Tom Heintjes: Hello, and welcome back to another episode of the Economy Matters podcast. I'm Tom Heintjes, managing editor of the Atlanta Fed's Economy Matters magazine, and today I'm sitting down with Scott Frame and Kris Gerardi, financial economists and policy advisers in the Atlanta Fed's research department. We recently published a working paper by Kris and Scott, as well as their co-authors James Conklin
and Haoyang Liu , about the role of subprime loans in the home price boom we saw last decade. The paper's title is intriguing—it's "Villains or Scapegoats? The Role of Subprime Borrowers in Driving the U.S. Housing Boom," and today they're going to discuss their research with us. You've each been on the podcast individually, but it's great to have you both here at the same time. So, welcome back to the podcast, Scott and Kris.
Scott Frame: Thanks, Tom.
Kris Gerardi: Thanks a lot, Tom. We're really psyched to be here.
Photo: David Fine
Heintjes: Scott and Kris, as I mentioned, the title of your paper is "Villains or Scapegoats? The Role of Subprime Borrowers in Driving the U.S. Housing Boom," and in it you address the belief that an expansion of mortgage credit to subprime borrowers was a principal driver of the housing price boom we saw in 2002 through 2006. But contrary to conventional wisdom, you found that the house price boom and the subprime boom occurred in different places, different regions of the country. I guess you could say the subprime boom had an alibi—it was somewhere else at the time?
Gerardi: Well, that's a good way of putting it, Tom. The paper shows that during the housing boom of the early to mid-2000s, areas with high growth in mortgage lending—and specifically, purchased mortgage lending—to subprime borrowers (which, in the paper, we define to be households with credit scores below 660) occurred on average in places that didn't experience high house price growth. So, for example, the so-called "sand states"—Arizona, Florida, and Nevada—that saw a huge growth in prices during the boom, didn't see a corresponding large expansion in mortgage credit to subprime borrowers. Instead we see high subprime mortgage growth in the Ohio River valley and many of the midwestern states where house price growth was really low.
Heintjes: Well, that's really interesting, and in reading your paper I saw that the narrative—correct me if I'm wrong, but the narrative seems to go something like this: The expansion of mortgage credit to subprime borrowers in the early to mid-2000s inflated the housing bubble, and this was responsible for the subsequent financial crisis and recession. So you're saying that we should not necessarily just buy into that narrative?
Gerardi: Well, the narrative has recently come under attack—not just from our paper, but by several academic studies—and it's looking increasingly fragile. So to be specific, there are a few recent papers that have shown that there was a more or less uniform expansion of mortgage credit across the entire income and credit score distribution at the national level, which really calls into question this idea that subprime credit specifically played an outsized role in inflating the bubble. Our study begins with the premise that subprime could still have played an important role in driving the bubble, if it was in fact concentrated in the geographic regions that experienced the big price gains. But, if this isn't the case, and there isn't a positive relationship between subprime growth and house price growth, I think the subprime narrative really begins to fall apart, and that's exactly what we find. So I'd say yes, we should be quite skeptical of the narrative at this point.
Heintjes: Right. Well, you know, this relationship between the growth in subprime lending and the house price boom seems like commonly accepted wisdom at this point. Was it believable because it seemed so logical, even if the logic, as you show, had some serious defects? Or was it because they occurred simultaneously?
Frame: That's a good question, Tom, and I don't really think there's one right answer. It was likely a confluence of events. Subprime mortgage defaults played a central role in the beginning of the financial crisis in 2007 and 2008, and so they were on everyone's mind. There were lots of anecdotes about borrowers who obtained subprime mortgage loans that they obviously couldn't repay in the long run, and there were also allegations of predatory lending in the subprime mortgage space. So given these events, I don't think it was such a huge leap to assume that subprime inflated the bubble. However, as economists, we're always taught to be skeptical of theories without hard, empirical evidence.
Heintjes: The whole "correlation is not causation" school of thought?
Gerardi: Exactly.
Frame: A great way to put that, yes.
Heintjes: Well, let me ask you to step back. As researchers, what occurred to you that there was something even here to look into? You mentioned that there had been some skepticism about this, so did that prompt you to look into this a little more deeply?
