Those of us who are econ geeks will know about the Pelzman Effect. Regulations that supposedly make us safer (say, seatbelts or cycling helmets) don't actually make us safer as behaviour changes to take account of the new safety. Almost as if there's what we consider to be an acceptable risk to take and reducing it in one manner just allows us to be silly in another so as to maintain that risk we're comfortable with. What I didn't know (but better econ geeks than I might have done already) is that there is a Reverse Pelzman Effect.
Exploiting an interesting natural experiment, the authors of that paper are able to show that we should abolish driving licences. The various States of Mexico found that bribery was impossible to avoid when attempting to gain a licence. So, to varying degrees, they changed their issuance system, some deciding simply not to have them any more. So, of course, death rates from car accidents went up, didn't they?
Erm, actually, no, they didn't. Those places that didn't bother with licences any more, allowing absolutely anyone at all to get in and drive, saw no change in such death rates any different from those that had now (well, hopefully) incorruptible issuance systems.
Yes, the death rate on Mexican roads is indeed appalling: but it's no worse in those areas that demand licences than it is in those that don't nor vice versa. Thus drivers' licences are not needed and we can abolish them.
I would have known this if I'd ever bothered to listen to Pater. I only remember it now because he's reminded me. The basic lesson of driving is always to drive as if every other driver is drunk, mad or incompetent. Or all three. And as it turns out, when we really do recall this because no one has to take a test then we do adjust our driving habits to take account of their madness or incompetence.