consum er willingness to pay for in-flight service and comfort levels: a choice experim ent

6
Consumer willingness to pay for in-flight service and comfort levels: A choice experiment Kelvin Balcombe a , Iain Fraser b, * , Liam Harris c a University of Reading, Reading, UK b Department of Economics, University of Kent, Wye Campus, Wye, Kent TN25 5AH, UK c Imperial College, London, UK Keywords: Charter airlines Choice experiment Mixed logit abstract In the competitive aviation market as a result of the emergence of low cost carriers, charter airlines have had to reconsider their approach to service provision. Specifically, the reduction in service and comfort levels offered by the low cost airlines provides charter carriers with an opportunity to differentiate their product based on the quality of the offering. To consider this strategic option we employ an on-line choice experiment to examine consumer choices with respect to the bundle of services on offer when deciding to purchase a flight. With these data we use the Bayesian methods to estimate a mixed logit specification. Our results reveal that in principle passengers are willing to pay a relatively large amount for enhanced service quality. Ó 2008 Elsevier Ltd. All rights reserved. 1. Introduction The competitive landscape of the European aviation industry has changed significantly since deregulation in 1992 as a result of price pressures due to fierce competition from the growth of low cost carriers (LCC). The success of LCCs results from an uncom- promising ‘‘no-frills strategy’’ that minimizes costs per by only offering a modicum of comfort in terms of legroom, seat dimen- sions, not installing window blinds, headrests, seat-pockets or seat reclining mechanisms (Franke, 2003). Furthermore, they charge passengers to check baggage into the hold, which simplifies ground handling, airport operations and turn-around times, plus food, drinks and in-flight entertainment provision are supplied on a ‘‘buy-on-board’’ basis (Barrett, 2003). The associated ancillary revenues form a significant part of the business model; for Ryanair (2006) 15% of revenues in 2006. However, because the LCC business model is based on maxi- mizing utilization of aircraft (two return journeys per day); to keep them in the air for the longest possible time, flight range is limited to 3 h. Thus, UK based LCC operations only fly to European desti- nations (Francis et al., 2006). In response, existing carriers, espe- cially charter airlines (CAs) have engaged in significant fleet redeployment outside of the 3-h range. For example, operations to Spain and the Balearic Islands have been reduced, with destinations such as Turkey, northern Africa and the Canary Islands being given greater emphasis (Bieger and Wittmer, 2006). 1 However, the extensive cost-cutting of LCCs presents an opportunity for CAs to differentiate away from the ‘‘commodity’’ travel product on offer (Dennis, 2007). We examine how CAs might differentiate their products. Specifically, we consider which attributes of in-flight cabin comfort and service may have a value to consumers by analyzing consumer willingness to pay (WTP) for in-flight attributes that are derived from an Internet delivered choice experiment (CE). With a CE survey respondents choose between alternative options that contain a number of attributes of different levels. Our CE was designed to consider air traveler choice as it relates to on-board service and comfort levels for CAs for a flight of between 4 and 5 h. The issue of service provision and ticket price has been previ- ously examined, for example, by Lee and Luengo-Prado (2004) who considered how legroom impacted the price paid by consumers. For one airline passengers did appear willing to pay more for additional legroom, whereas for another they found the opposite. Espino et al. (2008) and Martin et al. (2008) used a CE undertaken in Spain in which passengers were presented six attributes (price, penalty for ticket changes, food (sandwich, hot meal or a la carte), comfort (i.e., legroom), frequency and reliability) relating to two virtual airlines * Corresponding author. E-mail address: [email protected] (I. Fraser). 1 Generally CA ownership is heavily concentrated with the largest players belonging to large vertically integrated tour groups. Vertical integration means that a large proportion of European CA seats are provided for inclusion in such tour packages (Bieger and Wittmer, 2006). Contents lists available at ScienceDirect Journal of Air Transport Management journal homepage: www.elsevier.com/locate/jairtraman 0969-6997/$ – see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.jairtraman.2008.12.005 Journal of Air Transport Management 15 (2009) 221–226

Upload: humayun-khalid

Post on 11-Sep-2015

215 views

Category:

Documents


2 download

DESCRIPTION

Consum er willingness to pay for in-flight service and comfort levels:A choice experim entConsum er willingness to pay for in-flight service and comfort levels:A choice experim ent

TRANSCRIPT

  • tmaproaost alityineht.

    1. Introduction

    The competitive landscape of the Eurogulatiotition fLCCs reminimrms o

    to 3 h. Thus, UK based LCC operations only y to European desti-nations (Francis et al., 2006). In response, existing carriers, espe-cially charter airlines (CAs) have engaged in signicant eetredeployment outside of the 3-h range. For example, operations toSpain and the Balearic Islands have been reduced, with destinations

    (2008) and Martin et al. (2008) used a CE undertaken in Spain inwhich passengers were presented six attributes (price, penalty forticket changes, food (sandwich, hot meal or a la carte), comfort (i.e.,legroom), frequency and reliability) relating to two virtual airlines

    * Corresponding author.

