Transcript
Charles Davidson: Welcome to another Federal Reserve Bank of Atlanta ECONversation. I'm Charles Davidson with the Atlanta Fed's public affairs department. I'm here today with Atlanta Fed research economist Chris Cunningham. Chris, thanks for your time.
Chris Cunningham: Hey, Charles. Thanks for having me.
Davidson: Yes, sure thing. And thank you for joining us, and please keep in mind that throughout our conversation, you're free to ask questions by sending your queries to the email address events@atl.frb.org. We'll get to some questions in the last ten or so minutes of our show today, so please—if you have something you're curious about, by all means let us know.
So Chris, we're going to talk about research that you, along with a couple of coauthors, have conducted on hydraulic fracking and how that affects home values—and, more particularly, mortgage default rates. And so let's start, Chris, though, with a quick primer on what fracking is, and then maybe you could talk a little bit about why you guys chose to research the impact fracking has on mortgage defaults.
Cunningham: Sure. So fracking—and I'm probably not going to get this quite right, I'm not a geologist or a petroleum engineer—but fracking is a process by which lots of water and sand and some chemicals are injected underground into certain geological formations to break them up and release the hydrocarbons, the natural gas or the oil, trapped within them, and then that is brought back to the surface. And it's a technology that came on really quite quickly about 10 to 15 years ago and arguably transformed the energy market in the U.S. So it's a big experiment, and economists like natural experiments, and so I and many other researchers have been drawn to it.
Davidson: Right, okay. But Chris, can I ask you—is there a deeper purpose behind this research than purely what happens to home values in areas where there's lots of fracking? Is there a broader lesson?
Cunningham: Sure. So first of all, it is a legitimate policy question. A number of states and cities and countries have banned the process—it is a controversial one—and so we do have some interest in understanding how it affects housing markets. Housing markets are kind of a nice barometer of how people feel about certain public goods or amenities. But, yes, in addition to that, it's an economic shock that lets us look at how housing markets—including housing markets that were in the middle of a housing bust in the Great Recession—how they might respond to an employment shock. And so it tells us something more broadly about how people behave, and how housing markets behave, in response to employment growth, for example.
Davidson: Right. So we view housing values as a proxy for general attitudes about something—in this case, fracking?
Cunningham: Yes, it's a go-to barometer of public sentiment.
Davidson: Sure, okay. Now, fracking has become pretty widespread across the United States—I think you guys cite a stat that more than 15 million Americans live within a mile of a fracking well. Is that about right?
Cunningham: I think so. I'm just cribbing this from the Wall Street Journal—we used that to motivate the paper. In any case, these wells are close to homes, and how they impact housing markets is kind of a central question.
Davidson: Sure, okay. So why focus this work on Pennsylvania?
Cunningham: Well, a couple of reasons: one, Pennsylvania was relatively dense, had a lot of housing—
Davidson: Okay, so now we've got a slide up here...
Cunningham: This is the growth of unconventional wells over time in Pennsylvania.
Davidson: Okay, and "unconventional" equals "fracking"?
Cunningham: It equals "fracking," thank you.
Davidson: Okay, cool.
Cunningham: So we had a lot of established houses, housing markets—a lot of mortgages that were originated before fracking was known to be viable—and we can look at these houses, some that ended up being next to a fracking well and some that ended up not being next to a fracking well, and hopefully control for everything that might otherwise impact those markets and really tease out the impact of the wells on the housing market.
Davidson: Right. So basically then, an area that experiences a fracking boom is a good laboratory to study what happens when there's a sudden surge in economic activity, then.
Cunningham: Yes, that's the hope, at least.
Davidson: Okay, all right. So one of the major findings was that increased fracking—in Pennsylvania, in this study area—significantly decreased mortgage default rates, right? And I think, in fact, you guys write that living in a ZIP code where there is fracking is roughly equivalent to a pretty good boost in your FICO score. What are the dynamics at work there?
Cunningham: So we're using a mortgage servicer dataset. We have a large fraction of all of the private-label mortgages—what would tend to be some of the riskier mortgages originated before the housing bust—and we have a lot of the variables—all of the variables—that were used in the underwriting process. So we have FICO scores but also a lot of controls for the nature of the loan, whether it was a prepayment penalty, whether it was an adjustable rate, balloons—that sort of thing.
So we can control for those, and then if we look at these mortgages over time, we see them originated, and we follow them until they either are paid off and people get a new mortgage or move their house, or they're still there, or if they default. And so we can look at the effect of fracking on the propensity of default, and at the same time control for FICO scores. And yes, in our data it looks like having a fracking well start up near you is consistent with turning a "620 borrower"—which is typically a threshold for subprime—into somebody with a FICO score above 700. They think that it lowers default by about that much, so it's a pretty striking finding.
Davidson: Yes. It would seem that living near a fracking operation may be unappealing. I mean, there's noise, there's the possibility of various types of pollution—which one would assume would make a home less valuable. But I think you even cite research that suggests that very phenomenon. However, your work seems to indicate that the picture may be a little more complicated, right?
