Urban Road User Charging Online Knowledge Base
What Are The Policy Implications?
Prediction exists to inform policy and aid decision-making processes. Unlike some of the other themes of this report, it does not itself have policy implications for the success of road pricing schemes. However, there will be policy implications leading from previous prediction work related to the technical practice of approaches to prediction and to the findings about the performance of road pricing.
There is increasing complexity of prediction work and associated models over time, in response to increasing awareness of the complex relationships that govern the outcomes of transport policy innovations and to increasing demands for accurate predictions to feed into political decision-making.
At present, there is no set standard for approaches to predicting the impacts of road pricing schemes. The evidence from research studies and real applications suggests that considerable variations exist in the scope and approaches adopted. In the UK, government guidelines are moving towards standardising the range of responses that predictive modelling exercises need to include, but this does not yet mean that results from different situations can be considered directly comparable. The degree of variation is much greater where international comparisons may be desired. This presents a challenge for all those involved in the prediction process and may have particularly significant impacts for transferability.
Nevertheless, results of previous prediction work have consistently suggested benefits from road pricing schemes, regardless of the approaches adopted. However, alternative modelling approaches have produced rather variable predictions about the performance of particular scheme designs. Cordon pricing, which has traditionally performed well in models with coarse levels of spatial detail and limited transport networks, has tended to show much lower benefits in tactical network models due to the provision of greater opportunities for avoidance behaviour. Therefore, experience suggests that policy-makers should exercise caution when interpreting outputs from any individual prediction exercise and should seek to draw knowledge from a broader base.
Questions that frequently arise when prediction is discussed include whether it may have become over-reliant on complex models, whether the nature of the outputs (including levels of coverage and accuracy) are fully understood by all those who use them and whether it is worth the cost in terms of time and money. There are obvious attractions in using simple models that are easy to understand and relatively cheap to implement. There is also likely to be an important relationship between the revealed accuracy of model predictions and the levels of trust that people are prepared to place in future modelling activities. Where well founded doubts exist over the ability of models (or other prediction approaches) to produce reliable outputs for particular indicators, it is potentially unwise to attempt to estimate them in that way as it may only reduce respect for the prediction process in the longer term. In Stockholm, model-based predictions for the road pricing trial were generally not trusted, but turned out to be more accurate than expected (Eliasson, 2009). On the other hand, each proposal to implement a road pricing scheme represents an opportunity to acquire data to improve understanding of the processes at work leading to more accurate predictions. However, as already argued, external political factors may ultimately dictate the nature of prediction activities undertaken and expectations about the uses of outputs.