Built environment impact on travel behavior in the context of China : evidence from transit-oriented development areas in Shenzhen

The economic growth and urban spatial expansion in China have led to the rapid increase in private car ownership and automobile-dependent travel behavior. Along with the expansion and upgrading of urban rail transit networks, transit-oriented development (TOD) has been widely recognized as a land-use strategy of relieving over-reliance on automobile and encouraging transit patronage. Accumulated studies have found the differences between TOD and non-TOD communities in terms of the travel behavior of residents, whereas some literature has documented the effects of individual built environment factors on travel behavior. However, focus on neighborhoods in planned TOD areas is generally lacking. This study aims to examine the effects of built environments on travel behavior in the context of TOD in Chinese cities by using survey data from five selected TOD areas in Shenzhen. The subjects of inquiry are as follows: 1) whether built environment factors (e.g., density, land-use mix (LUM), and access to transit) affect travel behavior (e.g., mode choice, travel frequency, and vehicle kilometers traveled (VKT)) in the context of China; 2) the extent in which these built environment factors affect travel outcomes after controlling for socio-demographic variables. A cross-sectional household survey was conducted in 59 neighborhoods across 5 TOD areas in Shenzhen, China, in December 2013. The density, LUM entropy, and number of bus stops and bus lines within 500 m of the 59 neighborhoods were calculated on the basis of the GIS map of Shenzhen. The binomial logit model (BLM) was used to estimate household car ownership; the ordinary least square (OLS) was used to calculate VKT; the Poisson regression model (PRM) was adopted to measure walking and transit frequency; the conditional logit model (CLM) was employed to evaluate commuting mode choice. For each of these models, the comparison of “basic” and “expanded” models reveal the effects of land-use factors on travel behavior after controlling for socio-demographic variables. The results from four pairs of empirical models confirm the effects of built environments on shaping travel behavior. In terms of built environment variables, the study concludes the following: 1) LUM is the most powerful predictor of the travel behavior of residents; 2) Transit access factors effectively encourage vehicle alternative mode choices; 3) Non-residential density factors are more important than residential density factors in generating desirable travel behavior. Residential density is positively related to VKT per capita. With regards to travel outcome perspectives, the findings include: 1) non-work trips are more influenced by built environment factors than work trips; 2) private car ownership is primarily associated with socio-demographic factors such as personal income and household size. VKT per capita is more related to built environment characteristics than car ownership. These findings imply that: 1) increasing residential density may no longer be the first priority of TOD in the context of China. Mixed land use development should be more emphasized on TOD planning; and 2) TOD is an effective but insufficient measure to reduce vehicle ownership and travel demand to a large degree. It has to be coupled with other types of policies such as new plate quota (NPQ) or driving bans.published_or_final_versionUrban Planning and DesignDoctoralDoctor of Philosoph