Content uploaded by Lennart Adenaw
Author content
All content in this area was uploaded by Lennart Adenaw on Nov 21, 2022
Content may be subject to copyright.
Available via license: CC BY-NC-SA 4.0
Content may be subject to copyright.
ANATION-WIDE EXPERIMENT:FUEL TAX CUTS AND ALMOST
FREE PUBLIC TRANSPORT FOR THREE MONTHS IN GERMANY -
REPORT 5 INSIGHTS INTO FOUR MONTHS OF MOBILITY
TRACKING
Lennart Adenaw
Technical University of Munich
TUM School of Engineering and Design
Chair of Automotive Technology
Boltzmannstrasse 15, 85748 Garching
lennart.adenaw@tum.de
David Ziegler
Technical University of Munich
TUM School of Engineering and Design
Chair of Automotive Technology
Boltzmannstrasse 15, 85748 Garching
david.ziegler@tum.de
Nico Nachtigall
Technical University of Munich
TUM School of Engineering and Design
Chair of Automotive Technology
Boltzmannstrasse 15, 85748 Garching
nico.nachtigall@tum.de
Felix Gotzler
Technical University of Munich
TUM School of Engineering and Design
Chair of Automotive Technology
Boltzmannstrasse 15, 85748 Garching
felix.gotzler@tum.de
Allister Loder
Technical University of Munich
TUM School of Engineering and Design
Chair of Traffic Engineering and Control
Arcisstrasse 21, 80333 Munich
allister.loder@tum.de
Markus B. Siewert
Munich School of Politics and Public Policy
TUM Think Tank
Richard-Wagner-Straße 1, 80333 München
markus.siewert@hfp.tum.de
Markus Lienkamp
Technical University of Munich
TUM School of Engineering and Design
Chair of Automotive Technology
Boltzmannstrasse 15, 85748 Garching
lienkamp@tum.de
Klaus Bogenberger
Technical University of Munich
TUM School of Engineering and Design
Chair of Traffic Engineering and Control
Arcisstrasse 21, 80333 Munich
klaus.bogenberger@tum.de
November 21, 2022
ABS TRACT
In spring 2022, the German federal government agreed on a set of measures that aim at reducing
households’ financial burden resulting from a recent price increase, especially in energy and mobility.
These measures include among others, a
nation-wide
public transport ticket for 9 EUR per month
and a fuel tax cut that reduces fuel prices by more than
15 %
. In transportation research this is an
almost unprecedented behavioral experiment. It allows to study not only behavioral responses in
mode choice and induced demand but also to assess the effectiveness of transport policy instruments.
We observe this natural experiment with a
three-wave
survey and an
app-based
travel diary on a
sample of hundreds of participants as well as an analysis of traffic counts. In this fifth report, we
present first analyses of the recorded tracking data. 910 participants completed the tracking until
September, 30th. First, an overview over the
socio-demographic
characteristics of the participants
arXiv:2211.10328v1 [econ.GN] 18 Nov 2022
APREPRINT - NOV EMB ER 21, 2022
within our tracking sample is given. We observe an adequate representation of female and male
participants, a slight
over-representation
of young participants, and an income distribution similar to
the one known from the "Mobilität in Deutschland" survey [
1
]. Most participants of the tracking study
live in Munich, Germany. General transportation statistics are derived from the data for all phases of
the natural experiment – prior, during, and after the
9 EUR-Ticket
– to assess potential changes in
the participants’ travel behavior on an aggregated level. A significant impact of the
9 EUR-Ticket
on modal shares can be seen. An analysis of the participants’ mobility behavior considering trip
purposes, age, and income sheds light on how the
9 EUR-Ticket
impacts different social groups
and activities. We find that age, income, and trip purpose significantly influence the impact of the
9 EUR-Ticket on the observed modal split.
1 Introduction
In transportation research, it is quite unlikely to observe or even perform real-world experiments in terms of travel
behavior or traffic flow. There are few notable exceptions: subway strikes suddenly make one important alternative
mode not available anymore [
2
,
3
], a global pandemic changes travelers’ preferences for traveling at all or traveling
collectively with others [
4
], or a bridge collapse forces travelers to alter their daily activities [
5
]. However, in 2022 the
German federal government announced in response to a sharp increase in energy and consumer prices a set of measures
that partially offset the cost increases for households. Among these are a public transport ticket at 9 EUR per month
1
for
traveling all across Germany in public transport, except for long-distance train services (e.g., ICE, TGV, night trains), as
well as a tax cut on gasoline and diesel, resulting in a cost reduction of about 15 % for car drivers
2
. Both measures
are limited to three months, namely June, July and August 2022. As of mid June, more than 16 million tickets have
been sold
3
, while it seems that the fuel tax cut did not reach consumers due to generally increased fuel prices and oil
companies are accused of not forwarding the tax cuts to consumers 4.
