Effects of New Urban Rail Transit on Travel Behavior: A Natural Experiment

Abstract

Over the past few decades, the increasing popularity of using private vehicles has degraded the urban environment and reduced the quality of life for city dwellers. To cope with growing travel demand and reduce society's dependence on private vehicles, many local and regional governments around the world have heavily invested in urban rail transit systems, or plan to do so, to promote sustainable travel patterns. However, there is an ongoing debate about the effectiveness of urban rail transit investments, which depends on the extent to which rail transit can promote sustainable mobility and, in particular, convert car users to rail transit users. In response to this debate, many empirical studies examined the effects of rail transit on individual travel behavior.

There are several research gaps in the methodology and results of current empirical studies remain to be addressed. First, most studies have used cross-sectional data collected at a single point in time to examine this relationship, which cannot provide robust causal inference. Second, although the natural experiment has theoretical advantages over cross-sectional analysis, few studies have compared these two approaches to test the robustness of the cross-sectional studies. Third, among a handful of natural experiments, most of them defined the treatment and control groups using a single distance threshold to new rail stations, which might misestimate the impact range of urban rail transit systems. Fourth, few studies have explored the potential mechanisms behind the link between urban rail transit and travel behavior, particularly how built environment features and individuals’ travel attitudes mediate such link.

To address these gaps, this study used a new rail line in Wuhan, China, as a natural intervention to examine the effects of urban rail transit on individuals' travel behavior and to explore the potential mechanisms of the effects. To accurately capture the impact range of the new rail line, two treatment groups and one control group were assigned based on multiple distance thresholds. The first treatment group consisted of residents living within 800 m of the new rail stations, and the second treatment group consisted of residents living within 800-1600 m of the stations. The control group consisted of residents living within 1600-2400 m of the stations. First, paired t-tests were used to test whether the travel behavior of residents significantly changed before and after the introduction of the new rail line. Second, difference-in-differences (DiD) analyses were used to examine the impact of the new rail line by comparing the changes in travel frequency and duration for each travel mode (including rail, bus, car, e-bike, cycling, and walking) and total trips among different groups. Third, multilevel regression models were used in the cross-sectional analysis based on the follow-up data to examine the association between urban rail transit and travel behavior, and the results were compared with those of the natural experiment. Fourth, a structural equation modeling (SEM) approach was used to disentangle the interaction between the new rail line, the built environment, transit attitudes, and travel behavior.

There are four key findings from this study. First, the natural experiment analysis found that the opening of the new rail line increased rail use and reduced car driving and walking for both work and non-work trips. But its effects on bus, e-bike, cycling, and total trips were insignificant. This suggests that the increase in rail transit use after the introduction of the new rail line came at the expense of driving and walking. Second, the study found that the impact thresholds of the new rail line were 800 m for most travel modes, except for the frequency of rail trips for non-work purposes and walking trips for work purposes, whose thresholds reached 1600 m. Third, although the model of the cross-sectional analysis controlled for various potential confounders, it appeared to overestimate the effects of rail transit on cycling and total work trips compared to the natural experiments. Fourth, following the opening of the new rail line, residents near the stations perceived improvements in the surrounding built environment and increased their preference for rail, which in turn affected travel behavior. More specifically, for the built environment, improvements in transit and accessibility acted as mediators in promoting the switch from driving and walking to transit use, while walkability did not have a significant impact. In terms of travel behavior, the new rail line inspired residents' preference for rail, which further promoted rail use, but failed to influence driving or walking.

This study also has important policy implications for urban transport planning. In order to increase the efficiency of investment in urban rail transit and promote sustainable mobility, an integrated urban planning strategy that coordinates urban rail transit transportation planning and land use policies (such as land use diversity and pedestrian-friendly environment design) should be adopted.

Date of Award	11 Aug 2023
Original language	English
Awarding Institution	City University of Hong Kong
Supervisor	Yi LU (Supervisor) & Jiandong Peng (External Supervisor)