Impacts of urban rail transit on local air pollution : evidence from path analysis of 104 Chinese cities and quasi-experiments in Shenzhen and Hong Kong

Abstract

Substituting urban rail transit for private vehicles is commonly considered essential in abating traffic-related air pollution, but urban rail transit itself may function as an indirect source of pollution by encouraging intense urban development and increased energy use. In this thesis, I explore the nexus between urban rail transit and local air quality at different spatial/temporal scales, focusing on the following three quasi-independent but highly interconnected questions: (i) What is the underlying mechanism of rail-pollution effects?; (ii) What causes spatial heterogeneity in the rail-pollution effects?; and (iii) What causes temporal heterogeneity in the rail-pollution effects?

My first analysis presented in Chapter 2 focuses on the first question and examines the underlying mechanism of rail-pollution effects across 104 Chinese cities, taking a structural equation modeling approach. The results show that metro density negatively affects CO, NO2, and PM2.5 concentrations through indirect channels, presenting net elasticities of -0.027, -0.016, and -0.015, respectively. Metro density also presents confounding impacts on vehicle kilometers traveled, in terms of sign. These findings suggest that two opposing effects—traffic diversion and creation effects—are mingled in reality, making overall effects fuzzy. A detailed path analysis demonstrates that both effects are significant, but the traffic diversion effects (-0.278) tend to be largely offset by the traffic creation effect (0.122), showing a net elasticity of -0.156.

The second analysis discussed in Chapter 3 addresses the second question and captures spatial heterogeneity in air quality effects of metro network expansions in Shenzhen. The difference-in-differences model estimation results indicate that metro network expansions in October 2016 led to a 6.0% decrease in daily mean local PM2.5 levels. This air quality benefit tends to be greater in places with high population density, diverse land use, walkable communities, better access to transit, as well as rail plus property projects. These findings reflect that the transit-oriented development characteristics encourage public transit ridership and enlarge the air quality benefits of urban rail transit.

My third analysis in Chapter 4 focuses on the last research question and applies a generalized difference-in-difference-in-differences model to investigate temporal heterogeneity in the air quality effects of MTR shutdowns in Hong Kong. The pollution-generation effects of MTR shutdowns are substantially larger during rush hours on working days (18.9%) than any other time of the week (6.9%), reflecting the substitution effects of MTR on road traffic for road traffic commuting trips to avoid congestion. In addition, pollution-generation effects tend to decline over time, and NOx concentrations drop by 0.1% per hour during the shutdown period. This reflects the adaptation tendency of local commuters against interrupted MTR services.

In sum, my thesis research finds that urban rail transit has significant effects on air quality with two opposing effects—traffic diversion and creation effects, and the rail-pollution link holds primarily through indirect channels involving a set of intermediate variables explaining transport and built environment characteristics. Overall, traffic-diversion effects tend to be stronger than traffic-creating effects in Chinese context, but their magnitude is subject to spatio-temporal heterogeneity.