Let’s begin with our discussion of student loans, homeownership, and the macroeconomic implications. I joined the Board in 2012, having worked in the mortgage underwriting division of Fannie Mae for a couple of years. Already back then, the question was whether the student loan market was going to pose challenges. Could student loans become the same problem as mortgages were during the global financial crisis? Would households have increasing difficulties servicing these financial obligations?
To help address these questions, the Federal Reserve Board created a unique data set which combined individual credit bureau records with educational records of the same individuals. Using these data, we were able to start analyzing the student loan market early on. For instance, the relationship between student loans and homeownership was getting a lot of attention during 2013–2014. The housing market was still in the early stages of recovery and homeownership rates were down relative to the pre-crisis levels, especially for young individuals (Fig. 1). The overall homeownership rate fell by about four percentage points between 2005 and 2014, while homeownership of young individuals (between ages 24 and 32) fell by nearly nine percentage points. At the same time, student debt has been increasing and the average outstanding loan balance was about $7000 higher in 2014, in real terms, than it was in 2005. The natural question to ask was: “What is the contribution of student loans to the disproportionate decline in homeownership among young people?”
During the post-crisis period, there were many news reports documenting the economic hardship associated with student debt. Many graduates discussed their difficulties to repay, inability to buy a home, or financial challenges with starting a family. The media reporting was in large part corroborated by what could be inferred from surveys conducted by institutions such as Fannie Mae and Rutgers University. Many young people seemed to be agreeing that student loan debt was hitting them hard. However, when we started to think about the relationship between student loan debt and homeownership more carefully, the links were less clear-cut. On the one hand, college graduates should financially benefit from education over the lifecycle. It is well-known from existing research that these returns tend to be quite high on average. On the other hand, earnings can be quite low shortly after college exit. Graduates need to start repaying their student loans while—at the same time—they may be interested in applying for a mortgage. The student loan payments are likely to weigh on the ability to qualify for mortgage credit. Student loan payments could slow saving for the down payment, and credit scores could be affected adversely as well, especially if student loan borrowers miss some of the payments. Overall, it was not clear ex ante what would be the connection between student debt and homeownership in practice.
Let me now outline some basic stylized facts emerging from the data, which we discuss in detail in Mezza et al. (2014). If you look at the homeownership rates over the life cycle (Fig. 2), people who went to college, with or without debt, attain much higher average rates of homeownership than people who didn’t go to college. As you can see in Fig. 2, for those who did not attend college, the homeownership rate is initially higher than for college goers, with or without student loans. However, by their early thirties, college goers end up with much higher rates of homeownership than those without any post-secondary education. Overall, it appears that going to college—with or without student loan debt—still is the path to homeownership.
However, an important consideration is that college is becoming increasingly expensive. This has contributed to the increase in aggregate student debt. Some of this debt increase has been driven by higher enrollment, but the price effect is important. Figure 3 presents tuition data from the College Board. Although tuition generally tends to be lower at public 4-year schools than it is, for example, at private non-profit schools, it has risen far more. Tuition and fees at public 4-year colleges have basically tripled over the past 30 years. For other types of schools, such as private not-for-profit and public 4-year colleges, it has about doubled.
Let us now consider some thought experiments motivated by the analysis in a peer-reviewed publication (Mezza et al. 2020a) and a related Feds Note (Feiveson et al. 2018).
The first thought experiment is as follows: if we compared two individuals who were identical in all aspects but the amount of student loan debt, how would their homeownership differ? Our research finds that for young adults who went to 4-year public colleges, a $1000 increase in student loan debt accumulated during prime college-going years implies a lower homeownership rate by about 1.8 percentage points during their mid-twenties. This estimate may seem to imply a meaningful reduction in the probability of attaining homeownership for young borrowers. In practice, however, we think that this reduction only implies a transitory delay. We estimate that the extra $1000 in student debt delays homeownership by about 4 months. This finding is broadly consistent with the stylized facts presented in Fig. 2, in which people who attended college without student loan debt tended to attain homeownership a bit faster. But by age 35, all college goers experienced very similar rates of homeownership, debt or no debt.