Cunningham: Yes, that's right. So there are obviously lots of real and perceived disamenities with fracking, and here's a picture from...
Davidson: And so "disamenity" means the opposite of "amenity?" [laughter] Okay.
Cunningham: Yes, the opposite of "amenity." The process of putting one in is really quite industrial—there are lots of trucks and diesel engines and emissions from running those at the time. When it's done you've probably cut down a forest, or a field in which there'll be a wellhead, maybe some tanks, pipes on your land, but—
Davidson: So we could just—I'm sorry, but—is this one well, basically, that we're seeing here?
Cunningham: Yes, this is just one well being installed.
Davidson: Okay, so it's a pretty extensive operation.
Cunningham: That's right. Now, obviously, that's only there for a relatively short period time while they're putting in the well. But even when that's gone, there will be some infrastructure left, and you are putting holes into the ground and injecting water and some chemicals—some of which were not disclosed, at least initially. And so there is some existing work that suggests that people living very near wells—especially people who rely on well water—that those houses fell in value when those wells came online.
And our work does not contradict that—that could still be true—but the way that the paper was written was in order to get a very clean analysis. To find the effect of fracking on houses, they had to look at a very small geography—so places right next to a well, versus places near wells but not right on it. And so while that's very clean, it basically absorbs or removes any kind of positive benefits that might be coming in the form of employment or general economic activity. Our analysis operates at a little bit different/larger geographies—ZIP codes or counties—and when you look at that wider area, the net effect for houses appears to be positive, at least over the timeframe that we look at.
Davidson: Right. So Chris, a couple of times you've mentioned "controlling for certain variables." I wanted to ask you just to explain, especially for viewers, what that means. I think what you're talking about there is making sure that you're measuring exactly what you want to measure, and say, "Maybe this could be caused by something else"—sort of the "causation versus correlation" question?
Cunningham: Right. So, economists—at least, we like to think that we're thinking very carefully about causality, and it's a difficult problem. It's relatively easy to pull the data and say, "Okay, well, we see wells here, we see house prices or mortgage defaults here," and correlate them. But sometimes that can be confusing or confuse the story.
So for example, one of the things that we find—one of the exercises we do is an attempt to use the underlying geology to explain where the fracking wells occur. So this is a figure that shows the depth and thickness of the geology layer. We actually took those measurements, and we use just that geology to predict where the wells show up on the surface. And the hope is that—you know, these sedimentary beds were laid down millions of years ago, when pieces of Africa had not yet been attached to North America—that's going to be uncorrelated with what's going on with the housing markets on the surface.
And we do this because there is a lot of discretion on allowing the wells to happen on the surface. Some local governments will ban it. Even if it's not barred by regulation, if you're a relatively affluent neighborhood you may not want to tolerate those wells there, and the drillers would go elsewhere. And so one of the conclusions of our work is that when you use this statistical method—when you control for the geology—you actually get much bigger positive impacts of fracking on housing markets, suggesting that it was the areas that were more hard-pressed going into the Great Recession that were more amenable to fracking taking place. So it's almost an order of magnitude bigger effect once you use this. It's called the "instrumental variables" approach. So that's kind of at the heart of the analysis.
Davidson: Right. So Chris, going back to the question of larger implications from work like this: what would be a policymaker's lesson from this? Say we have a housing downturn down the road. Any lessons to be drawn from this in a case like that?
Cunningham: Well, to inform policy you would need to look at many studies and different flavors, so this is just one small piece of that. But one thing you might draw from it is that it looked like the effect of fracking had particularly positive effects when it was operating through the employment market, so it looked like right around the timing of putting new wells in, which has this employment surge, that seemed to have the best effect on mortgages, and by extension...
Davidson: Okay, so that's when you'd see the most jobs created, at that point.
Cunningham: Yes, it's less from just the royalty payment flows, or even forward looking measures, like when the permits were drawn. And we also do some attempts to disentangle that channel, so we look at actual employment flows, for example. So one takeaway would be that, should there be another housing market crisis or downturn, interventions that were more targeted towards the employment market—perhaps more generous unemployment insurance, or a targeted mortgage relief to those with job losses—might have been more efficacious, or would be more efficacious, than some of the interventions we'd considered instead, which were more targeted towards changing the terms of the loan, or the rate. This is consistent with work of one of my coauthors, Kris Gerardi, who is an economist here, called the "double trigger" hypothesis: that you really needed to be underwater, you needed to have a house that declined in value relative to the mortgage, but, too, you had to have some kind of shock—you basically had to lose your job.
Davidson: Sure, okay. Well, Chris, I'm not sure this was a central part of your research here necessarily in this particular project, but can you talk a little bit about just how much of an economic boom Pennsylvania saw as a result of fracking in the Marcellus formation that you guys explored? Is it lasting, or is it more transitory?
Cunningham: So again, we're mostly mining data around the housing markets. It's harder to extrapolate a broader picture. I will say that fracking wells seem to be associated with employment gains, some house price gains, decreases in unemployment. There's obviously other work in the development economics literature that suggests that, on the whole, resource booms and busts can have mixed or even negative long-run effects—the "resource curse"—and I can't speak to that, per se. I will say that we followed these existing mortgages over a fairly long period of time, and there didn't appear to be a hangover—at least with our initial analysis dataset. It could be that some subsequent ones suffered, but on the whole it looks like it was pretty beneficial to Pennsylvania.