For the Munich metropolitan region, Germany, we designed a study comprising three elements. The three elements
are: (i) a three-wave survey before, during and after the introduction of cost-saving measures; (ii) a smartphone app
based measurement of travel behavior and activities during the same period; (iii) an analysis of aggregated traffic counts
and mobility indicators. We will use data from 2017 (pre-COVID-19) and data from shortly before the cost reduction
measures as the control group. In addition, the three-wave survey is presented to a German representative sample. The
main goal of the study is to investigate the effectiveness of the cost-saving measures with focus on the behavioral impact
of the
9 EUR-Ticket
on mode choice [
6
], rebound effects [
7
,
8
], and induced demand [
9
]. Further details on the study
design and the first results can be found in our four previous reports [10, 11, 12, 13].
In this fifth report, we first provide an assessment of the sociodemographics of the sample group actively using our
smartphone tracking app in
Section 2
. In
Section 3
, we supply aggregated transportation statistics from the tracking data
that has been recorded from beginning of the experiment to end of September 2022.
Section 4
analyzes the mobility of
the participants regarding their
spatio-temporal
activity behavior taking into account the different
socio-demographic
groups among our participants. The results of the exploratory data analysis presented herein, are discussed in
Section 5
.
2 Tracking Participants
This section elaborates on the
socio-economic
characteristics of the 910
GPS-tracked
participants who took part in
the study from
May 24, 2022
via the “Mobilität.Leben” smartphone application available on iOS and Android devices
and recorded at least one trip until
September 30, 2022
. All
socio-economic
parameters were
self-reported
by the
participants within the survey phase of the study. For general results of this phase, the interested reader is referred to
our previous report [
11
]. This evaluation especially focuses on the tracked participants, constituting a subsample of all
surveyed participants.
1https://www.bundesregierung.de/breg-de/aktuelles/9-euro-ticket-2028756
2https://www.bundesfinanzministerium.de/Content/DE/Standardartikel/Themen/Schlaglichter/
Entlastungen/schnelle-spuerbare-entlastungen.html
3https://www.tagesschau.de/wirtschaft/unternehmen/neun-euro-ticket-135.html
4https://apnews.com/article/politics-business-germany-prices-deb85a000d63cd57b76446d9c90c3e18
2
APREPRINT - NOV EMB ER 21, 2022
2.1 Age Distribution of Participants
Figure 1
shows the relative age distribution of male and female participants within our tracking sample. For reference,
the demographic age distribution for Germany in 2022 is depicted. Comparing the age distribution of our study group
with the German age distribution, the following observations can be made: for male participants between 40 to 80
years, the participant distribution aligns well with the German age distribution, whereas participants of both genders
with an age between 20 and 40 are
over-represented
. Women between the age of 65 and 80 are
under-represented
in
comparison to men in the same age group. Participants younger than 18 do not occur as a minimum age of 18 is required
to participate in the study. Due to the general availability of data within all relevant age groups, we conclude that with
appropriate sample weights we can make our tracking sample representative from an age and gender perspective in
future analyses. Since this report is focused on a descriptive analysis of the recorded data, we leave an extrapolation
using such weights to later research.