So there is some good news emerging from the data: the impact of smaller changes in student debt on homeownership appears to be mostly transitory. However, since the estimates above focus on an incremental change of $1000 in student debt, one needs to consider how these estimates scale up based on the significantly larger observed debt increases. Therefore, in our second thought experiment (which is described fully in Mezza et al. 2020a), my co-authors and I use realistic changes in the levels and the distribution of student loan debt accumulated by young college goers between 2005 and 2014 to explore the effect of the increased debt on homeownership. Our estimates suggest that the increases in student loans observed between 2005 and 2014 have delayed homeownership by about a year on average. Put differently, a person aged 24–32 years in 2014 would be about a year behind in attaining homeownership than a similar person in the 2005 cohort. This average 1-year delay in attaining homeownership by young adults in turn translates into a roughly 2 percentage point reduction in the homeownership rate of the young between 2005 and 2014. To put this finding in context, the homeownership rate for young people declined by nine percentage points between 2005 and 2014, as I showed in Fig. 1. Therefore, our estimates suggest that increases in student loan debt between 2005 and 2014 explain about two percentage points of this nine-percentage-point decline, or about one-fifth. These findings lead me to conclude that student debt has been an important contributor to the decline in homeownership, but not its central cause—at least thus far.
Before turning to our third thought experiment, it is helpful to recall from the above discussion that student debt indeed tends to reduce homeownership on a temporary basis, and this effect likely works through diminished access to mortgage debt. This motivates our next question “What could be the impact of higher student loan debt on other credit markets, especially credit cards, auto loans and installment loans?”—a question we have started to explore in Mezza et al. (2020b).
The results are interesting in many ways. While increased student loan debt reduces borrowing in more tightly underwritten credit markets (such as those for mortgages and credit card debt), all else being equal, it leads to additional borrowing in credit markets with easier credit standards (such as those for auto and other nonhousing collateralized debts). This analysis suggests that increased student loan obligations result in differential effects on total borrowing by market segment by interacting differentially with the demand and supply of credit. We are still assessing how to interpret these results. Our current (and still preliminary) view is that in credit markets with more stringent underwriting, increasing student loan debt likely leads to a contraction in credit supply (that is, either through a reduction in entry to the credit market or through a reduction in credit limits). In contrast, in credit markets where underwriting is less tight, higher levels of student debt can lead to new demand for debt. As such, the ready availability of credit in these markets mitigates any potential contractionary effect that student loan debt service might have on borrowers’ spending.
This brings us to the third (and final) thought experiment: “What is the impact of rising student debt on the economy?” We explore this question in Feiveson et al. (2018). Figure 4 highlights the sharp increase in student loan originations over time. The average annual originations of student debt increased from about $20 billion in the early 1990s to roughly $60 billion since the turn of the century. Meanwhile, the total student debt increased from about $340 billion in 2001—already a meaningful number—to a sizeable $1.3 trillion in 2016. The debt stock is now at about $1.6 trillion.
It therefore makes sense to ask a hypothetical question “What would have been the impact on the economy if student loan originations had stayed at the 2000 level?” Lower debt would imply lower debt service costs and would open room for additional consumer spending. We find that even under some aggressive assumptions, it is difficult to come up with large GDP or consumption effects. Lower student loan originations would have boosted GDP growth by less than 0.05 percentage points in any year since 2001.
Without a doubt, student loans can be a big strain on individuals and households. Karen Dynan’s presentation has shown that a number of people have difficulties repaying these obligations. However, our thought experiment suggests that, at the macroeconomic level, the growth drag from higher student loans appears to be quite small, at least for now. We do not yet have adequate data to establish a clear picture. But my current thinking is that, while student debt is not yet a major drag on the aggregate economy, this could change if each new generation becomes more and more indebted, and incomes are not keeping up with the increases in the debt service. This should be the main concern going forward.
I will next turn to financial stability considerations, also discussed in Feiveson et al. (2018). The question is, “Can student loans become the next mortgage crisis? Can they cause a severe damage on the financial system?” To answer this, it is worth recalling that 90% of the student loan debt is government-guaranteed. Impaired student loans would therefore impact the government balance sheet, but they would not really undermine financial stability. Moreover, and what is very interesting, is that the subpopulation of borrowers who are struggling most to pay student loan debt account for a small fraction of overall debt in the economy. For example, looking at mortgage debt, the people who are at-risk student loan borrowers, account for about 2% of the aggregate mortgage balances.Footnote 1 Looking at the auto debt, the fraction is about 8%. Consequently, lenders do not appear to face much indirect exposure through loans to borrowers currently having trouble paying their student loans.
In sum, even with student debt, going to college is still a good path to homeownership and other benefits. I firmly believe that. Data show that until now, higher student debt has implied just a small delay in attaining homeownership. The macroeconomic risks have been manageable. However, these risks could increase going forward if student loans continue to grow rapidly without the corresponding increase in incomes.