A regression model with ARIMA errors to investigate the frequency and severity of road traffic accidents

Abstract

In this paper, models are developed to explain and forecast the frequency and severity of accidents in Belgium. The objective of this study is to enhance the understanding of the developments in road safety by studying the impact of various explanatory variables on traffic safety. It is investigated whether the number of accidents and victims is influenced by weather conditions, economic conditions and policy regulations. The model is used to predict the frequency and severity of accidents for a 12-months out-of-sample data set. Monthly Belgian data from January 1974 to December 1999 are used in the model, and predictions are made for the year 2000. Using a regression model with ARIMA errors, the impact of variables on aggregate traffic safety is quantified and at the same time the influence of unknown factors is captured by the error term. The results show a significant effect of weather conditions and laws and regulations on traffic safety, but there seems to be negligible statistical impact of economic conditions. The model can easily be used to forecast traffic safety, as can be seen from the reasonably good fit obtained on a 95% confidence level.