Built environment, car ownership and PM2.5: Stronger causal estimates from a quasi-experiment

Abstract

The causal relationship between the built environment, car ownership, and travel-induced pollutant emissions remains obscured by residential self-selection (RSS) bias. This study leverages China's unique housing demolition and resettlement program to conduct a quasi-experiment, analyzing how the built environment impacts car ownership and travel-induced vehicle-related PM2.5 emissions in Shanghai. By employing structural equation modeling (SEM) on data from 5370 households relocated via government-assigned housing (i.e., replacement housing), we circumvent RSS bias and isolate the built environment's causal effects. Results reveal that higher residential density and land-use mix decrease car ownership but increase PM2.5 emissions to a certain extent, while increased road network density and bicycle/pedestrian connectivity reduce both car ownership and emissions. Car ownership, influenced by socio-demographics, the built environment, and transport services, mediates the relationship. Findings highlight the dual role of transportation and urban planning: extremely dense, mixed-use areas may inadvertently elevate emissions without green transport modes, whereas pedestrian-friendly designs and robust public transit reduce car dependency, hence decreasing emissions. This study provides actionable insights for policymakers to align urban and transport development with sustainability goals, advancing empirical evidence in a high-density Asian metropolis and addressing gaps in regional literature.