Flatrate Bias in Public Transportation - Magnitude and Reasoning

Abstract

If customers can choose between a pay-per-use and a flat fare option they tend to prefer the flat fare option even if they do not reach the break-even point. The phenomenon has been shown for services like mobile communication, fitness studios or all-you-can-eat buffets and has been labeled as flatrate bias. The reasons for this bias have been attributed to different effects like taximeter effect, insurance effect, flexibility effect or convenient effect. We show that there is a flatrate bias for public transit fares too - customers prefer season tickets like monthly or yearly tickets even if they don't reach the break-even point. We are using mobility data to asses the magnitude of the flatrate bias regarding customers involved and amount of fares payed in excess. In order to understand what the underlying effects are we are conducting a survey measuring the strength of each suspected effect. Results show that the flatrate bias can be addressed mostly to an insurance and convenience effect. A sound understanding of the flatrate bias is important when it comes to the design of modern tariffs in Automatic Fare Collection Systems. Modern tariffs allow fine grained price differentiation and could substitute the flat fare options including their economical benefit for the transport agency.

Full text

Wirtz, Vortisch, Chlond: Flatrate Bias in Public Transportation: Magnitude and Reasoning
Flatrate Bias in Public Transportation: Magnitude and Reasoning
Matthias Wirtz (corresponding author)
Rhein-Main-Verkehrsverbund Servicegesellschaft mbH,
Am Hauptbahnhof 6, 60329 Frankfurt am Main, Germany
Tel: +49 69 27 307-348
Email: mwirtz@rms-consult.de
Peter Vortisch
Institute for Transport Studies, Karlsruhe Institute of Technology
Otto Ammann-Platz 9, 76128 Karlsruhe, Germany
Tel: +49 721 6084 2255
Email: Peter.Vortisch@kit.edu
Bastian Chlond
Institute for Transport Studies, Karlsruhe Institute of Technology
Otto Ammann-Platz 9, 76128 Karlsruhe, Germany
Tel: +49 721 6084 2257
Email: Bastian.Chlond@kit.edu
5,600 words (including references)
1 figures, 5 tables (6 * 250 words = 1,500 words)
Total: 7,100 words (including abstract, text, cover page, references, and figures)
1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27

Wirtz, Vortisch, Chlond: Flatrate Bias in Public Transportation: Magnitude and Reasoning
ABSTRACT
If customers can choose between a pay-per-use and a flat fare option they tend to
prefer the flat fare option even if they do not reach the break-even point. The phenomenon has
been shown for services like mobile communication, fitness studios or all-you-can-eat buffets
and has been labeled as flatrate bias. The reasons for this bias have been attributed to different
effects like taximeter effect, insurance effect, flexibility effect or convenient effect. We show
that there is a flatrate bias for public transit fares too - customers prefer season tickets like
monthly or yearly tickets even if they don't reach the break-even point. We are using mobility
data to asses the magnitude of the flatrate bias regarding customers involved and amount of
fares payed in excess. In order to understand what the underlying effects are we are conduct -
ing a survey measuring the strength of each suspected effect. Results show that the flatrate
bias can be addressed mostly to an insurance and convenience effect. A sound understanding
of the flatrate bias is important when it comes to the design of modern tariffs in Automatic
Fare Collection Systems. Modern tariffs allow fine grained price differentiation and could
substitute the flat fare options including their economical benefit for the transport agency.
2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16

