Changes in the predictors of transit ridership in post-COVID-19 US metropolitan areas

Abstract

This study investigates changes in transit ridership predictors pre- and post-pandemic across 35 United States metropolitan areas. Using extreme gradient boosting and data spanning January 2019 to June 2023, we identify a shift in the key predictors of transit ridership from internal factors in the pre-pandemic era to external factors in the post-pandemic era. Socioeconomic factors, gasoline prices, telecommuting, and polycentric development collectively contribute more to post-pandemic ridership than vehicle revenue mile (VRM), fare, transit coverage, and service areas, marking a reversal from the pre-pandemic era. Moreover, we uncover threshold and interaction effects unique to the post-pandemic era, including positive interactions between the proportion of African American residents and poverty rates, carless households and gasoline price, VRM and polycentricity. Notably, the rate of carless households moderates telecommuting’s impact on ridership. We recommend that transit agencies account for post-pandemic shifts in ridership predictors to ensure accurate forecasts and to adjust service provision and policy prioritization accordingly.