random taste variation, the independence of irrelevant alternatives (IIA) property, and the correlation between unobserved components and alternatives In contrast, the probit model assumes that error terms follow a normal distribution, making it a more flexible option However, a notable drawback of the probit model is that in certain situations, unobserved factors may not conform to this normal distribution.
Mixed logit models allow for unobserved factors to follow any distribution, incorporating both heteroskedasticity and correlation, while the other component is independently and identically distributed (iid) extreme value Notably, the first component can adhere to any distribution, including non-normal distributions Adler et al (2005) utilized the mixed logit model to create an itinerary choice model, while Warburg (2005) employed both multinomial logit and mixed logit models to analyze passenger flight choice behavior.
Researchers have developed various discrete choice models tailored for specific applications, often integrating concepts from other models For instance, a mixed probit model can be derived by analyzing observed components similarly to mixed logit, but with a normal distribution instead of an extreme value distribution Understanding the motivations and derivations of these models enables researchers to select the most appropriate model for their specific research objectives Additionally, both revealed preference and stated preference surveys play a crucial role in this process.
Surveys analyzing customer behavior primarily fall into two categories: revealed preference (RP) and stated preference (SP) surveys RP data reflects actual consumer choices in real-world settings, offering the advantage of genuine behavioral insights However, it poses challenges for trade-off analysis and is inadequate for modeling new market alternatives According to Yoo and Ashford (1996), RP data has three main limitations: insufficient variation in key variables for statistical calibration, difficulties in estimating trade-off ratios due to variable correlations, and the need for large sample sizes to gather enough observations Consequently, this method is less frequently employed in customer choice modeling For instance, Carrier (2008) utilized RP data from booking records, excluding non-booked travel options, while Escobari and Mellado (2014) gathered data from online travel agencies, focusing on posted prices and inventory changes to analyze flight demand.
The use of Stated Preference (SP) surveys offers significant advantages over Revealed Preference (RP) data by allowing researchers to understand hypothetical consumer responses and willingness to pay (Collins et al., 2012) SP surveys can also explore consumer choice behavior regarding non-existent alternatives However, they have limitations, such as potential respondent disinterest or bias in decision-making, as individuals may not act as they state (Warburg, 2006) Despite these challenges, SP surveys are widely employed for modeling choice behavior, with studies by Adler et al (2005) analyzing trade-offs in air itinerary choices and Collins et al (2012) investigating air traveler behavior through interactive stated choice surveys Additionally, research by Wen and Lai (2010) and Proussaloglou and Koppelman (1999) further demonstrates the application of SP data in examining passenger choices among air carriers.
To effectively address the limitations of revealed preference (RP) and stated preference (SP) data, it is essential to develop estimation techniques that integrate both data sources Utilizing both methods is recommended, as RP is valuable for forecasting demand and realistic applications, while SP serves a critical role in system planning.
In their 2012 study, Atasoy and Bierlaire utilized a mixed dataset of revealed preference (RP) and stated preference (SP) to enhance the model of itinerary choice This combination of data allowed for a more accurate estimation of price elasticity within the demand model.
Empirical review
Numerous studies have explored various facets of airline choice behavior, with notable research by Basar and Bhat (2004), Hess and Polak (2005), and Pathomsiri and Haghani (2005) focusing on airport selection in multi-airport regions Additionally, some papers delve into broader travel aspects, such as Ndoh et al (1990), who investigate both airport and route choices, and Furiuchi and Koppelman (1994), who analyze passengers' destination and airport selection Furthermore, research by Chin (2002), Algers and Beser (2001), Proussaloglou and Koppelman (1999), and Yoo and Ashford (1996) emphasizes air traveler choices beyond just airport selection.
The multinomial logit model is commonly used in studies of air travel choices, while other research, including works by Ndoh et al (1990), Furiuchi and Koppelman (1994), and Pels et al (2001), employs the nested logit model to analyze the complex and spatial decision-making of air travelers In contrast, studies that address behavioral aspects or effects of air travel choices often utilize the mixed multinomial logit model, as demonstrated by Hess & Polak (2005) and Pathomsiri & Haghani (2005).
