Kick the Reform Habit

Understand one fact about the president's speech in Buffalo August 22 and the White House's plans to reform college financial aid: President Obama's proposed tie of financial aid prospects to ratings relies on an overcomplicated, implausible set of mechanisms to accomplish a simple task. The admirable goal: target financial aid on the colleges and universities that make it easier for students of moderate means to attend and complete undergraduate programs.

We should not be surprised that the president's goal is mixed up with the fantasy of an algorithm that can judge the merit of colleges and universities. Ratings and rankings are the crack cocaine of today's generation of education reformers and have been since the Reagan administration concocted a "wall chart" attempting to compare state educational performance. Our current president, his White House advisers, Secretary of Education Arne Duncan, former Florida Governor Jeb Bush, and many others cannot get the idea out of their heads -- that if we just find the right (magical) formula, we can push the education system to perform better.

Sometimes algorithms are important and helpful, and I understand why dreams of a perfect educational judgment system are so appealing. Brookings Institution researchers Matthew Chingos and Beth Akers have the best short description of the potential benefits of a ratings system for higher education, written from the perspective of two algorithm advocates. Alas, they assume both the existence of meaningful, comprehensive, nonmanipulable data, and the ability of an algorithm to spot high and low performers with accuracy. We know from both college ranking systems and the history of elementary and secondary school accountability that such attempts generally fail. Cedar Reiner and Timothy Burke have explained some of the concerns I have about applying blind faith in bad formulas for higher education... and I am afraid that the president's promise, "We'll figure out the right formula in the next two years!" is as comforting as other similar claims have been. Such a system is inevitably a Rube Goldberg machine.

But we do not need to feed politicians' dependency on school and college ratings. Mr. President, tear down that wall chart! You can accomplish the same goal with much simpler, more robust tools -- and even better, you would not need Congress to amend the Higher Education Act to improve the federal government's financial aid system. Here is my Five-Step Program to break politicians' addiction to ratings systems, at least in higher education:

1. Distribute a large chunk of college aid directly to colleges and universities based on Pell graduates, if not exactly as proposed. The basic idea is good: institutions that effectively serve poor students should have some support to continue and expand their work. But we do not know whether it would be best to have a flat payment per graduate or weight it by financial aid received by the student. There are some potential advantages of weighting the reward by the size of Pell Grant received while enrolled, and possibly other financial aid: addressing transfer issues in a rational (if not perfect) way, giving some advantage to institutions with the poorest students, and giving institutions a significant incentive to help students keep Pell grants and other aid year-to-year. Instead of making one decision at the federal level, Congress could distribute funds to states, tell them they must distribute funds based on associate or bachelors degrees earned by those who received Pell grants, and let states partially or fully weight those rewards based on federal and/or state and institutional financial aid received by the graduates. Letting states determine if the rewards are weighted by financial aid may appeal to governors and state legislators, as well as allowing states to include state and institutional financial aid in the weights.

2. Cap student loans not just by students but also by institutions, with the cap tied to the number of recent graduates who carried federal student loans at the institution. For example, if the cap for four-year colleges was hypothetically set at $10,000 times the total loan-carrying graduates in the past three years, then an institution with a consistent 40 percent graduation rate would have a much lower loan cap for all its students than an institution with a consistent 60 percent graduation rate. (For example, a college that admits 1,000 students who take outloans every year would graduate 400 of them annually, 1,200 every three years, with a total loan cap of $12 million. If it graduated 600 students carrying loans instead every year, it would have a total loan cap of $18 million.) Institutions with lower graduation rates would either have to have lower net costs, raise their graduation rates, or stop recruiting students who need large student loans. 

Capping loans at an institution by the absolute number of loan-carrying students would not need a difficult-to-calculate graduation rate but would realistically address the capacity of an institution to educate students. The formula could be generous at the start and focus on limiting loans for the worst actors in higher education, those whose business plans rely on both the federal financial-aid system and also the gullibility of prospective students.

3. Cap not only personal student loans but also loan-forgiveness and so-called PLUS loans available to parents and graduate students, to circumvent individual loan caps. A comprehensive all-system cap that sets annual, per-degree, and lifetime loan and forgiveness caps would end the structure that has allowed Georgetown Law School to game graduate student PLUS loans so that neither Georgetown nor its students pay (in net) anything to the federal government. A comprehensive family-based cap on college-related loans would also address concerns about the exploitation of students and their families by tuition-dependent colleges and universities of all sectors. This subject is sensitive because of the claims by many tuition-dependent colleges and universities that they serve the public interest even while graduating a small minority of their students. The latest round of debates on PLUS loans and private historically black colleges and universities has gone to HBCUs, with what appears to be largely a reversal of the Obama administration's efforts to tighten credit-worthiness criteria for parents. The solution to the dilemma of the nonselective tuition-dependent college should not be the exploitation of families but the direct support of valuable institutions, which is why two more steps are necessary.

