Mostly just wanted to point out that there are two numbers that drive how many new cars are sold:
1) Increase in total number of vehicles in use (relatively level 2008-2013, a quick google didn't find newer data)
2) Old vehicles taken off the road being replaced by new ones
The net effect of a particular person buying a slightly used car over a new one is basically nothing. The big picture only changes when people are choosing to keep an old car on the road for longer.
One of the side effects of emissions controls introduced in the late 90s is that cars are way more reliable. Basically these days most cars can easily achieve 200k miles or more without heroic labor.
I can't find a free source of the data, but I know in my state 10+ year old auto registrations spiked from 2008 onwards.
I think what GP is getting at is if you start optimizing first for reuse, there's a lot of cars we wouldn't need to build.