Moreno (2006) employs the multinomial logit model to analyze airline choice for domestic flights in São Paulo, based on a survey of 1,923 passengers at São Paulo-Guarulhos International Airport (GRU) and São Paulo-Congonhas Airport (CGH) The study posits that airline selection is influenced by a tradeoff among flight cost, frequency, and airline performance It evaluates three variable types: cost variables (lowest and highest fares), flight frequency variables (including connections, travel periods, and days of the week), and airline age as a performance proxy Findings indicate that the lowest fare is the most significant factor in airline choice, with older passengers prioritizing airline age more than younger ones Similarly, Nason (1981) conducts a stated preference survey, utilizing the multinomial logit model to explore airline choice based on service attributes and passenger characteristics.
Prossaloglou and Koppelman (1995) utilize a revealed preference survey to analyze airline choices among passengers departing from Dallas and Chicago, identifying schedule convenience, reliability, fares, city pair presence, market presence, and frequent flyer program membership as key independent variables in their multinomial logit model Their findings indicate a positive correlation between the attractiveness of carriers and market share with frequent flyer programs Similarly, Nako (1992) concludes that frequent flyer programs significantly enhance airline demand among business travelers In a later study, Prossaloglou and Koppelman (1999) employ a logit model to assess passengers' choices regarding airlines, flights, and fare classes, positing that air travelers act as rational decision-makers seeking the highest utility Their research incorporates variables such as fare class, fare price, carrier market presence, service quality, frequent flyer participation, and flight schedules, while employing distinct models for business and leisure travelers based on stated preference data collected through a two-tier survey process The results reveal notable behavioral differences, with leisure travelers exhibiting greater price sensitivity and less time sensitivity compared to business travelers, who prioritize frequent flyer programs and are willing to pay more for preferred airlines.
Pels et al (2001) employ distinct models for business and leisure travelers, revealing minimal differences between the two groups Utilizing the nested logit model, the authors analyze passenger preferences regarding airports and airlines.
This research thoroughly examines nests defined by both airports and airlines, utilizing survey data from the San Francisco Bay Area for an empirical analysis.
In 1995, it was noted that airlines face two categories of competitors: those operating at the same airport and those at different airports This distinction is crucial, as access time to the airport significantly impacts both leisure and business travelers.
Understanding passengers' flight choice behavior is crucial for predicting air travel demand, as highlighted by Warburg (2005) This research aids carriers in developing effective pricing strategies and forecasting demand on new routes In a 2001 stated preference survey involving 119 business and 521 non-business passengers, respondents detailed their most recent domestic flights, providing valuable insights into travel preferences.
The study explores 10 binary choices between actual flights and hypothetical flights, presenting 10 itinerary alternatives with identical departure and arrival locations Warburg (2005) argues that a universal choice set does not exist, highlighting that travelers have varying flight itineraries, leading to unique choice sets for each passenger To analyze the behavior of two groups, the research utilizes both the multinomial logit model and the mixed logit model.
According to Koppelman (1999), the multinomial logit model reveals that business travelers prioritize time sensitivity, whereas leisure travelers focus more on fare costs Additionally, the analysis indicates that men exhibit a greater sensitivity to fare compared to women.
The study by Yoo and Ashford (1996) examines the flight choice behavior of Koreans who have taken long-distance international flights exceeding 10 hours Utilizing a logit model for both revealed preference (RP) and stated preference (SP) data, the researchers conducted surveys at Kimpo International Airport in Seoul, with RP data collected in October 1993 and SP data in August 1994, ensuring equal sample sizes for both datasets The findings indicate that passengers are willing to pay more for Korean airlines compared to foreign carriers, with Korean residents exhibiting a higher willingness to pay than foreign residents Similarly, Escobari and Mellado (2014) analyze international flight demand using a unique dataset that includes flight choices, prices, and characteristics of non-booked flights, sourced from an online travel agency.
Expedia.com offers 317 one-way, non-stop flights between New York and Toronto from December 19 to 24, 2008, operated by six carriers The study highlights that a 10% increase in ticket prices for a 100-seat aircraft leads to a decrease in demand by 7.7 seats, indicating a significant price sensitivity among travelers during this period.
