The shifting sands of observation

I just happened to come across an interesting paper about mortgage defaults from Peter Ganong and Pascal J. Noel, “Why Do Borrowers Default on Mortgages?”

The results of the paper strengthen the case that the Great Recession caused the housing default crisis rather than the other way around. I will discuss this below. But, the most interesting thing about the paper is something that isn’t there. As you read my description of the paper below, see if you can identify the dog that isn’t barking.

The abstract:

There are three prevailing theories of mortgage default: strategic default (driven by negative equity), cash-flow default (driven by negative life events), and double-trigger default (where both negative triggers are necessary). It has been difficult to test between these theories in part because negative life events are measured with error. We address this measurement error using a comparison group of borrowers with no strategic default motive. Our central finding is that only 6 percent of underwater defaults are caused exclusively by negative equity, an order of magnitude lower than previously thought. We then analyze the remaining defaults. We find that 70 percent are driven solely by negative life events (i.e., cash-flow defaults), while 24 percent are driven by the interaction between negative life events and negative equity (i.e., double-trigger defaults). Together, the results provide a full decomposition of the three theories underlying borrower default and suggest that negative life events play a central role.

I think this comports with other research that has shown that homeowners do not easily strategically default. They are generally willing to hold on well into negative equity territory.

And, so, what this paper seems to confirm is that, whether a homeowner has positive or negative equity, it is overwhelmingly something like a negative income event that will trigger a default. In general, if a homeowner can make the payments, they will. And they will commit to keeping a house even if, in practical terms, future marginal equity gains worth tens or hundreds of thousands of dollars would simply be recovery of collateral value on the mortgage.

This has implications for broader theories about the financial crisis and the Great Recession. The take that Scott Sumner and I, among others, have laid out, is that tight monetary policy, not a housing bust, caused the recession.

Given a simple choice between (1) a recession caused a default crisis or (2) a default crisis caused a recession, #2 has captured the conventional wisdom, but #1 is closer to the truth.

The difference between those two choices is large. As I have written about, endlessly, really the Great Recession happened for one reason - because the country had become fatalistic about it. Our policy choices were handmaidens for it. Any policies meant to alleviate a cyclical reversal - any policies meant to increase construction or prices, or slow down defaults - were cheating. A comeuppance was inevitable, and we needed to take our medicine.

But, the medicine was the problem. Our policy choices created a recession that led to a default crisis. Neither the crisis nor the defaults were inevitable. We could have chosen to avoid most of the damage.

What We Have Forgotten

That brings us to what isn’t in this paper. Have you noticed it?

In the conventional wisdom, a credit bubble led to an oversupply of overpriced homes, demand cratered when the bubble burst, defaults spiked on all of the poorly underwritten mortgages, prices collapsed in neighborhoods where foreclosure sales accumulated, and eventually, all those processes led to recession and crisis.

The main source of those triggering defaults, in the conventional wisdom, isn’t listed in the 3 categories in the paper above (negative equity, negative life events, or a combination of the two). The main source of triggering defaults was that unqualified borrowers had taken on loans that they never could have afforded. This was the dominant narrative before, during, and after the Great Recession.

This factor is nowhere in the paper! The primary mythology of the crisis has disappeared from the discussion! This has happened so completely that it isn’t included in any way, as an alternative explanation, as a control factor, or as far as I can tell, even as a footnote. It’s just gone.

Here are some reminders, if you need them, of the importance of the idea in the popular narrative:

Rate resets were blamed for a foreclosure crisis before a foreclosure crisis even happened. Figure 1 is a type of chart that was all over the place in 2007 and 2008. Figure 1 is from this January 2008 article, which conveys some of the fatalism about a housing “correction” that was dominant at the time. The thing that was supposedly going to happen was that homeowners, in their current financial condition, would not be able to afford the mortgages they had taken on.

Figure 1

Of course, rates declined sharply, especially by the time most foreclosures happened in 2009 and 2010. Some research on defaults has found that rate resets had little relevance to defaults. That is mostly because when rates were still high, borrowers could still refinance. By the time that defaults were important, rates had declined. By the time complex mortgages originated in 2006 and 2007 were resetting, rates were the least of borrowers’ problems.