4.0% 2.0% 0% 2.0% 4.0%
Population share
15
20
25
30
35
40
45
50
55
60
65
70
75
80
85
Age in years
Male (study)
Female (study)
Male (Germany)
Female (Germany)
Figure 1: Age distribution of tracked participants vs. age distribution for Germany in 2022
2.2 Gender Distribution of Participants
We asked participants to report their gender as one of three categories: male,female, and diverse. The sample contains
approximately equal numbers of female and male participants (422 female, 484 male and 4 diverse), as shown in
Figure 2.
female - 46.4%
diverse - 0.4%
male - 53.2%
Figure 2: Gender distribution of tracked participants
3
APREPRINT - NOV EMB ER 21, 2022
2.3 Income of Participants
We evaluated participants’ income levels using threshold definitions already known from the Mobilität in Deutschland
(MiD) [
1
] study, as shown in
Figure 3
. The evaluation shows that with a share of
35.5 %
most households have a
net income of more than
5,500 C
per month, followed by two groups of
21.2 %
each, of households with incomes
between
4,000 C
and
5,500 C
and between
2,500 C
and
4,000 C
, respectively.
10.5 %
of the sample households have
between
1,500 C
and
2,500 C
at their monthly disposal. The number of participants in lower household income groups
is significantly smaller with 9.5 % living on less than 1,500 C per month. 2.2 % did not disclose their income.
N.A.; 2.2%
0 to 1500 EUR; 9.5%
1500 to 2500 EUR; 10.5%
2500 to 4000 EUR; 21.1%
4000 to 5500 EUR; 21.1%
over 5500 EUR; 35.5%
Figure 3: Income of tracked participants
2.4 Dwelling
Participants of the tracking experiment mainly live in the south east of Germany, where the study was first announced,
as shown in
Figure 4a
. A majority resides in the greater Munich area and most within the city of Munich and its central
and thus most interconnected public transport (PT) tariff-zone (zone M) of the local PT agency
MVV
. (
Figure 4b
).
Concludingly, PT is easily available to the tracking participants.
100
200
300
400
500
Nr. of Participants
(a) Complete sample group
MVV zone M
MVV service area
Administrative boundaries of Munich
5
10
15
20
25
Nr. of Participants
(b) Study participants in Munich area
Figure 4: Residences of sample group
4
APREPRINT - NOV EMB ER 21, 2022
3 Transport Statistics
In this section, an overview over aggregated transport statistics (traveled distances, modal split) is presented together
with an update on participation rates. All analyses are performed based on tracking data recorded between
May 24, 2022
and
September 30, 2022
. This includes the total time range from beginning of June until the end of August during
which the
9 EUR-Ticket
was valid in Germany. During this observation period,
910
individuals have recorded a trip
with the provided smartphone application at least once. In
Figure 5
the amount of active participants per week can be
seen. In it, a participant is considered to be active in a given week if they recorded at least one trip in that week. It can
be seen that more than
700
participants have been active each week after the first week of the study. During the first
week of the experiment, around
300
participants were recorded. The first week of our experiment was also the last
week before the
9 EUR-Ticket
was introduced. After a high increase of active participants in the second and third
week participation peaked in the fourth week with around
820
active users. During the following weeks, the amount of
participants slowly decreased with 700 active participants in the last week of our observation period.
Figure 5: App usage: Number of participants who recorded at least one trip during the previous week
A total of
670,096
trips was recorded within the observation period assessed in this report. Based on participants’
feedback via the smartphone application,
4982
trips were recorded incorrectly due to technical errors. The remaining
665,114
trips have a total length of
6,550,250 km
. Of these, walking was the most prominent mode of transport with
372,430
reported trips, followed by individual transport, e.g. car, with
108,944
trips and
101,265
trips by public
transport. Trips by bike were recorded in
80,088
cases and by airplane and all other forms of transport in
786
and
1601
respectively. While the share of trips conducted by plane is low, accounting for only
0.1 %
of all recorded trips,
distances travelled by plane account for
21.2 %
of the recorded mileage. Due to this disbalance, we exclude air travel
from the following analysis to not skew our results. We also disregard all recorded trips with
non-specified
forms of
transport. Our used smartphone application can accept feedback from users. They can correct the detected mode of
transport, mark a trip as incorrect, or merge two trips as one. These corrections were made for
36,151
trips. It is also
possible to confirm the recorded trips. This has been done by our participants for 430,199 trips.