Wirtz, Vortisch, Chlond: Flatrate Bias in Public Transportation: Magnitude and Reasoning
INTRODUCTION
In most service sectors price discrimination is used and the service is transacted at dif-
ferent prices by the provider. The price segments can be based on attributes of the consumer,
the geographical area, time of consumption or they can be based on the amount of consump -
tion. A prominent example for price discrimination is the public transportation (pt) sector. For
local pt services in Germany and most other countries in Europe there are almost always
quantity based price differences. One option is based on a pay-per-use price where the cus-
tomer has to pay each time he uses the service. The second option is a flat fare price where
customers pay once for the service and can then use it as often as they prefer without addition-
al cost. A prominent example of the former option is as single ticket and a season ticket of the
latter.
This type of price discrimination is nowadays very common for service sectors like the
mobile communication market, the internet access market or the gym market. For these mar-
kets it has already been observed that customers when having the option between a pay-per-
use tariff and a flat fare tariff tend to opt for the flat fare tariff. This holds true even if they do
not reach the break-even point for a flat fare tariff. Research on this tendency for flat fares
goes back to the 1980s when the telephone market in the USA added a pay-per-use option to
the former flat fare only system (1). The tendency for flat fares has been labeled flatrate bias.
A considerable number of researches addressed the issue of the flatrate bias. The ma-
jority focused only on attributing the bias to the underlying effects without quantifying its
magnitude. Since we are focusing on the magnitude of the flatrate bias as well as it's causing
effects we concentrate the literature review on work that focused on both aspects too. An com-
prehensive overview of all the researches done can be found in (2) or (3).
Most of the quantitative investigations are based on stated preference data. For cus-
tomers of an electronic newspaper the flatrate bias is analyzed by (2). They conducted an on-
line survey with 147 participants and assessed that there is a 25 % increased willingness to
pay for a flat fare tariff. Main reasons were attributed to the insurance, taximeter and flexibil-
ity effect. In the market of mobile communication several studies have been conducted (3, 4,
5). The results for the increased willingness to pay for a flat fare tariff varies between 10 %
and 15 %. As main effects were identified the insurance, flexibility, taximeter, convenience
and the overestimation effect. Similar results were achieved when analyzing the delivery ser-
vice market by Nunes (6) or the internet access market by Lambrecht and Skiera (7).
Only few researches on the flatrate bias are based on revealed preference data. DellaV-
igna and Malmendier (8) analyzed data of the US health clubs with information on both the
contractual choice and the day-to-day attendance decisions of 7,978 members over three
years. On average customers with a monthly flat fare tariff of over $70 pay a hypothetical
price of $17, even though they could pay $10 per visit using a 10-visit pass which results in a
70 % overpayment. As major explanation for this is a overestimation of future efficiency or
future self-control stated.
3
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39

Wirtz, Vortisch, Chlond: Flatrate Bias in Public Transportation: Magnitude and Reasoning
The focus of our research is twofold: firstly we assess the magnitude of the flatrate
bias in public transportation. Therefore we use mobility data from two different data sources.
An algorithm is developed in order to find the best tariff option for the reported mobility
which then is compared to the actual choice made by the individuals. The results show that
the overpayment due to the flatrate bias could be as high as 11.8 %.
Secondly we elaborate which effects are causing the flatrate bias and estimate the
strengths of these effects. Based on a literature review we formulate the hypothesis that the
flatrate bias is caused by four different effects. The existence of these effects are then tested in
a survey. Using a logit regression to model the influence of each flatrate bias effect on the
choice behavior it is shown that the insurance and the convenience effect are major contribut-
ors to the flatrate bias. Finally, a discussion of the results concludes the paper.
The results of this research contribute to a better understanding of the flatrate bias in
public transportation. With the advancement of Automatic Fare Collection Systems (AFC-
Systems) the possibilities in tariff arrangements have been multiplied. One option is the auto-
matic selection of a tariff retrospectively in order to always offer the price minimal option for
the customer. With a solid understanding of the flatrate bias appropriate tariff arrangements
can be developed.
DATA
We used two data sources for revealed mobility data and one data source for stated
preference data.
The first data source of revealed mobility data consists of trip data recorded by the
AFC-System of the regional transport association KreisVerkehr Schwäbisch Hall GmbH in
Germany. The transport association operates roughly 100 routes and 250 vehicles in the south-
western part of Germany. It's AFC-System has been renewed in 2006 and has had 6 Million
transactions since then.
We use trip data from the second half of 2008 containing 7,500 customers. By that
time the AFC-System was used only for non season tickets and was operated as a check in
check out system recording each trip that was made. In total roughly 370,000 trips were recor-
ded and used for analysis.
In order to cover the mobility behavior of season ticket holders a second data source of
revealed mobility data was used. We used household survey data from the Rhein-Neckar re-
gion which is located in the south-western part of Germany too. The household survey was
designed as a one week panel survey comparable in it's design to the German Mobility Panel
(9). The most important fact is that people report in detail their mobility over one week.
We used data from 2007 and 2008 which consisted of 565 persons. We included only
those persons above 18 who reported to posses a season ticket and who at least used public
transportation once during the reporting period. This resulted in a total of 52 persons.
The third data source that we used is stated preference data from a survey which was
part of a lager research project regarding modern tariffs in AFC-Systems. It was carried out in
2012 in the city of Karlsruhe, which is located in the southwestern part of Germany. The 174
4
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40

