Worse urban environments increase transit use under peripheralization: Causal evidence from a natural experiment in Shanghai

Understanding how changes in the built environment causally shape travel behavior remains a central challenge in urban studies, particularly in contexts of large-scale resettlement and urban peripheralization. This study exploits a quasi-random housing allocation system in Shanghai as a natural experiment to identify the causal impacts of built environment degradation on commuting mode choices. An integrated analytical framework combining double machine learning and causal mediation analysis is developed to disentangle direct effects and indirect pathways operating through household transportation assets. Using survey data from 5967 households across resettlement communities, the empirical results show that built environment degradation significantly suppresses active travel while, counterintuitively, increasing reliance on public transit. Mediation analysis further reveals that these effects are jointly shaped by adjustments in private car ownership and non-motorized vehicle ownership, with distinct and heterogeneous transmission mechanisms across different built environment dimensions. In particular, declines in land-use mix, accessibility, and transit service provision weaken local walkability while reinforcing reliance on public transit under constrained mobility conditions. These findings challenge conventional expectations derived from Western suburbanization and point to a context-specific mechanism in which spatial mismatch, combined with institutional constraints on car ownership, reshapes mode choices by limiting households' ability to adapt through motorization. By integrating causal inference with a natural experiment, this study provides new insights into the behavioral consequences of housing resettlement and offers policy implications for coordinating land use, transport provision, and social equity in rapidly urbanizing megacities.