Ukpere et al (2012) explore the factors influencing airline choice in Nigeria's domestic air transport through a revealed preference survey Using a Likert scale questionnaire, they gather data on socio-economic characteristics such as sex, age, and marital status, alongside airline attributes like comfort, on-board service, fare, frequency, crew behavior, and monopoly power Their findings, analyzed with a nested logit model, reveal that these variables significantly impact passengers' airline selections at the selected airports, leading to recommendations for airlines to adopt competitive pricing and differentiate their services In contrast, Adler et al (2005) conduct a stated preference survey online, involving 600 individuals who recently purchased domestic air tickets, to examine the trade-offs in itinerary choices Key itinerary characteristics assessed include airline carrier, airport, fare, flight times, on-time performance, and the discrepancy between expected and actual arrival times Their analysis using a mixed multinomial logit model indicates that these service characteristics significantly influence decision-making; however, the study does not account for demographic and trip-related factors affecting individual choices.
This chapter presents the research methods, including the identification of the airlines attributes that may affect traveler’s choice, the data collection methods, and the analytical model.
Stated preference method
The stated preference (SP) method, as highlighted by Whitaker et al (2005), is a crucial tool for understanding decision-making behavior and forecasting demand for various airline services While traditional demand analysis relies on revealed preference (RP) data, which reflects actual choices made in real-world scenarios, this approach has notable limitations Primarily, RP data collection can be costly, and it does not accommodate new alternatives that may emerge in the market, restricting its applicability for future developments.
Wen and Lai (2010) argue that while revealed preference methods gather data from actual choices, they may not accurately reflect passenger decision-making, as travelers often overlook various attributes of airlines In contrast, the stated choice approach effectively evaluates how individuals would react to hypothetical scenarios involving different alternatives and their attributes This methodology has gained traction in analyzing airline selection and other decision-making issues.
It is said that in research of travel behavior, there are two types of stated response (Hensher,
In survey research, respondents first identify their preferences among alternatives using various scales, such as rating scales and rank ordering scales Rating scales, including Likert and 1 to 10 scales, capture both quantitative and qualitative attributes, while rank ordering scales allow individuals to express their degree of preference by ranking alternatives Studies by Warburg et al (2006) and Adler et al (2005) exemplify the use of rank ordering scales in airline choice surveys Additionally, respondents are asked to select their top choice from the listed alternatives, known as the first preference choice task Understanding response strategies at the outset of a stated preference (SP) survey is crucial, as it influences the resulting data.
This study employs a first preference choice task in its survey, where respondents select from three airlines—Vietnam Airlines (VNA), Vietjet Air (VJ), and Jetstar (BL)—based on airfares for a specific route Hensher (1994) highlights the advantage of stated preference (SP) data, as it mirrors revealed preferences by reflecting how individuals make choices after evaluating various alternatives This methodology has been utilized by researchers like Wen & Lai (2010) and Hong (2010) For instance, Wen & Lai (2010) presented air travelers with a choice set of three carriers for the Taipei-Tokyo route and four for the Taipei-Hong Kong route Similarly, Hong (2010) conducted an SP survey where respondents chose among British Airways, Air France, and Easyjet.
Questionnaire and survey process
The survey questionnaire, detailed in the Appendix, is divided into three sections The first section gathers demographic information and the primary purpose of the trip The second part asks respondents to assess the quality of airline services, including staff attitudes at check-in, flight attendants, in-flight food and drink, seat space, and on-time performance, with an option for those without prior experience to indicate "I have never used this service before." Lastly, the survey presents fifteen hypothetical scenarios involving routes from Tan Son Nhat Airport to various domestic airports, as shown in Table 3.1 Respondents are asked to select an airline based on the listed airfare for each route, disclose their intended purpose of travel, and indicate the maximum ticket price they would be willing to pay If they believe they will never visit a particular destination, they can select "I will never go there" and skip to the next scenario.
Table 3.1 Summary of hypothetical scenarios in survey:
Route From Sai Gon To Operation of airline
Note: x: having at least one flight in a day
A pilot test was conducted at an air ticket agency to identify the key factors influencing customer decisions when purchasing air tickets Eighteen customers who had recently bought tickets were surveyed to list the factors affecting their choices, which included fare, schedules, on-time performance, quality of staff service, and onboard seat comfort Following this, an online survey was conducted from October 16 to 23, 2016, using SurveyMonkey to design the questionnaire The survey link was distributed via email and shared on social media platforms like Facebook and Zalo, targeting air travelers who had previously flown with at least one of the airlines: VNA, VJ, or BL It is important to note that due to customer loyalty, many respondents may have only flown with their preferred airline, making it challenging to find individuals who have experience with all three airlines.