Frame: Well, we can't actually claim credit for the initial finding. Our co-author, Haoyang Liu—who has recently become an economist at the New York Fed—was the one who first discovered the pattern in the data. He was puzzled by it, and he came to see us to ask about whether we had seen a similar pattern, as we'd been working with this kind of data for quite some time. We were able to verify his findings and decided that they were really important to present to the public.
Heintjes: Well, I don't want to get too far into the weeds here—and, as usual, your paper includes a great deal of math that's over my head—but I want to talk briefly about your methodology. How did you go about conducting this research?
Gerardi: I think it's actually a relatively simple, straightforward analysis: we put together a county-level dataset on house price growth and subprime mortgage credit growth that covers the period of the housing boom in the early to mid-2000s, and then we basically tried to determine, using quite simple regression methods, whether there was a positive relationship. In other words, do we see evidence that counties which experienced high house price appreciation also experienced large growth in purchased mortgage originations to subprime borrowers? The answer turns out to be "no," and in fact, if anything, we actually find a negative relationship, so that the places that experienced the highest growth in subprime mortgage credit experienced the lowest price appreciation.
Heintjes: Interesting. Well, I noted that you incorporated FICO, or credit scores, into your methodology. Can I ask you, what's the advantage of using FICO scores as opposed to using reported income?
Gerardi: That's a good question. I think there are a few reasons for focusing specifically on credit scores. So there's really no consensus on the exact definition of a subprime mortgage, and so one's forced to take a stand: researchers have used different definitions based on the type of lending institution—so subprime mortgage lenders, for example—the type of borrower, and the characteristics of the mortgage products themselves. In the context of the determinants of housing prices, economic theory tells us that it's the borrowers and the margin between renting and owning that really impact price dynamics, which suggests that an income or credit score definition is really most appropriate. Traditionally, "subprime" referenced borrowers with poor credit histories, which made FICO scores a natural variable to focus on. In addition, precise measures of household income are really largely unavailable in the datasets that we have at our disposal. In contrast, we have good data on FICO, and credit scores more broadly.
Heintjes: Sure. Well, let me ask you guys: what effect did the subprime boom actually have on the housing market?
Gerardi: That's another really good question, Tom, which unfortunately we aren't going to be able to provide a really precise answer to. It's a really difficult question to address because there were likely effects in both directions—that is, the housing boom likely affected underwriting standards in the subprime mortgage market, as higher expectations of future housing values made mortgages appear less risky to lenders, and of course, higher expectations of price growth likely increased demand for mortgage credit by households themselves. Distinguishing between the effect of the subprime boom on prices and the effect of prices on the subprime market is an incredibly hard problem, which many researchers have really grappled with but few have successfully overcome.
Our belief is that there were likely effects in both directions, but the exact magnitudes of those effects have yet to be credibly established. So in our eyes, the contribution of our study is to largely rule out the theory that the expansion of subprime mortgage credit played a first-order role in driving the housing bubble.
Heintjes: Right. Well, in your paper you showed that credit to home buyers grew across the entire spectrum of creditworthiness, not just subprime. Was this broad-based expansion a factor behind the housing price boom?
Gerardi: It's certainly possible, but it's also consistent with the house price boom fueling demand for housing and mortgages across the entire credit score distribution. We know that expectations of house prices became extremely high during this period, and there's a lot of evidence showing that speculative activity in the market grew really rapidly, so I would argue that this is probably more consistent with the theory that house price growth fueled the large increase in the demand and supply of mortgages.
Heintjes: Well, you also note the importance of making a distinction between a credit expansion to risky borrowers and a credit expansion of risky products. Again, I need to stay in the shallow end here, but can you explain the notion at work with this concept?
