By now we all know the story of the 2007 housing crisis: too many subprime borrowers were given too much credit, which led to a spike in defaults and foreclosures. In the years since the crisis, however, a different narrative has come to light. In a recent HCEO working paper, FI network member Stefania Albanesi and her co-authors Giacomo DeGiorgi and Jaromir Nosal show that credit growth during this period was concentrated primarily in the prime segment, not the subprime population. Furthermore, they find that the rise in mortgage delinquencies during the crisis was driven by mid- to high-credit-score borrowers, who also defaulted disproportionately on their mortgages.
Albanesi and her co-authors analyzed anonymous credit files, which include information on debt holdings, delinquencies, public records, and credit scores, from the credit-reporting agency Equifax. They then used this data to “examine the evolution of household debt and defaults between 1999 and 2013.” Their findings depart from much of the previous literature on the housing market crash.
In seminal work on the topic, Atif Mian and Amir Sufi look at the creditworthiness of subprime borrowers, broadly defined as borrowers with low credit scores, based on their 1996 and 1997 credit scores. Albanesi et al. show that this is not an appropriate way to rank borrowers to assess their default risk at the time of borrowing. At any given time, they point out, people with low credit scores will be disproportionately young, as they are just starting to build their credit history. Mian and Sufi’s approach, the authors write, may confound “an expansion of the supply of credit with the life cycle demand for credit of borrowers who were young at the start of the boom.” Albanesi notes that this is a main reason why her team found different results.
“What we do instead is to rank individuals by a recent credit score,” she says. The study finds that borrowing for individuals with low credit scores was “virtually constant during the boom.” Whereas, between 2001 and 2007, credit growth was “concentrated in the middle and at the top of the credit score distribution.” The paper shows that the rise in defaults was also concentrated in the same population, the mid-credit-score borrowers. This finding led the researchers to wonder what was causing this rise, as historically these borrowers don’t default.
To find out, Albanesi and her co-authors honed in on speculative activity in the housing market, which was possible as the data showed how many first mortgages a borrower has. They looked at the behavior of borrowers they divide into two groups: investors and non-investors, the former of which is classified as someone with two or more first mortgages. The paper offers four reasons why real estate investors may have had higher default rates. For one, mortgages for non-owner occupied housing typically have higher interest rates to compensate for the higher risk of default, which makes them more expensive for borrowers. Additionally, only primary residences are protected by personal bankruptcy, which increases the probability of foreclosure on investment properties. The authors also note that if investors are “motivated by the prospect of capital gains, they have an incentive to maximize leverage,” or borrow heavily, in order to increase their potential gains. Lastly, they argue that losing a primary residence is more psychologically and financially costly than losing a rental or investment property.
The data on mortgage originations leading up to the crash also points to the increase in investor activity. “The sizable rise in mortgage originations for prime borrowers early in the boom combined with the modest rise in the fraction of borrowers with first mortgages for that period suggests that most of the originations reflect refinancing activity or real estate investing,” the authors write. Basically, during the housing boom middle- and upper-class borrowers took advantage of low interest rates and rising housing values, switching to lower cost mortgages via refinancing or by purchasing investment properties.
“Most importantly, we find that the rise in mortgage delinquencies is virtually exclusively accounted for by real estate investors,” the paper states. For borrowers with only one first mortgage, the rise in mortgage delinquencies was very small in comparison. A divergence also emerged in foreclosure rates among investors and non-investors. During the boom years, foreclosure rates were similar for the two groups. But as the housing market crashed, investor foreclosure increased “by a factor of 4 for the lowest quartile, and by more than a factor of 10 for quartiles 2-4,” the authors write. While for non-investors, the rate “roughly doubles in quartile 1-2, and rises very modestly for quartiles 3-4.”
“These borrowers account for 15 percent of all mortgages at the height of the housing boom, but they account for more than 50 percent of all foreclosures during the crisis,” Albanesi says of the investors.
The findings in this paper offer policymakers a new set of questions to consider when trying to prevent another housing bubble. Albanesi suggests it may be time for a review of the current credit-scoring model that companies like Equifax and FICO use to determine creditworthiness. The models are not fully understood as they are proprietary, but Albanesi notes they all share one feature: “as long as you make all your payments, the more debt you have, the more your credit score increases.” During the housing crisis, for example, the credit scores of borrowers with three or four mortgages increased as long as people made timely payments. Current scoring models fail to fuilly account for the additional default risk associated with investment properties. Addressing the limitations of current credit scoring models and updating them to more accurately account for variations in default probabilities would help to diffuse the risk of a future housing crisis.
Albanesi also noted the U.S. could consider implementing a debt servicing requirement, which Canada and some European countries have in place. Such a requirement would impose an upper limit on monthly debt payments relative to the borrower’s income, thus preventing the issuance of mortgages that are too large and reducing the probability of default. “It seems that would be something very important to stabilize the housing market and prevent a huge amount of speculative activity, when rental rates do not rise proportionally with housing values,” Albanesi says.
The authors hope this research will help change the misconceptions that continue to dominate the narrative of the housing crisis. “Subprime borrowers didn’t play a large role, and they didn’t receive more credit than in prior years,” Albanesi says. Instead, it was borrowers with good credit history engaging in risky speculative activity were responsible for most of the defaults, and possibly for the disproportionate rise in housing values in prior years. If we want to avoid another housing boom and bust, this research helps us first make the crucial step of understanding the causes of the last one.
Thank you to Stefania Albanesi, Giacomo DeGiorgi, and Jaromir Nosal for their assistance in completing this article.