Observed impacts of the Covid-19 first wave on travel behaviour in Switzerland based on a large GPS panel

Abstract

In Switzerland, strict measures as a response to the Covid-19 pandemic were imposed on March 16, 2020, before being gradually relaxed from May 11 onwards. We report the impact of these measures on mobility behaviour based on a GPS tracking panel of 1,439 Swiss residents. The participants were also exposed to online questionnaires. The impact of both the lockdown and the relaxation of the measures up until the middle of August, 2020 are presented. Reductions of around 60% in the average daily distance were observed, with decreases of over 90% for public transport. Cycling increased in mode share drastically. Behavioural shifts can even be observed in response to the announcement of the measures and relaxation, a week before they came in to place. Long-term implications for policy are discussed, in particular the increased preference for cycling as a result of the pandemic.

Full text

Journal Pre-proof
Observed impacts of the Covid-19 first wave on travel behaviour in Switzerland based
on a large GPS panel
Joseph Molloy, Thomas Schatzmann, Beaumont Schoeman, Christopher
Tchervenkov, Beat Hintermann, Kay W. Axhausen
PII: S0967-070X(21)00015-9
DOI: https://doi.org/10.1016/j.tranpol.2021.01.009
Reference: JTRP 2490
To appear in: Transport Policy
Received Date: 16 December 2020
Accepted Date: 23 January 2021
Please cite this article as: Molloy, J., Schatzmann, T., Schoeman, B., Tchervenkov, C., Hintermann, B.,
Axhausen, K.W., Observed impacts of the Covid-19 first wave on travel behaviour in Switzerland based
on a large GPS panel, Transport Policy, https://doi.org/10.1016/j.tranpol.2021.01.009.
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition
of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of
record. This version will undergo additional copyediting, typesetting and review before it is published
in its final form, but we are providing this version to give early visibility of the article. Please note that,
during the production process, errors may be discovered which could affect the content, and all legal
disclaimers that apply to the journal pertain.
© 2021 The Author(s). Published by Elsevier Ltd.

Observed impacts of the Covid-19 first wave on travel behaviour in
Switzerland based on a large GPS panel
Joseph Molloya,∗,Thomas Schatzmanna,Beaumont Schoemanb,Christopher Tchervenkova,
Beat Hintermannband Kay W. Axhausena
aIVT, ETH Zurich, CH-8093, Zurich, Switzerland
bWWZ, Uni Basel, Basel, Switzerland
ARTICLE INFO
Keywords:
Covid-19
Corona
GPS
Mobility
Tracking-App
ABSTRACT
In Switzerland, strict measures as a response to the Covid-19 pandemic were imposed on March
16, 2020, before being gradually relaxed from May 11 onwards. We report the impact of these
measures on mobility behaviour based on a GPS tracking panel of 1,439 Swiss residents. The
participants were also exposed to online questionnaires. The impact of both the lockdown and
the relaxation of the measures up until the middle of August, 2020 are presented. Reductions of
around 60% in the average daily distance were observed, with decreases of over 90% for public
transport. Cycling increased in mode share drastically. Behavioural shifts can even be observed
in response to the announcement of the measures and relaxation, a week before they came in to
place. Long-term implications for policy are discussed, in particular the increased preference
for cycling as a result of the pandemic.
1. Introduction
The sudden onset of the Covid-19 pandemic in early 2020 dramatically altered the rhythms of daily life around the
world. Within the space of a few weeks, borders were closed, lockdowns imposed and local economies brought to a
stand-still in the attempt to contain the spread of the virus. Schools were closed, and children had their first experience
of remote learning. After decades of unfulfilled predictions of the move towards teleworking, firms were forced to
make the shift almost overnight.Those under lockdown orders were generally only allowed outside for shopping and
small amounts of exercise. As a consequence, mobility behaviour became unrecognisable.
Understanding the impact these restrictions have had on mobility has been challenging, given the pace of the
change. New policies were being implemented constantly; classic survey methods would not have kept pace with the
dynamic changes to people’s travel patterns. In this paper we present both the method and descriptive results from an
app-based mobility survey panel of over 1500 people which was recruited as the pandemic took hold in Switzerland
in mid-March. Uniquely, this panel was recruited from a previous mobility survey undertaken in autumn 2019. Thus,
a baseline behaviour for each participant in the study is available against which their mobility behaviour under these
extra-ordinary circumstances can be understood.
The paper is structured as followed. First a timeline of the pandemic as it developed in Switzerland is presented,
including the various restrictions and relaxations which occurred. The period up to the end of July is covered in this
paper. A brief overview of GPS-based mobility studies in the literature is presented, as well as other available sources
of Covid-19 mobility data. In the methods and data section, the recruitment and app are presented. The results section
presents some of the key insights from the mobility tracking. In the discussion the implications of these findings for
transport policy in the near future are discussed.
1.1. Covid-19 timeline in Switzerland
The first confirmed case of Covid-19 was registered in Switzerland on February 25, 2020. The onset followed
that in Northern Italy, which was one of the earliest hotspots outside China. The situation deteriorated quickly and by
March 9, over 100 people had been infected. At the time of writing, July 31 2020, 791,725 people have been tested in
∗Corresponding author
joseph.molloy@ivt.baug.ethz.ch (J. Molloy); thomas.schatzmann@ivt.baug.ethz.ch (T. Schatzmann);
beaumont.schoeman@unibas.ch (B. Schoeman); christopher.tchervenkov@ivt.baug.ethz.ch (C. Tchervenkov);
beat.hintermann@unibas.ch (B. Hintermann); axhausen@ivt.baug.ethz.ch (K.W. Axhausen)
ORCID (s):
J Molloy et al.: Preprint submitted to Elsevier Page 1 of 16
Journal Pre-proof

