The risk modelling research presented in this paper uses time series constructed on monthly data and over a long period, is designed to further the analysis of the development of road risk, by bringing together all the explanatory factors of risk and risk exposure that appear at an infra annual rate. The choice of monthly periodicity, which is the product of a compromise between the variability and availability of existing data and that of the determinants - weather and time of year conditions, traffic flow or its substitutes (economic activity, prices, and network development), behavioural variables and road safety measures - position this approach midway between the modelling of daily fluctuations and the modelling of long-term trends. We will present a version of the model limited to two types of network, main roads and toll motorways and then also extend to the whole network. We will first describe the structure of the model as a whole, and then go on to outline the main results: the outcome of tests of hypothesis related to econometric specification, the values for the indicators' elasticity to their determinants, showing their development over the period, and the simulations carried out in the short term. Lastly, we will discuss these findings and suggest avenues of enquiry for continuing this research. (Author/publisher)
Abstract