Induced traffic prediction inaccuracies as a source of traffic forecasting failure

• Published in: Transportation letters 2011-10, Vol.3 (4), p.253-264
• Main Authors: Antoniou, Constantinos, Psarianos, Basil, Brilon, Werner
• Format: Article
• Language: English
• Subjects: Demand; Density; induced traffic; Infrastructure; land use - transport interaction; Marketing; Mathematical models; traffic demand forecasting; Traffic engineering; Traffic flow; Transportation; Transportation engineering; transportation forecasting models
• ISSN: 1942-7867; 1942-7875
• DOI: 10.3328/TL.2011.03.04.253-264

In this paper, evidence from the literature on the inaccuracy of forecasts from transport planning models is presented and its impact on capital infrastructure planning decision-making is demonstrated. Empirical evidence suggests that forecasts used for major planning decisions have been found to be rather inaccurate (when comparing forecast volumes with actually realized volumes after time passed). A meta-analysis of the relevant literature provides insight into the expected magnitude of these deviations as well as the main contributing factors, including induced demand.Furthermore, evidence of these irregularities in the case of two major expressway infrastructure projects in Southern Greece is presented, providing insight about a typical example of a country with a developing freeway network. Data from the immediate impact zone of the Attica Tollway in the metropolitan area of Athens and the neighboring interurban Elefsina-Korinth motorway are used to demonstrate that the projects resulted in considerable changes in the land use patterns and density, both in terms of residential and commercial development, resulting in the generation of additional traffic flows.Having demonstrated the need for re-thinking how long-term traffic is forecast, suitable recommendations for promising directions are made. Dealing with uncertainty is one key aspect that can be easily incorporated to existing forecasting tools. Furthermore, the need for more detailed and area-specific models, e.g. through the integration of activity-based modeling, specific mobility patterns and demands are also outlined.