Abstract Manager

The COVID-19 pandemic brought significant changes to travel demand patterns in North America, leading to greater emphasis on shared and safer transportation modes, and changes in their levels of service. Given the growing importance of multimodality in the post-pandemic era, understanding the interactions between the supply and demand of various modes during and after the pandemic is crucial. This study addresses two key gaps: first, the limited research on the interaction between supply and demand across different modes, and second, the challenge of analyzing pandemic data due to the complexity and fluctuation of demand patterns. The study examines the impacts of changes in supply, or level of service, on travel demand across multiple transportation modes, while accounting for fluctuations in ridership patterns. 

The methodology focuses on both elasticity and cross-elasticity of supply and demand. Elasticity measures how changes in the supply of one mode affect its own ridership, while cross-elasticity examines how changes in the supply of one mode influence the demand for others. For example, does expanding carsharing services in a region affect public transit and taxi usage, and by how much? The analysis includes various transportation modes in Montreal—bikesharing, free-floating and station-based carsharing, subway, taxi, bus, and private cars—using daily ridership and usage data from 2019 and 2024. For modes without changes in the level of service during the pandemic, previous Origin-Destination and travel surveys are incorporated for comparison. 

To address fluctuating travel patterns, change point detection is used to identify distinct phases for each mode’s ridership. Changes in supply during these phases, such as vehicle availability and operational capacity, are incorporated as input variables in time series and statistical models, with both binary and continuous variables tested. Two methods are proposed: Instrumental Variables (IV) for causal analysis and time series models with phase-specific inputs to capture the impact of supply changes on demand. The latter includes exogenous variables such as phase-specific binary variables, weather conditions, and day type. Elasticity and cross-elasticity are estimated through coefficients in linear time series models, while machine learning models are also employed, using SHAP values to analyze complex interactions. 

By testing scenarios with supply adjustments, this study provides insights into how expanding or limiting the service of one mode can affect the usage of others. These findings contribute to policy planning for multimodal travel demand and fostering better collaboration between transportation modes.