Cities are vital for economic productivity and innovation, fostering dense social connections through physical encounters. The diversity of these connections is a key predictor of economic growth and resilience to shocks. However, the COVID-19 pandemic and associated mobility restrictions significantly challenged the quantity and quality of urban encounters, raising concerns about their long-term social and economic implications. Previous research using large-scale location data has explored the nature and diversity of physical encounters, revealing that mobility behavior contributes significantly to urban segregation. The pandemic led to substantial lifestyle changes, including reduced trips to urban amenities and increased work-from-home arrangements, impacting not only physical and mental well-being but also the social fabric of cities. While studies have analyzed short-term mobility changes during lockdowns, little is known about the long-term effects on the diversity of urban encounters. This study addresses this gap by analyzing the longitudinal impact of the pandemic on the income diversity of urban encounters across three years, leveraging a large-scale, privacy-enhanced mobility dataset. The research aims to quantify the decrease in diversity, identify contributing behavioral changes, and explore the socio-demographic factors driving spatial heterogeneity in this decrease. The ultimate goal is to inform policies that balance pandemic response with the maintenance of diverse urban encounters.

Existing literature establishes a strong link between the diversity of social connections and economic productivity, emphasizing the importance of diverse networks for economic growth and community resilience. Studies using various data sources, such as call detail records (CDRs), credit card data, and social media, have investigated the characteristics of physical encounters in urban settings. These studies have highlighted the role of mobility patterns in shaping income segregation, demonstrating that mobility, rather than residential location alone, is a significant factor. The research has also shown that individuals tend to frequent places within their own socioeconomic status, with occasional visits to higher-income areas. Previous work has mainly focused on short-term mobility changes during the initial phases of the pandemic, analyzing the impacts of lockdowns on travel behavior, socioeconomic disparities, and disease spread. However, there is a gap in understanding the long-term effects of the pandemic on the quantity and quality of urban encounters and the diversity of social interactions.

This study utilizes a large-scale, privacy-enhanced mobility dataset provided by Spectus Inc., containing anonymized GPS location records for over one million mobile devices across four major US metropolitan areas (Boston, Dallas, Los Angeles, and Seattle) over a three-year period (2019-2021). The dataset is supplemented with location data from Foursquare, providing information on points-of-interest (POIs). To ensure data representativeness, post-stratification techniques were employed to account for regional and income level variations. Each user was assigned a socioeconomic status (SES) proxy based on the median household income of their home census block group (CBG), categorizing individuals into four income quantiles. Data filtering focused on users observed for more than 300 minutes daily and stays exceeding 10 minutes but less than 4 hours. These stays were spatially matched with nearby POIs to infer visit locations. The study employs two diversity metrics: *D_a*, measuring the income diversity at each place, and *D_i*, measuring the income diversity experienced by each individual. Both metrics quantify the evenness of time spent by people from different income quantiles. These measures are calculated for each two-month moving window, deseasonalized using 2019 monthly trends. To investigate behavioral changes causing decreased diversity, counterfactual analyses were conducted. Three hierarchical levels of behavioral changes were simulated: (i) reduction in total activity outside the home; (ii) changes in travel distances by income quantile; and (iii) microscopic changes in mobility behavior, including exploration and place preferences. These simulations involved creating counterfactual mobility datasets by randomly removing visits from pre-pandemic data to match the total visit duration during the pandemic. To analyze spatial and socioeconomic heterogeneity, a regression model was used to predict CBG-level income diversity and changes in diversity. The model included variables describing places visited, mobility metrics, and socio-demographic characteristics. Finally, the study examined the correlation between the stringency of COVID-19 policies (using the Oxford COVID-19 Government Response Tracker) and the decrease in income diversity.

The study's key findings reveal a substantial and persistent decrease in the income diversity of urban encounters during and after the COVID-19 pandemic. Across all four cities, experienced income diversity at places (*D_a*) and by individuals (*D_i*) decreased significantly compared to pre-pandemic levels (2019), even after aggregate mobility metrics recovered to pre-pandemic levels by late 2021. The largest decrease in diversity was observed in April 2020, reaching a 30% reduction in some cases. Even by late 2021, diversity remained approximately 10% lower than pre-pandemic levels. The decrease was consistent across all POI categories, with museums, leisure venues, transportation hubs, and coffee shops experiencing the most significant reduction. Counterfactual analyses indicate that the decrease in diversity was primarily due to a reduction in total activity outside the home during the initial pandemic waves. However, in the later stages, changes in exploration behavior and place preferences became the dominant factors. Social exploration – the probability of visiting a place where one's income group is not the majority – decreased significantly during the pandemic. Shifts in place preferences were also observed, with increased visits to routine locations like grocery stores and decreased visits to places like gyms and movie theaters. Analysis of spatial and socioeconomic heterogeneity revealed that areas with higher population density, a larger proportion of working-age populations, higher reliance on public transport, and larger movement ranges experienced a more pronounced decrease in income diversity during pandemic outbreaks. Finally, a strong negative correlation was found between the stringency of COVID-19 policies and the decrease in income diversity, suggesting a significant trade-off between pandemic control measures and the maintenance of diverse urban encounters.

The findings of this study demonstrate that the COVID-19 pandemic had a significant and lasting negative impact on the income diversity of urban encounters. Despite the recovery of aggregate mobility metrics, the decrease in diversity persisted, highlighting the importance of considering not only the quantity but also the quality of social interactions in urban spaces. The identified behavioral changes, such as reduced social exploration and shifts in place preferences, suggest that the pandemic may have altered long-term social habits. This decrease in diversity could have substantial cumulative effects on social capital, economic opportunities, and community resilience. The strong negative correlation between pandemic policy stringency and experienced income diversity emphasizes the need for urban planning and policy interventions that balance public health concerns with the maintenance of diverse and inclusive urban environments. Interventions should consider facilitating diverse encounters by reducing transportation costs and expanding access to public spaces in low-income areas.

This study provides crucial insights into the long-term impact of the COVID-19 pandemic on the income diversity of urban encounters. The findings demonstrate a significant and persistent decrease in diversity, even after the recovery of aggregate mobility. This decrease is largely attributed to behavioral changes, such as reduced social exploration and shifts in place preferences. The strong trade-off between pandemic control measures and income diversity underscores the need for policies that prioritize both public health and social inclusion. Future research could explore the long-term consequences of this decreased diversity on various aspects of urban life, including economic inequality and community resilience. Further research could also investigate the effectiveness of specific policy interventions designed to promote more diverse and equitable urban encounters.

The study acknowledges limitations related to data availability and the proxy nature of the metrics used. While efforts were made to ensure data representativeness and robustness, potential biases arising from data collection algorithms and the inability to fully capture the purpose and nature of encounters remain. The study focuses on income diversity, excluding other dimensions of social diversity, such as racial or ethnic diversity, which warrant future investigation. The geographical scope of the study is limited to four US cities, limiting generalizability. However, the findings provide valuable insights into the complex interplay between mobility, social interactions, and pandemic response measures.