Stability of model derived values and parameter estimation: Example of a simultaneous routeand mode choice model

Abstract

Current practise separates route choice from mode choice by treating the first choice in assignment models. It is obvious though, in particular for public transport assignment, that the implied valuations of the route choice and the mode choice should be identical, e.g. values of travel time savings or transfer penalties. This implies that the mode choice model parameter estimation is sensitive to the descriptions of the nonchosen alternatives, in general, provided by the assignment models with their apriori embedded parameters. The paper describes, tests, calibrates and demonstrates an iterative procedure to ensure the consistency of the parameters and the assignment results. For this purpose it develops a stochastic user equilibrium (SUE) assignment of connections, i.e. mode and route choice combinations. The SUE is based on an expanded nested logit structure, which properly accounts for similarities due to the connection overlappings.<br/>A realistic example, based on parts of the Swiss road and rail network underlines the need to adopt this approach in practical work, in particular as transport planning is increasingly charged with predicting behavioural responses to demand management measures.