Karthik K Srinivasan

This study formulates a disaggregate direct demand model of bus transit ridership while addressing the following substantive and methodological issues: (a) endogeneity and non-linearity of the influence of service frequency on ridership, (b) inter-route relationships such as competition and complementarity among routes within the bus transit network and with other transit networks (such as the metro/rail network), (c) relating spatially aggregated demand to disaggregate, stop-level catchment characteristics – although demand data are available only at an aggregation of stop-clusters, and (d) overlapping of catchment areas among closely spaced stops. The proposed model is applied to analyze bus transit ridership (boardings) during weekdays for morning peak period in Bengaluru, India. This study is among the first to develop a comprehensive direct demand model for forecasting bus transit ridership in an Indian city. Yet, the proposed conceptual and methodological framework and the findings from the study are general enough to be of use for transit planning in other cities of India and other countries. Transit agencies with spatially aggregate, fare-stage cluster-level ridership data can employ the proposed approach to examine the influence of disaggregate stop-level catchment characteristics on ridership. Additionally, transit agencies may utilise the proposed model to quantify bus ridership impacts of service network modifications, route alignments, and network connectivity/accessibility, while considering interactions with other transit networks. The empirical results suggest that while increasing service frequency increases ridership along low-frequency routes, the returns from increasing service frequency diminish as current frequency levels increase. Further, it is shown that route-level passenger kilometres, a variable commonly available with transit agencies, serves effectively as an instrument for addressing endogeneity between route-level service frequency and stop-route-level ridership.

This study examines the impact of bus service headway variability on bus transit ridership using direct demand models at different levels of spatial aggregation – route level and stop-route level – using transit demand and supply data from the city of Bengaluru, India. In addition, auxiliary models are developed to understand the determinants of service frequency and headway variability and to address the endogeneity of these service characteristics in the demand models. This is perhaps the first study in the public transit literature to compare and contrast the endogeneity and non-linearity effects of service frequency and headway variability in transit demand models using both a conceptual framework and empirical evidence from a large transit system. The empirical results offer evidence that variability in headways adversely impacts transit ridership and passenger-kilometres. The strength of the adverse effect increases with increasing variability. On the other hand, the influence of service frequency decreases with increasing frequency. Furthermore, it is shown conceptually and demonstrated empirically that ignoring the endogeneity of service variability results in an underestimation of its adverse effect on transit demand. On the other hand, the empirical results suggest that ignoring the endogeneity of service frequency would result in an overestimation of its beneficial effect. An important takeaway from these results and additional policy simulations is that transit agencies can potentially gain greater ridership and revenue by reducing headway variability rather than simply allocation more buses and crew to high-frequency routes.