Lockdown and adaptation: residential mobility in Spain during the COVID-19 crisis

Societies have become increasingly mobile during recent decades. Every day, millions of people travel across countries, between regions and within cities, an ever-increasing flow that is a defining characteristic of our era. However, extremely mobile societies have their weaknesses, and the COVID-19 pandemic disease has tragically revealed one of them. During the outbreak of a pandemic, every person on the move can be considered a risk and a potential vector of contagion (Iacus et al. 2020). Thus, lockdown policies that limit everyday mobility, commuting and residential mobility have been enforced in a variety of countries as the only possible strategy to stop (or slow) the spread of the disease, given the current lack of an available vaccine. Studies based on information from mobile devices provided by the primary telecom companies have shown the effectiveness of lockdown policies in reducing the daily mobility of the population (Askitas et al. 2020; Galeazzi et al. 2020; Santamaria et al. 2020). However, the current literature about the effects of lockdowns on changes of residence is scarce (at this time only one short study of the subject appears to have been published; Jones and Grigsby-Toussaint 2020).

As this article demonstrates, in the Spanish context the impact of residential mobility on the current coronavirus crisis is as important as that of daily mobility for three main reasons. First, its spatial range is wider and usually involves movement between towns, cities or regions, creating the risk of new outbreaks in areas that had controlled the spread of the disease. Second, residential mobility during the COVID pandemic is exceptional, novel and, to a certain extent, an unknown quantity. Customarily, residential mobility is explained through life course changes (Coulter et al. 2016) or triggering events related to family, work and housing. However, in the present exceptional situation, ‘lockdown mobility’ responds to different triggers: economic problems caused by COVID-19, feelings of fear and loneliness or problems related to cohabitation that has become unbearable due to confinement (Jones and Grigsby-Toussaint 2020). Third, according to the context and time, this phenomenon can be illegal in part, as it operates under the radar of official data sources that may, therefore, remain unaware of these changes of residence. Although some legal reasons to move might be at play, many individuals and families risk a fine by moving without a valid reason to do so.

This paper analyses residential mobility in Spain during the lockdown decreed by the government in March 2020, based on an online survey designed with four goals in mind: (1) to discover the triggers of residential mobility during the lockdown; (2) to analyse the reasons for moving despite the restrictions; (3) to study the differences in mobility within the population; and (4) to explore the potential threat posed by those moving from/to a household where symptoms of COVID-19 have been reported.

Residential mobility is usually defined as a decision taken when life course events create an imbalance between one's current dwelling and one's housing preferences (Clark 2017). Family changes such as leaving the family home, the birth of children (Michielin et al. 2008), divorce (Feijten and van Ham 2007) or the transition to the empty-nest stage or widowhood (Bloem et al. 2008) have been postulated as the primary push-and-pull factors. Changes in professional careers (Mulder and Hooimeijer 1999) and lifestyle choices (Ærø 2006) are also relevant. Thus, mobility can be a proactive behaviour undertaken to enhance residential conditions, achieve a certain lifestyle or prepare for a change in household composition. Alternatively, it can be a reactive conduct aimed at alleviating a problematic situation directly related to housing conditions or to other aspects of the mover's life.

However, the emergence of new residential preferences due to changes in life courses is not alone a sufficient cause for mobility (Coulter et al. 2016). The mobility decision can be a ‘black box’ for researchers, in which the configuration of preferences and their materialization in a mobility decision is mediated by a context that has a double scale (Mulder and Hooimeijer 1999): a micro context, which includes the limitations and resources of individuals; and a macro-structural context, including the housing market and economic or social (and currently health) situation. It is in this double context of social inequalities and the pandemic that residential mobility during the lockdown in Spain must be understood.

The COVID-19 lockdown has brought about the emergence of specific push-and-pull factors, which are generally different to the triggers of mobility in a context of ‘normality’ (Jones and Grigsby-Toussaint 2020). To cite but a few, pandemic-related triggers include loneliness and psychological problems derived from the situation (Devaraj and Parel 2020); difficulties related to forced coexistence with other members of the household (Rogers and Power 2020); and unemployment or other economic problems arising from the paralysis of the national economy (Conde-Ruiz 2020). Nevertheless, these triggers are not equal to all social groups, nor do they manifest themselves in the same way. Fernández-Prados et al. (2020) have found important differences in social resilience against the disease and its consequences. In this macro context of pandemic, economic crisis and mobility restrictions, the micro context of resources and limitations becomes even more important. While residential mobility can mean a chance to spend the lockdown in more amenable conditions for better-off households, for poorer individuals and families, moving is an obligation, a response to the economic, health or, at times, psychological problems arising from a situation.

