Enhancing rail direct demand models with competition between ticket types using contributions from economic theory and market research

Direct demand models estimated to ticket sales data have for many years provided evidence on how key variables influence rail demand in Great Britain. Nonetheless, there has been relatively little estimation of demand models disaggregated by ticket type which would provide own and

cross-price elasticities that can inform the pricing of different ticket products. We here report such enhanced models estimated on large data sets and exploiting the relationships of economic theory within a demand system. In addition, a complementary market research exercise is undertaken that itself provides ticket specific own and cross-elasticities and which also supports the estimation of models jointly based on actual and stated behaviour. We conclude that the demand systems approach can recover robust ticket specific own-elasticities but that there are econometric difficulties in estimating cross-elasticities even using supporting economic theory, so that cross-elasticities between tickets are better deduced from these own-elasticities than estimated.

This contrasts with the convention in the railway industry in Great Britain where own and crosselasticities are deduced from recommended conditional elasticity evidence. Market research also has a role to play and provides own-elasticities that, as is common, are rather larger and crosselasticities that are a little larger than those derived from ticket sales analysis. A key feature of this work is to reconcile those two approaches by scaling the market research elasticities using ticket sales data. This further supports our conclusion that generating robust own-elasticities and deducing cross-elasticities from these is currently the most fruitful method of obtaining a full set of own- and cross-elasticities for different ticket types within a demand system framework, and that this approach is superior to the conventional single equation approach.