Figure 6: Modal share based on the traveled distance. From top to bottom: data derived from MiD [
1
] for metropolis
regions; recorded data with our smartphone app before, during, and after the 9 EUR-Ticket period
To assess the impact of the
9 EUR-Ticket
on the general travel behavior of our sample we divided the reported data
into three parts: the first part includes all trips recorded before the
9 EUR-Ticket
was introduced on June 1, 2022. The
second part contains all recorded trips between June 1, 2022 and August 31, 2022. In this time period the
9 EUR-Ticket
was valid. The last and third part includes all trips between September 1 and September 30, 2022 and represents
the time period directly after the
9 EUR-Ticket
.
Figure 6
reveals the modal share of our sample group in terms of
traveled distances. Herein, modal shares are depicted for all phases of the experiment. For reference, modal shares for
5
APREPRINT - NOV EMB ER 21, 2022
metropolitan regions as taken from the MiD survey are shown. To allow for a more concise presentation, we aggregate
the means of transport as implemented in the smartphone app in the following manner:
•Public Transport: subway, light rail, regional train, train, bus, tram
•Bicycle: bicycle, bikesharing, e-bicycle, kickscooter
•Car: car, e-car, motorbike, taxi, uber, carsharing
•Walk: walk
A comparison of the 2017 modal shares from the MiD study to our data reveals principle differences between the data
sets. While MiD reports
32.4 %
public transport,
5.4 %
bicycle, and
59.5 %
car use while indicating
2.7 %
of walking,
we observe a higher use of public transport and a lower car use within all phases of the experiment. Setting the modal
shares in the three phases of our experiment side by side, it can be seen that the introduction of the
9 EUR-Ticket
had
no clear impact on the share of distance covered by bicycle. This remained almost constant at
6.7 %
percent and
7.2 %
percent. After the
9 EUR-Ticket
program ended, the share decreased to
5.9 %
, most likely due to decreasing
temperatures (the mean temperature in Munich was
20.2°C
in the time range before and during the
9 EUR-Ticket
and decreased to
14.2°C
afterwards)
5
. In contrast, the recorded share of walking clearly decreased from
5.4 %
of the
total recorded mileage before to
4.7 %
during the
9 EUR-Ticket
albeit almost constant weather conditions. This could
indicate that the hurdle to use public transport for smaller distances diminished for our sample group with the cheap
ticket.
The observed share of individual transport and public transport usage behaves the other way around: before the
9 EUR-Ticket
was introduced, the share of individual transport was
49.4 %
. It then decreased by
6.6 %
during
the time the ticket was valid. In the observed
time-frame
after the
9 EUR-Ticket
the distance share of individual
transport increased by
6.2 %
to a total share of
49.0 %
; slightly less than before the
9 EUR-Ticket
. Initially, our sample
group used public transport for
38.5 %
of the recorded distance. This share increased by
6.8 %
to
45.3 %
with the
9 EUR-Ticket before it decreased to 40.0 % when the 9 EUR-Ticket program ended.
Figure 7: Average traveled distance in km per active user, transport mode and day
In
Figure 7
the total travel distance per day and mode of transport for an average person from our sample group is shown
for working days and
non-working
days comprising Saturdays, Sundays, and public holidays. Note that from hereon no
data from before the
9 EUR-Ticket
is shown, because the relatively small sample size in this data set caused by a smaller
number of participants and a shorter observation
time-frame
appears to be unsuited for lower level disaggregations.
Instead, we compare the results of our sample group with results from the MiD study.
Overall it can be seen that if a person of our sample group is active on a
non-working
day the average travel distance
with all modes of transport is equal or higher than on an average day during the week. The only two exceptions are the
bicycle and the public transport usage in the period after the
9 EUR-Ticket
. The walking distance remained almost
constant over the two time periods. During the week our participants walked
2.5 km
and on
non-working
days between
2.8 km
and
2.9 km
. It is noteworthy that MiD reports significantly smaller walking distances, probably due to the
fact that people were not tracked but surveyed. The distance our sample group traveled by public transport decreased
clearly after the
9 EUR-Ticket
period. On an average working day by
4.4 km
and even more on a
non-working
day on
average by
7.9 km
. Nevertheless, the distance traveled by individual transport did not increase by the same amount.
