Evaluating a Replacement Ferry for the Isles of Scilly Using a Discrete Choice Model Framework
Marco Kouwenhoven, Charlene Rohr, Stephen Miller, Andrew Daly, RAND Europe, UK/NL; James Laird, Institute of Transport Studies, University of Leeds, UK
This paper describes a model to predict changes in travel demand and mode shifts for travel to the Scilly Isles. Interesting aspects of the modelling include the joint usage of SP/RP data, modelling frequency, and calculation of consumer surplus.
Access to the Isles of Scilly to and from the British mainland is currently provided by three modes of transport: a sea ferry, a helicopter service and a fixed-wing aircraft service. It is anticipated that both the passenger ferry and the freight vessel will soon come to the end of their economic life and will require replacement. In response, Cornwall County Council, on behalf of the Penzance to Isles of Scilly Route Partnership, are preparing a Major Bid Submission for capital funding support for improved transport links to the UK Department for Transport. As part of this bid, a robust Cost Benefit Assessment was required to quantify whether a replacement ferry between the Isles and Penzance is justified.
As part of this assessment a travel demand model was developed to examine travellers? responses and quantify their benefits. Because of the potential loss of ferry services, this model needed to address two possible traveller responses: modal shift changes and changes in total travel demand. The latter responses were represented through changes in journey frequency.
Both stated preference (SP) and revealed preference (RP) data were collected from existing travellers to the Islands to develop the transport models. The strength of SP data is in deriving the relative importance of the different aspects of service (price, crossing time, comfort, etc.). However, to derive elasticities and forecasts it is necessary also to estimate the absolute scale of response and in this respect RP data is required. The RP and SP data were used jointly to estimate the mode choice models, taking explicit account of different error variation in each data source. In the model estimation procedure correlation between alternatives, through nested model structures, and ?inertia?, i.e. preference for the observed mode, were explicitly tested. Stated intentions data, exploring how travellers? reported frequency of travel varied with differing ferry service scenarios, were used to estimate the relationship between travel frequency and travel accessibility.
Validation of the model was undertaken through examination of the resulting values of ferry travel time (and other service quality characteristics) and through examination of the ferry price elasticities, for each travel segment. The resulting final models for staying visitors, for daytrip visitors and for residents were used to build a forecasting tool that predicts the demand and revenue for each mode and the resulting consumer surplus for the coming 60 years.
An important consideration for forecasting was the issue of capacity restraint, where it was found that in a number of future scenarios the demand for ferry travel predicted by the models exceeded the annual ferry capacity. In these cases, the demand forecasts were adjusted by a ?shadow price? that reduced the overall demand to the capacity of the boat. The formulation of the shadow price term influences who is able to access an over-capacity service and therefore has a direct influence in the calculation of traveller benefits, although the shadow price itself is excluded from the benefit calculations.
The main objective of the modelling was to quantify traveller benefits for the Cost Benefit Analysis of the varying ferry options, which were undertaken in full conformance with the principles and practices set out by the UK Department for Transport. General UK appraisal guidance recommends that the ?Rule of Half? for calculation of traveller benefits; however, the rule breaks down if a new mode is introduced or an existing mode becomes redundant. The latter case is relevant here. In situations like these estimation of the traveller consumer surplus requires additional analysis using the travel demand model as an estimate of the demand curve. Specifically, in this study, the consumer surplus for each travelling group was calculated using the exact integral of the demand function, incorporating both the mode choice and trip frequency components. By using the demand function to calculation the consumer surplus, usual terms like time savings and changes in fares were incorporated in the consumer surplus calculation. Additionally, terms incorporating time scheduling and quality benefits, both in terms of the boat and harbour-side improvements, which are not normally incorporated in normal UK transport appraisal, were also included. This was a further advantage of using the integral of the demand curve in the appraisal process.
The study is unique as a multi-modal study in addressing links to an isolated island community, with modal choice between sea vessels, fixed wing aircraft and helicopter.
Association for European Transport