    1 Generally CA ownership is heavily concentrated with the largest playersbelonging to large vertically integrated tour groups. Vertical integration means thata large proportion of European CA seats are provided for inclusion in such tourpackages (Bieger and Wittmer, 2006).

    Contents lists available at ScienceDirect

    Journal of Air Transport Management

    lse

    Journal of Air Transport Management 15 (2009) 221226E-mail address: [email protected] (I. Fraser).sions, not installing window blinds, headrests, seat-pockets or seatreclining mechanisms (Franke, 2003). Furthermore, they chargepassengers to check baggage into the hold, which simplies groundhandling, airport operations and turn-around times, plus food,drinks and in-ight entertainment provision are supplied ona buy-on-board basis (Barrett, 2003). The associated ancillaryrevenues form a signicant part of the business model; for Ryanair(2006) 15% of revenues in 2006.

    However, because the LCC business model is based on maxi-mizing utilization of aircraft (two return journeys per day); to keepthem in the air for the longest possible time, ight range is limited

    willingness to pay (WTP) for in-ight attributes that are derivedfrom an Internet delivered choice experiment (CE). With a CEsurvey respondents choose between alternative options thatcontain a number of attributes of different levels. Our CE wasdesigned to consider air traveler choice as it relates to on-boardservice and comfort levels for CAs for a ight of between 4 and 5 h.

    The issue of service provision and ticket price has been previ-ously examined, for example, by Lee and Luengo-Prado (2004) whoconsidered how legroom impacted the price paid by consumers. Forone airline passengers did appear willing to paymore for additionallegroom, whereas for another they found the opposite. Espino et al.offering a modicum of comfort in tehas changed signicantly since dereprice pressures due to erce compecost carriers (LCC). The success ofpromising no-frills strategy that0969-6997/$ see front matter 2008 Elsevier Ltd.doi:10.1016/j.jairtraman.2008.12.005pean aviation industryn in 1992 as a result ofrom the growth of lowsults from an uncom-izes costs per by onlyf legroom, seat dimen-

    such as Turkey, northern Africa and the Canary Islands being givengreater emphasis (Bieger and Wittmer, 2006).1 However, theextensive cost-cutting of LCCs presents an opportunity for CAs todifferentiate away from the commodity travel product on offer(Dennis, 2007).

    We examine how CAs might differentiate their products.Specically, we consider which attributes of in-ight cabin comfortand service may have a value to consumers by analyzing consumer 2008 Elsevier Ltd. All rights reserved.specication. Our results reveal that in principle passengers are willing to pay a relatively large amountfor enhanced service quality.Consumer willingness to pay for in-ighA choice experiment

    Kelvin Balcombe a, Iain Fraser b,*, Liam Harris c

    aUniversity of Reading, Reading, UKbDepartment of Economics, University of Kent, Wye Campus, Wye, Kent TN25 5AH, UKc Imperial College, London, UK

    Keywords:Charter airlinesChoice experimentMixed logit

    a b s t r a c t

    In the competitive aviationhad to reconsider their aplevels offered by the low cproduct based on the quachoice experiment to examdeciding to purchase a ig

    journal homepage: www.eAll rights reserved.service and comfort levels:

    rket as a result of the emergence of low cost carriers, charter airlines havech to service provision. Specically, the reduction in service and comfortirlines provides charter carriers with an opportunity to differentiate theirof the offering. To consider this strategic option we employ an on-lineconsumer choices with respect to the bundle of services on offer whenWith these data we use the Bayesian methods to estimate a mixed logit

    vier .com/locate/ ja i r t raman

  • ying between Las Palmas de Gran Canaria and Madrid. Bothstudies found that the WTP vary depending on the type ofeconomic model estimated. In general the multinomial logit (MNL)

    preserves the normality assumption. Finally, the error term es,j,n isassumed to be extreme value distributed, independent of x0 and

    violate other modeling requirement as well as being behaviorallyinappropriate if we assume that individuals responses vary inde-pendently of socio-economic characteristics.

