The COVID-19 pandemic prompted governments worldwide to implement containment measures, including restrictions on human mobility. The effectiveness of these measures varies significantly depending on socioeconomic conditions, particularly in large urban areas of low- and middle-income countries. This study focuses on Bogotá, Colombia, a city characterized by significant income inequality, traffic congestion, and a substantial informal workforce. Existing research indicates that aggregated mobility data can inform public health actions and that socioeconomic disparities influence responses to lockdown policies. This paper aims to characterize the mobility network of Bogotá before and during the pandemic, using smart card validations from the integrated public transport system, and to analyze the relationship between changes in mobility and socioeconomic indicators like informality, poverty, and socioeconomic strata (SES). The study uses this data to provide evidence on the impact of these policies on urban mobility and how it relates to socioeconomic conditions of the city's population. This is important since the effect of these policies on urban mobility in Bogotá had not been addressed before.

Numerous studies using mobile phone or social media data have shown significant changes in mobility during lockdowns in high-income countries. Similar trends have been observed in some middle-income countries like China and Brazil. However, fewer studies have specifically investigated the impact of socioeconomic conditions on these mobility changes. Existing research suggests that socioeconomic factors can generate differential responses to lockdown policies, highlighting the unequal distribution of the pandemic's socioeconomic consequences. In Colombia, while there were improvements, Bogotá was still characterized by high income inequality and informality prior to the COVID-19 pandemic. This paper seeks to add to the existing body of research by specifically analyzing the mobility impacts of COVID-19 lockdowns in a large urban area within a middle-income country, examining heterogeneity of mobility change based on socioeconomic factors. This contributes to the understanding of the differential impact of containment measures based on various socioeconomic factors and characteristics.

The study utilizes smart card validation data from Bogotá's integrated public transport system (including TransMilenio and SITP) to analyze mobility flows between January and July 2020. Mobility networks were constructed by counting trips between stops, aggregating data over working days (Tuesday-Thursday) to control for holiday effects. The week of February 3rd served as the baseline representing normal mobility. Mobility ratios, calculated as the ratio of mobility flow in a given week to the baseline week, were used to analyze mobility changes. The study also incorporates socioeconomic data from the Census, including socioeconomic strata (SES), the share of informal workers, and a multidimensional poverty index, aggregated to the zonal planning unit (UPZ) level. These data were used to examine the correlation between mobility changes and socioeconomic indicators. A gravity model was employed to assess the effect of socioeconomic conditions on mobility flows between UPZs, considering population, distance, SES, informality, and multidimensional poverty. Poisson estimation was used to address heteroskedasticity and zero-valued flows. To assess mobility change at a municipal level, Facebook's Movement Range Maps (“Change in Movement” metric), combined with the multidimensional poverty index aggregated by municipalities, were used.

The study observed a general decrease in mobility flows following the implementation of social distancing interventions. However, the magnitude of the decrease varied significantly based on socioeconomic conditions. Mobility reductions were more pronounced in areas with higher SES, while areas with higher poverty and informal worker proportions experienced less reduction in mobility. The analysis of mobility networks showed high connectivity between different areas of the city, even during lockdown, suggesting resilience in the network's structure. While overall mobility decreased drastically during lockdown (to 16.8% of pre-lockdown levels), this reduction was not uniform across socioeconomic groups. The gravity model revealed that before the lockdown, better socioeconomic conditions at the origin were associated with higher mobility flows, and the effects of socioeconomic conditions in the destination were more pronounced. After the lockdown, better socioeconomic conditions in the origin were associated with a higher mobility reduction. The results indicated that zones with higher poverty, larger shares of informal workers, and lower SES had more moderate mobility reductions than zones with better socioeconomic conditions during the lockdown. The differences in mobility changes between the lower and higher socioeconomic groups increased during partial lockdown. The Facebook data analysis supported these findings, showing a correlation between lower mobility reductions and higher poverty levels across municipalities in Colombia.

The findings highlight the unequal impact of lockdown policies on different socioeconomic groups in Bogotá. The inability of lower socioeconomic groups to significantly reduce their mobility suggests challenges in achieving effective social distancing. The resilience of the mobility network underscores the need for long-term solutions to address structural issues in Bogotá's transport system. The socioeconomic disparities in mobility changes during the lockdown emphasize the importance of targeted income support policies to mitigate the unequal burden of pandemic containment measures. The study's findings are relevant to urban planning and policy interventions aiming to reduce inequality and improve access to essential services.

This study demonstrates that while lockdown measures effectively reduced overall mobility in Bogotá, their impact was unevenly distributed across socioeconomic groups. Lower socioeconomic groups experienced less mobility reduction, highlighting the need for comprehensive social support policies to ensure effective social distancing. The resilience of the city's mobility network underscores the importance of addressing long-standing urban planning challenges. Future research should investigate the specific reasons for persistent mobility in lower socioeconomic areas and explore further the potential of integrated data sources to help in more effective crisis response and recovery strategies.

The study's reliance on smart card data from the public transport system limits its scope to public transport users, potentially underrepresenting the mobility of individuals using other transportation modes. The aggregation of socioeconomic data to the UPZ level may mask some within-UPZ heterogeneity. Furthermore, the study does not directly address the causal relationship between socioeconomic factors and mobility, focusing primarily on correlation.