Measuring when Uber behaves as a substitute or complement to transit: An examination of travel-time differences in Toronto

Abstract

Policymakers in cities worldwide are trying to determine how ride-hailing services affect the ridership of traditional forms of public transportation. The level of convenience and comfort that these services provide is bound to take riders away from transit, but by operating in areas, or at times, when transit is less frequent, they may also be filling a gap left vacant by transit operations. These contradictory effects reveal why we should not merely categorize ride-hailing services as a substitute or complement to transit, and demonstrate the need to examine ride-hailing trips individually. Using data from the 2016 Transportation Tomorrow Survey in Toronto, we investigate the difference in travel-time between observed ride-hailing trips and their fastest transit alternative. Ordinary least square and ordered logistic regressions are used to uncover the characteristics that influence travel-time differences. We find that ride-hailing trips contained within the City of Toronto, pursued during peak hours, or for shopping purposes, are more likely to have transit alternatives of similar duration. We also find discrepancies in travel-time to be often caused by lengthy walk- and wait-times for transit or because of transfers. Our results further indicate that 30.61% of ride-hailing trips in our sample have transit alternatives of similar duration. These are particularly damaging for transit agencies as they compete directly with services that fall within reasonable transit expectations. We also find that 26.87% of ride-hailing trips would take at least 30 minutes longer by transit. In light of these findings, we discuss recommendations for ride-hailing taxation structures.