Long Term Changes in Transport Demand

Long Term Changes in Transport Demand


C Riff Brems, T Jensen, DTU Transport, DK


Formulation of a micro based approach enriches the identification and forecasting of changes in long term demand. Based on Danish pseudo-panel data from 1975 to 2009 the paper presents the most important long term effects.


The primary purpose of this paper is to present a context for identifying causes for changes in long term demand. Some changes are related to changes in population and location, some to changes in economy, and some to the quality of the transport system. The paper will include some indications of the importance of these causes based on Danish travel survey data and register data over a 35-year period from 1975 to 2009.

Determining the changes in transport demand on long term is not only important with respect to congestion and thereby for the profitability of infrastructure investments. It is also important with respect to necessary initiatives to reduce emissions from the transport sector. Traditionally, forecasting demand on long term is based on a macro approach, which relates demand to GDP, car prices etc. However, with the data now available long term demand can be based on a micro or semi-micro approach, which will enrich the future forecasts of long term changes in transport demand.

One element of the development of a national transport model for Denmark is a sub-model for forecasting long term changes in transport demand. The basis for this model is a pseudo-panel data set of the entire Danish population. The data set is created as part of the work with the national model, and it combines year-by-year register data of the Danish population with information of home and work locations, household income and car ownership with travel survey data (sample data) describing number of trips by trip chain, purposes, modes, travel times and distances, etc. The survey data are collected in 1975, 1981 and continuously from 1992 onwards with the exception of 2004/5. Finally, more general information like fuel cost and car prices are included dependent of the type of car(s) in the household.

With a pseudo-panel data set it is possible to combine the advantages of cross-sectional and time series data, and thereby it is possible to identify changes in different population groups over time. An example from a preliminary analysis shows that transport demand is higher for a 50-year old born in the 1950?es than for a 50-year old born in the 1930?es. The preliminary results show that we are going to underestimate the long term demand, if we assume that the travel behaviour of a 50 year-old in 2030 is the same as the behaviour of a 50-year old today.

The methodology for isolating different effects on long term demand based on a group of models estimated on panel or pseudo-panel data. Greene (1997) describes two different approaches with either fixed time or group effects and a third approach with a dynamic model with first-order effects. All types will be tested in the estimation process.

By September the estimation of the abovementioned models are completed, whereas the re-estimation and implementation of the identified effects in a discrete choice context will follow. For use in forecasting the results will be combined with the use of a prototypical population as described in e.g. Daly?s presentation at PTRC in 1998 and the documentation for the Dutch national model v. 5.0.


Association for European Transport