Should Students Borrow?

[Secretary of Education Betsy] DeVos issued the final repeal of an Obama-era rule aimed at holding low-quality education programs accountable by forcing them to prove their graduates were able to repay their student debt.” CNBC, July 1, 2019.

As summer ends, 17 million students head to college. Nearly 40 percent are borrowing to do it. Over the past 15 years, the number of people with student loans nearly doubled and now stands at roughly 45 million―that’s about one in four adult Americans. Furthermore, the amount of debt has virtually tripled to just over $1.5 trillion, so the average loan is something like $35,000. Politicians, reporters, and researchers characterize this as a crisis. Is it?

Our answer is yes. While the majority of those with student loans are able to handle them, a large fraction cannot. Despite the strong labor market, nearly 7 million people, or more than 15 percent, are over 90 days late on payments or in outright default (see here). Since virtually all of this debt comes through federal programs, the government has created the problem. What should policymakers do about it?

We should emphasize at the outset that education is like any other investment. In the same way that a business will borrow to invest in buildings, equipment or training, an individual will borrow to invest in obtaining skills. Both the firm and the prospective student have their eye on a future payoff―one is profit from a larger, more efficient, business, while the other is higher income from working in a profession that requires expertise. As economists would say, investment in physical capital is much like investment in human capital. Given similar attitudes toward risk, the decision to borrow depends on whether the expected return from the resulting investment exceeds the interest rate on the loan. We will come back to this shortly.

Turning to some data, the following chart of overall household debt since 2003 reveals several important patterns. First, mortgages (shaded blue) are the dominant form of borrowing, accounting for roughly two-thirds of household debt throughout this period. Second, student loans have increased dramatically, from 3 to 11 percent of the total (red line). As Akers and Chingos show, this increase in student debt reflects increases in college enrollments, as well as policy decisions by state governments to raise tuition and by the federal government to expand student loan programs.

Household borrowing (Trillions of dollars or percent of total loans, quarterly), 2003-2019

But, debt is only a concern if the borrower struggles to pay it back. Is there a problem? To answer the question, we can look at the distribution of the debt. As it turns out, 80 percent of borrowers have debt that is less than $50,000. To see what that means, first note that the interest rate on student loans is around 4½ percent. If this is a 20-year, fixed-rate, loan, monthly payments are $316. That’s roughly 6% of monthly income of the median U.S. household―similar to the share that the average person spends dining out. For the majority of borrowers, this does not seem overly burdensome. So, for the typical borrower, it doesn’t look like a big problem. (As Akers and Chingos point out, students in graduate and professional schools are typically the ones incurring debt in excess of $50,000.)

However, not all student borrowers appear to be making sound financial choices. Two important facts point to poor decisions. The first concerns for-profit schools. As Scott-Clayton documents, nearly half of student loan borrowers from for-profit colleges end up in default. This is in stark contrast to what happens with four-year public and private not-for-profit schools, where the equivalent numbers are 11.4% and 12.5%.

Even some traditional not-for-profit schools appear to be rather poor investments. The following figure shows the distribution of the 20-year net return on investment for completing a four-year college education, accounting for financial aid. Three things strike us about this picture. First, in-state college (the red line) is a good deal. Assuming long-run inflation of around 2% (the Fed’s objective), the real student loan rate is in the range of 2% to 3%. Setting the threshold for an acceptable return at the higher level of 3% (the dashed vertical line in the figure), all but 10 of the 495 public schools are worth the investment for the average student (leaving aside a risk premium). Second, out-of-state public schools (the blue line) look very much like the private schools (the black). We would guess that this is no accident, as private schools and public schools seeking out-of-state students are in direct competition. Third, we note that there is a lower tail here where borrowing is not expected to pay off. For private schools, 88 of 872 schools generate returns below the 3% floor. And, for out-of-state public schools, the number is 41 (again, out of 495).

Distribution of 20-Year Net Return on Investment to College, 2018

Source:  Payscale  and authors’ calculations. The website data, which account for financial aid, are based on the educational cost and on the wage distribution across age cohorts all at time  t , so the reported return can be thought of as real, rather than nominal.

Source: Payscale and authors’ calculations. The website data, which account for financial aid, are based on the educational cost and on the wage distribution across age cohorts all at time t, so the reported return can be thought of as real, rather than nominal.