Finally, residential mobility can involve both long-term changes, with lengthy spells of stability, as well as short-term movement, which is typical of certain stages in the life course and specific occupations. Youth, being a student and the early stages of a professional career are all linked to frequent changes of residence. However, as Clark et al. (2017) underline, for most people, moving is an extraordinary event, while staying is the pattern that dominates their lives. The COVID-19 pandemic and its consequences have created an extraordinary context in which the lack of a clear horizon of events in the near future fosters short-term movement and puts longer-term decisions on hold.

In summary, previous works on residential mobility have provided methodologies and explanations for the decision-making process, making it possible to understand what determines changes of residence during the current medical crisis. However, the type of movement observed in this context and the underlying reasons are different to the ones usually described in the literature. Mobility in this new context is largely reactive, often triggered by an entire new set of reasons related to both the disease and the lockdown, reveals greater inequalities between social groups and is characterized by short-term decisions. In short, this new phenomenon is more complex and volatile than mobility in regular situations.

The data source for this study is a Spanish nationwide online survey: The Survey on the Effects and Social Changes Caused by COVID-19 (Duque-Calvache et al. 2020). This is a unique source, as it brings together a number of issues that are essential for the analysis of mobility: change in residence, reasons for the change, destination and the characteristics of the individuals who move (additional information on the variables used and their correspondence to the questionnaire can be found in the online appendix, Table OA1). The survey was accessible in the form of a self-administered online questionnaire available from 24 April to 26 May 2020, and collected data from a total of 3,103 individuals from all over the country. One of the particular distinctions of this data source is its time frame, as it includes both the stricter periods of the lockdown and the first two stages of the de-escalation of restrictions as well. Other studies have either been unable to look at the initial weeks or do not cover the later stages, thus missing out on part of the process.

As the sample is self-selected, it therefore does not exactly correspond to the structure of the Spanish population. Specifically, the main biases in the sample are the overrepresentation of young people and students (both high mobility groups), and the underrepresentation of the elderly. Although this kind of bias is found in many online sample surveys (and is frequently explained as a result of the digital divide), it nonetheless needed to be corrected to ensure solid, reliable results with regard to the whole population. In order to do this, we took the population structure from a recent official, nationwide survey sampling (INE 2020) of 65,000 households and calculated the proportion of the population in each group by gender and age (in 5-year intervals), and then assigned weights to the data to match our benchmark numbers. We limited our sample to individuals between 20 and 69 years old, as we only had a limited number of respondents below and above those ages, which could have distorted the results. The final composition of the sample analysed in the paper comprises 2,895 respondents. The comparison between the population and the sample compositions, as well as the weights per group, can be found in the online appendix (Table OA2). Although we corrected their impact on the overall population with this operation, our study still benefits from the overrepresentation of high mobility groups, providing a better, more detailed understanding of residential mobility during the lockdown.

We analysed the data combining single variable tables, cross tables to study differences in mobility and reasons to move by social groups, in addition to a multiple variable model to explain the decision to move during the lockdown. We chose a binary logistic regression to maximize the reliability of the results, as multinomial models could excessively split the sample. The model makes it possible to evaluate and compare the importance of different explaining factors while controlling the rest of the variables. Finally, we calculated the average marginal effects (AMEs) of the different variables to more easily compare their effects.

Despite all the new laws, media messages and political discourse asking Spaniards to reduce mobility (and at times forbidding it), according to our data, 9.4% of the population moved during the lockdown or the less strict confinement stages. This figure might suggest that residential mobility can be hazardous to public health and must be considered when planning future lockdown situations. However, beyond the magnitude, to understand the scope and importance of mobility, it is necessary to further investigate the triggers and the individuals participating in it. Who moves and why?

In order to address these questions and obtain a fuller picture of the profile of the individuals who moved, we developed a logistic regression model, a technique often used in the analysis of residential mobility that makes it possible to control the interaction effects of each explaining factor. Recent studies on COVID-19 in Spain (Fernández-Prados et al. 2020) have used the same approach to measure social resilience. Our dependent variable concerned whether the respondent moved (or did not) during the period of study. An overview of the independent variables included in the model can be found in the appendix (Table OA4).

