Mobility detected by mobile phones
The COVID-19 pandemic in Spain. First wave: from the first cases to the end of June 2020
Monographs from the National Atlas of Spain.
Virtually everybody owns a mobile phone nowadays, most of which are smartphones. Citizens interact with them regularly, making and receiving calls, opening data sessions to check e-mails, searching for information on the internet, participating in social networks, etc. Mobile phone operators store this activity data (both calls and data sessions) for billing purposes, and this information may provide valuable raw material for studies on mobility. Mobile phone activity data provide real-time information on the user’s location at any point in time and on any trips they make. Analysing these data allows monitoring processes over time, such as the mobility of the population during the different phases of the pandemic.
Mobile phone activity data include the user identifier, the (anonymised) phone number that rings up, the number that receives the call (if any), the duration of the call or data session, the mobile communications antenna to which the phone connected and the time at which it connected (day, hour, minute and second). The position of a mobile phone may be inferred by the location of the antenna to which it is connected. However, this information is useful for knowing that the mobile phone is within the coverage area of a specific antenna, yet it does not provide information on the exact coordinates of the device as an error is to be expected depending on the density of antennas. The accuracy in urban areas with a high density of antennas is rather high, i.e. a few hundred metres. However, the error in rural areas may rise to several kiTU??lometres.
How a user has been moving may de deduced from analysing the position of a mobile phone over time. A mobile phone may sometimes stay in the same place. However, it changes position other times, i.e. connects to a different antenna, what means that the user has changed location. These changes in position show both usual mobility (e.g. moving from home to work) and occasional mobility (e.g. going on holiday). Other variables may also be inferred from mobile phone activity data. For instance, the place of residence of the mobile phone’s user is linked to the location where the device has got more activity at night. These data may be cross-referenced with different data sets to obtain detailed information on the user. For example, cross-referencing the user’s place of residence with maps depicting the income level of the census sections provides information on the socioeconomic status of the user.
Mobile phone activity data are always treated anonymously and published in aggregate form (for example, by municipalities or provinces) to guarantee user privacy. Therefore, information on mobility obtained from processing these data refers to groups of people rather than to specific individuals. Various statistics on the mobility of the population may be acquired, such as how many trips people make throughout the day as well as the origin, destination, length, time and distance of those journeys, etc. This information may also be extracted for any given period, allowing the analysis of changes in mobility over time.
Information on mobility obtained from mobile phones was used in many countries to monitor the degree of compliance with the limitations on mobility during the pandemic, e.g. the Indicators on Mobility accomplished by the Spanish Ministry of Transport, Mobility and Urban Agenda , following the state of alarm in March 2020, which was the source used to prepare this section.
Variables and phases selected to assess the impact of the pandemic on mobility from mobile phones
The first wave of the pandemic utterly changed the habits of citizens with regard to travelling. Mobility patterns were modified by the legislation approved by the public authorities in March 2020 and the rising awareness of the population. Given that coronavirus is transmitted through social interaction, the fight against the pandemic focused on keeping social contact to a minimum by locking down the population; in other words, issuing strict limitations on mobility that entailed confining citizens to a specific area (region, province, municipality, borough, neighbourhood and to their homes). Relating the data on mobility to the epidemiological data confirmed, for example, that the increase in social interaction in the summer and over the Christmas period of 2020 was closely connected to the subsequent second and third waves of the pandemic
Schools and shops were closed (except those selling essential goods, such as pharmacies and supermarkets) and walking on the street was forbidden during the two weeks following the enactment of the state of alarm (14-29 March 2020). As an exception, citizens were allowed to go to work (although home office was recommended where possible), to healthcare centres and to buy goods in pharmacies and supermarkets. The first extension of the state of alarm (from 30 March to 12 April 2020) was the most restrictive time, having all non-essential activities suspended. During the second extension (from 13 to 26 April 2020), citizens were allowed to return to work in non-essential activities if home office was not feasible. During the third extension (from 27 April to 10 May 2020), children under 14 were allowed to walk within 1 km of their home, yet always with the supervision of an adult. In the second week of the third extension (from 02 to 10 May 2020), phase 0 of the downscaling scheme began, allowing citizens to participate in non-contact outdoor activities, i.e. walking and playing sports. From 10 May 2020 onwards, the different phases of the downscaling scheme gradually enabled opening non-essential activities and shops and travelling from one province to another under certain conditions. This downscaling scheme ended by revoking the state of alarm on 21 June 2020, allowing travel throughout the country.