    The Bayesian approach to estimation of the ML is implemented

    3. Survey design and implementation

    K. Balcombe et al. / Journal of Air Transport Management 15 (2009) 221226222s,j,n

    uncorrelated across individuals or choices.The re-parameterisation in Eq. (2) means model estimation is

    being performed in WTP space. Typically, the ML is estimated inpreference space such that we would rst estimate marginal utili-ties. The resulting MRS is derived from these estimates. However,by estimating our model in WTP space, all the MRS is estimateddirectly. It has been found that this approach can signicantlyreduce the instability associated with WTP estimates in preferencespace. The issue of WTP instability in preference space has beenfound to be particularly problematic when the price or cost variableis not bounded above zero. In existing studies it has been commonpractice to x the payment coefcient that is an ad hoc approach toresolve the instability especially as xing the price coefcient may

    2 Other examples of CEs used to assess other aspects of consumer choice in termsspecication yields higherWTP estimates than themixed logit (ML)especially for additional legroom, i.e., V33 versus V15. In terms ofon-board food they found that moving from a sandwich to a hotmeal yield a WTP of V5 for both model specications. Overall theyclaim that the ML is statistically preferred to the MNL.2

    An ML model specication is used here employing the Bayesianmethods for estimation. There are a number of advantages ofemploying a Bayesian approach to estimate the ML. Here weemploy the Bayesian methods developed by Balcombe et al.(in press). We begin by describing our ML model specication andour Bayesian methods. We then describe our survey instrumentand how it was developed. In Section 4 we present our results andin Section 5 we conclude.

    2. The mixed logit model

    For model development we let xj,s,n denote a k 1 vector ofattributes from the CE presented to the jth individual (j 1,., J) inthe sth option (s 1,., S) of the nth choice set (n 1,., N). Nextdene Uj,s,n as the utility that the jth individual attains from xj,s,n. Inaddition, let yj,s,n be an indicator variable that is equal to unity if thejth individual chooses the sth option within the nth choice set, andzero otherwise. Next assume that an individual j attains utility fromthe sth choice in the nth choice set from the utility function

    Uj;s;n x0j;s;ntbj es;j;n (1)

    where bj is a (k 1) vector describing the preferences of the jthindividual and t(.) is an appropriate transformation of the param-eters. Within the ML literature an important transformation hasbeen one which also has the researcher to estimate the ML in WTPspace, as opposed to preference space which is the conventionalapproach. This is achieved by employing a re-parameterisation ofthe form:

    tbj t1

    b1j

    1; t2

    b2j

    ;..; tk

    bkj

    0 (2)

    that means that the quantities t2b2j;..; tkbkj are the marginalrates of substitution (MRS) with the numeraire being the rstattribute, which will always be the price or cost attributewithin theCE. We assume that the bj is normally distributed, and that the re-parameterisation t(.) does change the distribution, but t(bj) bjof air travel include Hess and Polak (2005), Rose et al. (2005), Hess et al. (2007) andTeichert et al. (2008).The selection of applicable attributes is central to the effec-tiveness of the CE. It is important to consider all applicable andrelevant attributes, while keeping in mind the available resourcesin terms of time and sample size. Increasing the number of attri-butes and levels greatly increases the size of the full-factorialmodel. To identify and focus on the attributes which are mostimportant in consumer preferences between different airlinecomfort offerings, focus groups were set up to conduct exploratoryresearch and attempt to gain a consensus to this end. In addition,interviews were conducted with airline management pertaining tothe viability and feasibility of the levels of attributes which were tobe included in the design. The use of industry specic interviews inconjunction with focus groups at the exploratory stage of attributeand level generation yielded the set of attributes employed.

    As can be seen from Table 1 the selection of attributes and levelshas focused on seat comfort, meal provision, entertainment, barservice and ticket price. The focus has been on the in-ight expe-rience on a ight with a given length of 4.55.5 h. Punctuality andairport facilities (e.g., lounges) were not included as they deviatedfrom this focus.4 The brand image of an airline was not included forthe same reasons, but also due to the vague and ill-denedperceptions likely to be associated with such an abstract term. Thechoices about the levels which these attributes should take weremade with signicant input from airline industry professionals, toensure viability.

    3 A detailed description of algorithm, in this case Gibbs with a Metropo-lisHastings (MH) Step is provided in Balcombe et al.

    4 As a referee indicated, another reason for excluding lounges is that they areby simulating the posterior distribution of the mean and variance/covariance of the preference parameters {bj}. In the Bayesiananalysis the algorithm used to undertake the simulation is referredto as the the sampler.3

    As is necessary in the Bayesian econometrics we have paidparticular attention to the performance of the sampler to ensureconvergence has been achieved. To begin with the sampler isinitialized at a starting point but is allowed to run for a periodknown as the burn-in before draws are recorded. After the burn-inphase to test for model convergence we initially observe the valuesof the parameters sequentially generated by the sampler. If ourmodel is correctly specied and performing appropriately ourparameters should move away from their initial starting points andby the time that the burn has nished they should be stable aboutamean. Tomonitor the serial dependence in the sampled values weestimate the autocorrelation coefcients for the sequential valuesgenerated by the sampler. Highly dependent draws mean that verylarge samples may be required in order to accurately map theposterior distributions. In accordance with standard practice, inorder to reduce problems of dependence we thin the sample byrecording only every kth value (k> 1) generated by the sampler. Wealso formally test for convergence by employing a modied t-test(taking account of dependency that remains) for which the nullhypothesis is no-difference between the rst and second half of thesampled values (with a sub-set of values removed from themiddle).likely to correlate with the quality of seat on the plane and such could be a source ofmulticollinearity in the resulting analysis.