We developed the model in a four-step process, adding new variables by blocks (demographic, socioeconomic, housing conditions and lockdown related variables) to check whether the model improved after they were included. The standard tests (pseudo R², improvement over null model, Akaike Information Criterion – AIC – and Bayesian Information Criterion – BIC) showed slight increases in the explanatory power of the model after each step. For the sake of brevity, we only include the final model, which contains all the variables (Table 1).

The model has limited predictive power, but this is common with regression models that analyse residential mobility (Duque-Calvache et al. 2017). More importantly, the model shows some interesting differences with regard to the usual explanations for residential mobility. In our survey, the movers tended to be young, and each year lived decreased the likelihood that they moved during the lockdown. However, the effect of age was not linear, as older respondents were more likely to move than middle-aged individuals. These two facts combined demonstrate the diversity amongst the movers in our sample. Respondents living alone and in non-family households were more prone to have moved. On the contrary, children anchored families to their homes in a context of improvisation and uncertainty. By occupation, the only group that showed a pattern significantly different to the reference category (professionals) were students, an expected result due to this group's high mobility rate in any circumstance, including the COVID-19 context. The cancellation of teaching in all Spanish universities for the remainder of the 2019–2020 academic year (announced in April 2020) also allowed many students to request special permission to return to their family homes, creating a window of opportunity for movement in a short period of time.

Renters showed a pattern of low mobility in the study that is highly surprising, given that this group always has a higher tendency to move in regular residential mobility studies (Clark et al. 2017). As some of the renters in the sample were forced to move due to housing defaults, it is likely that the rest of the group had a very low possibility of moving. The difficulties inherent in finding a new place to live during a lockdown probably forced many renters to postpone or discard any plans to move. Respondents who telework were more likely to move, but telework is both a condition of possibility (I may move) and a reason to move (I need a better workspace). Finally, after controlling the rest of the variables, the only problem related to the pandemic outbreak and its consequences that increased the likelihood of moving was housing default.

Our questionnaire asked the respondents about their reasons for moving, allowing multiple choices in addition to an open-ended response box to enter any reasons not included in the predefined categories. The most common reply (46.7%) was the search for greater comfort. The individuals forced to stay at home were quite concerned about housing conditions, the amount of space available or the environment in which their homes were located. At times, mediocre housing conditions could become unbearable – especially for families with children – with the lack of space and privacy creating new problems related to cohabitation. Of course, a move to an improved situation requires having a better house available, and its frequency was uneven across the social classes, as addressed below. The second most often cited reason (45.9%) was the need to be near and care for loved ones (mostly family), a recurrent motivation for residential mobility and immobility in Spain even before the pandemic (Clark et al. 2017). Family is a key social institution in the country, not only with regard to providing care and emotional support in everyday life, but also as a last resort in critical situations. The role of families in southern European countries (Spain, Portugal, Italy, Greece) has been described in numerous contexts, from care and contact with the elderly (López-Doblas et al. 2020) to young adults leaving the family home (Fuster et al. 2020). Finally, some respondents entered individual reasons that compelled them to move in the open-ended box. We labelled this category, which applied to 17.2% of the sample, ‘forced mobility’, which was not included amongst the predefined categories (and thus may be underestimated).

The personal economic problems that accompanied the onset of the pandemic and the containment measures include being fired, either temporarily or permanently, and a serious reduction in income. At times, these reductions led to rent and mortgage defaults. Although the Spanish government tried to ease payment problems by granting low-interest loans, the policy has not been as effective as expected (Conde-Ruiz 2020). In our data sample, most housing problems had either been solved by personal agreement or by the owner or renter leaving their home.

Although motivations at times overlapped, most respondents chose only one reason for moving. One notable exception was the combination of proximity and comfort, to the point that we defined this as a new joint category in the analysis. In this case, the decision to move was related to both a desire for improved housing conditions and to providing or receiving care. At an emotional and psychological level, the arrangement probably met the needs of both sides, the supporter and the supported.