Figure 3.1 The screen of the online survey
Attributes of airlines
Customer satisfaction is a crucial factor in the service industry, significantly influencing customer retention According to Fornell et al (1994), higher customer satisfaction leads to fewer complaints, which in turn reduces the costs associated with addressing service failures Zeithaml and Bitner (1996) identify several elements that contribute to customer satisfaction, including pricing, terms and conditions, product and service quality, and personal characteristics While service quality is vital for customer loyalty, Lee and Cunningham (1996) note that customers often weigh the trade-offs between costs and benefits, highlighting the importance of price Athanassopoulos et al (2001) further suggest that customer behavior can manifest in three ways in response to satisfaction: remaining with current providers, engaging in word-of-mouth, or switching providers In the airline industry, factors such as price, service quality, schedule times, flight frequency, aircraft types, and seating capacity also play significant roles in influencing airline selection, as detailed in various research studies.
Price Cost of a route (return fare) Continuos data Warburg (2005)
Cost of a route (one-way fare) AUD1600,
Average fare for each route Higher price;
Fare of the chosen flight Continuos data Adler et al (2005) Frequency of airline Number of flights/route/day Wen & Lai (2010)
Number of direct flights in the travel day
Flights per day Pereira et al (2007)
Number of flights per week Yoo & Ashford
Percentage of on time flight itinerary 50%, 60%, 70%,
Percentage of on time flight itinerary 50%-99% Adler et al (2005)
On time service schedules Sometimes delay,
Seat space on board Seat pitch 31", 32", 34" Collins & Hess
(2012) Passenger's evaluation of seat Very uncomfortable
Comfort No; yes Pereira et al (2007)
Check in service Passenger's evaluation of check in service
Very uncomfortable Comfortable enough Very comfortable
Kindness of employees Not very polite and friendly Very polite and friendly
Model specification
This study utilizes the Random Utility Model proposed by Manski (1977), positing that air travelers act rationally to maximize their utility Passengers are inclined to choose the airline that offers the greatest utility, which can be expressed in a specific mathematical form.
𝑈 𝑖𝑛 = 𝑉 𝑖𝑛 + 𝜀 𝑖𝑛 = 𝛼 𝑛 + 𝛽 𝑛 𝑋 𝑖 + 𝜀 𝑖𝑛 Where U: Utility level of passenger
V: Portion of utility (observed utility), and 𝑉 = 𝛼 + 𝛽𝑋
𝜀 : Error terms (unobserved utility) X: vector of explanatory variables i : Passenger i n = 1, 2, 3 denoted for Vietnam Airline, Vietjet, and Jetstar, respectively
It is reasonable to assume that the actual of choosing airline n is 𝑌 𝑖𝑛 , so:
𝑌 𝑖𝑛 = 1, if 𝑈 𝑖𝑛 is maximum or 𝑈 𝑖𝑛 > 𝑈 𝑖𝑚 (m = 1, 2, 3, and m ≠ n)
Let 𝑃 𝑖𝑛 = Pr( 𝑌 𝑖𝑛 = 1) be the probability of choosing airline n, and probability of individual i choosing carrier n is calculated as below:
An air traveler selects airline n when the condition 𝜀 𝑖𝑚 < 𝑉 𝑖𝑛 − 𝑉 𝑖𝑚 + 𝜀 𝑖𝑛 is met This indicates that the distribution of error 𝜀 𝑖 allows for probability estimation The multinomial logit model relies on the assumption that the error terms follow an identical and independent distribution of extreme values.
(Gumbel distribution) According to Train (2009), the function of probability could be rewritten as below:
In a multinomial logit model, the total probability of selecting three airlines by an individual equals 1, but identifying each probability independently is not feasible (Gujarati, 2011) To simplify this, one airline is typically designated as the reference or base choice, with its coefficients set to zero For instance, if the third airline is chosen as the base (α3 = 0 and β3 = 0), the probabilities for the three airlines can then be calculated accordingly.