Wirtz, Vortisch, Chlond: Flatrate Bias in Public Transportation: Magnitude and Reasoning
participants were selected by a random-address method and interviewed in person using a
computer (CAPI). The participation in the survey was encouraged by three different incent-
ives which the participants could choose from: a shopping bag, a USB memory stick or a gift
certificate.
The sample was stratified by age in order to account for the different response rates of
elderly people. The sample represented the basic population of Karlsruhe very good with re-
gard to sex, household size, number of cars per household and proportion of season ticket
users.
MAGNITUDE OF THE FLATRATE BIAS
As mentioned above it is common practice for pt operators to offer pay-per-use tariffs
as well as flat fare tariffs. The decision whether or not to sign up for a flat fare tariff must be
made by pt users in advance. A retrospective tariff change is not possible. Therefore the cus-
tomer needs to forecast somehow his future usage level and decide which tariff option suits
him best. If only ticketing costs are considered two possible errors could be made when mak-
ing the decision:
1. Customer opts for a flat fare tariff but does not reach the break even point. It would
have been better to choose a pay-per-use tariff in this case.
2. Customer opts for a pay-per-use tariff but reaches the break even point for a flat fare
tariff. In this case it would have been better to choose a flat fare tariff.
So in order to proof the existence of the flatrate bias of pt users we need to show that
there is a difference in the size of the two before mentioned errors: the number of pt users
buying a season ticket and not reaching the break-even point should be larger than the number
of single ticket users reaching the break-even point for a season ticket.
Being able to compare these numbers two facets have to be considered: One important
aspect is the variability in the day to day mobility behavior. Several studies have highlighted
this variability and shown that it includes the travel mode choice -- the outcome of this variab-
ility has been labeled as multimodal travel patterns (10, 11). The mode choice depends very
much on the circumstances the individual finds himself confronted with. This holds true espe-
cially for individuals using public transit, bicycle and going by foot. Therefore when calculat-
ing reliable figures for the usage frequency of public transit on the individual level it is essen-
tial to capture this variability.
The other important aspect concerns self reported usage frequencies. The usual pro-
cedure of asking participants for their usual mobility behavior doesn't work here because of
overestimation of pt trip frequency. As shown by Lathia and Capra (12) this overestimation is
pronounced for weekdays where on average pt users are off by 1.77 pt trips per travel day.
Method and Data used
The solution is to use mobility data on the trip level spanning a period long enough to
capture the variability described above. We use mobility data from the AFC-system of the re-
gional transport association KreisVerkehr Schwäbisch Hall GmbH for non season ticket hold -
5
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39

Wirtz, Vortisch, Chlond: Flatrate Bias in Public Transportation: Magnitude and Reasoning
ers and form the household survey of the adjacent Rhein-Neckar region for season ticket hold-
ers. Unfortunately we had to use mobility data from slightly different regions of southern Ger-
many since the AFC-system data of the regional transport association KreisVerkehr
Schwäbisch Hall GmbH was lacking data for season tickets and vice versa.
(a) (b)
FIGURE 1: Utilized simplifications
For both data sources we implemented an optimal-price algorithm which calculates the
combination of ticket options resulting in the lowest price. As ticket options the very common
and for all participants available types were used: single trip ticket, day ticket and monthly
ticket. We excluded all special monthly tickets like job tickets because the data didn't include
pricing information for special tickets. Several simplifications were utilized in order to keep
the calculation effort moderate:
The data we used only included the origin and destination of each trip and not the route
chosen. Since different route options might be priced differently we could not determine
the trip price unambiguously. An example is given in figure 1 (a) where there are two
route options to get from origin O to destination D. The one over Y crosses 2 zones the
other one 3 zones resulting in different prices for the same origin and destination. In these
cases we used the most inexpensive option. Since this affects only a minority of trips we
estimate the possible influences as negligible.
Although day tickets are valid for 24 hours and consequently can span over two days, we
curtailed them to be valid on one day only. This reduced the flexibility of day ticket usage
considerable -- see figure 1 (b). Without further analysis we assume that this causes the
algorithm to find solutions which cost more or the same as the best solution but not less.
As a consequence we might overestimate the cost for the best ticket option.
The aforementioned simplification was applied to monthly tickets too, which are only
valid for one calendar month only.
Season tickets in Germany are bound to the tariff zones specified on the ticket. If one
wishes to travel beyond the zones included in the season ticket, conjunction tickets have
to be purchased. Since the mobility survey data only included information on whether the
6
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32

