Abstract
Keywords: Road safety, Crash frequency, exposure, econometric model, time series.
The road safety is a complex system involving a number of interrelated factors, such as the characteristics of road infrastructure, road user behaviour, technological evolution of vehicles and the legal framework. The analysis of the influence of these variables at a macroscopic level requires the use of powerful econometrical models. During the last two decades, important efforts have been carried out in this field of research. One of the most important initiatives has been the DRAG family of models (Demand for Road use, Accidents and their Gravity), successfully applied in several countries and regions: Quebec, Germany, France, Stockholm, Belgium, California, and Norway (2). The main feature of DRAG models is a three-layer recursive structure for the explanation of road safety, including road use, accident frequency and accident severity as separate equations, and taking into account the main factors of influence. DRAG also allows for modelling flexible functional form, by using Box-Cox transformations, and avoids misspecification due to autocorrelation and heteroscedasticity.
INSIA has carried out a research project with the aim of applying the DRAG methodology to the analysis of the evolution of road safety in Spain. The I-DE (Interurban DRAG España) model uses the three-layer structure typical of DRAG models, with one equation in exposure layer, two equations in the accident frequency layer (fatal and injury accidents), and two equations in accident severity layer (morbility and mortality). A broad set of explanatory variables has been used in each equation. These variables may be divided into the following groups: (a) exposure, (b) fuel prices, (c) vehicle fleet, (d) laws and regulations, (e) enforcement, (f) road infrastructure, (g) weather conditions, (h) drivers, (i) alcohol and speed, (j) reasons for travel, (k) special events, (l) data aggregation. Special attention has been paid to the influence of technological evolution of vehicles and of primary and secondary safety improvements. The following variables have been tested: mean age of vehicle fleet, percentage of vehicles older than ten years, percentage of vehicles equipped with ABS, percentage of vehicles equipped with ESP. Monthly time series have been collected for the period 1990-2004.