Road Safety Across the Globe - An Econometric Analysis of Key Risk Factors

Abstract

This research work explicitly aims at empirical assessment of the enforcement of key risk factors (speeding, drink driving, non-compliance to seat-belt and helmet use, and child restraints) that contribute to the occurrence of road traffic crashes. The research is attempted using country level road safety data from World Health Organization (WHO) and International Road Federation (IRF). A statistical analysis of global and regional road safety legislation has been performed. Based on paramount transportation science indicative of several aspects in which road agencies perceive the enactment and enforcement of key risk factors, two separate econometric models are also developed. Effectiveness of enforcement level of speed limit law is modeled using random parameter ordered probit modeling framework. Past research affirms considerable correlation between driver speeding and drink driving behavior. Hence, bivariate ordered probit model is also developed in an attempt to simultaneously investigate factors influencing effectiveness of enforcement levels of speed limit and drink diving law respectively. The results of statistical analysis of the level of enforcement of safety legislation revealed helmet usage to be the most enforced legislation (32.25% of countries having high level of enforcement) and child restraint to be the least enforced legislation (only 11.17% of countries having high level of enforcement respectively). Several instances were observed (in many countries) where there existed considerable back-log in road agency practices in enforcing key risk factors. Fatalities per thousand registered vehicles (FPTRV), gross national income (GNI) per capita, hospital beds per hundred thousand population, presence of child restraint law, and policy for promoting walking and cycling were all observed to have consistent effect(s) on effectiveness of enforcement level of speed limit law. Moreover, effectiveness of enforcement level of helmet and seatbelt law, road safety audit of new roads, and presence of lead road safety agency were observed to have xi heterogeneous effects (random parameters) on response variable. The results of bivariate ordered probit model revealed that the effects of fatalities per thousand registered vehicles, hospital beds per hundred thousand population, and policy for promoting walking and cycling are simultaneously consistent both for effectiveness of enforcement level of speed limit and drink driving law. Also, the bivariate model encapsulates the effect of several other agency related variables on two response outcomes. The considerable discrepancies in road agency’s enforcement programs elucidate the need of analyzing national disaggregate datasets regarding road safety enforcement programs. In an attempt to exploit this finding, analyzing time series decomposition of individual key road safety variables may be helpful primarily through demonstration of autoregressive integrated moving average models.