Research Idea Scope
Using data from one state, conduct a statistical analysis to show the degree to which vehicle miles traveled and 1) accident rates and 2) cost of accidents correlate.
Auto insurance rates in the U.S. are not tied to miles driven, except in very limited ways (such as low-mileage discounts) and with the exception of experiments in a couple of states (Progressive in Minnesota, GMAC using ONSTAR). Some analysis has argued the fairly obvious point that charging by the mile could significantly reduce miles driven (perhaps by as much as 10%, especially in areas where insurance rates are high). A first step to showing companies and regulators the sense of such a rate innovation would be to show that such rates are actuarially valid. A statistical study could demonstrate the degree to which accident rates (particularly high-cost accidents) and miles driven are correlated. The correlation would not say that other rating factors should be disregarded — rather the correlation should adjust for other factors such as location, age, and driving record. Then rates could be set in which the per-mile rate would reflect these other factors. Miles driven can be checked either at an annual safety inspection or via an on-board electronic device similar to On-Star.