Time-Series Models of Aggregate Road Risk and Their Applications to European Countries

Abstract

This paper aims to describe how time-series analysis of road risk has been performed at the national level in Europe since Smeed's seminal study of 1949. The first part of the paper surveys European applications of time-series analysis to road safety since the beginning of the 1980s. A historical overview of the various approaches followed and the different types of model that have been used to analyse changes in road risk are given, referring to the foregoing historical account. The last part of the paper presents recent modelling conducted in the framework of the EU FP6 project “SafetyNet—Building the European Road Safety Observatory”, which ran from 2004 to 2008, with the aim of gathering harmonized data bases from the member states and performing a comparative monitoring of trends. Recommendations for using dedicated models which handle time dependency when applied to road safety were given. Applications to a number of national datasets, including France, the Netherlands and Greece, have revealed different ways in which risk exposure can be included in the models in order to conduct a comparative analysis of trends. Research directions for extending these comparative analyses are given.

View correction statement:

Erratum

Acknowledgements

The author thanks her partners in the SafetyNet Work Package 7 on ‘Data Analysis and Synthesis’, and especially the coordinators Emmanuelle Dupont and Heike Martensen for their comments and suggestions, as well as the anonymous referees for their constructive comments on the initial version of the paper.

This paper was translated by Kevin Riley.

Notes

The concept of exposure to risk (or risk exposure), developed in the field of epidemiology, has easily been extended to the field of road safety. Risk exposure is measured by the number of units in a population exposed to a given risk during a given period of time. In the field we are considering, the units exposed to accident risk include motor vehicles (or their occupants), two-wheelers (or their users) and pedestrians, and the time the units are exposed can appropriately be replaced by the distance travelled, as the different types of road users are in movement. In this paper, we shall refer to risk exposure as a general concept; however ,the variables chosen for measuring it in the applications will differ depending on the availability of the data (either the population, the vehicle fleet, the traffic volume or fuel consumption when data on the traffic is not available, etc.). Among them, the most widely used measure at national level is the mileage—or number of vehicle-kilometres travelled. An overview of the existing risk exposure data for European countries can be found in Yannis et al. (2005).