Casualty Rate Prediction for Oil Tankers

A model of oil tanker casualties is presented which permits an expected casualty rate for each tanker to be calculated based on its age and casualty history. These expected rates are shown to be good predictors of both the actual casualty experience and the probability of total loss. The model is based on the findings that 1) the casualty behavior of an individual tanker follows a Poisson distribution and 2) the Poisson parameters for all tankers of a specific age follow an exponential distribution. As a result, Bayes' theorem can be used to calculate each tanker's expected casualty rate given its age and casualty history. Keywords: Individual Risk Rating Plans