Wirtz, Vortisch, Chlond: Flatrate Bias in Public Transportation: Magnitude and Reasoning
participant owns a season ticket or not, the type of the ticket and the exact area of validity
of the season ticket was not given. Therefore as an assumption we assumed that the sea-
son tickets are valid for the most frequently used trip pattern. Depending on the reported
mobility it might be more cost efficient to buy a season ticket for a trip pattern that is not
the most used one and buy additional conjunction tickets in order to cover the whole trip
distance. We assume that the aforementioned configuration might be applicable only in
exceptional cases. Furthermore the additional handling of conjunction tickets seems very
cumbersome and we expect that this is practiced only in very rare situations.
Results
The result is separated according to the two possible types of errors (table 1). In total
12.2 % of all customers who use a single ticket tariff exceed the break-even point for a flat
fare. 13.2 % of all trips in this group are done by these customers. Due to this economically
deficient decision the fare payment in this group raises by 3.4 %.
TABLE 1: Magnitude of error made when choosing the fare type
Customers Share of Overpayment
Type of error No [%] trips [%] [%]
Customers with single ticket tariff but exceeding
the break-even point for a flat fare
915 12.2 13.2 +3.4
Customers with flat fare tariff but they don't
reach the break-even point for a flat fare
15 29.0 12.6 +11.8
Fisher's exact test <0.0001 0.5095
In the group of the customers opting for a flat fare tariff there are 29 % who didn't
reach the break-even point for the flat fare. They contribute 12.6 % of the trips and the fare
payment in this group is raised by 11.8 % due to their non optimal decision. The differences in
the number of customers not choosing the most cost efficient tariff option between the two
groups is significant using the Fisher's exact test (13). Therefore the results clearly show that
there is a flatrate bias for flat fare tariffs in public transportation too.
It should be kept in mind that the additional fare payments in the group of flat fare
users cannot be solely attributed to the flatrate bias. There are individuals in the group of cus-
tomers using the pay-per-use tariff option who don't make the most efficient decision. So there
is a quantity of pt users in both groups who are effected by random errors when choosing the
tariff option.
7
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26

Wirtz, Vortisch, Chlond: Flatrate Bias in Public Transportation: Magnitude and Reasoning
REASONING FOR THE FLATRATE-BIAS
Based on the research done in other service sectors as presented in the introduction we
formulated the hypothesis that the flatrate bias in in the sector of public transportation is
caused by the following four effects:
Insurance effect: The insurance effect describes the attitude of customers to avoid
variations in price. This stability is appreciated because it makes budget planning easier. The
dodging of exceptionally or unexpected high costs comes with the detriment of not saving
money when usage was below the break-even volume. Even if over- and underspending might
balance over time customers tend to prefer flat fare tariffs. As shown by Kahneman and Tver-
sky (14) people value mentally the agony of overspending more severe than the excitement of
saving money. Therefore the effort to avoid overspending is pronounced.
Taximeter effect: Each time a service is utilized pay-per-use tariffs show the resulting
cost quite clearly to the customer. On the other side flat fare tariffs have to be paid only once
and the customer is not confronted with the price during usage anymore. In accordance to the
theory of mental accounting by Thaler (15) people have different accounts for their spending.
The accounts themselves but also the time points when outgoings are made are valued differ-
ently. So if the time point of the outgoing is not related to the time point of consumption it is
assumed that customers are more likely to appreciate the service provided. In contrast when
the time point of payment is combined with the time of consumption it is assumed that cus-
tomers don't value the service as positively.
Convenience effect: A sound decision of a tariff can only be made if all necessary in-
formation for that decision is available. Since advertised information is not all-embracing ad-
ditional information needs to be collected and read by the customer which is time consuming.
If this effort is not made studies by Kling and Van der Plög (16) show that the probability is
higher to opt for a flat fare tariff in comparison to a pay-per-use tariff.
Self-discipline effect: The self-discipline effect describes the effort of people to
change their behavior. The actual behavior might not be regarded as desirable and one seeks
to get mental assistance as formulated by Wertenbroch (17). With regard to public transporta-
tion people might strive to increase their usage level of public transportation. When opting for
a flat fare tariff and paying the price in advance this might be a motivation to get the most out
of the deal and use public transport more often.
All four effects are assumed to have a direct influence on the flatrate bias.
Survey Design
In order to quantify the impact of these four effects we used stated preference data
from a survey which was part of a lager research project regarding modern tariffs in AFC-Sys-
tems.
The survey section related to the flatrate bias consisted of two parts. In the first part
the strength of the presumed flatrate effects were measured. The measurement was done indir-
8
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38

