The COVID-19 pandemic significantly altered human behavior, with potentially lasting impacts on workplace and residential mobility. These mobility patterns are intrinsically linked to urban issues like employment, transportation, and urban restructuring. Understanding how job and housing dynamics shifted during and after the pandemic is crucial for informing future urban development. Existing literature examines pandemic-related changes in daily travel behavior (reduced commuting and leisure travel, increased private car use), urban activity intensities (decreased workplace activity, increased residential activity, rise in teleworking and e-commerce), and jobs-housing relocation (job losses and gains, changes in working from home preferences). However, these studies often lack long-term perspectives, fail to consider the interdependence of job and housing relocation, and may suffer from survey biases. This study addresses these gaps by employing a comprehensive, longitudinal analysis of mobile phone data.

Previous research highlights the pandemic's impact on daily travel, showing a shift towards private car use and a decrease in public transport. Urban activity intensity decreased in workplaces and increased in residential areas due to widespread teleworking. Studies on jobs-housing relocation reveal job losses and gains across sectors, influencing workplace mobility. New motives for residential relocation emerged, including reduced income, psychological needs, and a desire for low-density living. Existing research acknowledges the role of socioeconomic factors like gender, age, and income in shaping mobility patterns and exacerbating inequalities. However, the literature lacks long-term comparative studies using large-scale, unbiased datasets that integrate job and housing relocation and commuting behavior.

This study utilizes an event study approach using ten months (April and November of 2018-2022) of citywide cellular signaling data from Beijing. The data includes timestamped location information, sociodemographic attributes (age, gender), and an affluence index. The study focuses on commuters within the sixth ring road. The period is divided into nine semi-annual phases: three pre-pandemic and six post-pandemic. Relocators (those with changed home, work, or both locations) are identified for each phase. Key relocation patterns analyzed include changes in home and work locations relative to the city center (represented by ring numbers of Beijing's ring road system), changes in housing price, population density, and accessibility of public services (public transport, schools, healthcare, green spaces). Commute time measures jobs-housing relations. Velocity and acceleration models are used, controlling for pre-relocation jobs-housing characteristics, sociodemographic attributes, and seasonal fixed effects. Data processing involved identifying home and work locations using stay times and a working-resting timeframe, filling in missing affluence index data using a Naive Bayesian algorithm, and integrating data with spatial data on population density, housing prices, and accessibility to various amenities.

The study reveals several key findings regarding the impact of the COVID-19 pandemic on relocation patterns in Beijing. Regarding home relocation, while there was an initial outward flight immediately after the outbreak, a long-term, mild inward relocation trend emerged post-pandemic. This trend varied across income groups: middle-income home relocators experienced a significant decrease in their pre-pandemic suburbanization trend, while high-income home relocators showed a new, mild suburbanization trend. Analysis of housing features suggests that middle-income relocators prioritized improved accessibility to public services, particularly schools and healthcare, while high-income relocators moved to locations with lower population density. For work relocation, a persistent outward movement of workplaces was observed following the initial increase of workers teleworking, reflecting accelerated decentralization of employment. This trend held true for both middle and high-income groups. The pandemic also impacted the jobs-housing relationship. A deceleration of jobs-housing separation was observed for all relocator groups, indicating an improvement in jobs-housing balance. This improvement was most pronounced for middle-income home movers and high-income work movers. Robustness checks, including models without seasonal fixed effects, models with individual weights, and Poisson regression models, supported the baseline findings.

The findings challenge the assumption of solely negative consequences from the pandemic, suggesting unexpected positive impacts on urban structure. The deceleration of suburbanization for middle-income individuals suggests a potential for increased inner-city vitality. The accelerated decentralization of employment may lead to a more polycentric city structure. The improved jobs-housing balance offers prospects for reduced congestion and carbon emissions. These results highlight the interplay between public health concerns and socioeconomic factors in shaping post-pandemic relocation patterns, informing urban planning and policymaking.

This study provides valuable long-term insights into the pandemic's impact on urban relocation patterns. The findings suggest a potential for positive structural changes in urban development, including more vibrant city centers, dispersed employment, and improved environmental sustainability. Future research could investigate similar patterns in other cities and countries, considering a wider range of socioeconomic groups and incorporating detailed travel mode data.

The study's limitations include potential confounding factors besides the pandemic (e.g., housing policies, the Russia-Ukraine conflict), sample characteristics not fully representative of Beijing's population (potentially over-representing middle-to-high-income workers), and the city-specificity of the findings (Beijing's strict pandemic control measures may not be generalizable). The study also infers relocators' motives based on available data rather than directly assessing them.