This study is an attempt to provide better estimates of the size (and sign) of the aggregate accident externality from driving. To begin, we choose a dependent variable, insurer costs, that is dollar-denominated and captures both accident frequency and severity; we also analyze insurance premiums as a dependent variable. We are concerned with aggregate effects across the full spectrum of driving in a given state. Our central question is whether one person’s driving increases other people’s accident costs.
They find a large externality, and they advocate a large tax.
It's worth thinking about the costs of auto accidents. I think that the non-pecuniary costs of dealing with insurance adjusters, body shops, and so on, or fairly significant. Of course, I should internalize some of that and drive less frequently and more carefully because of it (we're talking "in theory," right?).
But when I think of externalities from auto accidents, I think of the huge traffic jam I was in this morning, due to an accident.
How much would you pay other people to stay off the road? Are tolls the answer?