Constantine Yannelis & the State of Student Loans

May 03, 2018
Student loans

Student Loans

In 2010, student loan debt had surpassed credit card debt – taking second place behind mortgage debt, totaling at $825 billion. Fast forward eight years and this number have almost doubled to $1.5 trillion. To make things worse, five million Americans defaulting on student loans. The average student graduating in the class of 2017 owes $39,400 – what is more concerning is the number of people taking on substantial student loans ($50,000 or more). In 1992, only 2% of people who have student loans owed $50,000 or more. In 2014, this number reached 17%. This has many adverse implications for the individuals that take on these large student loans as well as the economy. I had the opportunity to speak with Professor Constantine Yannelis, a researcher from NYU’s Stern School of Business who uncovered these findings.

Can you give a little background on yourself?

I’m an Assistant Professor of Finance at the NYU Stern School of Business. I did my Ph.D. in Economics at Stanford, and most of my work revolves around household finance- in particular, I work on student loans and human capital investments.


What do you teach at NYU?

I teach undergrads primarily with a few MBAs. I teach corporate finance. So I do a little bit of research in corporate finance, but primarily I work on household finance, specifically on student loans.


A lot of your work seems to revolve around personal finance and student loans. How did you become interested in these topics?

I began researching student loans early on in graduate school. My interest in student loans started because this was an area in desperate need of data-driven, economic analysis. There are a tremendous number of open questions regarding the student loan market. Key questions about policy design depend on empirical parameters- for example, the sensitivity of earnings to student loans balances or alternative student loan repayment plans. However, there just aren’t a lot of estimates of these important, policy-relevant parameters. I saw a need for analysis here, and I’ve been working on this for several years. I think there are enough open questions in this area that I’ll be busy for many years finding the answers.


Based on the research that is going on, it looks like there are a lot of issues that have been brought to light.

Certainly. In the past couple years, the increase in information has been tremendous. Even so, it’s changing very rapidly, and so things that were true five or ten years ago may not be true today. For example, now about a quarter of borrowers, who own half the balances, are in some form of income-driven repayment plans. A lot of new information has raised more questions, chiefly about optimal policy design.


You published an article linking a decrease in home prices to student loan defaults, particularly related to the Great Recession. Can you talk about that?

That paper was part of a subset of my broader research agenda. The aim of that paper and another article was to determine what accounts for the rise in student loan defaults. Over the past decade and a half, we saw student loan defaults approximately double. It turns out, two factors explain about three-quarters of that increase in student loan defaults. One is a change in the composition of borrowers, so there are more borrowers at institutions with higher default rates. The second factor that explains much of this increase in default has been the impact of the Great Recession. In that paper, we used home prices effectively as a proxy for a sharp labor market decline during the Great Recession. We show that approximately a quarter of the increase in student loan defaults between 2006 and 2009 is associated with the decline in home prices.


There’s a study that found that 83% of millennials born between 1980 and 1998 attribute an inability to buy a home to their student loans. Do you think this could be a vicious cycle that’s happening?

It’s certainly possible. When you think about the effects of student loans on home ownership, there are two effects of student loans, and they go in the opposite direction. There’s one effect that I think most of the media has been focusing on, and that’s the direct impact of higher debt levels. If I have $50,000 of debt relative to no debt or minimal levels of debt, that’s likely to impact my credit score and make it difficult to qualify for a mortgage because lenders look at the debt-income ratio. On the other hand, student loans are not only increasing debt balances. People take loans for a reason, and often that reason is to provide liquidity. So there’s also a direct liquidity effect that taking a student loan can allow individuals to have more cash on hand and make down payments. I have a recent paper with co-authors, and we find actually that this liquidity effect dominates. We weren’t expecting this result at all, but we find that increases in student loan limits lead to a rise in home ownership rather than a decrease which is often the story we see in the media.


You’re saying that when fewer student loans are available, more home loans are available?

I’m saying the opposite. We were very surprised to have this finding. My prior was that either there would be no effect or that we would find that higher loan limits decrease home ownership. I think that’s most people’s prior, and that’s what you see typically in the media. We find the opposite: that increasing student loan limit leads to higher rates of home ownership, and this effect is immediate. We find evidence that this appears to be driven by the most liquidity-restrained borrowers. For example, the effect is larger for borrowers coming from lower income families, and the effect is also larger in the aftermath of the financial crisis when mortgage credit was harder to come by. What we’re seeing is that liquidity may be constrained for young people early on in the life cycle, but as student loans increase, students can take out student loans and use the cash to meet down payment requirements.


That’s interesting.

