Unequal changes in commuting patterns across socio-economic strata in response to pandemic restrictions

Abstract

Commuting patterns are a central component of urban dynamics and many societal activities. Exogenous shocks, such as a pandemic, might drastically modify them inducing heterogeneous variations across socioeconomic strata. Here, we quantify changes in work commuting patterns in Bogotá, Colombia during three different periods of the COVID-19 pandemic: pre-pandemic (2019), COVID-19 restrictions (2020), and partial reopening (2021). To this end, we use anonymized mobile phone data to infer home and work locations from recurring nighttime and weekday connection patterns, and to build daily commuting metrics. We aggregate mobility flows by administrative boundaries and socioeconomic strata. Additionally, we enrich the dataset with a range of other variables such as territorial vocation (i.e., urban versus rural), demographic information (i.e., population density) and, as a proxy for digital infrastructure quality, geolocated Speedtest measurements from Ookla. We find a marked reduction of commuting during restrictions in 2020 and a strong recovery in 2021, but with persistent heterogeneity across socioeconomic strata. Indeed, while commuting declined similarly across income groups during restrictions, groups of the population in the lower-income bracket rebounded faster to pre-pandemic levels. On the contrary, we find that groups in the higher-income bracket managed to keep higher stay-at-home behavior. Regression analyses reveal that territorial characteristics and disparities in digital connectivity significantly contribute to these differences, suggesting that infrastructure investments could help mitigate mobility-based inequalities.

Figures (13)

• Figure 1: Spatial distribution of socioeconomic stratification and changes in commuting patterns across residential and workplace areas. (A) Map of Bogotá showing residential blocks colored by socioeconomic stratum (SES) (low in blue, medium in yellow, and high in purple). Black lines indicate local planning unit (UPL) boundaries. (B) Reduction in commuting relative to the $2019$ baseline, stratified by residential SES (left) and work area SES (right) during COVID-19 restrictions ($2020$) and partial reopening ($2021$). (C). Workplace segregation patterns across periods. Left: Distribution of segregation indices by workplace SES for pre-pandemic ($2019$), COVID-19 restrictions ($2020$), and partial reopening ($2021$) periods. The segregation index ranges from $0$ (i.e., perfect socioeconomic mixing) to $1$ (i.e., complete segregation). Right: Variation in segregation relative to the $2019$ baseline, showing percent change in segregation by workplace SES. In all plots, the box boundaries represent the interquartile range ($IQR$) between the first and third quartiles ($Q1$ and $Q3$), the line inside the box indicates the median, and the whiskers extend to $1.5$ times the $IQR$ from the quartiles.
• Figure 2: Internet connectivity patterns and regression coefficients. (A) Home download speed distributions across residential SES groups. (B) Relationship between home download speed and speed differential (home-work) by SES level. Different markers and colors represent SES groups. (C) Regression coefficients with $95\%$ confidence intervals, ranked by absolute magnitude. Red/green indicates significant effects (negative and positive) at the $5\%$ confidence level; gray indicates non-significant effects.
• Figure 3: Scatter plot comparing population fractions estimated from Meta Data for Good high-resolution density maps (x-axis) and fractions of users associated with home antennas (y-axis), aggregated by administrative unit. Marker size represents unit area ($km^2$), and color indicates population density. The Pearson correlation ($0.88$) confirms strong agreement between the two independent population proxies.
• Figure 4: Distribution of the residential population traveled distance to work location per residential SES and year.
• Figure 5: Reduction in commuting relative to the $2019$ baseline, stratified by residential SES during COVID-19 restrictions ($2020$) and partial reopening ($2021$) for traveled distance to work.