We recorded an increase on working days by
0.9 km
and on
non-working
days by
1.9 km
. This means that the total
distance traveled by an average participant decreased especially on
non-working
days. After the
9 EUR-Ticket
ended
5Calculated based on the public available LMU weather data for Munich https://www.meteo.physik.uni-muenchen.de/
DokuWiki/doku.php?id=wetter:stadt:messung
6
APREPRINT - NOV EMB ER 21, 2022
on
August 31, 2022
the distance traveled by public transport decreased to
20.7 km
on working days and to
19.1 km
on
non-working
days. The distance traveled by means of individual transport increased to
22.1 km
on working days.
On
non-working
days the traveled distance by individual transport increased to
33.5 km
per user and day. The total
recorded distance per user and day amounts to
52.7 km
during and
48.4 km
after the
9 EUR-Ticket
on working days
and
65.8 km
and
57.7 km
on
non-working
days. In contrast, MiD reports an average travel distance of
37 km
per day
and person as well as lower average distances for each individual mode. This difference is partly based in the fact that
MiD values were determined for all participants including those who did not report any trip on a given day. Since our
data considers only the mobile part of our sample population,
i.e.
persons that moved at least once during the day, we
adjusted the MiD reference values in
Figure 7
based on the MiD mobility rate. Nevertheless, a significant difference
remains in all categories as well as in total mileage. The reasons for this gap may be subject to future research.
4 Mobility Behavior Analysis
In the subsequent section of this report, we investigate the mobility behaviour of the sample group. Thereby, we focus
on the effects the
9 EUR-Ticket
on the modal shares of different demographic groups, described by age and income, as
well as varying trip purposes. In contrast to previous sections, we do not analyze the period before the introduction of
the
9 EUR-Ticket
in this section. This is due to the further subdivision of trips within the phases, e.g. by trip purposes,
which results in significant differences in remaining total trip quantities and finally would lead to very small sample
sizes for certain segments within the phase before the
9 EUR-Ticket
. Instead, we use the MiD [
1
] as a reference and
compare the results of our sample group to the national average for German metropolises. We do only consider trips
conducted within the geographic borders of Germany.
In our analyses of the modal splits, we calculate the respective shares based on traveled distance by respective modes.
We apply the concept of main trip purposes (Hauptwegezwecke) from MiD, which allocates the complete distance
of an intermodal trip to the means of transport with the highest share of the total trip length. In order to facilitate
concise graphic representations of the results, we aggregate trip purposes available in the smartphone application in the
following way:
•home: home
•work/education: study, work
•leisure: sport, eat, family and friends, leisure
•errand: assistance, medical visit, errand
•shopping: shopping
•unknown: unknown, other
For the same reason, we aggregate means of transport in the manner described in
Section 3
. Due to the relatively
small share of walking trips in respect to the total distance and for the sake of clarity and simplicity, this mode is
not considered in the subsequent analyses. Values are normalized to the total distance covered with the remaining
three means of transport. In order to account for the known differences in mobility behavior between working and
non-working
days [
14
], we strictly distinguish between said working days and weekends and public holidays (once
more grouped as non-working days).
An overview over the general mobility behavior of the tracking group is supplied in
Figures 8a and 8b
. The purpose
was either automatically inferred by the smartphone application based on e.g. points of interest or manually assigned by
participants. As only around
5 %
of all recorded trips and activities are manually adjusted [
11
], unknown locations
make up a large number of tracked trips. When looking at subsequent statistics on trip purposes, note that roughly
half of all trips recorded were assigned the label unknown.
Figure 8a
shows the distributions of distances of trips with
different purposes. Trips with purpose work/education exhibit both the longest distance and the largest dispersion, with
the 3rd quartile extending up to
40 km
. Trip types errand and shopping show similar distributions in regard to trip
lengths. Both are relatively short with medians around
4 km
and smaller dispersion compared to the other purposes. A
second cluster with similar distance characteristics consists of purposes home,leisure, and unknown. Their medians are
located around 6 km. The 3rd quartile whisker reaches 30 km.
