Forecasting road fatalities by the use of kinked experience curve

According to the World Health Organization, more than one million road traffic deaths occur every year throughout the world. In order to cope with this challenge, many countries have established quantified road safety targets backed up with comprehensive safety strategies. Road safety targets need to be based on reliable forecasting methods. Following the pioneering work by Elvik (2010), this paper attempts to develop such forecasting models for 13 OECD countries based on the data available from 1970 to 2007. Deploying the methodology of both classical and kinked experience curves, we obtained the averaged experience slope of 55% from the kinked experience curve in contrast to 68.6% from the classical experience curve. The averaged standard deviation and R calculated also show better fit to the data from the use of the kinked over the classical analysis. For the two simulated forecasting periods, we, then, calculate mean absolute percentage error (MAPE) to measure forecasting accuracy. In comparing the MAPEs calculated from the kinked versus the classical models, we find that forecasting accuracy for the kinked models is again significantly higher. Finally, we use our kinked models to forecast the road fatalities for 13 countries through 2030. JEL classification: R41; R48