Student loan forgiveness could become the next Obamacare website
A well-intended policy might be undone by implementation
If the Affordable Care Act is the greatest policy legacy of President Obama, the rollout of the Healthcare.gov was his greatest administrative failure. The site was so dysfunctional it became a late-night comedy punchline. It also led to enormous emergency investments to fix the problem, which had some long-term benefits like the creation of the US Digital Service.
Imagine if Obama knew all of this ahead of time, and could have made those investments before the problem occurred?
Joe Biden is currently in a similar situation. We know enough now to be worried about how a potential student loan forgiveness program could easily become an administrative disaster.
If I’m engaging in a little bit of hyperbole by invoking the Obamacare website (although I’m not the first to do so), its because implementation of student loan forgiveness has gotten almost no attention. Discussions have centered on the politics of the forgiveness, the amount forgiven, and who will benefit from such forgiveness. Biden is trying to find a balance between the left wanting big loan forgiveness and political pushback from the right, and so seems to be leaning toward a modestly-sized and means-tested approach. But the resulting inattention to the implementation process could significantly hamper the reach of a policy the Biden Administration is betting a lot of political capital on.
Two approaches to equity: means-testing vs. universal access
The logic for means-testing is that it is unfair to give loan forgiveness to high-income earners. Indeed, the argument against student loan forgiveness from the right is horizontal equity: why give college graduates a benefit that non-college grads in otherwise similar (or worse) situations won’t receive, or past or future grads cannot access?
Here is the problem with that argument. Means-testing begets administrative burdens. Which means that some people, maybe many people, especially the people with fewest resources and most in need, will not receive benefits. This is the administrative burden Catch-22: those most in need struggle most with the procedural barriers they must satisfy to receive help. From this perspective, universal access can also be justified on equity grounds, in that it guarantees access to a program that the most needy will disproportionately benefit from.
Michael Stratford mapped out what a means-tested option would look like. It’s not great. Means-testing requires the Department of Education to verify individual identity and eligibility — (lets say anyone with an outstanding loan and income less than $125K) and reduce their outstanding loans by the specified amount (lets say $10K). By one estimate, this would eliminate debt for about one-third of borrowers, cutting half of debt for another twenty percent. But such estimates only hold if everyone who is eligible gets the loan forgiveness.
All this assumes that the Department of Education has up-to-date borrower income information.
It doesn’t.
Ok, so Education can get that information from IRS, right?
No it can’t, thereby killing the chance to automatically forgive the loans.
The law limits the IRS from sharing data, even with other parts of government, without explicit permission of the individual or exceptions specified in statute. The prohibition on data sharing reflects strong privacy concerns in the US that hamper the ability of government to reduce burdens on the public. This is a long-term problem.
Data-sharing between government actors can reduce administrative burdens. But it’s far harder for the federal government to share data than you might expect.
So means-tested loan-forgiveness eliminates the prospect of automated loan-forgiveness. We are in the world of application processes, and administrative burdens. Thus a policy choice made for equity purposes might result in an administrative process that makes it harder for the most needy.
What the process would look like?
With a means-tested program, eligible beneficiaries have to know about the benefit, get on the Department of Education website, set up a Federal Student Aid account and password, and complete some sort of application form. This might not sound like much, but basic administrative tasks like this can be enough to dissuade people from claiming benefits.
If the administrative process itself is confusing and frustrating, it could eliminate much of the political goodwill Biden is trying to create. For instance, the Public Service Loan Forgiveness program has been so onerous that many eligible borrowers have given up. To its credit, the Biden Administration has made real strides in fixing this and other loan forgiveness programs.
The absence of automated forgiveness also makes the implementation process more complex from the perspective of government. An application process has to be established, and choices about how it works need to be made. Additional staff have to be found to answer questions from the public. According to Stratford:
Scott Buchanan, executive director of the Student Loan Servicing Alliance, which represents the companies hired by the Education Department to manage student loans, said it would be “very difficult if not impossible” for the department and loan servicers to implement a sweeping loan forgiveness program before the midterm election, regardless of how it’s structured.
Once on the website, applicants would have to verify their income status. There are two basic options
The applicant uses the IRS Data Retrieval Tool allowing them to pull their tax data.
The Department of Education accepts a “self-attestation”: the applicant states they have earned less than the income cut-off.
The Data Retrival Tool option sounds feasible, but has some drawbacks when you look into it. It assume everyone has IRS tax information. Some people don’t. About 9-10% of people don’t submit tax returns, and are not required to if their income falls below a certain income level, meaning this tends to be a very low-income group who are disconnected from the tax system. This has been a problem with the expanded Child Tax Credit, which was automated and seamless for all bar those without tax information, who also tended to be those who would most benefit from the program.