The ratio of probability of choosing airline 1 and 2 over probability of choosing airline 3 (the base) is known as the odds ratios:
Taking the natural log of (*) and (**), the log of the odds ratios are called the multinomial logit model, which have forms:
This study examines the relationship between various independent and controlling variables and carrier choice, with X representing the vector of these variables The independent variables include price, flight frequency, and routes, which are hypothesized in the survey's third section to influence respondents' decisions Additionally, controlling variables, such as respondent demographics and evaluations of airline service, are gathered from the first two sections of the questionnaire A comprehensive list of the variables utilized in this research is provided in Table 3.4.
In November, flight prices and frequencies were analyzed based on real-time data from airlines, with the survey focusing on the average price for each route offered by different carriers.
2016 whereas the frequency of flights is the actual number of flights that each carrier has in a day Table 3.3 is the summary of value of independent variables
Table 3.3 Prices and numbers of flights by routes of carriers
Route From Sai Gon To Price (100,000 VND) Number of flights
Vietnam Airline Vietjet Jetstar Vietnam Airline Vietjet Jetstar
Type of variable Variables Denotation Unit Description
3 = Jetstar Independent variables Price pricevn 100.000 VND Airfare of VNA pricevj 100.000 VND Airfare of VJ pricebl 100.000 VND Airfare of BL
The frequency of airlines plays a crucial role in determining the number of daily flights for specific routes For Vietnam Airlines (VNA), the daily flight frequency is denoted as freqvn, while VietJet Air (VJ) is represented by freqvj Additionally, Bamboo Airways (BL) has its own daily flight frequency indicated as freqbl Understanding these metrics is essential for travelers seeking optimal flight options.
Marital status single (Dummy) 1 = Single Education schoolyear Years Number of schooling years Income income Million VND Average income per month Occupation job_emp (Dummy) 1 = company employees
The on-time performance of various airlines shows that VNA's previous flights departed punctually, while VJ and BL also demonstrated timely departures in their recent operations.
Type of variable Variables Denotation Unit Description
Seat space seavn_ufr (Dummy) 1 = Seat space of VNA is uncomfortable seavj_ufr (Dummy) 1= Seat space of VJ is uncomfortable seabl_ufr (Dummy) 1= Seat space of BL is uncomfortable
The check-in service experiences across various airlines indicate a common issue with staff demeanor Specifically, the VNA, VJ, and BL airlines have received feedback highlighting that their check-in counter staff are perceived as unfriendly This negative interaction at the check-in process can impact overall customer satisfaction and should be addressed to enhance the travel experience.
This chapter presents the summary of data collected form the stated preference survey as well as the regression results of multinomial logit model.
Data
Table 4.1 outlines the characteristics of social demography, revealing that out of 135 respondents, 13 did not complete the survey and were excluded from the data Each participant evaluated 15 scenarios to select an airline, with Figure 4.1 illustrating airline preferences for various destinations Notably, Ha Noi, Da Nang, and Phu Quoc emerged as the most sought-after travel locations, with over 98% of respondents indicating plans to visit Vietnam Airlines (VNA) recorded the lowest selection rate across all routes, with no choices for the Sai Gon – Tuy Hoa and Sai Gon – Chu Lai routes due to the absence of flights Consequently, VNA was excluded from these scenarios In contrast, VietJet Air (VJ) dominated airline preferences in most scenarios, particularly for the Sai Gon to Ha Noi and Da Nang routes, where over 60% of respondents favored VJ However, for routes like Da Lat and Hue, Bamboo Airways (BL) was the preferred choice among respondents.
Figure 4.1 Airline Choice for Destinations
(Notes: HAN: Ha Noi, DAD: Da Nang, VII: Vinh, CXR: Nha Trang, DII: Da Lat, HUI: Hue, THD: Thanh Hoa, BMV: Buon Me Thuot, PXU:
Pleiku, PQC: Phu Quoc, HPH: Hai Phong, TBB: Phu Yen, UIH: Quy Nhon, VDH: Dong Hoi, VCL: Chu Lai)
The survey, which included 122 respondents, comprised 84 females (68.85%) and 38 males (31.15%), with a majority being single (approximately 70%) Conducted online, the respondents are predominantly young, with an average age of 27; the youngest is 20 and the oldest is 46, although two individuals chose not to disclose their age Educationally, 83 respondents (68%) hold a university degree, while 32 (26.23%) possess a master's degree, and only one individual completed high school Employment data reveals that 83 respondents (about 68%) are company employees, whereas only one is self-employed Regarding travel preferences, over 40% of respondents indicated they typically travel by air for both leisure and business purposes Additionally, most respondents reported a monthly income ranging from 8 to 10 million.
