Highlights • Examine the reasons contributing to the increase in ridehailing demand. • Explore the shift from traditional taxi services to TNC services. • A joint framework of negative binomial and multinomial fractional split model is used. • Proposed model can be utilized for predicting future ridehailing trends.

Abstract The proposed study contributes to our understanding of the ongoing transformation of ridehailing market by examining the New York City Taxi & Limousine Commission data from a fine spatial and temporal resolution. We examine taxi zone based demand data from NYC for each month and explore the reasons contributing to (a) the increase in ridehailing demand and (b) the shift from traditional taxi services to Transportation Networking Company (TNC) services. The first component – taxi zone ridehailing demand - is analyzed adopting a negative binomial count model. The second component - share of traditional and TNC services demand - is analyzed using a multinomial fractional split model. The two model components are stitched together in a joint framework that allows for the influence of repeated observations as well as for the presence of common unobserved factors affecting the two components. The model estimation considered a comprehensive set of independent variables including transportation infrastructure variables, land use and built environment variables, weather attributes, and temporal attributes. Several performance measures were generated using the joint model for estimation and validation datasets. A prediction exercise is conducted to illustrate how the proposed model system can be utilized for predicting future ridehailing trends. Finally, an elasticity exercise is conducted to estimate the influence of independent variables on the ridehailing market.