Observed impacts of the Covid-19 first wave on travel behaviour in Switzerland based on a large GPS panel
Switzerland, over 35,000 people have been infected with the virus and more than 1,700 have died, which translates to
infection and death rates per 100,000 inhabitants of 38 and 2.4, respectively.
On March 16, the Federal Council declared an extraordinary situation, which gives it widespread competencies
that allowed it to order the introduction of uniform measures in all cantons. Non-essential businesses were closed,
along with schools, recreational facilities and public parks. Only food stores, post offices and healthcare institutions
were not affected. Furthermore, checks on the borders to Germany, Austria, France and Italy were introduced. Entry to
Switzerland from its four large neighbours was only possible for Swiss citizens, persons holding a residence permit for
Switzerland and those who have to enter Switzerland for work-related reasons. Employers were urged to reorganise the
working hours of their employees to avoid rush hour travel. Wherever possible, home-office was to be implemented.
To minimise the risk of infection, the Federal Office of Public Health (FOPH) recommended avoiding the use of public
transport wherever possible. People who had symptoms of respiratory disease and people over 65 years of age were
told to not use public transport. National and regional passenger transport services were maintained to support the
functioning of the economy and society. International passenger transport was maintained to countries with open
borders.
The number of confirmed Covid-19 cases in Switzerland increased rapidly. On March 20, more than 4,800 infec-
tions and 43 fatalities were reported. The canton of Ticino on the border to Northern Italy was hit especially hard by
the corona virus. The Federal Council decided to step up measures and forbid gatherings of more than 5 people in
public places, and an inter-personal distance of 2 meters was mandated for groups with fewer than 5 people. However,
unlike in neighboring countries as for example in Austria, Belgium, France, Italy, Luxembourg and Spain, people were
allowed to leave the house during the lockdown. Public transport continued operation, though at reduced frequency.
The peak rate of daily infections and Covid19-related deaths was in late March and early April. The statistics were
subject to weekly fluctuations and showed fewer cases towards the end of the week. On April 5, Switzerland reported
a total 158,000 Covid-19 tests, 15% (21,100) of which were positive, and 559 deaths, one of the highest incident rates
in Europe. After that, infections started to decrease as a consequence of the containing measures. Despite the fact
that the measures put in place to combat the virus were being followed well by the public and were having the desired
effect, the Federal Council extended the measures until April 26.
On April 27, hospitals were able to resume all medical procedures, including non-urgent procedures, and a first
group of businesses was allowed to re-open (garden centers, florists, hairdressing salons and cosmetic studios). The
Federal Council’s strategy to emerge from the lockdown was structured into three phases, with transitions depending
on the number of new infections, hospital admissions and deaths as well as hospital occupancy rates. In a first phase
from April 27 to May 11, measures were eased on businesses where a low level of direct contact is possible, where
precautionary measures can easily be put in place, and where there will be no significant movements of people. As
opposed to other countries, there was no general obligation for healthy people to wear a mask. Keeping a safe distance
and washing hands was seen as the most effective protective measures. Still, sick people were advised to stay home in
isolation.
On May 11, Switzerland moved to phase 2 by further easing measures as the spread of the corona virus had
continued to slow. Even though rules on hygiene and social distancing still applied, most types of businesses opened
again, including restaurants, along with mandatory school for students up to grade 9. Museums and libraries were
allowed to re-open and sports activities in small groups of up to 5 people were permitted as well. In parallel, restrictions
on entering Switzerland were relaxed and scheduled public transport services increased significantly. Furthermore, the
Federal Council decided to introduce contact tracing for new infections with Covid-19.
Entering phase 3 of the emergency plan on June 6, all events of up to 300 persons and spontaneous gatherings for
up to 30 people were allowed again. High schools and universities were able to resume, all leisure and entertainment
businesses plus tourist attractions re-opened. Switzerland reported a cumulative total of 30,988 infections, 1,663
Covid-related deaths and only 16 new cases. The number of new infections, hospitalisations and deaths continued to
fall despite measures being eased, and stabilised at a low level. The extended powers of the federal government expired
on June 19.
From June 22 onwards, the start of the fourth phase, most of the measures put in place were lifted and only the ban
on large-scale events remains in place until at least the end of August. Interestingly, the Federal council reduced the
minimum distance that should be kept between two people from 2 metres to 1.5 metres and strongly recommended to
wear a face mask when using public transport if it is not possible to maintain the necessary distance. It also lifted the
recommendation to work from home, leaving the decision to the employer, which is required to protect the health of
staff by putting in place appropriate measures. In contrast to the first wave, the prime responsibility in the event of a
J Molloy et al.: Preprint submitted to Elsevier Page 2 of 16
Journal Pre-proof

