Estimation of the crash rates of vehicles considering vehicle age, crash period and vehicle cohort

What will the future crash rate be of a vehicle built today? Or one built 5 years ago, or in 3 years time? Such questions are relevant to the costs and benefits of new safety technology, and to understanding future crash risks more generally. They are also a key to distilling the effect of vehicle safety from other factors in road safety. Crash rates have a few important dimensions; when viewed along these dimensions, the underlying patterns in the rates provide a means of estimating present and future crash rates of vehicles. These dimensions are:
·         the age of the vehicle, because it relates to factors including vehicle kilometres travelled, typical driver age etc.,
·         the period (calendar year) being considered, because it relates to factors including prevailing levels of compliance, changes to infrastructure, the characteristics of other vehicles in the fleet, and
·         the cohort (year of manufacture) to which the vehicle belongs, as it relates to levels of crashworthiness and primary safety.
Analogous age-period-cohort data are also used in public health research to study trends in the rates of disease. In this paper, age-period-cohort vehicle data are assembled from mass crash data from NSW. The data are used to impute crash rates by cohort and by period. Rates are imputed for passenger vehicles built between 1999 and 2009 and by projection to 2020. The results suggest that large reductions in crash rates may be expected for future cohorts of vehicles.