Econometric Modelling and Forecasting of Freight Transport Demand in Great Britain
S Shen, A S Fowkes, D H Johnson, A E Whiteing, ITS, University of Leeds, UK
This study applies up to date advanced econometric models to the analysis of road and rail freight transport demand in Great Britain (GB).
Empirically derived estimates of freight transport demand elasticities and accurate forecasts of future demand are important for freight planning and policy making. Econometric models can not only forecast future demand but can also explain economic or business phenomena and increase our understanding of relationships among variables. This study applies up to date advanced econometric models to the analysis of road and rail freight transport demand in Great Britain (GB). This study is being conducted as part of the ?Green Logistics? project, funded by EPSRC.
The advanced econometric models applied in the study are as follows:
PA: the Partial Adjustment model;
ReADLM: the reduced Autoregressive Distributed Lag model;
VAR: the unrestricted Vector Autoregressive model;
TVP: Time-Varying-Parameter (TVP) model;
STSM: Structural Time Series model.
Such models have shown their advantages in previous empirical studies in other economic fields. The traditional OLS static regression model is used as a comparator.
The empirical analysis is carried out based on the annual time series data on GB road plus rail freight tonne kilometres moved, covering the period 1974-2006. This is done at both aggregate and disaggregate (by commodity group) level. The empirical estimates of freight demand elasticities are included in the paper. As part of the investigation, individual models are estimated for each commodity group, (as well as for total road and rail freight), over the period 1974-1998, and forecasting performance is assessed using data for the period 1999-2006. The latter assessment is based on the Mean Absolute Percentage Error.. One- to four-year-ahead forecasts are generated and the forecasting performance of alternative models is evaluated both across commodity groups and over forecasting horizons.
The preliminary empirical results suggest that for forecasting, in most cases the STSM turns out to be the best among the competing models, followed by the reduced ADLM and the TVP model. The traditional regression model (Static Model) exhibits the poorest performance. For estimating elasticities, however, it is important to consider multicollinearity between GDP and trend, in which case it is necessary to consider other models such as the PA model which has advantages in this respect. This paper presents elasticities of freight transport demand by commodity with respect to GDP.
Keywords: freight demand, econometric model, elasticity estimation, forecasting comparison
Association for European Transport