5 Empirical analysis of the acceptance of flexible time-range tickets
5.3 Quantitative examination of hypotheses
5.3.8 Market simulation based on individual utility functions
Based on the individual utility values it is now possible to calculate the take-up rate of flexible time-range tickets as a function of the ticket parameters (time of information, given discount, etc.). In order to avoid the problems which could arise from using a probabilistic share of preference model to forecast consumer choices, the expected market shares were calculated by applying the ‘first choice’ method on the individual utility values.467
464 Backhaus / Erichson / Weiber (2010), p. 356ff; Orme (2000), p. 2f
465 The attribute importance is calculated by comparing the difference between the largest and the smallest part-worth value of a particular conjoint attribute against the sum of the absolute differences of all included attributes. For instance, the absolute difference between the part-worth value of a direct flight (1.15) and a connection flight (-0.67) would be 1.82, which corresponds to 30% of the sum of the difference of all attributes.
466 Since the importance values are ratio data, an absolute value of 50% means that the price is more than twice as important in the decision process as the flight specification modalities.
467 Cf. Orme (2000), p. 6f. In probabilistic models, the prediction of which alternative an individual would choose is made based on probabilities derived from the aggregated utility scores. A major problem of this approach is the fact that new products ‘steal’ market shares proportionally from all existing alternatives.
However, as the following ‘blue bus / red bus’ example illustrates this is not a realistic assumption. In this example, a blue bus competes against the car in the modal split and achieves a market share of 50%. If now a red bus would also start operations on this route and would gain 33% market share, these percentage points would be equally subtracted from all other alternatives, although it is more likely to assume that the gained passengers would mostly be blue bus passengers. This problem is also called
During the conjoint exercise the respondents were also shown two fixed choice situations (i.e. exactly the same composition of attribute levels was shown to all respondents), which were not taken into consideration in the utility modeling process.
Instead, these ‘holdout’ tasks were used to assess the goodness of fit of the model by comparing the expected market shares of the defined product types with the actual decisions by the respondents. As shown in Table 33 the model achieved a very good performance with an average deviation of only three percentage points.
Holdout choice task 1
Option 1 Option 2 Option 3 Option 4
Specified non-stop flight: €125
Flexible connection flight, notice: 1 week,
time-range 6 hours.
€75
Flexible connection flight, notice: 1 week,
time-range 8 hours.
€75
Flexible ticket (non-stop OR connection),
notice: 1 week, time-range 8 hours. €150 Forecasted market
share of alternative 54% 42% 3% 0%
Actual % of
respondents 50% 41% 8% 1%
Deviation in
%-points 4% 1% 5% 1%
Holdout choice task 2
Option 1 Option 2 Option 3 Option 4
Flexible non-stop flight, notice: 1 week,
time-range 6 hours.
€75
Flexible connection flight, notice: 1 week,
time-range 6 hours.
€75
Flexible ticket (non-stop OR connection),
notice: 1 week, time-range 8 hours. €150
Flexible non-stop flight, notice: 1 week,
time-range 6 hours.
€150 Forecasted share
of alternative 94% 3% 0% 3%
Actual % of
respondents 88% 5% 1% 4%
Deviation in
%-points 6% 2% 1% 1%
Table 33: Evaluation of market-share model based on holdout tasks468
In order to be able to estimate the market share that certain flexible products could achieve, at first the currently prevailing products have to be described in order to be used as reference points later on. A test conducted by the German “Stiftung Warentest”
in 2009 found an average ticket price at the three major European network carriers of around €250 - €300 incl. taxes & fees for return tickets within Europe (Lufthansa: €249;
British Airways: €265; Air France: €308). At the same time comparable flights with the leading LCC could be purchased for roughly €150 (Ryanair: €144; Easyjet: €166).469
‘Independence from Irrelevant Alternatives’ (IIA). By using a ‘first choice’ model based on individual data, the IAA problem is eliminated, since new products (if they show a higher utility) directly take away market shares from the previously considered alternative (and not proportionally from all available alternatives).
468 With regard to the interpretation of the displayed market share values it has to be noted that only people who have accepted flexible tickets in their relevant set (77.5% of all respondents) have participated in the conjoint exercise.