Observed impacts of the Covid-19 first wave on travel behaviour in Switzerland based on a large GPS panel
renewed increase in Covid-19 cases now rests with the 26 cantons in Switzerland.
In response to an increase in new daily infections in mid-June and increased ridership on public transport, the
Federal Council made wearing masks compulsory on public transport throughout Switzerland starting from July 6.
Since mid-June, the number of new corona virus cases has been rising in Switzerland as infected persons have entered
the country from countries both within and outside the Schengen area. Consequently, travellers entering Switzerland
from certain regions must quarantine for ten days upon their return.
1.2. Tracking studies in the literature
The use of GPS tracking in mobility studies is becoming an integral data collection method in transportation.
Where traditionally, such surveys required participants to carry a GPS logging device which had to be returned at the
end of the study for the data to be collected, more recent methods allow users to install an app on their smartphone,
which uses the GPS receiver and other location services on the phone to detect the location. These apps run in the
background, and depending on the phone model, have a minimal effect on battery life, something which used to be
extremely detrimental to such studies. For example, in the original MOBIS study sample on which this study is
based, only 12.5% of participants who started tracking were lost over the 8 week study duration (Molloy, Tchervenkov,
Hintermann and Axhausen,2020).
In recent years there have been an increasing number of studies which have used either a separate GPS device
(Nielsen,2004;Wargelin, Stopher et al.,2012;Livingston,2011), or a smartphone app (Allström, Kristoffersson and
Susilo,2017;Greene, Flake, Hathaway and Geilich,2012;Safi, Assemi, Mesbah, Ferreira and Hickman,2015;Stopher,
Daigler and Griffith,2018;Nahmias-Biran, Han, Bekhor, Zhao, Zegras and Ben-Akiva,2018). This data is then usually
segmented into trips (or sometimes lower level segments) and activities, on which the mode and trip purpose are
imputed. The usefulness of this method is indicated by multiple sources that identify trip under-reporting in traditional
paper-based travel diaries (Janzen, Vanhoof, Smoreda and Axhausen,2018;Wolf, Loechl, Thompson and Arce,2003;
Stopher, FitzGerald and Xu,2007).
1.3. Other Covid-19 mobility data sources
The rapid onset of the Covid-19 pandemic meant that there was little opportunity to set up new tracking studies
to measure changes in mobility. However, there were a number of data sources which were already available. The
most prominent were those from Apple (Apple,2020) and Covid-19 Community Mobility Reports (Google,2020).
The Apple reports monitor the number of search requests by transport mode (driving, walking, public transport) as an
indicator for mobility. The Google reports use records of visits to land use types (i.e. parks), based on anonymised
data collected from google apps and mobile devices. Both these reports present only overall aggregate numbers and
do not include analysis of socio-demographic differences. For the baseline, the Apple reports take January 13, 2020,
and Google uses the period January 3 – February 6, 2020. Another available resource for Switzerland is produced by
Intervista AG, in collaboration with the Zurich Statistics Bureau (Intervista AG,2020). The Intervista sample consists
of 2561 persons (daily average) and is broadly representative, with socio-demographic attributes for the participants.
Data is also available since the start of 2020.
1.4. The impacts of pandemics on mobility behaviour
It has been widely acknowledged that transportation is a key driver in the spread of infectious diseases (Baroyan
and Rvachev,1967;Herrera-Valdez, Cruz-Aponte and Castillo-Chavez,2011). The important role of mobility in a
pandemic has been demonstrated for historical pandemics such as the Spanish flu in 1918 (Trilla, Trilla and Daer,
2008;Ammon,2002). For a comprehensive overview of studies exploring the link between transport and infectious
diseases, see Muley, Shahin, Dias and Abdullah (2020). However, the impact of measures to overcome pandemics also
have an impact on mobility. In the much of the recently published transport literature on the Covid-19 pandemic, the
focus has been drawing conclusions on the impact of restrictions on the spread of the disease, with little focus on the
specific impacts on the restrictions on mobility itself within the population. In the following paragraphs a few studies
which do focus on mobility are highlighted.
In the current pandemic, mobile data has played a key role in understanding the changes in both regional and global
mobility during the pandemic. Many countries are now using mobile data to understand the effectiveness of measures,
including Austria, Belgium, Chile, China, Germany, France, Italy, Japan, Spain, United Kingdom and the United
States (Oliver, Lepri, Sterly, Lambiotte, Deletaille, De Nadai, Letouzé, Salah, Benjamins, Cattuto et al.,2020;Yabe,
Tsubouchi, Fujiwara, Wada, Sekimoto and Ukkusuri,2020). In particular, Heiler, Reisch, Hurt, Forghani, Omani,
J Molloy et al.: Preprint submitted to Elsevier Page 3 of 16
Journal Pre-proof

Observed impacts of the Covid-19 first wave on travel behaviour in Switzerland based on a large GPS panel
Hanbury and Karimipour (2020) used real-time anonymised mobile phone data to understand the changes in mobility
behaviour as a result of the introduced measures in Austria during the first wave. They saw a doubling of the number
of persons with a radius of gyration (activity space) of less than 500m, and increased segmentation of the community
structure.
In the USA, Badr, Du, Marshall, Dong, Squire and Gardner (2020) found a strong correlation between reduced
mobility behaviour and decreased Covid-19 case growth rates. Furthermore, they show evidence that behavioural
changes were already observable days to weeks before movement restriction policies were implemented, indicating
that it was not just the introduced policy measures that restricted mobility, but the desire of persons to avoid the
pandemic.
All these data sets use data collected from mobile phone providers, which are less effective at capturing mobility
changes at local urban scales. Socio-demographic differences (e.g. across age, gender and household-size) can also
not be explored with this data. This is important for understanding how the mobility of different groups was affected
by the pandemic.
Other work has used travel surveys to explore the impact of the Covid-19 pandemic on mobility. Beck and Hensher
(2020) performed a household survey of 1457 respondents. The main conclusions of this paper focus on telework
(working-from-home) and suggest that there could be significant implications for localised transport networks. The
shift towards more flexible working conditions may also lead to a spreading of peak-hours.
Huang, Wang, Fan, Zhuo, Sun and Li (2020) used location data from Baidu in China to examine the impacts on
mobility and activity patterns as a result of the pandemic. They suggest that the government should promote cycling
as means of transport and the construction of bike lanes.
2. Methods and Data
2.1. Recruitment
Starting in September 2019, a sample of 5,375 persons living in Switzerland were recruited to a mobility pricing
field experiment called MOBIS (MObility Behaviour in Switzerland). The study required the participants to use an app
(available for both Android and iPhone) which used the phones location services to track their location, and identify
trip stages and activities. They were tracked for 8 weeks to investigate their response to a conceptual mobility pricing
scheme. 3680 persons completed the 8 weeks, with the last person finishing in January. They did not have to uninstall
the app afterwards, and could continue using it. At the start of March 2020, around 300 people were still using the
app. The tracking panel for the MOBIS-Covid19 study was recruited from those 3680 participants who completed
the MOBIS study. Of these participants, around 1,600 volunteered to reactivate the tracking app, Catch-my-Day,
developed by Motion-Tag. Table 1presents the distribution of the sample by different socio-demographics, showing
that the MOBIS-Covid19 sample is broadly representative. Our sample is slightly more educated and more likely to
be employed - but this is primarily since those over 65 years of age were not invited to the original MOBIS study.
Participants were only eligible to participate in the MOBIS study if they used a car at least 2 days a week. This also
skewed our sample towards car drivers, compared to the Swiss population.
Weekly reports have been produced online in three languages (English, German and French) since the start of
the MOBIS-Covid19 study, to which both participants and the public have access. Participants also have access to
a custom dashboard of their own mobility behaviour during the crisis, as long as they continue tracking. This was
designed as a form of non-monetary incentive to encourage continued participation.
J Molloy et al.: Preprint submitted to Elsevier Page 4 of 16
Journal Pre-proof