The reasons given by people in different socioeconomic groups initially subverted our expectations. Professional middle-class groups seemed to be motivated largely by proximity, as opposed to the mobility of working-class and low-educated groups, whose main motivation was comfort. This distribution does not correspond to the stereotypical view of middle classes as more individualistic and interested in their own comfort versus the family-focused working classes (Brun and Fagnani 1994; de Pablos and Susino 2010). These social class patterns are easier to understand if the emphasis is placed not on what they want, but on what they already have (or do not have) in their lives. The working classes tend to live closer to family networks (Clark et al. 2017), but in poorer housing conditions, while professionals and the middle classes live farther from their families and in nicer homes (de Pablos and Susino 2010). Hence, when faced with the lockdown, working-class respondents tended to solve their liveability problems, while middle-class individuals tried to decrease the distance from their families.

We also found diverging patterns amongst the non-employed. Retirees tended to seek comfort or proximity and were not affected by forced mobility. On the contrary, the unemployed and homemakers were very likely to move against their wishes, as expected due to their vulnerable situation. University students in Spain usually return to the family home at the end of the academic year and most do not leave their parents’ home until they reach their thirties (Fuster et al. 2019). Facing a lockdown and the virtualization of all their university activities (including classes and exams), these students merely returned early to the nest, where they found company and comfort (Fuster et al. 2020).

The role of economic problems in changes of residence is clear in the responses of individuals put on a temporary paid leave (70% of their usual salary) or fired. Indeed, it is likely that the loss of a job forced respondents to move. However, the individuals on paid leave cited comfort and proximity, perhaps because they perceived their current work status as temporary (it should be noted that at the time of our survey, the reduction in the number of businesses did not seem to be causing forced mobility, but as the situation extends over time, this may change dramatically). Finally, the pattern for teleworkers is similar to that of professionals, while the pattern of non-teleworkers is more like that of working-class and unemployed respondents. This indicates that teleworking is not an option for every worker and all kind of companies. White-collar workers, students and professionals can relocate and continue working, expediting their residential mobility.

One of the main consequences of mobility during the lockdown has been the reconfiguration of existing households, either temporarily or indefinitely. In a volatile, unstable and unfamiliar scenario, new living arrangements are easier to find and more diverse than usual and deserve further attention. Most residential changes in our survey involved the homes of relatives. If movement to the homes of friends or partners is included, more than seven out of ten respondents altered the composition of an existing household. In a moment of uncertainty and turmoil, when the location and control of outbreaks was beyond state control, the arrival of new individuals in a pre-existing household was undoubtedly a risk and public health problem.

However, the danger inherent in mobility in times of lockdown went beyond households. The second most highly preferred destination was a second home, usually located away from the respondent's primary city (and province) of residence in the case of Spain (Torrado et al. 2020). As a result, these moves likely involved travelling mid- to long distances, with consequences for the spread of the disease and control of outbreaks. Mobility to second homes was found to be linked to the search for comfort, as expected, but also to proximity to and care of loved ones, something explained by the location of many second homes in an individual's home town. A significant amount of second homes are low-value, inherited properties in rural, inland areas, but close to aging relatives (Leal 2006).

Mobility to the homes of relatives is also connected to proximity and comfort. Interestingly in this context, forced mobility does not seem to have motivated a return to the family home, a divergence from the usual experience of ‘boomerang kids’ in Spain (Donat and Berngruber 2018). Rather, forced mobility generally pushed people to rent their residences instead of seeking shelter with relatives or friends. Those who did move in with friends sought the comfort and comfort/proximity mix, perhaps trying to make the best out of the lockdown, or to share these difficult moments with somebody they trust. Moving to a partner's home was linked to comfort and proximity, but also to support in a situation of forced mobility. Finally, a number of respondents moved back to their primary residence when they had the chance, once the strict lockdown was over. The speed with which the situation changed – a matter of weeks – caught many families off guard, separating some of them while forcing others to overstay their planned visits.

Although the number of respondents who changed residence and also stated they had been in contact with the virus comprises only 1.17% of our total sample (but 12.5% of those who did move), this phenomenon is very important when attempting to understand the reasons for this movement and the consequent potential threat to public health. Although contact with the illness is self-reported in our questionnaire – and thus unreliable to a certain extent – the self-evaluation of symptoms was encouraged across the country in the worst moments of the first wave of the disease, due to the unavailability of tests for the population.