  • includes the AP plus a private screen built into the seats, whichallow personalized viewing and other media. This option repre-sents the current Premium Economy offering on some CAs long-

    Table 1Attributes and levels used in choice experiment.

    K. Balcombe et al. / Journal of Air Transport Management 15 (2009) 221226 223Seat pitch represents the distance between the back of a seatand the back of the seat in front. The 28inch level of seat pitchrepresents a typical CA seat pitch. This is very close to the legalminimum and is often a source of complaints and passengerdiscontent (Quigley et al., 2001). However, due to cost pressures,CAs may wish to minimize their unit costs by reducing the seatpitch to the legal minimum. This strategy is also often employed byLCCs, whose seat pitch is also kept close to the minimum.Increasing seat pitchwill lead to rows of seats having to be removedfrom the aircraft. The 31inch seat pitch level is representative ofBritish Airways Euro-Traveler, short-haul economy class. Anyperceived differential between British Airways and a CA or LCCmaybe captured here. The 34inch seat pitch conforms to BAs ClubEurope short-haul business product.5 The British Airwayscomparisons are merely representative and seat pitch can varyfrom aircraft to aircraft with the same airline. Respondents werepresented with an interpretation of the legroom associated witheach level in a pre-survey brieng.

    The 17inch seat width represents a typical CA seat, and is alsothe width of a British Airways Euro-Traveler-class seat. BritishAirways superior Club Europe-class seat width is 19inch; the upperlevel chosen here is 18.5inch. Increasing seat width can be evenmore costly than increasing pitch, due to the fact that the increaseddimension can result in a far larger number of seats needing to beremoved (i.e., one from each row).

    The levels of in-ight meal provision were chosenwith focus onshedding light on the viability of not servingmeals on the aircraft atall. Several airlines have changed their level of service provision tomimic the LCCs. However, they have subsequently re-introducedfree on-board food provision for a number of reasons includingnegative image (Dennis, 2007). Thus although it is commonlyassumed that eliminating meal provision greatly reducescomplexity of operations, and is incorporated in the LCC businessmodel this need not be a strategy that all airlines employ. Thesandwich option would be simpler in practical implementation,and is included as an intermediary, in order to assess both the

    Attribute Units Status QuoLevels

    Additional Levels

    Seat Pitch Inches 28 31, 34Seat Width Inches 17 18.5In-Flight Meal Level Hot Meal None, SandwichIn-Flight Entertainment Level Standard Standard plus Amenity Pack,

    Standard plus Amenity Packplus Own Screen

    Complementary in-FlightDrinks

    Level None Complementary Bar Service

    Ticket Prices Euros 300 285, 325, 400, 500potential to create ancillary revenue in the form of on-boardsandwich sales, but also to improve the interpretation of anywillingness to accept (WTA) having no meal. For example,Thomas Cook currently serves hot meals on its ights. The esti-mation of the levels of this attribute will reveal if the carrying,preparing and serving of a hot meal are valued by customers.

    The entertainment levels were chosen to represent the standardCA in-ight entertainment provision, consisting of overheadscreens showing pre-set programming. The second level adds anamenity pack (AP), whichmay include items for children, a blanket,a pillow or other typical airline amenity items. The third option

    5 These products apply to non-domestic inter-EU ights, the longer of whichapproach the ight duration in question here.haul eet services. Complimentary bar service is kept as a simplebinary yes/no option. On-board drinks sales can constitute signi-cant ancillary revenue. The added perceived value of complimen-tary drinks can be compared to this, to assess its potential to addvalue.

    Finally, the price attribute was set as a representative gure forthe 4.55.5-h ight duration presented to the respondent.Departing from the UK, this would include destinations in countriessuch as Egypt, Turkey, and as far aeld as Gambia. The actualdestinations used in the CE were Dalaman, Izmir and Antalya(Turkey); Sharm el Sheikh and Hurghada (Egypt); and Banjul(Gambia) with departures from Manchester and London Gatwick.Prices were researched on the basis of a booking in the summerseason made two months in advance.6

    Given the selection of attributes and their associated levels,employing an orthogonal fractional factorial design and takingaccount of dominated options 15 proles are derived. In this casea prole refers to a combination of different levels of attributes,with each prole taking on exactly one level of each attribute. Theaim was to construct a choice card based on each of the prolesfrom the orthogonal design. Each choice card is made up of threeproles. To construct the choice cards each of the proles weretaken as the rst value on each card. Then to generate a secondprole, the shift method was used. Then, to generate the nal setof choice cards with three proles a status quo was added to eachwhich represented a standard offering for a CA seat. The status quooption is 28inch seat pitch, 17inch seat width, a hot meal, standardin-ight entertainment, no bar service and a price of V300. Finally,a dont know option was added. This option allows the respon-dents to provide an answer that does not force an inappropriatechoice. Thus, it enables the elicitation of a more realistic view of therespondents true preferences. In the analysis we present wetreated dont know responses as a zero choice: that is they werenot included for estimation purposes.

