Exploratory: Sustainable Cities for Citizens

The COVID-19 pandemic is changing the world in unprecedented and unpredictable ways. Human mobility is at the epicenter of that change, as the greatest facilitator for the spread of the virus. To study the change in mobility, to evaluate the efficiency of mobility restriction policies, and to facilitate a better response to possible future crisis, we need to properly understand all mobility data sources at our disposal. This post regards a work dedicated to the study of private mobility sources, gathered and released by large technological companies. This data is of special interest because, unlike most public sources, it is focused on people, not transportation means. i.e., its unit of measurement is the closest thing to a person in a western society: a mobile phone. Furthermore, the sample of society they cover is large and representative. On the other hand, this sort of data is not directly accessible for anonymity reasons.

We consider the use of private data sources (Google and Facebook) for assessing the levels of mobility in Spain. By doing so, we draw conclusions on two fronts. First, on the behavior and particularities of private data sources. And second, on how mobility changed during the COVID-19 pandemic in Spain.

Regarding private data sources, we have shown the differences between using an absolute measure (like Facebook) and a relative measure (like Google). Both of them have limitations when used in isolation. The former lacks a contextualization of its values, while the latter depends entirely on the baseline used. When used together, they provide a visualizing of mobility where consistent patterns can be easily identified (as presented later in this section). For specific purposes, using a single data source may suffice, as long as it fits the goal:

●      An absolute measure like Facebook's can be very useful for epidemiologic purposes, as it provides a pure measurement of mobility. That includes estimating number of contacts in a society, modeling the spread of the virus, and measuring the impact of policies on absolute mobility;

●      A relative measure like Google's can be very useful for socio-economic purposes, as it provides a contextualized measurement of mobility. That includes understanding the change caused by the new normality, and the economic impact of mobility restriction policies.

Regarding the analysis of Spanish mobility during the COVID-19 pandemic, we extract several conclusions. On one hand, data shows a huge mobility containment, sustained for a month and a half (March 15 to May 1st, approximately), very close to its theoretical limit (as represented by mobility during the hard-lockdown). This duration was sufficient to contain the spread of the virus and bring infection numbers down to traceable scale. In hindsight, the policies implemented in Spain seem appropriate and proportional to the severity of the situation. That being said, the role, timing and convenience of the hard-lockdown remains to be further discussed. We show a relatively modest contribution of this policy to mobility reduction. On the other hand, the hard-lockdown may have had an effect on prolonging adherence.

We identify mild differences between regions during the three months of restricted movement. Certain regions had a stronger adherence to confinement than others, mostly in relative terms. This may be caused by regional differences in pre-pandemic mobility, which is used as baseline for the relative measurement. A similar artifact are the inverted peaks of weekends, where a relative measure spikes down and an absolute measure spikes up. As demonstrated, this is the result of combining a measure relative to the weekday with an absolute measure.

We also saw significant differences among days. Weekends exhibit the highest volume of mobility reduction in absolute terms, even during the hard-lockdown, when work-related trips were forbidden for all except essential services. At the same time, weekends have the smallest mobility reduction in relative terms, indicating that the effort society had to make in this regard with respect to its previous patterns was smaller. Fridays and Sundays are particularly relevant days, the first because it represents the biggest change from normal behavior, the second because it represents the biggest absolute decrease in mobility. These particularities could be exploited for the general good.

Finally, we analyzed the new normality by looking at the weeks of de-confinement, up until June 27, a week after the state of alarm was lifted on the whole of Spain. In this period, we found Saturdays and Sundays to be already at pre-pandemic levels of mobility. In contrast, working days (Monday to Friday) still show significant differences. The new normality also shows differences between regions, particularly for working days. Regions with large metropolitan areas exhibit a reduction in mobility between 4% and 14% after restrictions were lifted. Indeed, the new normality is most new on urban working days.

The paper [1] describing this research has been recently published and it is available here: http://arxiv.org/abs/2007.07095

Written by: Dario Garcia Gasulla, Senior Researcher, HPAI group, Computer Science Department, Barcelona Supercomputing Center (BSC), dario.garcia@bsc.es

Revised by: Agnese Bonavita, Luca Pappalardo

[1] Raquel Pérez Arnal, David Conesa, Sergio Alvarez-Napagao, Toyotaro Suzumura, Martí Català, Enric Alvarez, Dario Garcia-Gasulla, “Private Sources of Mobility Data Under COVID-19”, 7 2020.