DADHANPQCCXR DII HUI BMV UIH VII TBB VCL PXU HPHVDH THD
Airline Choice for Destinations - Depart from Sai Gon
The survey assesses airline service quality, focusing on check-in services, cabin crew attitudes, and onboard food and drink, rated on a scale of not good, good enough, and very good (Figures 4.4 to 4.7) Additionally, respondents evaluate flight punctuality by reflecting on their experiences with three airlines (Figure 4.8) and their acceptance of airlines' on-time performance (Figure 4.9) For those who have not utilized any airline services, an option is provided to indicate their lack of experience.
A survey revealed that approximately 36% of respondents have never flown with Jetstar Pacific, the highest rate among three airlines In contrast, Vietnam Airlines received positive feedback, with most passengers rating its service quality as good or very good Notably, only 1.64% and 0.82% of respondents found the check-in and cabin crew service unfriendly, respectively Regarding punctuality, around 60% of respondents agreed that Vietnam Airlines flights depart on time, and approximately 74% considered its rate of delayed flights acceptable In comparison, Vietjet Air had the highest rate of unpunctual flights at 62%, with over 30% of passengers deeming its schedule changes unacceptable.
The survey asks respondents to indicate their willingness to pay for air tickets on various routes, such as from Sai Gon to Ha Noi Figure 4.8 illustrates the average prices respondents are prepared to pay for these flights.
Respondents' income levels in VND per month indicate a willingness to pay more for long-distance trips, such as those from Ho Chi Minh City to Hanoi and Hai Phong, while showing a preference for lower fares on shorter journeys, like the route from Ho Chi Minh City to Nha Trang.
Figure 4.3 Willingness to pay for routes
Bài viết này đề cập đến các thành phố và địa điểm du lịch nổi bật tại Việt Nam, bao gồm Hà Nội (HAN), Đà Nẵng (DAD), Vinh (VII), Nha Trang (CXR), Đà Lạt (DII), Huế (HUI), Thanh Hóa (THD), Buôn Ma Thuột (BMV), Pleiku (PXU), Phú Quốc (PQC), Hải Phòng (HPH), Phú Yên (TBB), Quy Nhơn (UIH), Đồng Hới (VDH) và Chu Lai (VCL) Những địa điểm này không chỉ thu hút khách du lịch bởi vẻ đẹp tự nhiên mà còn mang đậm bản sắc văn hóa và lịch sử của Việt Nam.
HAN DAD VII CXR DII HUI THD BMV PXU PQC HPH TBB UIH VDH VCL
M e an o f Wi lli n gn e ss to p ay ( 1000 VN D )
Notes: 122 respondents, in which 2 not reveal their age
Demographic Characteristics Number of respondents Percentages (%)
Both of Leisure and Business 53 43.44
Figure 4.4 Check-In Service Evaluation
Figure 4.5 Cabin Crew Service Evaluation
Figure 4.6 Food & Drink Onboard Evaluation
Figure 4.7 Inflight Seat Space Evaluation
Figure 4.8 On-time Performance Evaluation
Empirical results
According to Gujarati (2011), odds ratios quantify the preference between choices by comparing the probability of selecting one alternative over another A positive coefficient indicates that an increase in the variable enhances the likelihood of choosing option i over option j, thereby increasing the decision maker's utility Conversely, a negative coefficient suggests that an increase in the variable decreases the odds of choosing option i in favor of option j, indicating a preference for j over i.
In the multinomial logit model, relative risk ratios (RRR) are derived by exponentiating the multinomial logit coefficients (e^coef) These ratios indicate that a unit change in an independent variable is expected to alter the relative risk of outcome j compared to the base outcome by a factor corresponding to that parameter.
Table 4.2 presents the outcomes of the multinomial logit models, with Model 1 focusing solely on controlling variables Model 2 incorporates two airline attributes: price and frequency, while Model 3 further includes routes to analyze their impacts In these models, Jetstar Pacific (choice 3) serves as the reference point, with Vietnam Airlines and Vietjet Air represented as choice 1 and choice 2, respectively.