Observed impacts of the Covid-19 first wave on travel behaviour in Switzerland based on a large GPS panel
Table 1
Comparison of MOBIS-Covid19 (MOBIS) sample with the last national travel diary Mikrozensus (MZ) 2015
N %
Variable Value MOBIS MZ MOBIS MZ
No 13 2,587 0.8 6.2
Yes 1,480 31,824 90.4 75.8Access to car
Sometimes 144 7,584 8.8 18.1
Under 18 - 7,546 - 13.2
(18,25] 259 5,126 16.9 9.0
(25,35] 272 8,117 17.7 14.2
(35,45] 314 8,789 20.4 15.4
(45,55] 359 9,560 23.4 16.7
(55,65] 333 7,384 21.7 12.9
Age
66+ - 10,568 - 18.5
German 1,191 39,023 72.8 68.4
French 446 14,450 27.2 25.3Correspondence language
Italian - 3,617 - 6.3
Mandatory 96 9,954 5.9 19.3
Secondary 758 25,594 46.3 49.5Education
Higher Ed. 783 16,124 47.8 31.2
Female 828 28,928 50.6 50.7
Gender Male 809 28,162 49.4 49.3
1 209 19,439 12.8 34.0
2 546 20,222 33.4 35.4
3 331 7,400 20.2 13.0
4 417 7,133 25.5 12.5
Hosuehold size
5 or more 134 2,897 8.2 5.1
4 000 CHF or less 92 10,139 5.6 17.8
4 001 - 8 000 CHF 470 18,728 28.7 32.8
8 001 - 12 000 CHF 495 9,945 30.3 17.4
12 001 - 16 000 CHF 267 3,878 16.3 6.8
More than 16 000 CHF 168 2,593 10.3 4.5
Income
Prefer not to say 143 11,807 8.7 20.7
Employed 1,196 27,521 73.1 48.2
Self-employed 111 4,133 6.8 7.2
Apprentice 10 1,488 0.6 2.6
Unemployed 59 1,423 3.6 2.5
Student 83 2,114 5.1 3.7
Other 110 9,421 6.7 16.5
Main employment
Retired 68 10,990 4.2 19.3
Swiss 1,517 43,347 92.7 75.9
Nationality Other 120 13,743 7.3 24.1
J Molloy et al.: Preprint submitted to Elsevier Page 5 of 16
Journal Pre-proof

Observed impacts of the Covid-19 first wave on travel behaviour in Switzerland based on a large GPS panel
2.2. App based tracking
The Catch-my-Day app works similarly to other tracking apps. It runs in the background, collecting location data
from the phone, with the participant’s consent. The tracking can be turned on and off at any time. The data is transferred
to the server when the phone is connected to a WiFi network, where the location points are processed to detect the
transport stages, activity locations, mode of travel and where possible, trip purpose. The trip purpose detection is based
on the activity type nominated by the participant for other activities at the same location. The detection algorithms are
based on Openstreetmap data (OpenStreetMap contributors,2017) and available public transport schedules. Users are
able to validate the imputations made by the algorithms and make corrections.
Since not all activities have an assigned purpose in the data collected from the app, we impute the missing activity
purposes for the rest of the data using a random forest approach, using the reported activity purposes as the ground-truth
dataset.
2.3. Online surveys
The tracked data was supplemented by online surveys, which the participants were invited to participate in by email.
Initially, socio-demographic information about participants was collected in the original MOBIS study. Participants
were also asked in May to complete another 2 surveys, one during the lockdown, and one after, asking about their
working conditions and experience during the lockdown.
2.4. Sample weighting
As mentioned in above, our sample is skewed towards car-users in urban areas, due to the sample composition
of the original MOBIS study. To make some correcrtion for this, the MOBIS-Covid sample is weighted against the
original 22,000 participants which filled out the introductry survey in the MOBIS Study. This allowed a weighting
of the MOBIS-Covid results by age, gender, income, education, mobility tool ownership and accessibility. Weights
were calculated for each person-week in both the MOBIS and MOBIS-Covid study periods, against the sample present
in that week. This accounts for the changing sample composition over the weeks of the study. The weights are both
reasonable and stable across the study period, with the weights in the range 0.5-3.
3. Results
Across all segments of the population, a downward trend in the average travelled Kilomters per day was observed
before the lockdown measures were put in place on March 16. This indicates that the Swiss population preempted
the preventative measures, and were proactively taking steps to protect themselves from the virus by reducing their
mobility. This is also a consequence of the clear messaging from the Swiss federal government, which mostly gave
forewarning of at least a week before measures were implemented.
Mobility behaviour was most suppressed in the first week of the lockdown, however, the average daily distance and
the number of trips per day immediately started increasing, demonstrating the challenge of sustaining a suppressed
mobility demand over a long period through non-policed lockdown measures.
3.1. Impact of home-office and Kurzarbeit (Short-work)
During lockdown, 20% of participants reported being on kurzarbeit (Short-work). kurzarbeit is an industrial-
relations policy that exists predominately in German speaking countries, where a employee in the scheme has their
hours reduced by between 20 and 95% due to extreme economic circumstances. They are then paid the majority of
their usual wage, with the difference between hours worked and hours paid covered by government finances.
Surprisingly, being on kurzarbeit had little effect on the number of trips per day during the lockdown, compared
to those on normal conditions. The amount of home-office (telework) a participant was performing is much more
influential (see Fig. 2and Fig. 3). After the lockdown was lifted, the difference between the groups has reduced as
more of those who were allowed home-office have returned to the workplace. However, differences still persist, clearly
indicating that home-office is effective for suppressing travel demand.
3.2. The cycling boom
Significant changes in modal split were observed, particularly towards cycling, where a large increase in the av-
erage daily distance travelled was observed (see Fig. 1). The change in average daily distance is, on some weeks,
J Molloy et al.: Preprint submitted to Elsevier Page 6 of 16
Journal Pre-proof

