This seems like such an obvious solution and, with the advent of things like MOOCs, something that needn't be particularly costly. (Setting aside the fact that even 'costly' but effective rehabilitation is demonstrably cheaper in the long-run.)

As well as the obvious benefits of equipping offenders with skills that will make them more employable upon release it seems intuitively true that the struggle for self-improvement would make them all-round 'better people'. For those who argue that self-study isn't enough, another component could be pairing up inmates of differing attainment levels and having effectively having them teach one another -- again, the act of teaching and helping a stranger has got to be a positive experience.

Is there a big problem with any of this? Have studies been conducted that indicate my intuitions are incorrect? Have I massively underestimated the difficulties and costs associated with such a scheme? ...Or is it simply a combination of lack of imagination on the side of the administration, coupled with a strong urge for punitive justice on the side of the electorate?

If I recall right, education for education sake did not lowered recidivism much. It made prisoners better behaved while in prison (e.g. there was less violence and less problems in prison).

I suspect that part of the problem with practical education for prisoners is that everybody else have to pay for it. So, law abiding (or not caught yet) citizens get all jealous about the perk they are not getting.