What benefits might justify collectivizing student debt and how might they be measured? These are my last two topics, and perhaps the most important.
I submit that only benefits external to students could justify allocating to others the cost of a student's education. Otherwise the move is simply a transfer payment with transaction costs. I have heard some argue that subsidies would allow students to serve people of low or modest incomes, who cannot afford high rates. One could conceive of that as an externality taking the form of decreasing the number of transactions that do not occur by increasing the supply of lawyers willing to work for little. In standard terms, it reduces the deadweight loss relative to the current system.
For the moment, suppose that is true. The benefit so described takes the form of lessening the harm caused barriers to entry, in the form of UPL rules, that prevent legal assistants or others who do not have to incur law school debt at all from satisfying at least a portion of this demand. In other words, this argument claims the benefit of lessening the harm caused by a system lawyers impose and schools support in the first place. I personally find that a less than compelling claim to benefit.
But cogent criticisms of licensing requirements are very, very old, and there is no chance that licensing itself will go away, though I would do away with it altogther if I could. So I'll leave that topic and fall back to asking why subsidies might be expected to solve the very severe problem facing persons with little or no money but significant legal problems.
If subsidies are tied to income, by far the simplest way to do it, they offer no guarantee of such an increase in supply. Lawyers who are simply no good at picking cases or running them, and who make little because they are no good, are subsidized just as much as those working to ameliorate the deadweight loss caused by licensing. The increase in supply might in fact take the form of an increase in marginal or even hopeless cases, thus causing a net drain on social welfare. Maybe all subsidy recipients would work to reduce the effects of licensing, but I doubt it very much.
Since the Cahns wrote in 1966 there has been a substantial amount of commentary noting the need for legal help for poor people and those who are not poor but earn little and have few assets. I do not think the commentary has ameliorated the problem or led to significant solutions. In part that is because of the sometimes great distance between scholarship and practical reforms, in part that is because there is little that can be done working within the licensing system, a point the Cahns saw so clearly.
But open entry is not the only baseline to use when judging benefits (a good thing given that open entry is not going to happen). There has, however, been significant change in the supply of a limited number of legal services. It has not come from the bar but from Menlo Park, Mountain View and environs (Intuit is there.) Interactive software is a big step up from form books and you can't beat the economics: fix the data in a tangible medium one time and then have trivial costs per-user, as opposed to fix the data many times (in the heads of many lawyers). Software can't litigate (though it can radically lower the cost of doing so), so this is no panacea, but it is a significant data point that real change in providing legal services has come from outside the bar (with, in some cases, the steadfast opposition of the bar).
So perhaps the baseline should be that the benefits must exceed that which could be obtained through subsidizing technology aimed at providing service even a person making $20,000 a year could afford. It's still a second-best baseline relative to open entry, and it does not solve the problem of litigating wrongful eviction or termination, but it at least benchmarks the investment in relation to a mode of service that has done demonstrable good for the consumer segment about which many people are, quite rightly, concerned.
Measuring the benefit would still be problematic, I concede, but we do need to start somewhere and the proper baseline will help.