Table 2 compares the characteristics of the movers who had and who had not been in contact with someone affected by the virus, showing that those who moved despite the potential contagion were slightly younger. This could be a sign of irresponsibility amongst the younger population, an idea repeated countless times by the media during the summer in an attempt to create awareness of the risks of new outbreaks (or, as some argue, perhaps to transfer the blame for further problems to ordinary citizens). However, the survey suggests that the actual reason for this behaviour on the part of some is due to social inequality. Individuals who moved even when they had come into contact with the virus were largely workers and the unemployed living in precarious housing conditions, affected by the reduction in economic activity and who had lost their jobs or homes. The data hints at a clear trend: the change of residence was not a choice, but imposed by the circumstances of individuals and families. Recent information about a second wave in Spain (in early Autumn 2020) reinforces this interpretation, insofar as the disease is now affecting the more poverty-stricken neighbourhoods and districts (measures to confine these districts have already been enacted in Madrid as of the time of writing). Mobility may spread the virus, but the underlying problem is inequality.

The primary conclusion from our research concerns the complexity of decreeing a lockdown to control a pandemic. It is clear from the survey that restricting everyday routines, closing schools and universities and imposing teleworking in many jobs sends ripples through people's lives that may force, or motivate, them to change residence, thus creating a reaction of increased mid- to long-distance moves as people decide where, how and with whom they want to spend an uncertain period of time. Moreover, this phenomenon is to be expected in future pandemic outbreaks. One potential policy proposal to limit this type of movement would immediately restrict mobility when a lockdown situation is decreed, without prior announcement. However, decision-makers need to take into account the social problems that these rearrangements may alleviate or even solve, as shown in this paper. In our survey, people had many objective, important reasons to move: childcare, elder care, loneliness, psychological distress, unhealthy housing conditions, spaces ill-equipped for teleworking and so forth. Forbidding residential mobility should be considered a serious, potentially harmful measure. Our results show that movers are diverse and so are their reasons. Consequently, treating all residential mobility as nonessential (and hence, dispensable) is a problematic misrepresentation. If controlling the spread of a virus by prohibiting residential mobility creates new, serious problems for individuals and families, it could be a harmful measure and one that – if necessary – must be enacted with careful planning.

Second, social and cultural patterns matter, even in the midst of a pandemic outbreak. Family has long been a cornerstone of everyday life and mobility decisions in Spain, and this did not change during the lockdown. Life course stages, gender and social class still structure and guide residential behaviour and, as a result, the information provided by scholars today on the subject is important and useful, even in this uncertain scenario. Connecting our findings with the literature on residential mobility, catastrophic events create a special context in which residential priorities and needs are reordered. Although key variables kept their relevance, our survey proved the significance of new circumstances, such as temporary paid leave or a reduction of activity. Therefore, an adequate assessment of the impact of COVID-19 on residential mobility requires updating the usual models and explanations, finely tuning them by adding measures of volatile, specific factors created by the situation.

The third relevant idea in our paper concerns the crucial role of social inequality in the control of a contagious disease. Inequalities permeate and structure behaviour, particularly in difficult times. The middle classes in general have more opportunities during a lockdown (such a second residence, or teleworking). For all the classes, the dependent populations like children and the elderly require extra care during a crisis, but the solution to these needs varies between the social classes. Poorer citizens have suffered during the financial crisis caused by the virus, with consequent additional health risks for the population that have often forced them to move and readjust their lives. Rather than treating this group as a menace and vector of contagion and criminalizing their movement, the authorities should think of them as the primary victims, and pay extra attention to their needs and problems.

No potential conflict of interest was reported by the author(s).

Ángela Mesa-Pedrazas is a social researcher and PhD student in Social Sciences in the University of Granada. Her main area of interest is the social construction of public spaces in the cities. In a broader sense, her topics and professional experience include different social realities linked to spatial dimensions.

José Manuel Torrado holds a PhD in Sociology by the University of Granada and teaches in the Department of Sociology, University of Huelva. His research topics are urban sociology, demography and human geography, with a special focus on the analysis of residential behavior, flows of residential mobility and the study delimitation, and classification of spaces based on social and functional criteria.

Ricardo Duque-Calvache is a PhD in Sociology and Assistant Professor in the Department of Sociology in the University of Granada (Spain). His formation included some research stays in different universities in Europe and the United States. His research topics are focused in urban studies and demography, with special attention to residential behaviour, metropolitan areas, gentrification processes and, more recently, changes in mobility and everyday life due to COVID-19.