Introducing Mass Transit in Gothenburg - Relationships Between Travel Time and Mode Choice

Abstract

Transport is a vital service that allows us to move between places we need to be and makes our daily life puzzles possible. Transport planning has for a long time focused on car travel which is coupled with both congestion and environmental issues. It is necessary to promote a more sustainable mobility and one way to do that is to make better use of transit. To achieve a shift from car towards transit it must be a competitive alternative. One of the most influential factors on individuals’ choice of transport mode is travel time. Reduce a modes travel time and the utility of that mode will increase, and so will its market share. This thesis aims to investigate the relationship between travel time and mode choice by quantifying the effects of transit travel time through a meta-analysis approach. The literature search focus on, but is not limited to, studies of mode choice taking a multinomial logit (MNL) approach for reasons of interpretability. Elasticities derived from a sample of 16 studies conclude that a reduction by transit travel time by 1% can be expected to bring an increase of the transit mode share by 1.29%. Further, a 1% reduction in in-vehicle times of bus and metro are found to increase their shares by 0.59% and 0.33% respectively. The acquired elasticities are applied on a sketch complement to the transit network in Gothenburg, Sweden, to see what changes in mode split that can be expected if it were to be implemented. The potential travel time savings brought about by the proposed service are estimated between twelve activity centres in the city. The service is one of metro-standard and configurated based on a combination of local knowledge and transit ridership data provided by the local transit agengy. Results show an expected increase of the transit mode share in the range of 4.9-18.8 percentage units in the affected area and an increase of daily transit trips by as much as 20 000 (8%), perhaps even more. Ideally a meta-analysis gathers all previous studies of a certain topic. Although that is rarely a possibility, a larger sample of studies would have enabled more accurate and more diverse results to be derived. The decision to focus on MNL mode choice models might have impeded the final sample size and also the occurring modal distinctions through its inherent IIA assumption.