    Fig. 1 provides an example of the choice cards presented torespondents. This card includes various explanatory notes becauseit was the one used in the introduction to the survey respondentshow the survey would work. It includes information to improveresponse quality but without being so long as to detrimentallyaffect response rates. It can be seem that we did not label thestatus quo option. This removes potential labeling effects, andpotentially avoids anchoring the respondents choice. However, ifany respondents recognize this combination of attributes, or if thecombination already represents a favored combination for thesample, this may manifest itself in a bias toward its selection.

    The survey instrument was delivered via the Internet. It wasmade available to the public via the web site of a large travelcompany. We adopted this approach to survey delivery because itprovided the most cost effective means by which to generatea sample of sufcient size.

    In terms of survey delivery, potential survey participants wereinitially asked if they would be prepared to undertake the survey.This took the form of link on the web site. They were informed thatthe survey was being conducted for academic purposes and thatthe information provided would remain condential. If they agreedto participate they were then automatically directed to the surveyinstrument via an appropriate link. To ensure that the data

    6 In this CE we have not addressed the issue of how a price can vary depending

    upon time of purchase and associated seat availability. To include this aspect ofconsumer behavior would require a very different CE to that developed here.

  • take

    ld cnds,s in

    en 3t me

    also gives some detail about the different to th

    andviesying

    nspoThank you for your time, this survey should not

    You will be asked to select which flight you woucould be to a destination such as the Canary IslaAssume you are now booking the flight 2 month

    You will be faced with a series of choices betwedifferent levels of seat-pitch, seat width, in-flighservice.

    The following shows an example choice card. It attributes. Please review these and then proceed

    Seat Pitch is given in inches,

    Stmopla

    K. Balcombe et al. / Journal of Air Tra224collected from the survey avoided fatigue bias on the part ofrespondents in terms of responses being made, we blocked thechoice cards into three groups of ve. Thus, each respondent onlyneeded to undertake ve choice tasks. Each choice task appeared asa separate screen similar to that shown in Fig. 1.

    The sample consisted of 568 useable responses, comprising56% males with the vast majority of respondents being UKnationals (almost 90%). This indicates a slight bias in the samplein terms of males but not large. The current UK population gendermix is almost 50:50. Next, in our sample 63% of respondents hadno children, and 32% were single. Benchmarking these surveycharacteristics against the UK we can see that they are realistic.Currently in the UK there are estimated to be 33% single house-holds and 62% of households have no children. In terms ofeducation 22% of our respondents had some level of universityeducation (undergraduate and/or postgraduate). The distributionof income was relatively evenly distributed over the sample, with

    For each card, please select only one option. If ythe dont know box.

    28" 17" none standard plus amenity pack

    31" 18.5" sandwich private screen plus amenity

    28" 17" hot meal standard

    Seat

    Pitch

    Seat

    WidthIn-Flight Meal In-Flight Entertainme

    Seat

    Pitch

    Seat

    WidthIn-Flight Meal In-Flight Entertainme

    Seat

    Pitch

    Seat

    WidthIn-Flight Meal In-Flight Entertainme

    ranging from 28 to 34. 28 Seat Pitch will give anaverage person roughly one inchlegroom, if sitting upright. Increasing Seat Pitch to 31 or 34will give 3 or 6 inches extra legroom, respectively.

    An ampillow,in-flighpacks f

    The pindividpassengchoose

    This is the in-flight meal be served. Snacks can be eat on board in addition.

    Extra seat width is available on some options

    Fig. 1. Front pagee survey.

    ard entertainment includes and short programmes on overhead screens. more than 5-10 minutes.

    hoose on a 4 - 5 1/2 hour flight. This North- or Western Africa or Turkey. advance.

    airline tickets. They will each include al, in-flight entertainment and bar

    rt Management 15 (2009) 22122626% earning below 25,000, 16% earning over 50,000 anda sample average of 35,000. For the UK the average income fornon-retired households was 37,600 in 2006/07. Finally, theaverage age of respondent was 34 years old whereas for the UK itwas 39 in 2007.