A kernel density estimation of the start times of trips with different purposes is depicted in
Figure 8b
. Following Scott’s
rule [
15
], the kernel density estimation uses a Gaussian kernel with bandwith
1.8·n(−1/(6))
for
n
samples. With regard
to the temporal distribution of trip start times, a large peak for trip type work/education occurs in the morning around 8
o’clock as to be expected. In contrast,
home-trips
accumulate in the late afternoons and evenings. Trip types errand and
shopping occur in two peaks each, in the morning and the afternoon with a slight dip in between. Trip type shopping
is shifted approximately one hour towards the evening. Trips with purpose leisure increase continuously throughout
7
APREPRINT - NOV EMB ER 21, 2022
(a) distances of trips by purpose (b) temporal distribution of trips by purpose
Figure 8: General analysis of different trip purposes
the day and peak at 17 o’clock. Trips that were assigned to type unknown do not show clear similarities to one of the
aforementioned types and can thus not be reassigned on these grounds.
4.1 Modal Split for Trip Purposes
(a) working days (b) non-working days
Figure 9: Modal Splits for different trip purposes, calculated using respective distances
With questioning prevailing mobility habits and routines being named one of the greatest opportunities of the
9 EUR-Ticket 6
, we examine the change in modal shares of PT, bicycle, and car for different trip purposes in
Figures 9a and 9b
. It can generally be stated that PT usage is higher on working days than on
non-working
days
with differences of
5 %
to
10 %
for respective purposes. The modal splits of our sample group are approximately similar
to the one obtained from MiD on
non-working
days (with exceptions, especially errands) but significantly higher on
working days. Trips of purpose errands yield a substantially smaller share of PT and a larger share of car respectively,
compared to the other purposes both for working and
non-working
days. For all but one purpose, the share of PT
significantly increased during the
9 EUR-Ticket
period. Purpose work/education during the week is the only group
where no clear increase in PT is evident. The decreases in PT ridership range from single digits to 10 %. The share of
cycling is mostly constant, meaning most trips not conducted with PT after the
9 EUR-Ticket
period were replaced by
car. The largest drop in PT ridership can be observed for trip purpose leisure on working days. On
non-working
days,
type shopping’s PT ridership dropped the most.
6https://www.bundesregierung.de/breg-de/themen/deutsche-einheit/faq-9-euro- ticket-2028756
8
APREPRINT - NOV EMB ER 21, 2022
4.2 Modal split by income group
(a) working days (b) non-working days
Figure 10: Modal Splits for different income groups, calculated using respective distances
The second major goal of the
9 EUR-Ticket
was to relieve the burden on households whose cost of living had risen
sharply as a result of current price developments, which especially impacts households with lower incomes. We
therefore investigate the impact of the ticket on different income groups in Figures 10a and 10b.
For all income groups, PT usage is higher during the week than on weekends and generally exceed the one from
MiD. The results of our sample group illustrate how PT ridership generally decreases inversely proportional to income.
Furthermore, the desired effect of the
9 EUR-Ticket
(increased PT ridership) on working days likewise relates inversely
proportional to income. On
non-working
days, the picture is different. The largest drop in PT usage of approximately
20 % is to be observed for the group with the lowest income while the other groups show similar absolute decreases,
although from different starting points. The trips were entirely replaced by motorized individual transport. The share of
cycling is lowest for income group
0 - 1500C
, especially on working days. With one exception, PT ridership increased
on all occasions during the period of the
9 EUR-Ticket
. This exception is posed by the income group
more than 5,500 C
,
where the level stayed constant. While the share of cycling did not change significantly between phases, income group
1,500C - 2,500C
increased their share of cycling on working days after the end of the
9 EUR-Ticket
and replaced a
significant amount of trips conducted wit PT before.
4.3 Modal Split of Age Groups
Concluding the analysis of the
9 EUR-Ticket
’s impact on demographic groups, we explore the effects on the modal
splits of different age groups in
Figures 11a and 11b
. The data of our sample group reveals a general trend of
decreasing/increasing PT/car ridership, respectively, proportionally to age. This trend is stronger on working days,
while the modal split of age groups between 40 and 70 years on
non-working
days is approximately similar. Another
finding valid for all age groups is a higher PT usage on working days than on
non-working
days. Compared to MiD,
the share of PT is higher except age groups above 40 years on
non-working
days. Distances traveled by bicycle are
comparable, although the youngest age group distinctively exhibits the smallest relative share of active mobility on both
kinds of days.