Observed impacts of the Covid-19 first wave on travel behaviour in Switzerland based on a large GPS panel
Figure 1: Weekly change in average daily Km travelled
greater than 100%, which is well beyond what the seasonal causes would imply. The magnitude of the variations fluc-
tuates throughout the weeks, depending on the weather, which was excellent during the lockdown period, but variable
afterwards.
A large increase of the number of cycling trips per day was also observed, as can be seen in the temporal patterns
by time-of-day and type of day in Fig. 4. The red curve shows the average number of cycling trips started for each hour
of the day during the 2019 baseline period, whereas the blue curves show this for the last two weeks respectively. All
intermediate weeks are plotted in grey.
These temporal patterns indicate that cycling was primarily used as a leisure tool. First, the increase primarily
occurred on weekends, indicating a leisure objective. Second, although an increase in the number of trips was also
observed during the week, much of this occurred over midday, hinting that these trips were not conducted for commut-
ing purpose. Indeed, considering that a large number of commuting and station-access/egress trips did not take place
during the lockdown, this suggests that the use of cycling for other purposes has increased.
In Figure 5, the increase in mode share for different trip purposes is presented. The mode share compared to the
baseline period has increased for all trip purposes. The influence of the weather is visible across all purposes. The
increases were largest during the lockdown, but the trend has stabilised since then, at around a 75% increase over the
baseline period.
3.3. Socio-demographic variations during and after the lockdown
As indicated by the analysis of working conditions, particularly home-office capability, certain segments of the
working population were able to reduce their travel more than others. Fig. 6shows the reduction in the average daily
travelled kilometers by education level, as compared to the reference period in September/October 2019. Those with
J Molloy et al.: Preprint submitted to Elsevier Page 7 of 16
Journal Pre-proof

Observed impacts of the Covid-19 first wave on travel behaviour in Switzerland based on a large GPS panel
Figure 2: Trips/day by home-office arrangement
Figure 3: Trips/day by kurzarbeit status
a tertiary education (i.e. from a university or technical college) had a larger decrease in daily kilometers than the less
educated. Towards the end of the lockdown and post-lockdown, this difference became more pronounced. Those with
the mandatory education level experienced a rapid increase in their daily kilometers. This coincides with the return to
work for those in service industries, where professional degrees are not required. Travel per day for this segment has
plateaued in the post-lockdown period, indicating that a new daily pattern for this cohort has been reached.
The relaxation of the lockdown measures were met with a continued increase in travelled distance. All socio-
J Molloy et al.: Preprint submitted to Elsevier Page 8 of 16
Journal Pre-proof

Observed impacts of the Covid-19 first wave on travel behaviour in Switzerland based on a large GPS panel
Figure 4: Hourly bicycle trip counts
Figure 5: Increase in bicycle mode share by trip purpose
demographic groups saw a spike in the average daily distance as the lockdown measures were lifted, to varying degrees,
before a return to a trend similar to that observed during the lockdown.
The various adaptions to the lockdown measures by different household sizes is also insightful, as seen in Fig. 7.
Again, the total average weekly kilometers dropped by approximately 60%, with the reduction in travel correlated
with household size throughout the lockdown period. On one had, larger households had more travel constraints (e.g.
taking care of the children who now have to stay at home). Furthermore, those in smaller households, particularly
single person households had the smallest reduction in daily travel. It is important to note that no household had more
J Molloy et al.: Preprint submitted to Elsevier Page 9 of 16
Journal Pre-proof

Observed impacts of the Covid-19 first wave on travel behaviour in Switzerland based on a large GPS panel
Figure 6: Change in kilometers travelled by education level
than one person participating in the study, hence we cannot say how tasks such as shopping were redistributed among
the household members to reduce public exposure. It can also be reasonably inferred that those in larger households
had a larger incentive to reduce their daily travel. If one household member were to contract the virus, it would have
spread to the whole household. This ties to the idea of social pressure, with larger households resulting in a level of
normative behaviour within the household community.
The relaxation of the lockdown measures not only impacted travelled distance, but also had an effect on the average
number of trips per day and the activity space. Indeed, at the beginning of the lockdown, the number of daily trips
plummeted by nearly half and remained low during the entirety of the lockdown, whereas they have essentially returned
to pre-Covid numbers during the post-lockdown period (see Fig. 8). The evolution, however, was quite gradual. The
effect of the relaxation is much more evident on the activity spaces, as can be seen in Fig. 9. The lockdown caused
a reduction in the activity space to around 50𝑘𝑚2, which only slowly increased during the lockdown - indicating
that while people were travelling more, they were doing so within their local neighbourhood. Immediately after the
lockdown was lifted, the area of the weekend and holiday activity spaces increased. A short downward trend over
3 weeks followed, before the activity space started growing again. This post-lockdown downward trend was also
observed on weekdays, which one would not expect to be compatible with the rhythms of the 5-day work week. The
size of the activity spaces are now similar to what they were pre-Covid, both during the week and on weekends.
3.4. Change in road speeds
The general reduction in travel observed during the lockdown period also included a significant reduction in car
travel. With less vehicles on the roads, congestion levels dropped during the lockdown, and average travel speeds
across different trip-distance classes correspondingly increased.
Fig. 10 shows the effect of the Covid-19 crisis on median car travel speeds during the week, i.e. excluding weekends
and holidays. The black line indicates the baseline travel speeds from the baseline period in 2019. Car trips of over
500 km, which correspond to driving across Switzerland, or with an average speed of over 180 km/h were considered
unrealistic and removed from the analysis.
During the lockdown period from March 16 to May 11, an increase by up to 15 km/h in the peak-hour speeds was
observed, indicating a decrease in overall congestion. Reductions in congestion were observed during both morning
and evening peaks. However, no reduction was observed in the middle of day. Since the relaxation of the lockdown
J Molloy et al.: Preprint submitted to Elsevier Page 10 of 16
Journal Pre-proof

