Do Road Planners Produce More ‘Honest Numbers’ than Rail Planners? An Analysis of Accuracy in Road‐traffic Forecasts in Cities versus Peripheral Regions

Abstract

Based on a review of available data from a database on large‐scale transport infrastructure projects, this paper investigates the hypothesis that traffic forecasts for road links in Europe are geographically biased with underestimated traffic volumes in metropolitan areas and overestimated traffic volumes in remote regions. The present data do not support this hypothesis. Since previous studies have shown a strong tendency to overestimated forecasts of the number of passengers on new rail projects, it could be speculated that road planners are more skilful and/or honest than rail planners. However, during the period when the investigated projects were planned (up to the late 1980s), there were hardly any strong incentives for road planners to make biased forecasts in order to place their projects in a more flattering light. Future research might uncover whether the change from the ‘predict and provide’ paradigm to ‘predict and prevent’ occurring in some European countries in the 1990s has influenced the accuracy of road traffic forecasts in metropolitan areas.

Notes

1. Project promoters, forecasters and managers sometimes object to using traffic during the first year of operations as the basis for measurement. However, if this objection were to be followed, it would be virtually impossible to make meaningful comparisons of forecasted and actual traffic across large numbers of projects and across project types, geographical areas, and time periods because no common standard would be available. Moreover, for projects for which data are available on actual and forecasted traffic covering more than 1 year after operations begin, it turns out that projects with lower‐than‐forecasted traffic during the first year of operations also tend to have lower‐than‐forecasted traffic in later years. The methodology used in the studies based on Flyvbjerg’s database is already widely used in practice for measuring the inaccuracy of travel demand forecasts. (For a more detailed discussion on methods of measuring inaccuracy, see Flyvbjerg, 2005a.)

2. For 53 of the road links, data about the year of completion were available, whereas such information was missing for the remaining 31 projects. Information about ownership was available for 70 of the projects; among these, 63 roads are publicly owned, four are private and three are state‐owned enterprises. The remaining 14 roads, all located in the USA, are all toll roads, but information about their ownership is not available from the collected data.

3. However, the authors have not investigated whether or not this explanation fits with the actual situation in the corridors of the rail projects of the Flyvbjerg database.

4. Actually, two of the present authors have submitted a proposal for such a project to the Norwegian Ministry of Transport. A decision about whether the project will be funded is expected early 2006.

5. However, this belief is not very well founded (e.g. Engebretsen et al., 1998; Flyvbjerg et al., 2003, pp. 65–72).

6. For example, interviews made with members of the Norwegian Parliament’s Committee on Transport in the early 1990s showed that it was difficult to identify cases where the benefit–cost ratio had played any decisive role on the politician’s standpoint. Other circumstances, notably impacts for local business and trade, the opinions of the local electorate, and the type of project exerted stronger influence on the final decisions of the politicians (Nyborg and Spangen, 1993).

7. This control was made in the form of a full three‐way analysis of variance with backward elimination. In spite of non‐homogeneity of variance, the conclusion is so clear that the non‐homogeneity of variance cannot destroy it. The factors of the analysis are DK (Denmark or other European countries), toll (yes or no) and geography (metropolitan, non‐metropolitan eligible to Objective 1 or 2 funding, and other non‐metropolitan), but only nine of the 12 combinations are represented in the dataset. Due to this incomplete representation of the combinations, the three‐factor interaction cannot be estimated. For the two‐factor interactions, the following interactions are successively removed, DK versus toll with p = 0.568, DK versus geography with p = 0.210, and toll versus geography with p = 0.113. For the main effects, geography is removed with p = 0.612 and then toll is removed with p = 0.428. Only DK cannot be removed with p < 0.001.

8. For example, in Sweden the number of passenger‐km by cars was reduced from the year 1979 to 1980 and did not reach the 1979 level again until 1983. The number of tonne‐km of goods transported by trucks was even more strongly reduced and did not reach the 1979 level again until the mid‐1990s (Johansson and Nilsson, 2004). In Germany, the total volume of road traffic (measured in vehicle‐km) was reduced from year 1980 to 1981, but reached the 1980 level again already in 1982 (Tappe, 2001).

9. In the early 1980s, the Danish national‐level projections on which planning of concrete road projects was to be based operated with a high alternative assuming a 20% increase in car traffic between 1978 and 2000, and a low alternative assuming a 40% reduction over this period (Vejdirektoratet, 1982, p. 8). Although a considerable growth in truck transport was expected, the total traffic development was expected to be modest even in the high alternative. The low projections were partly motivated by economic pessimism, but also by changed political attitudes to road traffic, where a strengthening of public transport at the expense of private motoring was emphasized as a main transport–political goal (Vejdirektoratet, 1982, p. 4).

10. Instances of biased forecasts are not something for which only the planners should be blamed—it is rather a problem comprising a whole planning and decision‐making culture. The planners are not necessarily those who have constructed the transport models used to make the forecasts, and often the main source of bias is the politicians or builders. Nevertheless, the planners can hardly say that they are not aware of the criticism raised against faulty traffic models, nor can it be defended ethically to adhere to policy‐makers requesting skewed analyses.