Recent press reports have celebrated the development of automated or "self-driving" cars and the benefits they would bring, yet actual benefits may fall far short of this potential. Drivers are exposed to only a fraction of the costs of accidents that they cause, they may underestimate the risks of accidents, and automated features may work in only a fraction of all driving situations. Studies have found that some drivers respond to safety devices by driving more â€˜intensely,â€™ and some drivers refuse to accept even proven occupant protection devices such as seat belts. More drivers will resist the purchase and utilization of automated vehicle features that change their driving behaviour, especially if this reduces the speed or enjoyment of their driving. This paper analyzes the likely response of drivers to automated safety features. It suggests that benefits will be much less than forecast and will mostly occur a decade or two in the future because of driver behaviour and our likely reluctance to force drivers to submit to automated control in most driving situations. The shortfall in benefits will depend in part on the details of motor vehicle insurance policies and on public policies adopted by state, provincial and federal governments. There is an economic argument for policy encouragement of cost-effective accident-avoidance features but not for features that save time for the driver.
University of Toronto, Department of Economics, Working Papers #564