Observed impacts of the Covid-19 first wave on travel behaviour in Switzerland based on a large GPS panel
Figure 7: Change in kilometers travelled by household size
measures on May 11, peak-hour speeds have returned to almost pre-Covid-19 values, a sign that congestion is returning
to usual levels in peak periods. This is consistent with the fact that public transport use is still suppressed, while
car mileage is roughly back to the baseline (see Fig. 1). Post-lockdown, non-peak travel speeds during the day are
now lower than they were during the baseline period, particularly for the 20-50km trip-distance class. Viewed in
combination with Fig. 11, which shows the hourly counts for car trip departures in the two weeks post-lockdown, it
is evident that more car trips are being generated during the middle of the day than during the baseline period. This
is a natural consequence of the reduction in public transport usage, and has concerning implications for the coming
months if public transport demand remains suppressed.
3.5. Transport mode share shifts
The Covid-19 crisis also had had a clear impact on mode shares. While the overall distance travelled has recovered
since the lockdown and is now only 15% below pre-Covid values, public transport has been most affected. The daily
distance travelled on public transport modes is still 50% below the reference values, 8 weeks after the lockdown ended.
In contrast, the daily car distance is now essentially back to pre-Covid levels. The biggest winner has been cycling,
which has seen an increase in travelled distance throughout the crisis. However, it is also necessary to look at the
transport modal shifts that occurred.
The change in mode shares over the course of the Covid-19 crisis for different types of public transport subscriptions
- GA (National), Halbtax (50% discount) and other - can be best visualized with the help of ternary diagrams. Fig. 12
plots the shifts in mode shares by distance. The modes are grouped into the following categories:
•Motorized individual transport (car, motorbike, taxi, Uber)
•Public transport (bus, tram, ferry, metro, train)
•Non-motorized transport (walk, bike)
During the lockdown, a higher share of kilometers were performed using motorized individual and non-motorized
modes as compared to the reference period. After the lockdown, the share of public transport has increased and the
share of non-motorized modes has decreased slightly. The share of motorized individual modes it still greater than
J Molloy et al.: Preprint submitted to Elsevier Page 11 of 16
Journal Pre-proof

Observed impacts of the Covid-19 first wave on travel behaviour in Switzerland based on a large GPS panel
Figure 8: Average number of daily trips
Figure 9: Activity space (95% CI Ellipse)
during the reference period. The modal shifts were much larger for those with a full national subscription. Those
with other types of subscription - which are usually for a particular zone or connection - have not returned to public
transport in the same way that national ticket holders have.
J Molloy et al.: Preprint submitted to Elsevier Page 12 of 16
Journal Pre-proof

Observed impacts of the Covid-19 first wave on travel behaviour in Switzerland based on a large GPS panel
speeds-1.png
Figure 10: Road travel speeds
Figure 11: Car travel hourly counts
4. Discussion
The reductions in mobility have been drastic. The average daily kilometers travelled fell by 60% and 95% for car
and train modes respectively (see Fig. 1). Car travel has recovered slowly, back to the Autumn values within 2 weeks
of the lockdown measures being relaxed. However, public transport ridership at the same point was still low, at 20%
of pre-Covid levels, despite the importance of public transport as a commuting mode in Switzerland. This is likely a
result of many people still working from home. Indeed, further results show that the morning and evening peeks are
only slowly returning - and this return is dominated by car travel.
J Molloy et al.: Preprint submitted to Elsevier Page 13 of 16
Journal Pre-proof

