Forecasting Demand for High Speed Rail in the Two Main Corridors of Norway

Forecasting Demand for High Speed Rail in the Two Main Corridors of Norway


S Fluegel, A H Halse, Institute of Transport Economics, NO


We analyze the demand potential for high speed rail in Norway using a model combining stated preference and revealed preference data. The impact of different levels of service is assessed and the prospect for business is discussed.


As the current long distance rail network in Norway is far from offering high speed rail (HSR) opportunities, the level of service (LoS) of the conventional trains is rather poor with respect to speed and frequency. Consequently, the air and - for leisure trips - the car mode are the preferred modes of transport for long distance travel. Despite the difficult topography (fjords and mountains) and the relatively low population density in and between the biggest cities in Norway, HSR has been put on the political agenda.

Arguably the two most promising corridors for HSR are the ones between the capital Oslo and the two second largest cities Bergen and Trondheim. In 2010, there were close to 2 million one-way trips per year in each of the two corridors. A new HSR network could more than halve the current in-vehicle travel times of around 6 h 30 min. This potential time reduction is likely to make trains more competitive with aviation, especially for business trips, where the possibility of returning home on the same day seems an important factor for choice behavior. This paper presents concrete ridership forecasts for the two corridors.

Preferences for HSR relative to the current modes car, air, bus and conventional rail are inferred from a study combining data on revealed and stated preferences (RP-SP data). From a large scale RP study, respondents are recruited for a tailored SP study. Close to 900 respondents completed the self administrated web questionnaire of the SP study that includes mode choice experiments between the respondent?s current mode of travelling and hypothetical HSR options. From these binary choices, mode choice models (segmented by trip purpose) for all five modes are formulated and estimated. The mode choice models indicate that HSR is competitive (winning over 50% of the market) under reasonable assumptions about LoS (albeit the HSR-fare assumed will most likely imply a subsidy by the state). The HSR demand elasticity with respect to travel cost is found to be rather high for leisure trips and the demand elasticity with respect to travel time is found to be high for work-related trips. Other important variables are access/egress time, frequency and the share of ride spend tunnel. However, the marginal effects of tunnel share, which on the absolute level might well constitute between 40-70% of the entire HSR track, is found significant only for leisure trips.

Besides the calculation of mode-switching probabilities, the effect of an increased measure of accessibility due to the introduction of HSR is estimated using intentions data of trip frequency obtained in the questionnaire. Using a simple exponential model, this impact of accessibility on (additional) trip frequency is found to be significantly different from zero but rather low in value. This analysis indicates a moderate size of newly generated demand of HSR.

In order to make predictions at the network level, we utilize current and future origin-destination trip matrices for the base scenario (defined as scenario without HSR) estimated with the Norwegian National Transport Model (NTM5). These matrices are then aggregated into somewhat broader zones and adjusted using the ridership information gathered in the RP study. Then, based on the mode choice model and the estimated increase in trip frequency, we calculate ridership for HSR and the other transport modes under different specifications of HSR. Evaluation years are chosen to be 2024, 2043 and 2060. By systematically varying HSR-fares we derive a complete demand function for HSR and determine the revenue-maximizing fare.

The paper discusses the plausibility of the forecasts and critically examines the underlying assumptions. In an additional exercise further sensibility tests are performed by varying other LoS-variables like frequency and tunnel share.

In the last section, prospect for the business case of HSR in the bespoken corridors are discussed. Using external data on expected operation and investment cost, it is indicated that it will be hard to cover operation costs, not to mentioned investment costs. Finally, we discuss other impacts of HSR, in particular potential CO2-emission savings.


Association for European Transport