Frame: Yes, that's an important distinction to make, Tom. We know that underwriting standards were broadly relaxed during the housing boom, and this relaxation took place along borrower characteristics—such as providing loans to borrowers with lower credit scores and income levels—as well as mortgage characteristics, such as originating loans with lower down payment requirements, lower documentation requirements in terms of verifying income and assets, and also longer amortization schedules. As you know, in the United States the typical mortgage has a 30-year amortization period, so some of the mortgage products introduced during the last decade had longer periods—40-year amortization schedules—or they may even allow for some negative amortization for a period. So the focus of our paper is on the expansion of credit-based measures of borrower characteristics, and specifically we focus on credit scores, so we're focused on the borrower characteristics, not the product characteristics.
Heintjes: Gotcha, right. Well, Scott and Kris, this narrative of "subprime lending led to the house price boom" seems so baked into the way people think of that period that it would be a really difficult narrative to alter. When you discuss your findings with your peers, how are they initially greeted?
Frame: That's an interesting question, Tom. Actually, most of the researchers that we've interacted with about this have been fairly receptive to the findings. Much of this may reflect the timing of our study, given that our findings are consistent with some very recent papers that have called into question the conventional narrative that subprime fueled the housing boom. If we had released this paper, say, five to seven years ago the reception may have been much cooler.
Heintjes: Well, even if the subprime boom didn't directly contribute to the house price boom, is it fair to say that the subprime boom did, in a way, contribute to the crisis in the housing market after so many subprime borrowers defaulted on their mortgages? And I know that's a bit far afield from the paper we're discussing here, but would we be remiss in not acknowledging that aspect of the dynamic?
Frame: That's an important point. Our paper really speaks to the fact—or the lack thereof—of the expansion of mortgage credit to subprime borrowers on the rise in house prices during the housing boom. Our paper doesn't have anything to say about the subsequent crisis that was triggered when house prices began to decline sharply. There's lots of empirical evidence that shows that subprime borrowers were the first ones to default during the housing bust. They defaulted en masse, and these defaults triggered the financial crisis, which ultimately led to the Great Recession.
Heintjes: Right. Well, I mentioned that you're financial economists, and also policy advisers here at the Atlanta Fed, so let me ask you about the policymaking implications of your research: what are they, if any?
Gerardi: That's a tricky question, Tom.
Heintjes: Yes, I thought it might be. [laughs]
Gerardi: I'll take a quick stab at it. I think in the aftermath of the crisis, there has really been this big debate about the most effective policies and reforms to prevent a future bubble from taking place, and I think our paper suggests that policy prescriptions specifically intended to limit access to credit for marginal borrowers are unlikely to prevent a future boom or bubble from occurring. But we do know, and there's a lot of evidence coming out recently, that prime borrowers—so, borrowers with credit scores in the 700s and even 800s—also changed their behavior during the boom and defaulted en masse during the bust, which led to a lot of the GSEs' problems, for example. And so the point is that it wasn't just a problem specifically focused around subprime borrowers, so we need to be a little bit careful in terms of singling that group out going forward.
Heintjes: Right. I guess it's hard to isolate, because in a situation like that there's a great deal of contagion that goes on, where there's spillover effects. Well, Scott and Kris, I'm afraid I can't let you gentlemen leave the studio today without answering the question you posed in your paper's title: are the subprime borrowers victims or scapegoats? Who's going to take that one? [laughter]
Frame: I'll let Kris take that one.
Gerardi: In the context of explaining the underlying sources of the bubble itself, we'd have to say "scapegoats." I'll leave it at that.
Heintjes: Right. That's what I thought you would say, but it was interesting to hear you state it. And Scott, I assume you would agree with that?
Frame: I would agree with that, yes.
Heintjes: Okay, well, you're on the record, so thank you both. Well, this has been a fascinating conversation, and I want to thank you both for sharing your insights with us and taking the time to be here today.
Gerardi: Thanks a lot, Tom—this was great.
Frame: Yes, thank you, Tom.
Heintjes: And I should note that we will have a link to your paper on the website so people can take a deeper dive into your research, and I think it's well worth the time to do so. And I would encourage all people listening to this podcast episode to read it. It's a very interesting and well-researched look at events that still reverberate in the economy today, even though they occurred a decade ago.
And that brings us to the end of another episode of the Economy Matters podcast. Again, I'm Tom Heintjes, managing editor of Economy Matters, and thanks for spending some time with us today. Please come back next month for another episode.