4. Use the Experimental Sites waiver provision of Title IV (the portion of the Higher Education Act with the rules for most student aid programs) to let public and nonprofit colleges work together on the student-services and business sides of their work, creating consortiums that draw on common services for some critical supports. Using Experimental Sites to improve student services and tighten the business operations of colleges is far more likely to help students than using Experimental Sites for MOOCs. In May, Southern New Hampshire University President Paul LeBlanc complained that federal rules prohibited nonprofits from working together on bundled services, even as many used commercial (and more expensive) corporations such as Academic Partnerships. Now, LeBlanc thinks that Experimental Sites might allow that bundling of services in the nonprofit college world. While LeBlanc would like SNHU to provide bundled services for online, competency-based education, we can extend the concept to consortiums of nonprofits with common interests, such as relatively nonselective HBCUs, and encourage such consortiums to address issues that most directly affect student persistence and completion and that can reduce costs if shared between institutions.

5. Directly support consortiums of tuition-dependent colleges and universities, contingent on sharing of relevant data on student progress and completion. If the federal government provides technical assistance grants supporting both instruction and student persistence/completion efforts to consortiums of nonselective public and private colleges, it could make data-sharing a condition of such grants. This could be accomplished through the Fund for the Improvement of Post-Secondary Education (FIPSE).

Together, these five steps address the general goals the president is targeting but likely without needing the Higher Education Act to be amended. None of them require the assumption of finely tuned ratings, and none should require a huge amount of statistical work by institutions beyond what Title IV colleges and universities must track today. The greatest difficulty is likely in using the appropriations process for the first and last steps -- and my guess is that it is easier to persuade Congress to send money to the states to reward colleges that graduate poor and moderate-income students than to ask members of Congress to give up their FIPSE earmarks.

But as long as no statutory change is required, there is some hope of addressing the fundamental dilemma with allowing loans to prop up tuition-dependent colleges and universities. The controversy over HBCUs and student/family loans is one form of a broader question untouched by the ideas above: How can we address the needs of tuition-dependent private colleges that admit students with weaker records -- when it is difficult to separate exploitative institutions from institutions that have an historical record of serving the public good? In the last century, black college graduates have represented a disproportionate number of African-American professionals after college. That history does not justify indefinite indirect subsidies by the federal government through student and family loans, especially given the inability to discharge educational debt in bankruptcy. But it does justify an understanding that some private institutions have low graduation rates while meeting a legitimate public purpose. In that regard, HBCUs constitute the canaries in the coal mine for tuition-dependent institutions more generally.

Advocates of a federal ranking/rating system argue that they can adjust measures of student success by the difficulties of serving the institution's population. I am skeptical of that claim, in part based on the experience with K-12 accountability algorithms that repeatedly fail to demonstrate sensitivity or specificity in identifying weakly performing schools, and in part based on the substantial disconnect between the available database (IPEDs) and even the best social scientific attempts to quantify entering-student needs. My experience and common sense about statistics lead me to believe that a federal ranking/rating system would not be worth all the candlepower that the White House is going to put into it. Or fairy dust sprinkled into the computers.

Instead of trusting a magical-algorithm approach, we should provide some direct support to such institutions in a way that allows them to be a little more efficient business-wise, boosts capacity in supporting students, and provides a little more accountability. The last two suggestions above focus on those goals. For tuition-dependent private colleges and universities that are allergic to sharing records, they could join together in consortiums without getting federal assistance. Those that most desperately need direct support and technical assistance would access to resources, with the understanding it comes bundled with accountability and sharing of data.

I am highly skeptical that the president's desire for a rating system will do much good. But it is not enough for skeptics of college ratings/rankings to point out all their flaws. We need concrete policy alternatives, ways to get to the same end without them. The above is my not-so-modest attempt to tackle that goal. It eliminates the Perfect Algorithmic Mechanism (that is DOA anyway) in favor of simpler, more robust mechanisms. And for tuition-dependent nonprofits that have a claim of serving the public good by serving especially needy students, it has the option of providing modest direct support in return for sharing of data.

But my attempt may not be the best option. What is your proposal for targeting aid to the most needy students in a way that realistically could happen in the next few years?