Wirtz, Vortisch, Chlond: Flatrate Bias in Public Transportation: Magnitude and Reasoning
ect by presenting the participants describing sentences; for each flatrate effect one sentence
(table 2).
TABLE 2: Sentences used to describe the flatrate effects
Describing sentence
Insurance effect I prefer to stay on the save side and pay the flat fare price. So it will
never cost more.
Taximeter effect If I use the pay-per-use tariff I would always be tempted to forbear
from making another pt trip.
Convenience effect Usually it's not worth the effort to double check if the flat fare comes
cheaper.
Self-discipline effect I'm more willing to keep the car in the garage if I have paid the flat
fare.
The participants should then judge the strength of their agreement to these sentences
on a 5 five point Likert scale ranging from “strongly agree” to “strongly disagree”. The Likert
scale used included the indifferent answer “neutral”.
In the second part the participants were asked to put oneself in the situation of using
public transport on average 16 times per month. For this usage pattern they were asked to
choose between two different tariff options: one pay-per-use tariff where each trip including
transfers has a fixed price and each trip is billed separately and one flat fare tariff which is
valid for one month and allows an unlimited number of trips.
While the tariff options and the average usage of 16 times per month remained the
same the number of minimal and maximal usage in one month was varied (table 3). In total
four different scenarios were presented. For the price of the different tariffs a between subjects
design was used: the flat fare price was offered to one group of participants for 16 € per
month and for the other group it was offered for 20 €. The price for the pay-per-use tariff was
for both groups set to 1 € per trip.
TABLE 3: Choice of flat fare over pay-per-use tariff option
Trips Group 1 (20€) Group 2 (16€)
Scenario Min Max Share flat rate Overpayment [%]
1 0 26 38 % +10 % 51 %
2 0 32 64 % +16 % 75 %
3 10 24 77 % +19 % 91 %
4 10 36 95 % +24 % 94 %
9
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20

Wirtz, Vortisch, Chlond: Flatrate Bias in Public Transportation: Magnitude and Reasoning
Results
The minimum and maximum number of tips per month was varied so that the average
value is closer to the minimum value in scenario 1 and 3, in the middle between minimum and
maximum in the scenario 2, and closer to the maximum in scenario 4. In general the share of
participants opting for the flat fare option is higher in the second group since the price of the
flat fare is lower for this group. From an economical perspective there is no advantage for the
second group opting for the flat fare tariff since on average the cost will be the same for both
tariffs: 16 €. Nevertheless in all four scenarios the majority of participants decided to choose
the flat fare tariff.
For the first group a pure economical decision would result in choosing the pay-per-
use tariff option in all scenarios. On average 16 trips are made with pt which would cost 16 €
when using the pay-per-use tariff compared to 20 € with the flat fare tariff. Although the best
option would be to opt for the pay-per-use tariff many participants decide to use the flat fare
tariff and therefore the spending on tickets is higher than necessary. This surplus is given in
the column “Overpayment”. It reaches almost 25 % in the scenario 4 because almost all parti-
cipants opt for the flat fare tariff and this option is 25 % more expensive than the pay-per-use
tariff.
Already visible here is the influence of the usage thresholds, the minimum and maxim-
um number of trips per month. In scenario 3 the share of participants choosing the flat fare
option is higher than in scenario 2 although the average usage number in scenario 3 is closer
to the lower threshold.
TABLE 4: Spearman-Correlation matrix of the effect variables
IE TE CE SE
Insurance effect (IE) 1.000 0.339
(<0.001)
-0.131
(0.085)
0.246
(0.001)
Taximeter effect (TE) 0.339
(<0.001)
1.000 0,027
(0.7227)
0.190
(0.0121)
Convenience effect (CE) -0,131
(0.085)
0.027
(0.7227)
1.000 0.097
(0.2039)
Self-discipline effect (SE) 0,246
(0.001)
0.190
(0.0121)
0.097
(0.2039)
1.000
In order to model the strength of the effect variable we used a multivariate logit regres-
sion (18) that predicts the probability
P
of the participant to choose a flat fare tariff (FF):
10
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26

