Primary vs secondary tasks
Random thought exercise: Are you safer being driven by someone who is doing it as their job (e.g. taxi driver) or by yourself or a friend who drives on an as-needed basis?
The obvious first question: What is the accident rate for taxis, limos, truck drivers and regular cars? There's a suprisingly relevant result titled Taxicab and Livery Crashes in New York City 2004 which while a little dated does have some good numbers.
Crash rates are one-third lower for taxicabs and liveries than for other types of vehicles. The crash rate was 4.6 taxicabs involved in reported accidents per million miles traveled and 3.7 liveries involved in reported accidents per million miles traveled, compared with 6.7 vehicles involved in reported accidents per million miles for all vehicles in New York City, in 2004.
It would be easy to conclude that people who drive all day are simply better at it (see the 10,000 Hour Rule) but that wouldn't be the whole story. The paper goes on to identify a couple of other incentives which would seem to play a much more significant role.
A different explanation appears to lie behind the drop in livery crashes since 1999. Since 2001, insurance companies that underwrite livery vehicles have more carefully tailored auto insurance premiums to driving records. While the livery industry has for many years had an incentive to drive more safely and reduce insurance costs, the benefits would accrue only to the industry as a whole. Changes in insurance underwriting brought this relationship to the vehicle level, so that drivers with accidentprone records now bear the brunt of higher insurance premiums. This has apparently led to more careful driving among livery drivers
A great example of the real world impact of a policy refinement. In this case, the tighter alignment of goals (specific drivers rather that fleets/industry being held accountable) looks like a win all round. But is it?