9: The Road, Risk, Uncertainty and Speed

This paper, adapted from Gaudry and Vernier (2000) and Vernier (1999), demonstrates how one can identify the impact of fine road infrastructure features, both geometric and surface, on road speed and safety, taking due account of the simultaneous nature of the speed-safety relationship. This is done within a DRAG-type multi-level structure where accident frequency or severity and speed equations are estimated with Box-Cox transforms.

Our explanation of accident outcomes distinguishes, beyond pure randomness, between two (subjective) components of observed (objective) risk, namely calculated risk linked to speed, and uncertainty or “dangerousness”, in the manner of Frank Knight (1921). Both components are part and parcel of a new empirical measure of “perceived risk”, expected maximum insecurity EMI, that is demonstrably separable from other determinants of speed choice. In our structure, the first two groups of equations explain accident frequency and severity with discrete choice Logit models admitting of non linearity and choice-based sampling—permitting future completion of the database at low cost. The third and last group of equations also consists of non-linear flexible-form models and explains the mean and the variance of speeds.