Observed impacts of the Covid-19 first wave on travel behaviour in Switzerland based on a large GPS panel
Figure 12: Mode shares by distance during the Covid-19 crisis
The fall in the number of new corona cases in Switzerland since the introduction of the measures (Federal Office of
Public Health,2020) shows that restricting social interaction is a key requirement in dealing with the pandemic. The
home-office requirement aimed to reduce the possibility for disease transmission both in the workplace and on public
transport. Indeed, the daily number of public transport trips has fallen by around 90%.
An important question is to what extend home-office will be encouraged post-pandemic, and the implications for
transport policy. A reduction in peak-hour travel may reduce crowding and thus postpone or negate the need for further
infrastructure investment and dynamic demand measures, such as mobility pricing.
Furthermore, it remains to be seen if some of the behavioural shifts observed during this period will become
permanent. These shifts have been both positive and negative.
The increase in car usage during the day, with midday off-peak travelled kilometers now slightly above the 2019
baseline (see Fig. 11), suggests that people are driving more to avoid public transport. It remains to be seen if this
observed trend continues in the coming months. If and when workers eventually return to the workplace, this shift
from public transport to car usage would lead to additional congestion and be financially challenging for the public
transport operators. The obligatory wearing of masks was introduced on the July 6, in order to reduce transmission on
public transport. This will also be important for encouraging a return to public transport. Passengers may see public
transport as safe enough when masks are required. The alternative would have been other measures to increase public
transport capacity to allow social distancing, such as increasing the train length or providing more services. These
policy measures haven’t been implemented, with mask wearing deemed to be sufficient. Furthermore, the cost burden
is shifted to the passengers, who need to provide their own mask. If disposable masks are used, this effectively increases
the per trip cost of travel, which may further suppress public transport demand in the coming months. Tirachini and
Cats (2020) suggest that there is a possible risk that if the public transportation is seen as unsafe during the pandemic
due to crowding and a lack of social distancing measures, formed negative perceptions of public transportation will
become more prevalent and could persist after the pandemic, resulting in the formation of new undesirable habits.
Home-office will continue to play a role. Globally, many companies are continuing to allow or even encourage
home-office working (The most well publicised examples being the big tech companies). This is also the case in
Switzerland. In the post-lock period, we have already seen how congestion has increased in certain times of the day.
The Cycling boom observed since March also raises policy challenges for Switzerland. Other research has high-
lighted the negative effect of car traffic on willingness to cycle (Kaplan, Luria and Prato,2019), The lack of car
traffic during the lockdown undoubtedly made cycling more attractive. Sales of cycles increases dramatically, not just
in Switzerland (Matthias Heim,2020), but worldwide (Roger Harrabin,2020). Sustaining the observed increase in
cycling would be associated with health benefits (De Hartog, Boogaard, Nijland and Hoek,2010), but also with an
increase in accidents. The Covid-19 pandemic has shown a willingness in the population to take up cycling, and policy
measures have already been mooted or implemented in other jurisdictions to further encourage this trend (Keohane
J Molloy et al.: Preprint submitted to Elsevier Page 14 of 16
Journal Pre-proof

Observed impacts of the Covid-19 first wave on travel behaviour in Switzerland based on a large GPS panel
and Abboud,2020). In the countries there have been moves to introduce new temporary and permanent cycling in-
frastructure as a result of the pandemic. One of the key justifications is to simultaneously reduce demand for public
transport and try persuade public transport users to shift to active modes rather than use the car, which would increase
congestion. Such a policy has been suggested for Zurich, the largest city in Switzerland, and the initiative was approved
in the last elections in September 2020. This is a good step towards making these changes permanent.
In Handy, Van Wee and Kroesen (2014), the key research needs for promoting cycling are discussed. In partic-
ular, they note that even in pro-cycling cities, strong evidence around cycling behaviour is important for effectively
directly resources. To this aim, the wealth of data collected in the MOBIS-Covid19 study will prove invaluable in
highlighting areas which are underserved by cycling infrastructure. Lanzendorf and Busch-Geertsema (2014) identify
the importance of local government policy in increasing the cycling mode share. They also note that in the case studies
presented, the governments took advantage of favourable general conditions. With this in mind, governments should
take advantage of the observed cycling uptake as an opportunity to further develop their cycling strategies.
5. Conclusion
The application of app-based tracking, combined with online surveys of participants allowed the study of the
changes in mobility patterns caused by the Covid-19 pandemic. The study sample is broadly representative, with
a slight over representation of car owners. Important variations in travel reduction were observed across different
socio-demographic groups, which are obscured in other data sources. Finally, the implications for transport policy in
Switzerland are widespread, especially if further increases in congestion are to be avoided, and the observed uptake of
cycling to be made habitual.
References
Allström, A., Kristoffersson, I., Susilo, Y., 2017. Smartphone based travel diary collection: Experiences from a field trial in stockholm. Transporta-
tion research procedia 26, 32–38.
Ammon, C., 2002. Spanish flu epidemic in 1918 in geneva, switzerland. Eurosurveillance 7, 190–192.
Apple, 2020. Covid-19 Mobility Trends Reports. online, [https://www.apple.com/covid19/mobility], accessed 31/07/2020.
Badr, H.S., Du, H., Marshall, M., Dong, E., Squire, M.M., Gardner, L.M., 2020. Association between mobility patterns and covid-19 transmission
in the usa: a mathematical modelling study. The Lancet Infectious Diseases 20, 1247–1254.
Baroyan, O., Rvachev, L., 1967. Deterministic models of epidemics for a territory with a transport network. Cybernetics 3, 55–61.
Beck, M.J., Hensher, D.A., 2020. Insights into the impact of covid-19 on household travel and activities in australia–the early days of easing
restrictions. Transport policy 99, 95–119.
De Hartog, J.J., Boogaard, H., Nijland, H., Hoek, G., 2010. Do the health benefits of cycling outweigh the risks? Environmental health perspectives
118, 1109–1116.
Federal Office of Public Health, 2020. New coronavirus: Situation in switzerland. available online at: https://www.bag.
admin.ch/bag/en/home/krankheiten/ausbrueche-epidemien- pandemien/aktuelle-ausbrueche- epidemien/novel- cov/
situation-schweiz- und-international.html. Accessed: 2020-05-28.
Google, 2020. Covid-19 Community Mobility Reports. online, [https://www.google.com/covid19/mobility/], accessed 31/07/2020.
Greene, E., Flake, L., Hathaway, K., Geilich, M., 2012. A Seven-Day Smartphone-Based GPS Household Travel Survey in Indiana, in: TRB 95th
Annual Meeting Compendium of Papers, Transportation Research Board.
Handy, S., Van Wee, B., Kroesen, M., 2014. Promoting cycling for transport: research needs and challenges. Transport reviews 34, 4–24.
Heiler, G., Reisch, T., Hurt, J., Forghani, M., Omani, A., Hanbury, A., Karimipour, F., 2020. Country-wide mobility changes observed using mobile
phone data during covid-19 pandemic. arXiv preprint arXiv:2008.10064 .
Herrera-Valdez, M.A., Cruz-Aponte, M., Castillo-Chavez, C., 2011. Multiple outbreaks for the same pandemic: Local transportation and social
distancing explain the different" waves" of a-h1n1pdm cases observed in méxico during 2009. Mathematical Biosciences & Engineering 8, 21.
Huang, J., Wang, H., Fan, M., Zhuo, A., Sun, Y., Li, Y., 2020. Understanding the impact of the covid-19 pandemic on transportation-related
behaviors with human mobility data, in: Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery & Data
Mining, pp. 3443–3450.
Intervista AG, 2020. Mobilität in der Schweiz während der Covid-19-Pandemie (german). online, [https://www.intervista.ch/
mobilitaets-monitoring- covid-19], accessed 31/07/2020.
Janzen, M., Vanhoof, M., Smoreda, Z., Axhausen, K.W., 2018. Closer to the total? Long-distance travel of French mobile phone users. Travel
Behaviour and Society 11, 31–42.
Kaplan, S., Luria, R., Prato, C.G., 2019. The relation between cyclists’ perceptions of drivers, self-concepts and their willingness to cycle in mixed
traffic. Transportation research part F: traffic psychology and behaviour 62, 45–57.
Keohane, D., Abboud, L., 2020. Cycling lanes, wider pavements: How EU cities rethink public transport. Financial Times URL: https://www.
ft.com/content/09f183f3-57d9- 45b6-bf75- d4e43390f467.
Lanzendorf, M., Busch-Geertsema, A., 2014. The cycling boom in large german cities—empirical evidence for successful cycling campaigns.
Transport policy 36, 26–33.
J Molloy et al.: Preprint submitted to Elsevier Page 15 of 16
Journal Pre-proof