On working days, the age groups 30-39 and 40-49 both yield higher shares of PT usage after the
9 EUR-Ticket
although
the absolute increase is small. The opposite is true for all other age groups on working days. On
non-working
days,
the only group showing a higher PT usage after the period of the
9 EUR-Ticket
is the one below 20 years of age. The
largest drop of PT usage is to be found for the 20 to 29 year old participants on
non-working
days. Further comparably
large decreases are evident for the oldest two age groups above 60 years, on working as well as
non-working
days,
although a large share of participants is likely to be retired.
9
APREPRINT - NOV EMB ER 21, 2022
(a) working days (b) non-working days
Figure 11: Modal Splits for different age groups, calculated using respective distances
5 Discussion and Conclusion
This report presents first analyses of the data collected within the first four months of the tracking experiment of the
”Mobilität.Leben” study. As a baseline for the understanding and interpretation of all subsequent analyses, it assesses
age, gender, income, and dwellings of the tracking participants. Subsequently, general transport statistics, as well as
disaggregated statistics on activities and the usage of different means of transport are presented with a special focus on
modal splits. This section discusses our findings and the limitations of both the study and the employed methodology.
A first limitation of the study is constituted by the size of our sample group as well as the available
time-frame
for the
assessment of
pre-9 EUR-Ticket
behaviors. While participation rates have shown to be constantly high throughout
the reported
time-frame
, only one week of data before the introduction of the
9 EUR-Ticket
exists. In addition, due to
the successive
start-up
phase, far fewer participants submitted data during this first week than in later phases of the
experiment. Seasonal effects may also be present during all phases of the experiment since all data was recorded in
less than a year’s time. This lack of data, especially in the first phase of our experiment, motivated the decision to
leave out any data recorded before the start of the
9 EUR-Ticket
for more disaggregated analyses. In consequence
for purpose- or
subgroup-specific
statistics, there is no direct reference from before the
9 EUR-Ticket
from within
the study group. When analyzing the impact of the
9 EUR-Ticket
by
e.g.
a comparison of modal splits during and
after its introduction, potential backlashes from
re-increased
pricing or other compensatory mechanisms may also
influence mode choice without any indication in the presented data. Although the MiD survey can serve as an alternative
reference for
pre-9 EUR-Ticket
behaviors to some extent, large differences between MiD and our surveys can be
observed especially for modal splits. These differences appear to be in the same order of magnitude as the observed
influences of the
9 EUR-Ticket
indicating that no clear baseline of mobility behavior prior to the
9 EUR-Ticket
can be
established. Considering the differences in survey methodology of MiD and ”Mobilität.Leben” supports this argument.
Apart from the particularities concerning the first phase of the experiment, it should be noted that all the results presented
in this report are initially only representative of the group of participants observed. This group is – as shown in
Section 2
– relatively wealthy, a little younger than the German population, and includes more male than female participants.
Whether and to which extent observations regarding this study group can be extrapolated onto larger populations in and
outside of Germany remains unclear.
Another point worthy of discussion is the way trip purposes and means of transport are labeled, which are – for the best
part – not surveyed directly, but inferred from raw GPS trajectories supplied by the participants. This is done based on
the algorithms employed by the MOTIONTAG smartphone app. While tracking participants have the option to correct
any labeling manually and thereby directly reveal activity types and mode choice, only around
5 %
of all recorded trips
and activities are manually adjusted [11]. Approximately half of all recorded activities retain an unknown label.
Despite all limitations and reservations due to sample size, quality, and sample group composition, a clear positive
effect of the
9 EUR-Ticket
on the aggregated modal share on PT usage is revealed by our experiment for both working
days and non-working days.
The disaggregated modal splits shown in
Section 4
have revealed significant influences of trip purpose, income, and age
on the modal split and PT utilization respectively. Most notably, work and education trips during workdays were hardly
affected by the
9 EUR-Ticket
. In contrast, errands, leisure, and shopping activities were preceded by PT trips more
10
APREPRINT - NOV EMB ER 21, 2022
often during the
9 EUR-Ticket
program than afterwards. Independent of the experiment phase, increasing household
incomes correlate with decreasing PT usage and increased car usage in our sample. Additionally, the availability of
a reduced ticket price for PT appears to affect
low-income
households significantly stronger than their
high-income
counterparts especially on
non-working
days. Similarly, age presents to have an adverse effect on PT usage in our
sample. Nevertheless, the relative reduction of PT utilization after the end of the
9 EUR-Ticket
is especially large for
participants under 30 and over 60 years of age.
