As you may recall from my NY Times a piece a few years ago, one has to recognize that the regulatory problem is not just parametric optimization. In my example, the problem was that football helmets, which are actually quite good at protecting the head, result in MORE head injuries.
Interestingly, last year--2020--there were far fewer cars on the roads, because of social distancing, the shutdown of bars and restaurants, and the fact that many people did not commute to work at all.
The result: MORE TRAFFIC DEATHS. People drove much more aggressively on the nearly empty roads. To be fair, the total number of accidents did in fact fall by quite a bit, as you would expect. But the severity of the accidents that did happen? That was up sharply, especially outside of cities.
The point? Regulations have to take into account the likely response of citizens, based on expectations. You can't just twirl dials and pull levers when it comes to public policy.
(If you don't get the Peltzman reference....)