    4. Results and analysis

    We began our estimation of the data by identifying an appro-priate model specication. After a rigorous examination of the datawe estimated a model in which all our coefcients are random. Weemployed the log-normal distribution for the price coefcient andthe normal distribution for all other coefcients. This choice ofdistribution is in contrast to many of the existing classical studiesthat have employed the triangular distribution (Hensher andGreene, 2003), but unlike them, we estimate the data inWTP space.

    ou are unsure which you prefer, tick

    yes 300

    pack none 400

    none 300

    Please tick only

    one

    ntComplimentary

    Bar Service

    Ticket

    Price

    ntComplimentary

    Bar Service

    Ticket

    Price

    ntComplimentary

    Bar Service

    Ticket

    Price

    Don't Know

    enity pack can include e.g. a free headphones for use witht entertainment, and special or children.

    rivate screen includes an ual screen for each seat, soers can independentlywhat they watch.

    which willbought to

    This Price isfor a Returnflight with all taxes, fees etc included.

    Drinks can still be purchased onboard if there is no complimentary bar service

    of the survey.

  • Thus, the model specication estimated is;

    Ui b1Pricei b2iPitchi b3iWidthi b4iBari b5iAmenity Packi b6iScreeni b7iSandwichi b8iNo Meali b9iStatus Quoi ei (3)

    where Price is the price of the ticket, Pitch and Width are seatspecication characteristics, Bar is a drinks service, Amenity Packand Screen are on-board entertainment options, and Sandwichand No Meal are the food options. These parameters provide uswith a measure of how much our survey respondents are WTP to

    highest specication on offer relative to the status quo then the netWTP is V121.

    The only CE study which provides comparable estimates isEspino et al. (2008) who found that the MNL yielded higher WTPestimates than the ML. In the case of additional legroom the MNLyielded WTP of V33 and the ML V15. Our estimate is V22 which isclearly comparable. However, our estimate for the WTP fora sandwich is such that we cannot be condent that the parameteris not zero. We also nd that the WTP for having No Meal is muchhigher than Espino et al. report, although the duration of the ightexamined in our study is signicantly longer. It could also be the

    K. Balcombe et al. / Journal of Air Transport Management 15 (2009) 221226 225change their in-ight on-board travel services. Finally, the StatusQuo captures an option common to all choice sets that can beconsider the current default with respect to in-ight travel serviceprovision.

    In terms of computational performance the MCMC samplerworked very well for this particular model specication. The resultswe generated by employing a burn-in of 1000 draws, then under-taking a further 10,000 draws. Both the burn-in and the subsequentdraws used in the analysis are constructed by sampling only one inevery 100 draws so as to yield robust statistical results. Bothobservation of the sample and our convergence statistics indicatedthat our model converged.

    The rst thing to note about the results reported in Table 2 isthat the Price coefcient is the sign we would expected a priori.Second, as this model is estimated in WTP space these estimatescan be directly interpreted as WTP estimates. However, it isnecessary to note that some of the variables have been scaled aspart of the estimation process. The Price coefcient has beendivided by 100 and Pitch andWidth by 10. Thus, we can see that onaverage our survey respondents are WTP an extra V22 for addi-tional Pitch (leg space) and V29 for extra seat width. With respectto the other attributes we can see that respondents are notWTP foran Amenity Pack (AP) because of the negative sign, although for thisparameter the posterior mass is to the positive and negative side ofzero as indicated by the lower quartile estimate being negative andthe upper quartile estimate being positive, which means that weare not condent that this parameter is not equal to zero. Similarlythe sandwich parameter also has a posterior density both positiveand negative. However, respondents are prepared more for a Barservice and a screen. Finally, the negative sign on NoMeal indicatesthat survey participants are WTA the loss of this service but onlywith an associated reduction in the price of the ticket, i.e., V31. TheWTP for these changes attribute levels are V31, V37, and V31,respectively. Although these might appear to be rather large it hasto be remembered that once on-board an airline is a monopolyprovider of all services and as such consumers may well havebecome conditioned to expect these kinds of values for theseadditional services. It also needs to borne in mind that theseadditional amounts in and of themselves are relatively smallcompared to the overall price of the ticket. To place these values incontext if an individual was to purchase a ticket that included the

    Table 2MXL regression results (excluding socio-economic variables).

    Variable Mean Standard Deviation Lower Quartile Upper Quartile

    Price 1.337 0.484 0.991 1.592Pitch 2.187 2.872 0.236 4.121Width 2.886 0.755 2.372 3.402Bar 0.307 0.372 0.056 0.561AP 0.080 0.341 0.312 0.153Screen 0.373 0.477 0.049 0.695Sandwich 0.068 0.364 0.174 0.316No Meal 0.316 0.389 0.054 0.580

    ASC 1.042 1.265 1.889 0.188case that the estimate we report is for a reduction in the level ofservice provided, as opposed to a level higher than the status quo.

