Application of Quasi-induced Exposure to Estimate Relative Risk of Causing RTAs: a Case from Czech Republic

The EU Road Safety Policy Framework 2021-2030 sets the goal of 50 % reduction in the road traffic accident (RTA) related injuries and fatalities by 2030. This project aims to improve the understanding of the underlying risk factors of accidents, which is important for designing policies and measures aimed at reduction of RTAs, fatalities, and injuries. Specifically, our paper focuses on the role of driver heterogeneity with respect to the risk of causing an RTA. Using a detailed administrative database of over 1m RTAs from the Czech Republic (2011-2021), we (1) estimate the relative risks (RR) across the known risk factors, in particular age and alcohol consumption, and (2) in cooperation with Police Czech Republic we derive specific policy recommendations reflecting the results.

The key challenge in estimating risk factors from accident data is the unobservability of risk exposure. We address this by leveraging the quasi-induced exposure (QIE) approach, which uses the not-at-fault drivers as a representative sample of the driving population. We pay special attention to the assumptions behind QIE and their empirical verification.

Our preliminary results show that the risk that drivers under the influence of alcohol produce an 11.7 (95 % CI 6.0-25.2) times higher risk of causing a fatal RCT than sober drivers. The estimated RR for drivers with BAC > 0.1 % is 35.2 (CI 9.3-297.0). Compared to drivers aged 25 to 64, young (age 18 to 24) and elderly (age 65+) drivers are 2.5 (CI 2.0-3.1) and 2.7 (CI 2.1-3.4) times more likely to cause a fatal RCT, respectively. Our tests of the underlying assumptions validate the QIE methodology in our data for fatal accidents. However, the key RR estimates for less severe accidents are quantitatively and qualitatively similar to those for fatal accidents.
In addition to RR estimates, the QIE methodology also facilitates estimating the characteristics of the driving population and its variation across time and space. This information will be used by the police and other relevant agencies to improve the allocation of resources, preventive, and enforcement measures as well as information campaigns.

This research is a collaboration between the Anglo-American University, Faculty of Law of Charles University Prague, and the Czech Traffic Police Headquarters. It is funded by the Technology Agency of the Czech Republic (grant name: “A New Generation of Traffic Accidents Statistics for the Police of the Czech Republic”).

Peter Bolcha, Senior Lecturer, Anglo-American University, Prague

Peter Bolcha is a Senior Lecturer at Anglo-American University, Prague. He earned both his M.Sc. and Ph.D. in Economics at the University of Economics, Prague. He had published number of academic papers in the field of economics and applied econometrics.

He has worked on multiple projects focused on policy evaluation with Czech Ministry of Trade, Czech Traffic Police and energy company RWE.

Recently, he has been a team member of a 3-year project "Developing advanced tools for analysis of traffic accidents for Czech Police" and now leads the project “New Generation of Traffic Accidents Statistics for Police CR”, both funded by the Technological Agency Czech Republic.