Observed impacts of the Covid-19 first wave on travel behaviour in Switzerland based on a large GPS panel
Livingston, J., 2011. Atlanta regional commission-regional travel survey-final report. PTV NuStata, Austin, TX .
Matthias Heim, 2020. Die Schweiz hat umgesattelt (german). SRF (Swiss Radio and Television) URL: https://www.srf.ch/news/
wirtschaft/neues-mobilitaetsverhalten- die-schweiz- hat- umgesattelt.
Molloy, J., Tchervenkov, C., Hintermann, B., Axhausen, K.W., 2020. MOBIS: An 8-week mobility pricing field experiment using GPS tracking.
Presented at Mobile Tartu.
Muley, D., Shahin, M., Dias, C., Abdullah, M., 2020. Role of transport during outbreak of infectious diseases: Evidence from the past. Sustainability
12, 7367.
Nahmias-Biran, B.h., Han, Y., Bekhor, S., Zhao, F., Zegras, C., Ben-Akiva, M., 2018. Enriching activity-based models using smartphone-based
travel surveys. Transportation Research Record 2672, 280–291.
Nielsen, O.A., 2004. Behavioral responses to road pricing schemes: Description of the danish akta experiment, in: Intelligent Transportation
Systems, Taylor & Francis. pp. 233–251.
Oliver, N., Lepri, B., Sterly, H., Lambiotte, R., Deletaille, S., De Nadai, M., Letouzé, E., Salah, A.A., Benjamins, R., Cattuto, C., et al., 2020.
Mobile phone data for informing public health actions across the covid-19 pandemic life cycle.
OpenStreetMap contributors, 2017. Planet dump retrieved from https://planet.osm.org . https://www.openstreetmap.org.
Roger Harrabin, 2020. Coronavirus: Boom time for bikes as virus changes lifestyles. BBC URL: https://www.bbc.com/news/
business-52564351.
Safi, H., Assemi, B., Mesbah, M., Ferreira, L., Hickman, M., 2015. Design and implementation of a smartphone-based travel survey. Transportation
Research Record 2526, 99–107.
Stopher, P., FitzGerald, C., Xu, M., 2007. Assessing the accuracy of the Sydney Household Travel Survey with GPS. Transportation 34, 723–741.
Stopher, P.R., Daigler, V., Griffith, S., 2018. Smartphone app versus gps logger: A comparative study. Transportation Research Procedia 32, 135 –
145. Transport Survey Methods in the era of big data:facing the challenges.
Tirachini, A., Cats, O., 2020. Covid-19 and public transportation: Current assessment, prospects, and research needs. Journal of Public Transporta-
tion 22, 1.
Trilla, A., Trilla, G., Daer, C., 2008. The 1918 “spanish flu” in spain. Clinical infectious diseases 47, 668–673.
Wargelin, L., Stopher, P., et al., 2012. GPS-based household interview survey for the Cincinnati, Ohio Region: executive summary report. Technical
Report. Ohio. Dept. of Transportation. Office of Research and Development.
Wolf, J., Loechl, M., Thompson, M., Arce, C., 2003. Trip rate analysis in GPS-enhanced personal travel surveys. Transport survey quality and
innovation 28, 483–498.
Yabe, T., Tsubouchi, K., Fujiwara, N., Wada, T., Sekimoto, Y., Ukkusuri, S.V., 2020. Non-compulsory measures sufficiently reduced human
mobility in tokyo during the covid-19 epidemic. Scientific Reports 10, 1–9.
J Molloy et al.: Preprint submitted to Elsevier Page 16 of 16
Journal Pre-proof

Journal Pre-proof

ERROR: undefinedresource
OFFENDING COMMAND: findresource
STACK:
/ADBCFF+STIXGeneral
/CIDFont
/ADBCFF+STIXGeneral
[0 -1000 250 0 ]
1
Journal Pre-proof

Highlights
• Understanding of the change in mobility patterns in Switzerland using a large GPS
Panel during the Covid19 Pandemic
• Evidence for a large increase in cycling kilometers during the first lockdown which
was sustained into the summer
• Changes in mobility behavior varied across different socio-demographic groups
Journal Pre-proof