To summarise, restrictions on mobility were most stringent during the third and fourth weeks of the state of alarm and were then progressively lifted until they ended on 21 June 2020. They affected both the number of trips made and the distance travelled (due to the 1 km radius rule and to limitations on travel between municipalities, provinces and regions). Two indicators were chosen to analyse journeys: firstly, the number of daily trips, and secondly, the number of daily trips weighted by the distance travelled. This second indicator was applied at a provincial level to assess the percentage drop in travellers-km within each province (intra-provincial mobility) and between provinces (inter-provincial mobility). Finally, a third indicator was used to depict lockdown, i.e. the percentage of the population that left their mobility area.
The fall in the amount of trips and distances travelled had a clear impact on the two archipelagos (Balearic Islands/Illes Balears and Canary Islands / Canarias). With tourism at a strict standstill and limitations on both travel between the islands and to mainland Spain, the drop in mobility in the two archipelagos exceeded 90%. Other coastal provinces also registered significant falls, as did the main urban areas, led by Madrid, where the high proportion of office employment as well as the presence of a large administrative and service sector also contributed to the sharp fall in mobility during the most stringent week of restrictions.
Changes to mobility during the first wave of the pandemic
The amount of trips was severely reduced during the first wave of the pandemic, especially when more stringent restrictions remained in force, such as strict lockdown barring a few very specific essential activities. The graph on the Evolution in the number of trips during the first wave of the pandemic starts with the two weeks prior to lockdown (which may be considered as reference weeks) and ends on 28 June 2020 (the first week after the state of alarm was revoked). It reveals a sharp drop, from 140 million trips per day during the reference weeks to just over 60 million in the third and fourth weeks of lockdown (the strictest period), i.e. approximately 43 % of the daily trips before lockdown. Most of the population reduced their journeys to a minimum, and the amount of people not making any trips at all clearly raised. By contrast, certain groups carrying out specific jobs (especially delivery people) increased their trips due to the boom in e-commerce and the high demand for home-delivered food products. From then on, the number of trips gradually raised as the downscaling scheme progressed, reaching 120 million daily trips in the week right after the end of the state of alarm. However, this figure is still lower than the one registered prior to lockdown as certain habits remained, i.e. home office, online shopping, etc., as well as a higher degree of awareness and caution regarding coronavirus stayed, especially amongst the most vulnerable age groups. The evolutionary curve shows a clear weekly rhythm, with lower mobility at weekends and more trips during the week.
The travellers-km indicator shows a similar curve to the one on the number of trips, yet an even more pronounced drop-off is to be observed. This indicator shows the amount of trips weighted by the distance of the journeys. In sum, fewer trips were made whilst restrictions were in place, and those that were made covered less distance. The graph shows how the travellers-km indicator fell from approximately 1.4 billion during the days before the state of alarm was enacted to around 400 million in the third and fourth weeks of the state of alarm, i.e. it dropped by 72%.
The graph on the Population leaving their mobility area during the first wave of the pandemic shows how travelling was reduced, registering a sharp drop during the first four weeks in which the state of alarm remained in force. Interestingly, this indicator took longer to recover than the previous two indicators, with figures in June 2020 well below those recorded before the pandemic. Trips to other mobility areas were predominantly undertaken for work purposes, although some were made for leisure or family meetings. They were notably reduced during the state of alarm and, to a large extent, failed to recover during the downscaling phases.
The maps showing the evolution in intra-provincial mobility unveil how restrictions affected life during the first weeks of the pandemic. They reveal a maximum reduction in mobility in the phase affected by the most stringent restrictions and a gradual recovery in intra-provincial mobility in the run-up to the summer. The differences between provinces are few and far between, e.g. home office was widely implemented in provinces with a large proportion of jobs in the tertiary sector, whilst the need to physically go to work in the more agricultural and industrial provinces entailed higher levels of mobility. These differences between provinces are more remarkable in the first week, which shows a sharper drop in trips in Madrid and Barcelona than in the rest of the country. The map of the situation during the fourth week of lockdown shows the economic standstill throughout Spain, which translated to a massive drop in trips in all provinces.