Here is a quote from “All the Devils Are Here” (pg. 252):

Fall 2006. Larry Litton is a mortgage servicer.  In 1988, he and his father, Larry Litton Sr., founded Litton Loan Servicing, building it into one of the nation’s largest mortgage servicers.  Inevitably, they service a lot of mortgages for subprime originators. Litton also notices that early payment defaults are soaring.  The mortgage originators are freaking out and blaming him.  “The WMC guys are saying 'You suck,'“ Litton recalls.  He remembers thinking, “Maybe we're doing something wrong.” So Litton comes up with what he calls an “ultra-aggressive move”: hand delivering welcome packages to new homeowners, so there will be no confusion over where the mortgage checks should be mailed.  But when the Litton employees arrive at the newly purchased homes, they discover something truly startling.  “My people came back and said, 'Thirty percent of the houses are vacant,'“  Litton recalls.  In other words, borrowers who closed on mortgages had so little means to make even the first payment that they never bothered to move in.

Note, this description is dated very early - Fall 2006. Here is a chart of delinquencies over time. Aggregate delinquencies, even in subprime mortgages, were not particularly high yet, by then (Figure 2).

Figure 2

This assertion is extreme. 30% of these borrowers gave up before they even moved in and made a single payment. I have always thought there is something amiss with this story, but my point here isn’t to dissect it. My point is that the uncontroversial conventional wisdom about the housing bust was that borrowers couldn’t afford the mortgages they had, even before experiencing a negative life event.

There is the infamous Jim Cramer “They know nothing!” breakdown on CNBC in August 2007. Here’s part of what he said: “Fourteen million people took a mortgage in the last 3 years. Seven million of them took teaser rates or took piggyback rates. They will lose their homes. This is crazy!”

By February 2008, CNN was reporting:

For months, we've fretted about the Armageddon that will hit when subprime adjustable rate mortgages start resetting to much higher interest rates.

What's happening is even worse: Many of these loans are defaulting well before their rates increase.

Defaults for subprime loans issued in 2007 - none of which have reset yet - hit 11.2 percent in November. That represents perhaps 300,000 households, and is twice the default rate that 2006 loans had 10 months after being issued, according to Friedman, Billings Ramsey analyst Michael Youngblood.

Defaults are spiking well before resets come into play thanks to the lax lending environment of the past few years. Many borrowers were approved for mortgages that they had little chance of affording, even at the low-interest teaser rates .

This was the theory of the housing bust that was canonized before there was a bust. And, please be clear about what has happened here. The paper above doesn’t refute these accounts. These accounts - which dominated the perspective about what was happening - are so irrelevant to what was actually happening that this new paper about defaults doesn’t address it.

The core motivating myth about the inevitability of the housing bust, and eventually the financial crisis and foreclosure crisis has simply disappeared without anyone seeming to have noticed! At the same time that Ganong and Noel can rightly ignore the issue in their paper, you can go pull up a chair at your local neighborhood bar, mention to the fella on the stool next to you that it should be easier to get a mortgage, and wipe the spittle off your face after being passionately reminded of the last time we let people get mortgages they couldn’t afford.

Caveats: I am not enough of a statistician to fully deconstruct the Ganong and Noel paper, but it does seem like the framework they are using is not necessarily set up to catch interest rate motivated defaults. They are using “above water” defaults as a benchmark, and it is possible that those defaults could include rate-induced defaults. And, while the typical defaulter had negative income shocks, there were some who did not. As they mention, however, their data set cannot capture all income shocks.

I think the fact that, as far as I can tell, they have chosen not to discuss or control for rate-induced defaults, itself, is a sign of how thoroughly the idea has been refuted and departed from the debate about defaults.

There are clues within their data of the unimportance of rates. For instance, Figure A-2 includes the level of mortgage payments, and there is no discernible change in payments near the default periods. They also compare the subset of subprime mortgages to other mortgages, finding that defaults unrelated to negative shocks are only slightly higher than in other mortgages, suggesting that within subprime mortgages, additional defaults related to rate changes or initial affordability do not account for a large proportion of defaulters.

It appears that our understanding of the housing boom and bust has been evolving, though maybe it’s evolving without our noticing.