Wirtz, Vortisch, Chlond: Flatrate Bias in Public Transportation: Magnitude and Reasoning
logit P (FF )=β0+β1
⋅IE +β2
⋅(IE ×STH )+β3
⋅TE+β4
⋅(TE×STH )+β5
⋅CE +β6
⋅(CE ×STH )
+β7
⋅SE +β8
⋅(S E ×STH )+β9
⋅STH +β10
⋅PFF +β11
⋅MIN +β12
⋅MAX
As independent variables we used the four effect variables: insurance effect (IE), taxi-
meter effect (TE), convenience effect (CE) and the self-discipline effect (SE). The four effect
variables were coded from -2 for “strongly disagree” over 0 for “neutral” up to 2 for “strongly
agree”. In order to control for the influence of the different prices of the flat fare tariff and the
different usage thresholds given the price of the flat fare tariff (PFF), the minimum (MIN) and
maximum usage per month (MAX) were included as independent variables too.
Additionally we added a binary interaction variable “season ticket holder” (STH) for
the four effect variables. The value of this variable is 1 if the participant reported to use a sea -
son ticket in real life and 0 if not. We constructed the variable based on information from per-
sonal questions in the very beginning of the survey. Since correlations between independent
variables lead to biased results the Spearman-Correlation of the effect variables are shown in
table 4. The insurance effect shows slight correlations to the other effect variables while there
is almost no correlation between the other effect variables.
TABLE 5: Logistic regression on the choice of flat fare tariff
Variable Coefficient Standard
Error
Odds-
Ratio
(intercept) -2.310⁺1.259 0.099
Insurance effect 0.735*** 0.106 2.086
Insurance effect * season ticket holder -0.184 0.221 0.832
Taximeter effect -0.170⁺0.096 0.843
Taximeter effect * season ticket holder 0.206 0.182 1.229
Convenience effect -0.243** 0.091 0.785
Convenience effect* season ticket holder 0.530** 0.186 1.699
Self-discipline effect -0.058 0.111 0.943
Self-discipline effect * season ticket holder 0.432 0.266 1.540
Season ticket holder 0.358 0.450 1.430
Flat fare price -0.161** 0.054 0.851
Minimum usage 0.233*** 0.025 1.263
Maximum usage 0.161*** 0.029 1.175
*** significant on the 99,9 % level; ** significant on the 99 % level; * significant on the 95 % level; + significant on the 90 % level
McFadden Pseudo-R2=0.29
11
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17

