2025/09/03 by Toger, Marina, Türk, Umut, Östh, John +1
#Bayesian estimation #GSM-based mobility flow data #Greater Stockholm Area #Spatial econometrics #heteroscedastic spatial Durbin panel data model #spatial spillover effects
paper · doi:10.57938/7327aa62-ce8a-4840-9def-410a94845e16
This study applies a heteroscedastic spatial Durbin panel data model to investigate how sociodemographic and socioeconomic factors influence regional commuter mobility in the Greater Stockholm Area. Commuter mobility, defined as the flow of people to and from workplaces across regions and over time, is measured using high-frequency, high-resolution origin-destination data derived from mobile phone records, providing high-frequency, high-resolution insights into commuting patterns. The analysis uses a balanced panel of 681 regions from 2018–2024, incorporating an 18-nearest-neighbor spatial weight matrix to capture the topological relationships. Direct (withinregion) and indirect (spillover) effects are estimated using Bayesian inference, enabling robust interpretation of marginal effects in the presence of spatial lags in dependent and independent variables. Results show that spatial spillovers exert a more decisive influence than direct effects, with educational attainment and car ownership emerging as the most influential determinants of commuter mobility. According to total impact estimates, the demographic structure plays a comparatively minor, yet still significant, role.