These tabulations ignore the substantial differences in the 6-year graduation rates in these schools. If we assume that the expected return on investment equals the graduation rate times the return in the above figure, while the annual return for those who do not graduate is -1% (on average over 20 years), then the picture looks markedly worse: only half of private schools meet the 3% real return threshold.

Beyond the issue of graduation rates, there also is the wide variation in employment outcomes (and hence returns to investment) that depend on the field of study. As Hershbein and Harris point out in 2014, the average college graduate can expect to earn a lifetime income of roughly $1.2 million, double that of a high-school grad. However, expected lifetime earnings across college majors range from less than $800,000 to over $2 million! And, even then, as we will discuss shortly, there is quite a bit of risk associated with the choice of any career.

So, again, the average borrower appears to be making a sound financial decision. Is everyone? We can get some sense of the full distribution of outcomes by looking at information on loan repayment status in the following chart. The red bars show the number of borrowers who are delinquent or in outright default (with actual numbers in white). Note the slow rise from 2003 to 2011, and then the sudden jump in 2013. Importantly, roughly half of borrowers have rising balances through the period (this is the gray portion of the bar). For 2017 (the latest data available), only 37% of borrowers―that’s 16.6 out of 44.7 million people―have falling balances and are actually repaying their loans. The implication is that, while the average student borrower appears able to handle the consequences of debt incurred to finance their education, a large swath of the population does not.

Repayment status of student loan borrowers (millions), 2003-2017

What should policymakers do about this? Some people suggest switching to a system of universal free public college and cancelling student loan debt (see here). In our view, this would be an extremely inefficient way of providing support to those in need. First, as the rate-of-return data suggest, in-state public colleges are already a very good investment. Second, when you offer subsidies to everyone, regardless of their ability to pay, the wealthy receive the same benefits as those who need support to obtain a college education. And third, since graduate and professional school students are the ones with the largest loans—including many with high-income prospects—they will be the biggest beneficiaries of any such program. Beyond these economic arguments, we suspect that broad-scale debt forgiveness will sour attitudes, poisoning support for lending to those who really benefit.

Another popular suggestion is to provide additional information to help people make informed decisions. This means offering reports that bring together things like expected returns by school and major, along with graduation rates, to help people anticipate which investments are likely to pay off. This would surely help, and there are people out there trying to do it (see here). But, given the challenges associated with financial literacy―half of the population doesn’t understand the relationship between interest rates and inflation―better information (like cigarette package warnings) will only get us so far. Instead, we need to find ways to help people do the right thing.

Indeed, since most students borrow through federal programs, the government has a responsibility to help people make smarter decisions. Given the repayment problems, one clear possibility is to shift some portion of the government expenditure into outright grants. That is, rather than offer lower interest rates, provide cash. However, increasing funding directly to students may drive up college tuition even further.

Beyond increasing grants to the least wealthy, we see two things policymakers can do: stop making bad loans and help those who do borrow to manage the risks. On the first, we urge the government to stop subsidizing activities that make people worse off. Offering student loans to someone who has a high probability of default—due to their selection of a particular school or course of study—is not a way to improve lives. From this perspective, the recent rollback of federal rules holding colleges accountable for their students’ payback rates is a very disturbing move in the wrong direction (see the opening quote).

Turning to the second point, we need an appropriate safety net for students who do borrow. Career choice is risky and people can be unlucky. As we discuss in an earlier post, the standard deviation of real household income is high and rising. People who change jobs often suffer wage declines (see here). Furthermore, technology will surely have a bigger impact on some jobs than others. Judging which occupations will disappear and when is extremely difficult to predict for anyone, much less an 18-year-old entering college.

Since student loans mostly require fixed payments, the borrowers currently bear the associated income risk. An alternative, which we support, is a shift to contracts that are income-linked. Instead of a fixed nominal amount, the standard federal student loan contract would be a payment that is a fixed percentage of a person’s discretionary income. Such contracts exist among private parties, but they are not the default option. Moreover, because these contracts are complex and the providers are unregulated, there is a risk of exploiting borrowers.

To conclude, since the turn of the century, student borrowing has exploded. Student loans are generally a force for good. They help many people acquire skills when they otherwise could not afford it. In the large majority of cases, college remains a good investment, and most people are able to handle the debt they incur in obtaining an education. But, many cannot. Government programs, no matter how well intentioned, should not create hardship for millions of people and their families.