Wirtz, Vortisch, Chlond: Flatrate Bias in Public Transportation: Magnitude and Reasoning
The results of the model are presented in table 5. The most pronounced influence on
the flatrate bias are created by the insurance effect and the convenience effect. The insurance
effect is the most important reason for choosing a flat fare tariff. For each step on the Likert
scale towards an agreement with the presented statement it is twice as likely to opt for a flat
fare tariff. While the insurance effect seems to be more pronounced for non season ticket
holders, differences between both groups are not significant.
With regard to the convenience effect it was very important to have the interaction
variable season ticket included in order to see the different directions of influence. While for
season ticket holders the convenience effect has a positive impact on choosing a flat fare tariff
it has a negative effect for non season ticket holders. So it becomes obvious that the conveni-
ence effect is rather a general explanation for not opting for a economically justified tariff de-
cision. It shows the tendency of people to stick with their usual tariff choice and the avoidance
of reevaluating previous decisions.
The taximeter effect has only a minor effect on the flatrate bias. For participants
without a season ticket it is slightly significant and has a negative effect on the flatrate bias.
Although participants do agree to the annoying effect of having to pay each time they con -
sume the service they are less likely to opt for a flat fare tariff.
The self-discipline effect shows a positive effect for the group of season ticket holders
while it has only a neglectable effect on the non season ticket holders. Since the effect is not
significant for both groups it can not be proven if season ticket holders use the flat fare tariff
as motivational help for using public transportation.
The price of the flat fare tariff has a negative effect on the chances to opt for it which
is inline with the expected behavior. Both the minimum and maximum usage numbers have a
positive effect on the chances to choose a flat fare tariff. Our results show that the lower usage
threshold has a higher impact on the likelihood to choose a flat fare tariff: the odds-ratio for
the minimum usage number is 1.262 while for the maximum usage number it's only 1.174.
This is contrary to results from Gerpott (2009) and Nunes (2000) who analyzed the
flatrate bias in the mobile telephone market. They concluded that with an increasing ratio of
(maximum usage - break-even volume) to (break even volume - minimum usage) the probab-
ility of choosing a flat fare tariff increases too. This “ratio rule” would mean that only relative
changes between the usage levels and the break-even volume matter which is not sufficient to
describe the results we found. One possible reason might be the inclusion of a minimum us -
age level of zero in our scenarios.
We analyzed the effect of the socio-demographic attributes sex, age and employment
status on the choice of a flat fare tariff too. There were no significant differences regarding the
impact of the four effect variables on the flatrate bias. Nevertheless we found minor differ-
ences with regard to price sensitivity. For people above 60 years of age and female parti -
cipants in general the price of the flat fare tariff had a smaller influence on the tariff choice
compared to the other groups.
12
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39

Wirtz, Vortisch, Chlond: Flatrate Bias in Public Transportation: Magnitude and Reasoning
CONCLUSION
Previous studies in other service areas than public transportation have shown that there
is a tendency to choose flat fare tariffs even if the break-even point is not reached. Our results
clearly show that users of public transportation show a flatrate bias too. By analyzing mobility
data we showed that there is a significant difference between the number of customers using
the pay-per-use option but would be better off with the flatfare option and the number of cus-
tomers opting for the flat fare option but would be better off not to do so. The surplus on fare
payments due to the flatrate bias might be as high as 11.8 %.
We formulated a hypothesis claiming that the flatrate bias is created by for different ef-
fects: insurance effect, taximeter effect, convenience effect and self-discipline effect. Using
survey data we showed that the insurance effect and the convenience effect are responsible for
the flatrate bias. When implementing AFC-Systems with modern tariffs concepts should be
implemented to deal with these two effects properly.
As shown by the quantitative analysis of the flatrate bias the insurance and the con-
venience effects represent valuable utilities for the customer. When offering tariffs that fulfill
the customer needs with regard to these two effects then they are a clever way to provide ad-
ditional service the customer is willing to pay for. If a tariff system is introduced which does
not provide such service the lacking revenue needs to be compensated.
Since the taximeter effect has only a minor impact pt users should not be annoyed by
the pure fact that modern tariffs might base their price on the actual usage level. Furthermore
it became obvious that the minimum usage level plays a crucial role in the decision for choos -
ing a flat fare tariff. Especially if the minimum usage level means to not use the service at all.
The minimum usage level has a higher impact on the flatrate bias than the maximum usage
level.
Our study has only focused on the service area of public transportation. While keeping
the focus on the individual it might be interesting to see whether people behave consistently
with regard to the flatrate bias in other types of services where flat fare options are available.
Or are there intrapersonal variations in the flatrate bias depending on the sevice area under fo-
cus. As far as we know this question has not been answered yet.
ACKNOWLEDGEMENTS
This paper is base on a research supported by the Deutsche Forschungsgemeinschaft
(DFG). Additionally we very much appreciate the support from the transport association Kre-
isVerkehr Schwäbisch Hall GmbH for providing the trip data of their AFC-System.
REFERENCES
(1) Kenneth E. Train, Daniel L. McFadden, and Moshe Ben-Akiva. The demand for local
telephone service: A fully discrete model of residential calling patterns and service
choices. The RAND Journal of Economics, 18:109–123, 1987.
13
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35