That doesn’t mean that these other effects aren’t happening. A couple of other studies from the Federal Reserve have found that higher levels of debt caused by tuition hikes reduce home ownership. So their experiment is very different from ours because as debt increases from tuition hikes, the borrowers don’t have any additional liquidity or cash on hand, so you see this increase in debt, which of course has an adverse effect on home ownership. Our study is looking at the impact of the limit increase. So there’s the effect of increasing debt, but there’s also the effect of having increased cash on hand, and we find that that effect dominates.


The Wall Street Journal recently published an article about your findings. Saying that the number of students taking on more than $50,000 in debt has increased from 2% to 17% from 1992 to 2014.

That sounds about right. Effectively, the number of students and the share of students taking on very large levels of debt has increased quite sharply. Also, there’s a tendency to think that these borrowers don’t matter very much because the general earnings are high for these borrowers and default rates are very low. Keep in mind that the typical borrower who takes on very high levels of debt, they often completed school, they went to elite, expensive institutions and many times they went to graduate school. So the typical borrower with very large balances is doing quite well, but because these balances are so large, this small share of borrowers who aren’t doing well and defaulting account for a fairly large portion of dollars in default. The fraction of borrowers with very large balances who come from higher-risk, high-default institutions has also been increasing quite rapidly. To some extent, there’s been this decoupling of default rates and repayment rates. Many of these borrowers with large balances are finding ways to avoid repaying their loans without defaulting. For example, through income-based repayment, deferment programs, or public sector loan forgiveness. This has a potential to be costly.


You mentioned that people who take on these “Jumbo-loans are typically better prepared for the labor force. There’s another WSJ article saying that five million people are defaulting on their student loan repayments. What percentage of this is made up of people who take on these large student loans?

About half of all dollars in default come from borrowers with more than $25,000 in debt. About a quarter to a third come from borrowers with more than $50,000 in debt, despite the fact that only about 15% of borrowers who default have more than $25,000 in debt.


So, a large chunk of this is owed by a small number of people. How do you think these defaults will affect the broader economy in the years to come?

Certainly, student loan defaults are costly, particularly due to the life cycle profile of these borrowers who have student debt. If they default early on in the life cycle (which most borrowers who default do) this damages their credit history, and it may prevent them from being able to access credit later on in the future. That can affect consumption, it can affect GDP growth, and it can change future investments in human capital. That is a significant portion of Americans- there are well over 40 million borrowers with some student debt outstanding.


That’s a big number.

It’s the second largest source of household debt after mortgages.


Unlike other loans, student loans are taken, and you don’t necessarily take them on when you know that you have a high income.

Exactly. These information asymmetries that we see in all consumer credit markets are potentially much higher in the student loan market.


Where do you see this trend going? We have 40 million people in debt with about $1.5 trillion that we’re not sure will get repaid. Where do you see this going?

I don’t see any apparent trends that would lead to a reduction in student debt levels. So I think that debt levels will continue to rise in the future at rates similar to those that we’ve seen in the past. I don’t see any significant countervailing forces.


Is there a solution to this?

There aren’t many obvious solutions. The reason for this is that many proposals depend on parameters that we’re unsure about. The biggest thing I’d like to see in the next couple of years is more data availability and more research in the student loan space. I think the first thing to do is to increase research, have randomized control trials, and modeling of student loan programs, and then determine what the best policy solutions are. I think that’s my biggest hope for the next couple of years: that we can get more data and research out in this space to better answer these questions.


Do you think this is a bubble that’s happening?

There are a couple of different definitions of bubbles- one is through the failure of the transversality condition. I don’t think this is in any sense a rational bubble as there might be in asset pricing models. I also don’t think that we’re going to see a rapid reduction in student loan borrowing. The reason it’s hard to call this a bubble is because there’s nothing priced here. Classically, when we think of a bubble, we think of the price of tulips or the price of homes. But here, the prices are fixed.


Congress sets interest rates. We could come up with some ad hoc definition of a bubble, but I’m not sure what that would be. People could stop taking student loans, but I think that’s unlikely to happen because returns on education are quite high. I don’t think we can call anything about the student loan market a bubble, but that doesn’t mean that there isn’t a very significant impact of student loans on the U.S. economy. You have to keep in mind that those impacts are both negative and positive. Obviously, student loans exist to alleviate credit constraints and allow borrowers to make high-return investments in education. Of course, these productive investments are being made, and this increases GDP growth and individual incomes. On the other hand, there are potential concerns with debt overhang and credit constraints early on in the life cycle that are often focused on in the media.


Studies that show that the value of a college education vastly outweighs not taking on one and saving X amount in tuition or student loans. It’s still very valuable.

On average, the returns on education are high. The issue is that there’s quite a bit of heterogeneity. Right now, we don’t have any risk-based pricing in student loans, nor do we have any risk-based credit limits. That’s an area where more research could shed light.


About the Author

Sabrina Kite

Join our mailing list