Exploring the nonlinear association between bike-sharing and household car ownership in Shenzhen, China

Although many studies have explored the influences of bike-sharing on car use, few have emphasized its potential to reduce the growth in car ownership, which encourages car dependence. Using travel survey and bike-sharing data from Shenzhen, a city with rapidly growing car dependence, we developed gradient boosting decision trees models to capture the nonlinear relationship between bike-sharing and car ownership, while controlling for household and built environment characteristics, as well as other travel alternatives. The results suggest that bike-sharing is a strong predictor of car ownership, and they have a negative and nonlinear association. The association between bike-sharing and car ownership is also stronger in suburban areas than in urban areas. The findings highlight the potential of shared micromobility services in alleviating car dependence and underscore the importance of extensive spatial coverage in reducing car ownership. Furthermore, bike-sharing could be a viable alternative in mitigating car ownership besides built environment interventions and rail transit investments.