469 Cf. Stiftung Warentest (2009), www.test.de; The data was based on selected routes from German airports to Berlin, Barcelona, London, Mallorca, Paris, Rome and Vienna, which constitute rather competitive routes and therefore might even understate the overall average price level on all European O&Ds. However,
Since consumers in most O&D markets nowadays already have the chance to choose between non-stop flights and connection flights, the latter are to be included as well in the market share scenarios. To compensate the consumers for the additional travel time and the inconvenience of changing planes, these tickets are usually sold at a lower price than non-stop flights (a Canadian study found an average discount of 15% in comparison to non-stop flights).470 In the scenarios connection flights are therefore always priced one price step below the non-stop flights. Based on this market data three scenarios were build which are presented in Table 34.
Scenario name /
Offered flights & prices (one-way)
Very competitive route (incl. LCC)
Regular route
High-price route Reference products
(specified flight times)
Non-Stop flight € 100 € 125 € 150
Connection flight € 75 € 100 € 125
Newly added product Various types of flexible time-range tickets at different prices Table 34: Developed market scenarios
The aim of introducing several scenarios is to be able to estimate the acceptance of flexible time-range tickets, while they are offered next to a realistic set of already existing options. Furthermore, the variation of the price of competing products (as a function of the competition on a route), allows for an evaluation of the likelihood to select flexible products as a function of the relative discount to existing market alternatives.
5.3.8.1 Highly competitive route
In the first scenario, a highly competitive route is assumed, where the price of a non-stop flight does not surpass €100 and connection flights are even offered at a price of
€75 (all prices are given for one-way flights). If a flexible product would be added to this market it would be able to be the first choice of up to 63% of all people, who have previously accepted flexible time-range tickets in their relevant set (depending on the product characteristics with regard to the degree of uncertainty involved). To obtain overall market shares, it has to be considered that 22.5% of all respondents in principle refused to consider flexible tickets as a possible alternative in their flight selection process. Therefore, if the overall market is considered, the attainable market share of flexible time-range tickets in a very competitive market situation ranges between 6 – 49%. As shown in Figure 79, in general, flexible tickets which guarantee the buyer a non-stop flight, are able to attract more passengers, than tickets which do not define whether the flight will be a direct or an indirect flight. Furthermore, in both categories tickets entailing a notice period of one week are preferred to those, where the customers are only informed of their precise flight times one day prior to departure. The lowest
the figure is in line with the price level measured by the US Bureau of Transportation Statistics based on a large and representative sample of US domestic flight fares for the first quarter 2010 (US data is taken since no comparable European statistics exist, however, given the competitive structure the two regions are deemed comparable). Looking at all flights (network carrier and LCC) a price level of €232 was found (cf.
U.S. Department of Transportation (2010), www.bts.gov; the original value of $328 was converted to Euro using the exchange rate from March 2010 (0.72, cf. Oanda (2010), www.oanda.com).
470 Cf. Gillen / Hazledine (2006), p. 16
acceptance at what tim time-range share of pe price does willing to c Figure 80)
471 Own illust
472 Cf. Stiftun
e is achieve me the pass e of eight ho ve to spec n flight) is ab
9: Expected
Regular rou cond scena by two, corr ean routes
eople who a s not excee compromise
.
tration ng Warentest (
ed by ticke senger will ours. Howev ion flight, 8 time range 4%) would accept the
d 58%, whi e on a fixed
(2009), www.t
ts, which le depart eith ver, even th top flights an overall m
re of flexible
ecified one lmost exact turn ticket)
€100. If flex y would be a 8h time rang e, notice pe
accept the most prefer ich also sho
schedule a
est.de
eave it open her by non-his ticket typ
(and at the market shar
e tickets (ve
e-way flight tly to the av ).472 Furthe xible time-ra able to achi ge, notice p eriod 1 wee e least acce
rred flexible ows that a and cannot
n until one -stop or by pe, if offere e same pri re of 6%.
ery competi
t was price verage price ermore the
ange tickets ieve a mark period 1 da ek). If the p
epted flexib e ticket optio
certain frac even be lur
day prior to indirect flig d at a price ce of a ful
tive route sc
ed at €125 e charged b
scenario s would be ket share be ay) and 58%
rice is lowe le ticket. Ho on even at t ction of trav red by lowe
o departure ght during a e discount o
lly specified
cenario)471
5 (which, i by Lufthansa
included a added at a velers is no er prices (cf e,
Figur
5.3.8.3 Hi Figure 81 s one-way di impact of d
€125, €100 market sha
473 Own illustr
re 80: Expec
igh-price r shows the r
irect flight different pric 0 and even € ares betwee
ration
cted market
route results of th
is offered a ce levels, in
€75. In this en 9% and 6
share of fle
he market s at a price o n this scena environme 69%.
