A Novel Idea to Keep Students in College: Failure Insurance

January 10, 2010

Imagine a first-generation college student whose high-school preparation was less than ideal. She has just finished her first semester, and she realizes now that college is going to be tougher than she had hoped. She failed one course and struggled to earn C's in her other subjects. She worries that she'll eventually flunk out, and she wonders whether she should walk away now before she accumulates any more student debt.

But what if she could hedge her risks by buying a "failure insurance" policy that would reimburse her for a portion of her student-loan debts if she did flunk out? Would that make her more willing to stay for another semester?

Failure insurance might sound outlandish—but a well-designed insurance system could actually improve students' effort and their attainment rates, according to a working paper that was presented in Atlanta last week at the annual meeting of the American Economic Association.

The paper's authors—Satyajit Chatterjee, an economist at the Federal Reserve Bank of Philadelphia, and A. Felicia Ionescu, an assistant professor of economics at Colgate University—do not have a blueprint in their pockets for such an insurance program. What they put forward in their paper is something more abstract: a set of formal models that prove, in their view, that failure insurance might work.

The authors begin by presenting mathematical models of students' decisions to enroll in college and to put effort into their schoolwork. In the authors' view, those decisions are shaped by students' finances, their beliefs about their future earnings, and by the amount of misery—"disutility," in econospeak—that they suffer when they do academic work.

The paper, "Insuring College Failure Risk," concludes that it is theoretically possible to create failure-insurance policies that would hit a sweet spot. That is, the insurance policies' potential payoffs would not be so high that they would give students an incentive to shirk on their schoolwork, but would be high enough to make students more comfortable with staying in college and taking on more debt. Staying in college would allow some of them to develop better study skills and to put themselves on a path to graduation.

The authors speculate that the insurance might be offered as a component of the federal student-loan program, but they do not describe a concrete policy apparatus. The purpose of the paper is simply to argue that insurance can work in theory.

At this point it's impossible to say whether this working paper will be anything other than a blip in the scholarly literature. Many scholars, of course, emphasize the idea of easing students' debt anxieties in a more obvious way, by lowering tuition and increasing grant aid. And others have recently been exploring financial incentives for students to take full course loads and to earn high grades.

But regardless of whether it is ever put into practice, Mr. Chatterjee and Ms. Ionescu's provocation has already drawn attention from The Wall Street Journal, The New Republic, and the Student Lending Analytics Blog.

Mr. Chatterjee and Ms. Ionescu answered questions from The Chronicle by e-mail last week. (They composed their answers together.)

Q. What does your mathematical model suggest about why too few people attempt college and why too few people complete it?

A. In our numerical model, some students are discouraged by the financial risk associated with a student loan—namely, the student realizes that she may take out a loan but fail to earn a degree. This risk discourages some students from attempting college. Among students who do attempt college, some drop out after learning the effort required to earn a degree. The decision to drop out is also affected by the perception of financial risk. The greater the risk, the more likely it is that a student will drop out.

Q. As a practical matter, what would an optimal insurance system look like? How would the system minimize the incentives for shirking?

A. Practically speaking, we would expect our insurance scheme to work as follows: A student will be eligible for loan forgiveness if he or she has "failing" grades. Currently, a student must maintain a certain GPA in order to be eligible for a student loan. The same standards can be applied to determine whether a student has "failed" and is therefore eligible to collect insurance. In the event of failure (so defined), the student is forgiven a portion of his or her outstanding student loan. Since only a portion of the loan is forgiven, the student still bears some financial risk. This residual risk is like a deductible in a standard insurance contract and will lower the incentive to shirk.

Q. Are you hopeful that real-world institutions could actually produce the optimal insurance system that you modeled in this paper?

A. We don't have a good sense of this. However, here is a thought. In principle, our insurance scheme could be implemented by a school rather than the federal student-loan program itself. Since schools are in a better position to monitor effort, and therefore minimize the shirking problem, it is conceivable that our scheme might appear attractive to schools that are suffering from too many dropouts or failures.

Q. Why could an insurance system be superior to the kind of income-contingent repayment schemes that Milton Friedman favored?

A. It is superior in the sense that our scheme is not likely to be affected by moral hazard. The reason is that the gain from obtaining a college degree is pretty large in expectation, and most students will not want to stop working hard in college simply to get a portion of their student loan reimbursed when they fail. (This may not be true if the likelihood of failure is very high; for this reason our proposal may not be applicable to students with very low preparation for college—in our quantitative work, we exclude students with SAT's below 700.)

In contrast, we believe Friedman's proposal ties repayment to earnings. His scheme would lower repayment when income is low, which gives an incentive to cut back on effort (just to lower payment on the student loan).

Q. In your paper, you model students' study effort as a binary variable—that is, you assume that students are either working full-tilt or they're completely shirking their studies. But does that really make sense? Many students are highly motivated to finish college but are also juggling jobs and family responsibilities. Those students often take conscious or half-conscious risks with their schoolwork—trying to calculate the minimum amount of effort that will earn them a B or C. Could an insurance system move these students along a continuum of effort?

A. We do believe that it is sensible to model effort as a binary variable. We think that if a student finds it optimal to exert any effort in college at all, he or she will find it optimal to exert as much effort in college as possible. The reason, again, is the large college premium in earnings. However, it would be important to explicitly model continuous effort and make sure our intuition on this is correct. We plan to investigate this point in future revisions of the paper.

Your point relates also to something we ignore—which is how a student allocates effort across courses. Potentially, offering insurance against failure could alter the amount of effort allocated to a particular class and may even affect how many and what kind of courses a student takes. These issues would also be important to look into in future work.

Q. In your paper, you assume "that once students fail, they never attempt college again. … Once a student avails herself of insurance, she cannot re-enroll in college without repaying the indemnity with interest." Could you elaborate on that? Wouldn't that rule create a serious barrier for those students who move from institution to institution over a long period of years before they graduate?

A. Upon reflection, we think that an alternative arrangement would be better. Instead of requiring a student to pay back the indemnity with interest if she re-enrolls in college again, the insurance program could stipulate that once a student avails herself of insurance, she cannot get insurance again. She is, of course, free to go to another college but she will not be eligible for insurance against failure.