Mobility between provinces (see maps on inter-provincial mobility) followed a similar pattern to mobility within them (intra-provincial mobility). Again, the disparities between the provinces were more remarkable during the first weeks of lockdown and during the downscaling phases. By contrast, changes in mobility were more uniform during the two weeks with the most stringent restrictions, when a sharp fall in inter-provincial travel was registered in all provinces. The most significant contrast was to be found again in Madrid and Barcelona in relation to their neighbouring provinces. The two major cities also saw a remarkable reduction in travel to other provinces in all weeks assessed, with many business conferences and meetings taking place virtually. Conversely, as many of the people living in Guadalajara, Toledo, Segovia and Ávila are employed in Madrid, these provinces registered a higher proportion of inter-provincial trips, well above the national average. This situation was to be found again for Barcelona and the surrounding Catalan provinces and, to a lesser extent, for Seville (Sevilla), Saragossa (Zaragoza), Valladolid and Corunna (A Coruña) and their neighbouring provinces. The drop in inter-provincial mobility was particularly sharp in the islands due to their heavy reliance on tourism and to a general aversion to air travel.
Mobility between regions also reduced significantly. These flows, which were especially intense between neighbouring regions and between regions with a larger demographic and economic weight, were restricted during lockdown to a minimum and were only allowed due to duly justified reasons, such as work. As a result, the maps showing the trips made to the Region of Madrid (Comunidad de Madrid) and to Catalonia (Catalunya/Cataluña) from other regions reveal a very acute drop across the board. Furthermore, they depict a sharp decline in mobility between the Region of Madrid (Comunidad de Madrid) and Catalonia (Catalunya/Cataluña), with a significant fall in travel due to the drop in business trips. Finally, travel from the islands was clearly affected during the first wave of the pandemic due to the restrictions on air travel.
Other changes to mobility observed during the pandemic
Mobile phone data were used continuously throughout the pandemic to track mobility. Only data up to the end of June 2020 are taken into account for this monograph. However, data from later on have enabled observing, for example, how the holiday preferences of Spaniards changed during the summer of 2020 compared to previous years. In August 2020, people made fewer trips for holiday purposes and, in general, opted for destinations closer to home. In addition, Spaniards avoided air travel as much as possible (which was particularly detrimental to the islands) and chose instead destinations they could drive to, what gave them greater self-reliance should they need to return home in the event of an incident related to the pandemic. Tourism to urban destinations (such as Madrid and Barcelona) also registered a sharp drop, whilst trips to coastal or inland destinations, where many holiday homes are located, withstood the crisis well. Mobile phone data also eased monitoring Spanish patterns of travelling during extended bank holidays in autumn 2020, over the Christmas break in 2020 and during Easter holidays in 2021.
This same source of information has also enabled verifying that many citizens living in highly urbanised areas in Spain retreated at the beginning of the pandemic to their second homes (on the coast, in the mountains or in smaller towns) and stayed there until after the summer of 2020 because the risk of infection was deemed to be lower in these less populated places. Furthermore, it has enabled assessing the relationship between mobility restrictions, such as municipal, provincial and regional lockdowns, and the incidence rate of COVID-19, allowing limitations to be adjusted in line with the evolution of the main epidemiological indicators. In short, mobile phone data have provided useful information to the health authorities throughout the pandemic, allowing them to verify the degree of compliance with mobility restrictions and their relationship to the dynamics of the main epidemiological indicators.
Co-authorship of the text in Spanish: Juan Carlos García Palomares and Javier Gutiérrez Puebla. See the list of members engaged
Adaptation of the text and translation into English for this international version: Andrés Arístegui Cortijo (Translator in chief)
You can download the complete publication The COVID-19 pandemic in Spain. First wave: from the first cases to the end of June 2020 in Libros Digitales del ANE site.