Wirtz, Vortisch, Chlond: Flatrate Bias in Public Transportation: Magnitude and Reasoning
(2) Timo Schulze and Karen Gedenk. Biases bei der tarifwahl und ihre konsequenzen für die
preisgestaltung. Zeitschrift für Betriebswirtschaft, 75(2):157–184, 2005.
(3) Jan Krämer and Lukas Wiewiorra. Beyond the flat rate bias: The flexibility effect in tariff
choice. Telecommunications Policy, 36(1):29–39, 2012. ISSN 0308-5961. doi:
10.1016/j.telpol.2011.11.015.
(4) Torsten J. Gerpott. Biased choice of a mobile telephony tariff type: Exploring usage
boundary perceptions as a cognitive cause in choosing between a use-based or a flat rate
plan. Telematics and Informatics, 26(2): 167–179, 2009. ISSN 0736-5853. doi:
10.1016/j.tele.2008.02.003.
(5) Michael D. Grubb. Selling to overconfident consumers. American Economic Review,
99:1770–1807, 2009.
(6) Joseph C. Nunes. A cognitive model of people’s usage estimations. Journal of Marketing
Research, 37:397–409, 2000.
(7) Bernd Lambrecht, Anja; Skiera. Ursachen eines flatrate-bias - systematisierung und mes-
sung der einflussfaktoren. Schmalenbachs Zeitschrift für betriebswirtschaftliche
Forschung, 58:588-617, 2006.
(8) Stefano DellaVigna and Ulrike Malmendier. Paying not to go to the gym. American Eco-
nomic Review, 96(3):694–719, June 2006.
(9) Dirk Zumkeller, Oliver Lipps, and Bastian Chlond. The german mobility panel: Options,
limitations, and the complementary use of secondary data. In International Conference on
Transport Survey Quality and Innovation, Grainau, Germany, 1997.
(10) Claudia Nobis. Multimodality: facets and causes of sustainable mobility behavior. Trans-
portation Research Record: Journal of the Transportation Research Board, 2010(1):35–
44, 2007.
(11) Tobias Kuhnimhof, Ralph Buehler, Matthias Wirtz, and Dominika Kalinowska. Travel
trends among young adults in germany: increasing multimodality and declining car use
for men. Journal of Transport Geography, 24:443–450, 2012.
(12) Neal Lathia and Licia Capra. How smart is your smartcard?:measuring travel behaviours,
perceptions, and incentives. In Proceedings of the 13th international conference on Ubi-
quitous computing, UbiComp ’11, pages 291–300, New York, NY, USA, 2011. ACM.
ISBN 978-1-4503-0630-0. doi: 10.1145/2030112.2030152.
(13) William J. Conover. Practical nonparametric statistics. Wiley series in probability and
statistics : applied probability and statistics section. Wiley, New York, NY, 3. ed. edition,
1999. ISBN 0-471-16068-7.
(14) Daniel Kahneman and Amos Tversky. Prospect theory: An analysis of decision under risk.
Econometrica, 47(2):263–292, 1979. ISSN 00129682.
14

Wirtz, Vortisch, Chlond: Flatrate Bias in Public Transportation: Magnitude and Reasoning
(15) Richard H. Thaler. Mental accounting matters. Journal of Behavioral Decision Making,
12:183–206, 1999.
(16) John P. Kling and Stephen S. Van der Plög. Telecommunications Demand Modelling,
chapter Estimating local call elasticities with a model of stochastic class of service and us-
age choice, pages 119–136. North-Holland, 1990. ISBN 0-444-88539-0.
(17) Klaus Wertenbroch. Consumption self-control by rationing purchase quantities of virtue
an vice. Marketing Science, 17(4):pp. 317–337, 1998. ISSN 07322399.
(18) David W. Hosmer and Stanley Lemeshow. Applied logistic regression. Wiley series in
probability and statistics. Wiley, New York [u.a.], 2. ed. Edition, 2000. ISBN 0-471-
35632-8.
15