The study involved 122 respondents tasked with making decisions across 15 scenarios, resulting in a potential total of 1,830 observations However, if respondents indicated a lack of demand for travel from Sai Gon to a specific destination, they were not prompted to select an airline for that scenario Consequently, the analysis yielded 605 usable observations for the regression analysis, with detailed results available in the Appendix.
Dependent variable: choice Coef rrr P>|z| Coef rrr P>|z| Coef rrr P>|z|
Number of flight of VNA 1.036*** 2.817 0.000
Number of flight of VJ -0.274** 0.760 0.034
Number of flight of BL 0.307*** 1.360 0.008
On time Performance of VNA (Punctuality = 1) -0.408 0.665 0.250 -0.471 0.624 0.233 -0.451 0.637 0.252
On time Performance of VJ (Punctuality = 1) -0.121 0.886 0.812 -0.507 0.602 0.375 -0.460 0.631 0.418
On time Performance of BL (Punctuality = 1) 0.738 2.092 0.111 1.229** 3.419 0.017 1.239** 3.451 0.017
Seat space of VNA (Uncomfortable =1) -0.748* 0.473 0.054 -0.992** 0.371 0.021 -0.970** 0.379 0.023
Seat space of VJ (Uncomfortable =1) -3.067* 0.047 0.052 -5.055*** 0.006 0.006 -4.224** 0.015 0.016
Seat space of BL (Uncomfortable =1) -0.562 0.570 0.369 -0.706 0.494 0.315 -1.079 0.340 0.120
Check in Service of VNA (Unfriendly =1) 1.747*** 5.737 0.000 2.463*** 11.738 0.000 2.443*** 11.508 0.000
Check in Service of VJ (Unfriendly =1) 14.854 2,824,890 0.983 14.151 1,398,023 0.975 15.727 6,764,777 0.987
Check in Service of BL (Unfriendly =1) -0.673 0.510 0.296 -0.770 0.463 0.291 -0.662 0.516 0.353
Dependent variable: choice Coef rrr P>|z| Coef rrr P>|z| Coef rrr P>|z|
Number of flight of VNA -0.375*** 0.687 0.001
Number of flight of VJ 0.309*** 1.362 0.002
Number of flight of BL 0.139 1.150 0.305
On time Performance of VNA (Punctuality = 1) -0.750** 0.472 0.043 -1.140** 0.320 0.015 -1.078** 0.340 0.016
On time Performance of VJ (Punctuality = 1) 1.068*** 2.909 0.002 1.646*** 5.186 0.000 1.678*** 5.357 0.000
On time Performance of BL (Punctuality = 1) -0.588* 0.555 0.052 -0.823** 0.439 0.028 -0.894** 0.409 0.013
Seat space of VNA (Uncomfortable =1) -5.450*** 0.004 0.000 -7.594*** 0.001 0.000 -7.361*** 0.001 0.000
Seat space of VJ (Uncomfortable =1) -1.659*** 0.190 0.001 -1.751*** 0.174 0.005 -2.230*** 0.108 0.000
Dependent variable: choice Coef rrr P>|z| Coef rrr P>|z| Coef rrr P>|z|
Seat space of BL (Uncomfortable =1) 2.307*** 10.045 0.000 3.148*** 23.295 0.000 3.161*** 23.602 0.000
Check in Service of VNA (Unfriendly =1) 14.356*** 1,717,379 0.983 14.097 1,325,579 0.976 15.813 7,371,118 0.987
Check in Service of VJ (Unfriendly =1) -0.463 0.629 0.397 -0.726 0.484 0.281 -0.572 0.564 0.361
Check in Service of BL (Unfriendly =1) -0.533 0.587 0.374 -0.906 0.404 0.192 -0.833 0.435 0.221
Choice = 3: Jetstar Pacific (BL) - Base outcome
This research explores the relationship between airline choice and passenger characteristics, revealing consistent results across three models The study indicates that individual characteristics are statistically significant at the 10% level, with the exception of career Notably, male passengers and higher income levels positively influence the likelihood of choosing certain airlines, while age, marital status, and education level have negative effects Specifically, females are less likely to choose VNA or VJ over BL compared to males, while married individuals show a greater preference for VNA over BL when compared to singles, holding other factors constant.
Higher income levels are associated with a preference for VNA and VJ airlines over BL, as indicated by the positive coefficient of the income factor As illustrated in Figure 4.10, an increase in income correlates with a greater likelihood of choosing VNA and VJ, while the probability of selecting BL decreases.
