Or, to elaborate on the bonus idea, you could offer payments that go up as the proportion of released prisoners who don’t reoffend go up. So, $x for the first 10%, $x+1 for the next 10%, $x+2 for the next 10%, etc. (This idea has been proposed for private welfare-to-work organisations in the UK.)

At the moment, a certain amount of government money is spent on reform programmes, with reasonable success for both puppy-smackers and puppy-kickers**. Say the government wants to keep spending about the same amount of money**, but would instead structure it as a flat bonus for each released prisoner who doesn’t reoffend. Depending on the size of the bonus, there are three** possibilities:

1. The bonus is so low that the prisons don’t try to reform anyone.
2. The bonus is high enough for everyone to be reformed — but then the government’s overpaid to reform the puppy-smackers.
3. The bonus is large enough to try to reform puppy-smackers but not puppy-kickers. Inmates are released and puppies are kicked. The government may** have given insufficient incentive to reform puppy-kickers, given the cost of them remaining unreformed — and may also have overpaid to reform puppy-smackers.

One solution: pay a different bonus for puppy-smackers and puppy-kickers. But, in the real world, there are hundreds of ways of abusing puppies, and setting a value for each type of reform would be a bureaucratic mess.

Another would be to fund the current amount of reform as a minimum and then offer a bonus on top of that. Maybe this would work, maybe this wouldn’t, but would someone please think of the puppies?

**Gross oversimplification and/or wacky assumption. Little of this is necessarily true, of course, but I could make up plausible numbers for which it would be true if I actually knew anything about the subject.

So do I 😛

“If you had a non-recidivist bonus, you’d need safeguards against prisons abandoning all attempts to reform serious crims relatively likely to re-offend”

Why?

Why would a for-profit company be more risk-averse than the median voter? I have a hard time imagining it. Don’t we usually model firms as being risk-neutral? Doesn’t the median voter support policies that forgo massive amounts of wealth in hopes of protecting against risk?

As long as the prison is not more risk-averse than society. Although I guess you could just charge the size of the “fee” to take this into account 😛

It’s still a prison and offenders will be sent to it.