Acknowledgements
The authors would like to thank the TUM Think Tank at the Munich School of Politics and Public Policy led by Urs
Gasser for their financial and organizational support and the TUM Board of Management for supporting personally
the genesis of the project. The authors thank the company MOTIONTAG for their efforts in producing the app at
unprecedented speed. Further, the authors would like thank everyone who supported us in recruiting participants,
especially Oliver May-Beckmann and Ulrich Meyer from M Cube and TUM, respectively. This project is partially
funded by the Bavarian State Ministry of Science and the Arts in the framework of the bidt Graduate Center for
Postdocs.
References
[1]
Bundesministerium für Verkehr und digitale Infrastruktur. Mobilität in Deutschland Ergebnisbericht. Technical
report, Bonn, 2018.
[2]
Michael L Anderson. Subways, strikes, and slowdowns: The impacts of public transit on traffic congestion.
American Economic Review, 104:2763–2796, 2014.
[3]
Martin W Adler and Jos N van Ommeren. Does public transit reduce car travel externalities? quasi-natural
experiments’ evidence from transit strikes. Journal of Urban Economics, 92:106–119, 2016.
[4]
Joseph Molloy, Thomas Schatzmann, Beaumont Schoeman, Christopher Tchervenkov, Beat Hintermann, and
Kay W. Axhausen. Observed impacts of the covid-19 first wave on travel behaviour in switzerland based on a
large gps panel. Transport Policy, 104:43–51, 4 2021.
[5]
Shanjiang Zhu, David Levinson, Henry X. Liu, and Kathleen Harder. The traffic and behavioral effects of the
I-35W Mississippi River bridge collapse. Transportation Research Part A: Policy and Practice, 44:771–784, 12
2010.
[6]
Moshe E Ben-Akiva, Steven R Lerman, Steven R Lerman, et al. Discrete choice analysis: theory and application
to travel demand, volume 9. MIT press, 1985.
[7]
Lorna A. Greening, David L. Greene, and Carmen Difiglio. Energy efficiency and consumption — the rebound
effect — a survey. Energy Policy, 28:389–401, 6 2000.
[8]
Kent M. Hymel, Kenneth A. Small, and Kurt Van Dender. Induced demand and rebound effects in road transport.
Transportation Research Part B: Methodological, 44:1220–1241, 12 2010.
[9]
Claude Weis and Kay W Axhausen. Induced travel demand: Evidence from a pseudo panel data based structural
equations model. Research in Transportation Economics, 25:8–18, 2009.
[10]
Allister Loder, Fabienne Cantner, Lennart Adenaw, Markus Siewert, Sebastian Goerg, Markus Lienkamp, and
Klaus Bogenberger. A nation-wide experiment: fuel tax cuts and almost free public transport for three months in
germany – report 1 study design, recruiting and participation, 2022.
[11]
Fabienne Cantner, Nico Nachtigall, Lisa S. Hamm, Andrea Cadavid, Lennart Adenaw, Allister Loder, Markus B.
Siewert, Sebastian Goerg, Markus Lienkamp, and Klaus Bogenberger. A nation-wide experiment: fuel tax cuts
and almost free public transport for three months in germany – report 2 first wave results. 2022.
[12]
Allister Loder, Fabienne Cantner, Andrea Cadavid, Markus B. Siewert, Stefan Wurster, Sebastian Goerg, and
Klaus Bogenberger. A nation-wide experiment: fuel tax cuts and almost free public transport for three months in
germany – report 3 second wave results, 2022.
[13]
Allister Loder, Fabienne Cantner, Andrea Cadavid, Markus B. Siewert, Stefan Wurster, Sebastian Goerg, and
Klaus Bogenberger. A nation-wide experiment: fuel tax cuts and almost free public transport for three months in
germany – report 4 third wave results, 2022.
[14]
Regine Gerike and Angelika Schulz. Workshop synthesis: Surveys on long-distance travel and other rare events.
Transportation Research Procedia, 32:535–541, 2018.
[15] David W. Scott. Multivariate Density Estimation. Wiley, August 1992.
11
