From the comments, from Susan Dynarski on debt forgiveness

Zimmerman’s article does not make the case you claim. This RD captures the effect of going to a 4-year instead of a community college. The counterfactual is college attendance. This article is one of our key pieces of evidence that the quality of colleges matter. Put an student at a 4-year instead of a community college and you get the estimates discussed in the abstract.

Spending per pupil at four-years is about 75% higher than community colleges (historically it’s been even higher). https://hechingerreport.org/proof-points-states-and-localities-pump-more-money-into-community-colleges-than-four-year-campuses/

Here is the link, to my post on the educational returns to the marginal student. Please go back and read that for the context on Dynarski’s statement.

And from an email from David J. Deming, Harvard researcher in the area:

Sue is right that community college attendance is more common for students below the threshold in the Zimmerman paper. But many of them also attend no college at all. Table 4 shows that making the GPA cutoff increases years of 4 year college attendance by 0.46 and decreases years of community college attendance by 0.17. This implies that there is an increase in total college attendance – so the counterfactual is a mix of 2-year college and no college.

On the substance of the comment “do these people need debt forgiveness”? I’d say that ex ante they do not, but maybe some should receive it ex post. The Zimmerman estimate captures ex ante returns. Debt forgiveness is ex post. FIU’s grad rate was around 50% at that time, so the average return of 22% includes graduates and dropouts together. Ex post returns could be 44% for graduates and zero for dropouts.

Your idea of limiting debt forgiveness to dropouts was great. I wish that had been on the table. We’d worry about moral hazard if it became a forward-looking policy, but the Biden policy was probably not foreseeable in advance.

I would have also liked to see debt forgiveness focused on institutions rather than students. Forgive debt obtained at low-quality for-profit colleges. I would actually guess that college quality is a better predictor of lifetime wealth than current income. A person making $50k as a working adult 2 years after dropping out of University of Phoenix surely has lower expected lifetime wealth than a person who graduated from Harvard a few years ago and is making $50k in a public sector job.

My view is that decent returns to the marginal student still create problems for the Dynarski debt forgiveness argument. Overall the private returns to education are good. You can pack some of the problems into specific subgroups, but to the extent you do that the case for more debt relief targeting — much more targeting — rises rather steeply and rapidly.