I think you’ve misunderstood me; my apologies for not being clearer and more explicit. I’ll try to fix that below.
The question I was trying to address was: How much of the explanation can defect risk be? And my answer was: Not much more than the expected cost of the defects, which in turn is probably rather less than Pr(serious defects) * value of car without defects, which for not-very-old cars is empirically quite a small fraction of the cost of the new car.
(The “used price = new price / 2” case was just an example.)
Since the difference between new and used prices is a large fraction of the cost of the new car, therefore, it seems unlikely that defect risk is most of the explanation—as I said,
suggesting that hugely reduced price relative to new cars can’t be mostly about the risk of such defects.
The context was James Miller’s suggestion (if I understood him correctly) that defect risk might in fact be a large fraction of the explanation for the big difference in price between new cars and used-but-not-very-old cars.
I don’t think that’s wrong, but I have another suggestion: Car prices may be subject to a variation of Goodhart’s Law. Defects may not be that likely in used cars, but attempting to act as though they are not likely would create incentives that would make them become likely.
This might require precommitment or superrationality on the part of the consumers, but a lot of “irrational” consumer behavior can be modelled as rational precommitment, even if the consumer doesn’t consciously realize that’s what it is.
That’s a very interesting idea. I’m pretty sure it’s too sophisticated to be consciously part of the reasoning of more than a tiny fraction of car buyers, so if it’s an important part of the explanation it must be (as you suggest) unconscious—presumably as a result of some general-purpose unconscious tendency to over-penalize risks of that general sort. This suggests some interesting psychology experiments; I wonder whether they’ve been done.
I think you’ve misunderstood me; my apologies for not being clearer and more explicit. I’ll try to fix that below.
The question I was trying to address was: How much of the explanation can defect risk be? And my answer was: Not much more than the expected cost of the defects, which in turn is probably rather less than Pr(serious defects) * value of car without defects, which for not-very-old cars is empirically quite a small fraction of the cost of the new car.
(The “used price = new price / 2” case was just an example.)
Since the difference between new and used prices is a large fraction of the cost of the new car, therefore, it seems unlikely that defect risk is most of the explanation—as I said,
The context was James Miller’s suggestion (if I understood him correctly) that defect risk might in fact be a large fraction of the explanation for the big difference in price between new cars and used-but-not-very-old cars.
I don’t think that’s wrong, but I have another suggestion: Car prices may be subject to a variation of Goodhart’s Law. Defects may not be that likely in used cars, but attempting to act as though they are not likely would create incentives that would make them become likely.
This might require precommitment or superrationality on the part of the consumers, but a lot of “irrational” consumer behavior can be modelled as rational precommitment, even if the consumer doesn’t consciously realize that’s what it is.
That’s a very interesting idea. I’m pretty sure it’s too sophisticated to be consciously part of the reasoning of more than a tiny fraction of car buyers, so if it’s an important part of the explanation it must be (as you suggest) unconscious—presumably as a result of some general-purpose unconscious tendency to over-penalize risks of that general sort. This suggests some interesting psychology experiments; I wonder whether they’ve been done.