Thursday 10 September 2009

Drunk driving limits and risk aversion

John Key's National Government, elected partially due to public annoyance at Labour's "nanny state" initiatives, is talking about lowering the drink driving limit from 0.08 to 0.05.

David Farrar rightly asks, given that the vast bulk of drunk drivers killed in car crashes had well over 0.08, would the reduction in deaths among folks in the 0.05-0.08 range be worth the cost in terms of reduced enjoyment of nights out.

I've heard rebuttals of this question pointing out that most drivers who have wine with dinner are in fact under the 0.05 limit anyway and so would not be affected. This seems a nonsense though. It's difficult to know with certainty what your blood alcohol level is after any given quantity of beer or wine, so sensible folks who weigh heavily the costs of being caught over the limit will target a level sufficiently below 0.08 that they will not err and be over the limit. The wider the confidence interval around your point estimate, the lower will be your targeted level to avoid the risk of erring on the upside.

So, if the limit is dropped from 0.08 to 0.05 and most folks currently target somewhere around 0.04, their drinking, and their enjoyment of a night out, will drop considerably even though the dropped limit remains theoretically unbinding. And the folks who are happy to be caught for the 10th time driving on a suspended license at four times the legal limit will continue to ignore the limit.

Isn't it better to have a sensible limit with heavy punishment rather than a low limit that cannot have a punishment sufficient to deter the really dangerous folks? With a limit at 0.08, really dangerous anti-social deviants like this guy can have the book thrown at them; with a limit at 0.05, there would be no support for heavy punishment because the net is cast too wide.

No comments:

Post a Comment