    In terms of actual pricing by airlines for various on-boardservices, Dennis (2007) notes that Aer Lingus offers a full Irishbreakfast for V7 and a hot ham and cheese Panini for V5. Althoughthe price of the breakfast is signicantly less than the WTP for thehot meal we estimate there is a reasonably correspondence withour mean estimate for the price of a sandwich and the Panini. Thus,the WTP estimates are comparable with other results and data.

    The correlation coefcients are reported in Table 3 reveal thatthere is no consistent pattern. For example, we can see that havingmore leg space and seat width are negatively correlated. Thus, thereis no reason to assume that tall passengers who would prefer moreleg space demand wider seats. We can also see that extra leg spaceis positively correlated with having a screen and correlated with NoMeal. Interestingly, there is a positive correlation between seatwidth and having bar service, and a negative correlation betweenseat width and no meal.

    The socio-economic variables included in the analysis are Age,Income, Gender and Education. All variables have been included asdummy variables. In the case of Age and Income the data have beendivided at the sample mean; 34 years old and 35,000 per annum.

    These variables are included in the ML as

    bij a0 a1Agej a2Incomej a3Genderj a4Educationj ei (4)

    This model was estimated assuming the same random param-eter distributions. In terms of computational details this modelrequired a burn-in of 2000 draws, then undertaking a further10,000 draws. Both were constructed by sampling only one in every250 draws. This conguration was necessary given the higherdimension of the model compared to the basic ML specication andthe resulting problems of achievingmodel convergence. Our resultsfor this model are presented in Table 4.

    As can be seen in the table all the socio-economic variables havean impact on the various WTPs. Starting with Age we can see thatolder respondents are WTP more for aspects of seat comfortwhereas younger respondents are WTP more for an on-board barand screen. Thus, differences in Age indicate that older travelers aremore concerned with comfort and Younger travelers prefer very

    Table 3WTP correlation coefcients.

    Pitch Width Bar AP Screen Sandwich No Meal ASC

    Price 0.042 0.018 0.024 0.043 0.005 0.045 0.019 0.026Pitch 1 0.744 0.205 0.368 0.607 0.523 0.522 0.940Width 1 0.125 0.249 0.460 0.433 0.418 0.71Bar 1 0.077 0.032 0.185 0.077 0.233AP 1 0.202 0.135 0.187 0.374Screen 1 0.290 0.312 0.587Sandwich 1 0.412 0.505No Meal 1 0.502

    ASC 1

  • competition are viable strategies to deal with the signicantlyTable 4

    K. Balcombe et al. / Journal of Air Transport Management 15 (2009) 221226226Mixed logit results conditioned on socio-economic variables.

    Age Variable Young Old

    Mean stdv Mean stdv

    Price 1.019 0.425 1.179 0.496Pitch 2.028 3.662 2.702 3.602Width 2.773 1.656 3.738 1.709Bar 0.473 0.398 0.200 0.396AP 0.141 0.361 0.024 0.363Screen 0.408 0.596 0.188 0.594Sandwich 0.044 0.406 0.157 0.403No Meal 0.372 0.449 0.279 0.449ASC 0.991 1.451 1.559 1.434Income Variable Below Average Above Average

    Mean stdv Mean stdv

    Price 1.083 0.466 1.078 0.456Pitch 1.623 3.527 3.309 3.519Width 2.500 1.441 4.126 1.717Bar 0.351 0.416 0.375 0.429AP 0.138 0.359 0.021 0.360Screen 0.322 0.603 0.307 0.607specic forms of on-board service provision. In terms of Income wesee that higher income respondents are WTP signicantly higheramounts for seat comfort, e.g., Pitch Young V16 versus Old V33.Interestingly lower income respondents require a higher WTA nomeal compared to higher income respondents. When we considergender we see that females areWTPmore for seat width andmalesset pitch. Males are also WTP far more for an on-board entertain-ment screen and females require signicantly higher WTA for nomeal. Finally, whenwe consider educationwe can see that levels ofhigher education are related to lower WTP for seat pitch but muchhigher seat width. Also lower levels of education yield a muchhigher WTP for use of the bar.