Davidson: Right, okay. Well, a couple of audience questions here. It doesn't sound like you guys are necessarily trying to weigh in specifically on the benefits of fracking and whether they outweigh problems such as potential water contamination, earthquakes, and—as we discussed—just kind of a visual blight. I mean, this isn't a "fracking good/fracking bad" sort of exercise, is it?
Cunningham: Well, I certainly went into this paper with really no strong priors. I was happy to find either result, and I guess maybe I expected even to find more negative effects from fracking, given the existing literature. And it's gotten a lot of media coverage, too, that we probably all read. I guess I would say that I would be—having done the work, I'm more sympathetic to letting fracking continue than I was going into it. It's tough, because those wells are going to be there forever, and it's hard to make a real strong case that the future will value the risks of having drilled those holes through the geological layer. But yes, on the whole I think we should be at least a little bit more sympathetic to it.
Davidson: Yes. Okay, another audience question here: if you could sum up quickly why we should care about this, what would you say? I guess this goes back to the "broad lessons" kind of question.
Cunningham: Sure. Again, the big takeaway is—one big takeaway—that during the Great Recession, a lot of models actually predicted much higher rates of default. There were a lot of mortgages that were underwater that probably...if people were operating purely out of financial self-interest, the term in the housing literature is "ruthless default"—they would have defaulted. And in fact, many people kept current on those mortgages, stayed in their homes, and rode it out.
Davidson: So this would be where, "Okay, we'll cut cable TV, we'll take our kid out of private school, whatever—we'll make some even way more serious cuts than that, probably, in order to keep our house."
Cunningham: They did what they had to do, and often it was when they really couldn't do anything else, that that generated a default. Not in all cases. I mean, obviously there was some speculation, there were some hot markets, but in many cases people tried to stay current, and so some help along the way targeting those people, I think, is probably a good idea if we have to go through this again in the future.
Davidson: Yes. Well, that gets to the end of our audience questions here. Well, Chris, again, I think we covered this, but this notion of causality and correlation—can we just quickly recap again why is that so important in economic research? I think it's kind of self-evident. That really gets to the heart of really understanding what's happening in the real world, right?
Cunningham: Yes. Ideally we want to be able to run experiments in which—like in our case, houses or mortgages—you treat one group, you don't treat another. In some cases, you really can do that. There are some amazing datasets being created with targeted interventions—sometimes in the developing world, sometimes here. But in lots of cases, you just can't do that. We can't rerun those experiments. They're too expensive.
So we've got to try to find them that occur naturally. This technology for fracking becomes newly viable, a place like Pennsylvania has this great heterogeneity, this great variation in the distribution of resources, and we can recreate a plausible experiment. That is sort of the goal, and when we do that, then we learn—and we probably learn more than we would have just by looking at the statistics and cross-section.
Davidson: Right. So will you guys be continuing this particular research, Chris?
Cunningham: This might be the last paper, we'll see. We are not the only economists trying to mine the fracking experiment—
Davidson: No pun intended there. [laughter]
Cunningham: I walked into that, actually. But you never know. Sometimes some question will come up and then you'll...there is one particular question that we haven't really called out. There's actually a provision from Freddie Mac and Fannie Mae, the GSEs, that actually prohibits them buying a loan from a house that's near a well. So one of the nice things about this dataset in our experiment is we have houses that got the mortgage, and then got the well—the mortgage came before the well—and so you can actually look at the performance, and again we find that they seem to have performed better than they would have, but for the well. So that's a nice test of that rule, and to the extent that we're finding an effect that suggests that maybe that rule, with regards to wells, is not so efficacious. So we may go back and revisit this with a little bit cleaner, more targeted knowledge about the homes. Otherwise, there are always other papers to write.
Davidson: Yes, certainly. Well, Chris, one last item here, just for the viewers, I wanted you to just quickly describe your field of expertise as an economist, your research interests, and how this fits within that portfolio.
Cunningham: Sure. So generally, I'm an urban economist—I think about cities. Housing is obviously a big part of the services that cities provide. I also look at issues around local public finance, property taxes, and municipal budgets—and again, those always get back to housing as well. So this particular topic came up because many economists were looking at the fracking boom, and my coauthor Lily Shen, who is at Clemson now, was at Penn State, so she had a front door seat to it. She came to us and said, "Look, we 've got to do something on this question." But more generally, I think about issues around the production of housing, housing supply, and the provision of local public government.
Davidson: All right—well, Chris, thanks so much for your time today.
Cunningham: Thanks for having me, Charles.
Davidson: And you can find the actual research paper we're discussing today at frbatlanta.org, along with a wealth of other research and economic information. And we look forward to seeing you in August, for the next ECONversation. That ends our discussion for today. Thank you for joining us, and again, please visit our website at frbatlanta.org. Thank you.