Asymmetric modal interactions between demand-responsive transit and fixed-route transit: A spatiotemporal autoregressive approach

Demand-responsive transit (DRT) and fixed-route transit (FRT) play distinct roles in public transit systems, especially in rural contexts. Existing studies report conflicting modal interactions (substitution vs. complementarity) depending on service design, underscoring the need for precise estimation of modal interactions before drawing policy conclusions. This study employs a Spatio-Temporal Autoregressive model with eXogenous variables (STARX) to estimate DRT-FRT interactions using daily ridership data from rural Cheongju, South Korea, while explicitly controlling for spatiotemporal autocorrelation, built environment, and sociodemographics. The analysis yields two key findings. First, DRT and FRT exhibit asymmetric complementarity: DRT exerts a time-varying influence on FRTshort-term substitution transitioning to long-term complementarity at the system level-while FRT's aggregate influence on DRT is negligible. Locally, DRT's substitutive pressure on FRT is concentrated in elderly and high-employment areas across short-and medium-term horizons. Second, models that ignore spatiotemporal dependencies systematically overestimate interaction effects, inflating evidence of complementarity. Since DRT services typically operate with substantial public subsidies, such overestimation risk justifying excessive DRT expansion and inefficient allocation of public resources. By correcting spatiotemporal bias, this framework enhances the evaluation of transit sustainability and guides public investment in rural mobility.