The research reported here was conducted for the Passenger Demand Forecasting Council, a body representing the train operators and other railway organisations in Great Britain. The objective of the study was to conduct a quantitative analysis of aggregate rail ticket sales data to develop models to examine the effects on rail demand of changes to fare, timetable related service quality (time, frequency and interchange) and station accessibility. The focus of this paper is on the of station catchment areas and on the identification of the influence of catchment area socio-demographics on the demand for rail travel. as we are aware, previous models which have attempted to consider access and egress time have used the above constant elasticity form. However, it may be that constant access and egress elasticities are not the most appropriate functions. For example, it is not inconceivable that the rate at which rail trips per head decay is relatively minor within the urban area around the station but then falls off dramatically outside of the built up area. To examine this 'distance decay' effect four alternative access and egress functional specifications are considered, including the exponential decay function, the logit decay function, the Gompertz decay function and the quadratic decay function. The development of models to cross sectional data for London and non-London based non-season ticket flows has provided a number of interesting and significant findings :After accounting for correlation between fare and GJT, the cross sectional models generate sensible estimates of GJT elasticities of around -1.25 for non-London traffic, -0.96 for traffic to London and -0.70 for traffic from London; The models show very little statistical difference between including access/egress time inside or outside of GJT but there are good theoretical arguments that advocate specifying access/egress time outside GJT; A considerable effort has been spent examining the functional form of the access/egress decay effect which concludes that statistically and theoretically superior models can be achieved specifying more flexible inverted s-shaped decay functions as opposed to constant elasticity models; The specification of the population elasticity of demand is statistically and practically superior when specified in relation to the access/egress weighted station catchment area population rather than the population for an individual zone. For the covering abstract please see ITRD E135207.
Samenvatting