Motor insurance models and road-safety studies address the same underlying risk, but at very different scales. Insurance models typically predict annual claim counts from a small set of rating variables, such as age, mileage, vehicle characteristics, or past claims. Road-safety research, by contrast, seeks to understand mechanisms operating much closer to the crash itself, including speed, fatigue, distraction, road environment, and driving conditions. I recently uploaded on ArXiv a paper (From Rating Factors to Crash Mechanisms: A Multiscale Causal DAG Framework Linking Motor Insurance and Road Safety) that proposes a multiscale causal framework, organized around a literature-informed DAG, to connect these two perspectives. It shows that a predictive rating factor generally cannot be interpreted directly as a causal crash mechanism. Examples based on mileage and driver age illustrate why: the same actuarial contrast may be compatible with many different mechanistic explanations. Sharper conclusions will require data closer to the driving process, such as telematics, trip context, or linked crash and insurance-claim records…

Arthur Charpentier

Arthur Charpentier, professor in Montréal, in Actuarial Science. Former professor-assistant at ENSAE Paristech, associate professor at Ecole Polytechnique and assistant professor… Read more

View all posts