    5. Conclusions

    A CE has been deployed to examine consumer WTP for on-board/in-ight service provision and level of comfort. Our resultsindicate that product differentiation and avoidance of the intra-EU

    Sandwich 0.010 0.398 0.209 0.397No Meal 0.407 0.440 0.221 0.441ASC 1.206 1.456 1.258 1.450Gender Variable Male Female

    Mean stdv Mean stdv

    Price 1.293 0.480 0.824 0.298Pitch 2.997 3.575 1.513 3.590Width 2.652 1.645 3.826 1.668Bar 0.322 0.420 0.409 0.415AP 0.116 0.359 0.058 0.364Screen 0.463 0.586 0.137 0.585Sandwich 0.168 0.400 0.003 0.399No Meal 0.272 0.446 0.406 0.448ASC 1.366 1.456 1.071 1.459Education Variable School Higher

    Mean stdv Mean stdv

    Price 1.065 0.452 1.152 0.489Pitch 2.491 3.631 1.794 3.617Width 2.450 1.139 5.598 1.196Bar 0.447 0.392 0.079 0.386AP 0.073 0.365 0.147 0.365Screen 0.303 0.609 0.373 0.604Sandwich 0.096 0.406 0.092 0.410No Meal 0.301 0.449 0.422 0.449ASC 1.201 1.470 1.331 1.461increased competition within the LCC 3-h range. This survey hasaddressed which attribute levels are perceived to be valuable byconsumers, and has provided WTPs for specic levels of thesevalues. In addition, we have been able to identify which type ofcustomer is WTP for which specic type of service provision.

    Overall the results indicate that a revised provision of additionalon-board comfort and service levels yield a net WTP of approxi-mately V120. This may appear to be a rather large amount but itneeds to be remembered that price differences of this magnitudealready exist in the market for trips of this type when comparingdifferent airlines. Thus, it would appear that there is scope for CAsto consider the overall quality of their on-board service provisionand not simply follow the approach adopted by the LCCs andoperate a no-frills service. Indeed, theWTP plus theWTA estimatesfound for the reduction in food service provision indicate thatfollowing the LCC no-frills strategy need not be the only businessmodel to pursue. However, several CAs have already started to offerlonger-haul ights that embrace the no-frills approach. The successor otherwise of these services will provide important insights andactual market data information on the topic of quality and type ofon-board service provision demanded by consumers.

    Acknowledgements

    The authors thank two anonymous referees for their commentson an earlier version of this paper.

    References

    Barrett, S., 2003. How do the demands for airport services differ between full-service and low-cost carriers? Journal of Air Transport Management 10, 3339.

    Balcombe, K., Chalak, A., Fraser, I.M., in press. Model selection for the mixed logitwith Bayesian estimation. Journal of Environmental Economics andManagement.

    Bieger, T., Wittmer, A., 2006. Air transport and tourism perspectives and chal-lenges for destinations, airlines, and governments. Journal of Air Transport andManagement 12, 4046.

    Dennis, N., 2007. End of the free lunch? The responses of traditional Europeanairlines to the low-cost carrier threat. Journal of Air Transport Management 13,311321.

    Espino, R., Martin, J.C., Roman, C., 2008. Analyzing the effect of preferenceheterogeneity on willingness to pay for improving service quality in an airlinechoice context. Transportation Research E 44, 593606.

    Francis, G., Humpreys, I., Ison, S., Aicken, M., 2006. Where next for low cost airlines?A spatial and temporal comparative study. Journal of Transport Geography 14,8394.

    Franke, M., 2003. Competition between network carriers and low-cost carriers retreat battle or breakthrough to a new level of efciency. Journal of AirTransport Management 10, 1521.

    Hensher, D.A., Greene, W.H., 2003. The mixed logit model: the state of practice.Transportation 30, 133176.

    Hess, S., Polak, J.W., 2005. Mixed logit modelling of airport choice in multi-airportregions. Journal of Air Transport Management 11, 5968.

    Hess, S., Adler, T., Polak, J.W., 2007. Modelling airport and airline choice behaviourwith the use of stated preference survey data. Transportation Research E 43,221233.

    Lee, D., Luengo-Prado, M.J., 2004. Are passengers willing to pay more for additionallegroom? Journal of Air Transport Management 10, 377383.

    Martin, J.C., Roman, C., Espino, R., 2008. Willingness to pay for airline servicequality. Transport Reviews 28, 199217.

    Quigley, C., Southall, D., Freer, M., Moody, A., Porter, M., 2001. Anthropometric Studyto Update Minimum Aircraft Seating Standards. Report Prepared for the JointAviation Authorities.

    Rose, J.M., Hensher, D.A., Greene, W.H., 2005. Recovering costs through price andservice differentiation: accounting for exogenous information on attributeprocessing strategies in airline choice. Journal of Air Transport Management 11,400407.

    Ryanair, 2006. Ryanair 2006 Year-End Corporate Accounts. Ryanair. Available from:http://www.Ryanair.co.uk.

    Teichert, T., Shehu, E., von Wartburg, I., 2008. Customer segmentation revisited: thecase of the airline industry. Transportation Research A 42, 227242.

    Consumer willingness to pay for in-flight service and comfort levels: A choice experimentIntroductionThe mixed logit modelSurvey design and implementationResults and analysisConclusionsAcknowledgementsReferences