An Econometric Analysis of the Impact of Reliability on Passenger Rail Demand

The results of a study commissioned by the UK Department for Transport to investigate the impact of reliability on passenger rail demand using econometric analysis are described. Although econometric analysis of rail ticket sales data has made a major contribution to the understanding of UK rail demand, most notably with regard to external factors, fares and timetable related service quality, applications to reliability have been limited. As the main purpose of this study is to examine the impact of reliability in detail, it was essential to measure performance where reliability most directly manifests, i.e. at the origin-destination (O-D) level. A sample of 248 O-D pairs was chosen using the basic principles that the sample should include: a small number of train operating companies (TOCs); both London and non-London flows; situations in which there had been significant changes in reliability during the previous 5 years; and some variation in train service levels over the period. The data set covers 2002 to 2007, with 13 4-weekly periods per year. The number of trips and revenue by O-D was obtained from LENNON ticket sales data and generalised journey times (GJT) from MOIRA. In addition, socio-economic-demographic data at the Government Office Region level were obtained from ONS. Measures of reliability were obtained both on the service group level, the TOC level and the O-D level. The modelling framework, specifications and estimation procedures for the econometric model are described. The model was estimated for season and non-season tickets separately. For non-season tickets full-priced and reduced-priced were estimated separately. Since the model estimated is of the double-logarithmic form, the elasticities are obtained directly as the coefficients of the explanatory variables Reliability, as characterised by above metrics, is found to have a marginal, but statistically significant, effect on rail demand and the effects in relation to ticket type are discussed. Both static and dynamic specifications indicated that full fare patrons are more sensitive than other patrons to reliability. Further studies with a more substantial dataset are considered desirable