Figure 4.10 Predicted probability of airline choice and income
Pre d ict e d p ro b a b ili ty
Pr(VNA) Pr(VJ) Pr(BL)
The analysis indicates that as age and years of schooling increase, the likelihood of choosing VNA or VJ over BL decreases, suggesting that older individuals or those with more education are more inclined to select BL.
Figure 4.11 Predicted probability of airline choice and age
Figure 4.12 Predicted probability of airline choice and school year
Pre d ict e d p ro b a b ili ty
Pr(VNA) Pr(VJ) Pr(BL)
Pre d ict e d p ro b a b il it y
Pr(VNA) Pr(VJ) Pr(BL)
Figures 4.11 and 4.12 illustrate the correlation between the predicted probability of airline choice and factors such as age and school year As age and school year increase, the likelihood of selecting VJ and BL airlines also rises, in contrast to VNA.
On-time performance significantly influences the utility of airline alternatives, particularly for VJ and BL, as indicated by a 5% level of significance The positive coefficient for VJ's punctuality suggests that previous delays with VJ decrease the likelihood of passengers choosing it over BL Consequently, after experiencing poor service with VJ, passengers are more inclined to prefer BL.
Passengers show a preference for VJ over BL when they have experienced delays with VNA or BL, highlighting the negative impact of unpunctual flights Additionally, the on-time performance of airlines is notably significant at a 10% level when comparing VNA and BL, with the exception of VNA's performance in three specific models.
VJ in model 1 It may be understood that performance of VNA does not affect on the choice of VNA
The discomfort caused by limited in-flight seat space is likely to adversely affect passengers' preferences for certain airlines If travelers find the seating cramped or uncomfortable, they may choose to avoid that airline in the future This factor has a statistically significant impact at the 10% level, with the exception of the uncomfortable seat variable for VJ in the first panel of Table 4.2 When comparing VJ to BL, the negative coefficients for uncomfortable seat space indicate that passengers are more inclined to select VNA or VJ over BL if they perceive the seating on VNA or VJ as comfortable Thus, a comfortable seating experience enhances the likelihood of choosing VNA or VJ over BL.
This study investigates the connection between the quality of check-in services and the perceived utility of alternative airlines It is anticipated that poor service will negatively influence customer choices; however, all related variables are found to be statistically insignificant at the 10% level Consequently, the evaluation of check-in services by respondents does not significantly affect their airline selection decisions.
Model 2 employs the multinomial logit model to analyze airline attributes and control variables, revealing that airline prices are statistically significant at the 1% level, except for Vietjet in Panel 1 of Table 4.2 The negative coefficient for Vietnam Airline’s price suggests that an increase in its fares decreases the odds of choosing VNA compared to BL, as passengers tend to favor Jetstar when VNA raises its prices Similarly, Panel 2 indicates that a rise in Vietjet's airfare reduces the odds of selecting VJ over BL under the same conditions In contrast, Jetstar's positive price coefficients imply that higher BL ticket prices increase the likelihood of choosing either VNA or VJ over BL, assuming other factors remain constant Additionally, the negative coefficient for VNA's price highlights that if VNA’s fares rise, the odds of choosing VJ over BL decrease, leading passengers to prefer BL instead of VJ.
The frequency of flights, which refers to the number of daily flights on a specific route offered by each airline, plays a crucial role in passenger preference Generally, airlines with more flights are favored, as this provides passengers with greater options In a comparison between VJ and BL, VJ shows a positive correlation with the number of flights, indicating a preference for it Conversely, VNA exhibits a negative and significant relationship, suggesting that an increase in VNA's daily flights decreases the likelihood of choosing VJ over BL Similar results are observed when comparing VNA with BL.
Figures 4.13, 4.14, and 4.15 illustrate the relationship between airline pricing and the likelihood of selection As prices rise, the probability of choosing a particular airline declines Specifically, when Vietnam Airlines (VNA) raises its prices, the likelihood of selecting both VNA and Bamboo Airways (BL) diminishes, while the probability of choosing VietJet Air (VJ) increases This shift occurs because consumers tend to prefer purchasing tickets from VJ over BL when VNA's fares are elevated.
Model 3 is the multinomial logit model of controlling variables and categorical variable (routes)