exible ticket
imulation in of €150. In ario the flex nt the offere
s (regular ro
n the high-y order to b xible ticket ed flexible t
oute scenar
yield scenar be able to options are tickets are a
rio)473
rio, where a assess the e offered at able to gain a e t n
F
Of particul given disco and conne actual assi a strong in proportiona
Figure 81: E
ar interest t ount. If for i ection flight ignment on ncrease in t ately seen specifically are of 9%, a
a market s es that the y few peopl
ional discou
e likelihood en price re ction. Unles dly conside uration poin gered by ad
Usage of fl ng been co d whether t
Expected tak
though is th instance the
s during a ly being do the market
inferior ma y, a discoun a discount o share of 42
acceptanc le accept fl unts do not
to accept a duction is ss a certain ering flexibl
nt, after w dditional dis
exible tim nfronted wi they would
ke-up rate o
he differenc e least attra
time-range ne one day share if the rket share nt of 17% re
a flexible tim not followin n minimum
e tickets a which the counts decr
me-range ti ith the choi use flexible
of various fle
ce in the ac active flexib e of eight h y prior to de
e price is lo increase if elative to a s ults in a ma ows for an ollowing a lo e range tick
roportional
me-range ti ng a linear
discount is s an altern marginal in reases.
ickets for ice-based c e time-range
exible
time-ceptance ra ble ticket (w hours in the
parture), on owered from
the price i specified no arket share
acceptance ogistical fun kets. Howev gain in mar
ckets as a r but rathe s granted cu native at al
ncrease in
additiona conjoint exe e tickets ins
range ticket
ate as a fun which include
e relevant s ne can see t m €125 to € s further de on-stop fligh
of 35% and e of hypoth nction. At lo ver, after th rket shares.
function of r a logistic ustomers a l. There is acceptanc
al trips ercise the r stead of exis
ts
nction of the es non-stop set with the that there is
€100, with a ecreased to ht leads to a d a discoun hesis seven
or in order undertaking
Although st the actual respondent that these t
H8 Flex them to targ it is the cas While in ch the faced y different de study. In th
474Cf. Backh
475 Garrow (20 g even addi
Figure 82:
tated prefer consumpti ts who say tickets are a
xible time-ra
cussion
in chapter ent strategie heir restrict n the prevai . In this co get the low-se with pure hapter 4 flex yield manag esign option
he following
aus / Broszka 009), p. 252
38%
would use it inst xisting tickets a make additiona
Usage of
% of responde
additional tr itional trips
: Usage of fl
rence data on of a se
that they w able to indu
ange tickets
of result
2.5.5, netw es, which al tion-free pri iling study:
ntext, netw
f flexible t
ents; n=275 re possib
rips. In this if the condi
lexible time
concerning ervice474, o would use ce addition
can induce
ts
work airlines llow them to
cing (RFP) 51% of all work airlines
ent, without onal) price d ange tickets gine to make ev e trips than in t with such a tic
time-rang
espondents, w ble option whe
context, 73 tions of the
-range ticke
g future beh ne can, ba
flexible tick al demand,
e additional
s are under o counter th ) models. T
trips made s are lookin t having to discounts.47 s were pres e preceding
e by potent f flexible tim
r (2006), p. 14
ge tickets f
who accepted en looking for
3% said th ticket are a
ets for addit
avior often ased on th kets for add
which verif
demand.
heavy pres he price pre he success e by leisure ng for new cannibalize
75
sented as o sections o tial custome me-range ti
47 23%
nly use it instea ntly used tickets
for additio
flexible time-r flights)
at they cou appealing (F
tional trips
tends to ov he huge pr
ditional trip fies hypothe
ssure to find essure origi
s of this mo e travelers w
products, w e existing re
one possible of chapter 5 ers during a ickets for th
4 esis eight.
d new yield nating from odel is also were made which allow evenues, as
e answer to 5 examined