Its also not clear if the current version of the Data Retrival Tool, designed for FAFSA borrowers, would work, or whether the Department of Education and IRS would need to create a new agreement, which could take months. It’s not just that the IRS, like the Department of Education, is understaffed for this task. These sort of programs are not in their mission statement, and they tend to be reluctant to get involved in non-core tasks.
The fear with self-attestation is fraud: people will claim benefits they are not eligible for. To mitigate fraud, the applicant could be asked to make the promise under penalty of perjury, and subject to potential future audit. The Department of Education could also ask applicants to provide voluntary documentation, such as screenshot of their tax returns.
How big is the threat of fraud here? Not that great. First, this is not like unemployment insurance fraud where identity theft allows the stealing and redirecting of money to fraudsters. In this case, all that happens is that the balance on the Department of Education debt total declines by a fixed amount. No actual money is being sent out. So it’s not a target for crooks. Second, the federal government can verify via audits if misclaiming took place, and reclaim money.
What can Biden do? Best option: universalism
From a policy design perspective, I sympathize with the Biden administration’s desire not to forgive loans for high-income borrowers. Both fairness and costs are concerns. Federal Reserve economists estimated that a $75,000 income cap would reduce the cost of a $10,000 loan forgiveness plan from $321 billion to $182 billion. But this is still a lot less than other alternatives. A $50,000 loan forgiveness plan would cost $904 billion, which would decline to $507 billion with the same income caps. With a more generous $125,000 income cap, savings from means-testing will be much less.
From a policy implementation perspective, means-testing is a problem and Biden should pursue universal loan forgiveness. The equity benefits of reaching the most in need trumps the equity considerations of giving money to those who don’t need it. Simpler is better.
One basic concern I have is that the White House is not thinking about these tradeoffs, wrongly assuming that means-testing will be free of frictions. I don’t know what the empirical scale of the trade-offs looks like, i.e., how many and which group of borrowers will lose out because of means-testing, but Biden should before he makes any final decision.
One way of estimating the scale and the distribution of the effects of means-testing would be to compare the effects of Obama and Biden policies to forgive student loan debt for individuals with disabilities. The Obama plan sent letters to eligible beneficiaries telling them they could receive loan forgiveness, based on Social Security data that confirmed their disability status. Letters were sent to 387,000 eligible claimants to encourage them to apply. But the application process was still onerous, with a three-year reporting requirement for income documentation. According to the Department of Education, only about half of borrowers identified as eligible for loan discharge received the benefit. And about 41,000 borrowers who received loan forgiveness later had their loans reinstated because they failed to provide earnings information. A GAO report found that 98% of such reinstatements went to borrowers whose income was low enough to receive the benefit; it was the documentation requirements that got them. By contrast, the Biden Administration provided automatic loan forgiveness 323,000 borrowers with disabilities and removed income verification requirements. The gap between the reach and effectiveness of these two approaches seems like a good back-of-the-envelope proxy for the reach of universal vs. means-tested student loan forgiveness.
Second-best option: design to reduce burdens
Assuming universalism is not on the cards, the self-attestation solution is better, at least in the immediate term, because it can be implemented more quickly, reduces burdens, and for reasons explained above, faces a low threat of fraud.
Biden could help in other specific ways. Going back to the Obamacare website example, it makes sense to bring together the best execution team before things go wrong rather than afterwards. This might include tech specialists like the US Digital Service who can design application process, and ramping up hiring for people who can answer questions from applicants. The White House can also engage in publicity to reduce learning costs, and facilitate outreach via third parties who are trusted by hard-to-reach beneficiaries. Biden can push the IRS to be more involved than they would otherwise might want to be.
All of this requires more administrative work and leadership attention, reiterating the point that means-testing adds more complexity than going with the universal approach. It could be that the White House is on top of all of this, and I am worrying for nothing, but none of the coverage I have seen gives reason for confidence.
Updating our theories of how politics works
At the broadest level the Obamacare and student loan examples underlines that policymakers need to update their basic theories of how political acts generate political returns. In the most simple terms, the standard working theory seems to be:
Public benefits = recipient gratitude and goodwill
If policymakers were to think more carefully about implementation, they would revise their working theory to be a bit more complicated:
Public benefits that are easy to get = recipient gratitude and good will
Public benefits that are hard to get = much less gratitude and goodwill
Promised public benefits that are impossible to get = backlash
Biden is already going to face some backlash from those opposed to the idea of any type of student loan forgiveness. He should do all he can to avoid it from the people the program is intended to help.
Extra stuff:
Check out Annie Lowery, Dylan Matthews and Dara Lind talk about “the time tax” on The Weeds podcast
If you enjoyed this, chances are you will enjoy reading the archive with posts about the Biden administration’s efforts to reduce administrative burdens, as well as new research on the topic. Please subscribe if you have not already, and consider sharing.
Is there a way to get to means testing through the backdoor via personal income tax? Beyond a certain AGI the forgiveness is considered income?