This paper deals with a model for road risk, developed for the French Road Safety Observatory, under two contracts concluded between the Ministry of Transportation, the University and INRETS. According to the Drag-type approach, time series modelling including exogenous variables has been applied on a monthly basis to the triangle 'risk exposure/risk/gravity', for the period 1975-1999. The model related to the monthly number of injury accidents and fatalities, aggregated on the whole territory, and disaggregated by road network (main roads, motorways, secondary roads, urban network), is described. The exogenous variables measure risk exposure, the climatic conditions and the calendar configuration. The elasticity values of the numbers of injury accidents and fatalities with respect to these determinants are discussed, and compared with the values obtained in the literature. Examples are given of the road safety indicators evolution over the period, and the performance of the model is discussed. Special attention is paid to the climate influence. On both main roads and motorways, the two networks on which the traffic volume is measured, the climate effect is separated into two parts: a direct effect, due to modification of driver behaviour, and an indirect effect through risk exposure. The results obtained, related to temperature, rainfall and frost, are discussed. For the covering abstract see ITRD E126595.
Samenvatting