GSEs, Loan Performance and the Myth of the “Mortgage Meltdown”

by David Fiderer

“At the end of the day what really matters is losses,” said Mark Zandi of Moody’s Analytics. “Where are the losses?”

Zandi, speaking at a recent American Enterprise Institute Book Forum that delved into the root causes of the financial crisis, took a position at odds with others on the panel. Everyone else embraced the research and analysis of AEI mortgage expert Edward Pinto.

Pinto’s infamous 27 million mortgages

According to Pinto, the path to disaster was paved by the Federal Housing Enterprises Financial Safety and Soundness Act of 1992, which set affordable housing goals for the government sponsored enterprises. These goals, which were expanded over time, allegedly poisoned the well, and thereby caused a marketwide deterioration in credit standards.

By 2008 things got so bad that, according to Pinto, close to half of all U.S. mortgages were “subprime or otherwise high risk.” And these subprime-equivalent mortgages made the financial meltdown inevitable. However Pinto and his disciples refuse to discuss how those 27 million mortgages performed after 2008.

The Financial Crisis Inquiry Commission reviewed Pinto’s research extensively. But staffers could find no way to reconcile his risk categorizations to actual loan performance. The FCIC’s findings were soon echoed by the research of David Min, then of the Center for American Progress. Four years later, Zandi’s updated analysis confirms and validates the earlier assessments made by Min, and almost all FCIC commissioners.

The one FCIC holdout, the Commissioner who wrote a scathing rebuttal to the Majority Report, was Pinto’s colleague at AEI, Peter Wallison.

A library of Wallison/Pinto endorsements

“Whenever he had the opportunity, [FCIC Chair Phil] Angelides would point out that nine of the ten commissioners disagreed with my analysis,” writes Wallison in his book, Bad History, Worse Policy, published almost exactly two years ago. “Nevertheless, by the fall of 2011…virtually all the Republican candidates for president were saying in debates and elsewhere that the financial crisis was caused by government housing policy, with Fannie and Freddie at its heart.”

Now, four years after the FCIC Report, Wallison can point to a long list of mortgage experts who wrote books based on Pinto’s research. All of these authors, like Wallison, deem the FCIC analysis to be illegitimate and therefore unworthy of any direct response. As it happened, the recent AEI Book Forum was occasioned by the launch of Wallison’s latest volume on the subject, Hidden in Plain Sight, which covers the same ground as Bad History.

Wallison follows in the footsteps of professors at the University of Chicago, Columbia, Stanford, and four professors at NYU. Other mortgage experts, at the Hoover Institution, the Discovery Institute, the Mercatus Center, plus a former member of Britain’s Parliament, also wrote very similar tomes. The author of Hidden in Plain Sight can cite at least ten other books that rely on Pinto’s research to declare that the financial crisis was caused by affordable housing goals.

All these authors claim that close to half of all U.S. mortgages, or 27 million, were subprime-equivalent. And none of them take Zandi’s approach, which is to compare GSE loan performance with the rest of the market. In fact, Paul Sperry, of the Hoover Institute, trashed the FCIC because it called Zandi to testify as a “star witness.” (If I didn’t know better, I might think that there was some kind of coordinated messaging strategy, like that of a political campaign.)

But where are the losses?

As the one Pinto-skeptic in the room, Zandi proceeded to answer his own question, “Where are the losses?” As of year-end 2013, approximately $1 trillion in credit losses on pre-crisis loans had been realized. But the realized loss rate among different sectors varied considerably. Best in class were Fannie and Freddie, with a realized loss rate of 3%. Then came depository institutions, like banks, which had a realized loss rate of 6%. The strong outlier was private label mortgage securities, with a realized loss rate of 23%, seven times that of the GSEs.

These numbers are in line with Laurie Goodman’s 2010 projections, which showed a 24% overall loss rate on private 1st lien securities. And Zandi’s 2013 numbers are consistent with his year-end 2012 numbers, which showed private label losses as 51% of the grand total, and GSE losses as 14% of the nationwide total.

These lopsided disparities are confirmed over and over from data going back two decades. By any standard — delinquencies, defaults, loss severity — GSE mortgages perform exponentially better than the rest of the market, whereas private label mortgages perform exponentially worse. To state otherwise is to lie.

Wallison: Affordable housing created subprime securitizations

Not so fast, responded Wallison, who claimed that private label securities should not be viewed as severable from the GSEs. “Fannie and Freddie were very eager to have these eligible loans under the affordable housing requirements, and the private sector was assembling them,” he said. “They were creating pools of mortgages, all of which were eligible for the affordable housing requirements. And because of that, it looked as though the private sector was doing a lot more than Fannie and Freddie were doing. But in fact, the private sector it was simply selling to Fannie and Freddie in those years. And that is why, Fannie and Freddie’s share looked less in terms what they were putting out, and the private sector like [it was] more.”

Not really. During the halcyon days of 2005–2007, about $2.9 trillion private label single-family mortgage securities were issued, of which about 15% were purchased by Fannie and Freddie.

Moreover, it’s hard to see how all these subprime securities met affordable housing goals, since the 2000 regulations specifically exempted “B&C loans,” aka subprime, from affordable housing calculations (65 FR 65090). Moreover, the GSEs were constrained by “ability to repay” and other anti-predatory rules (65 FR 65085), which disqualified of a huge part of the subprime market from affordable housing goals.

In mortgage finance, subordination is everything

Also, you need to repeat Zandi’s question and go deeper. “Where are the losses,” as they pertain to private label securities? The simple and straightforward answer is the subordinated tranches, those rated below triple-A. In mortgage finance, if you don’t understand seniority and subordination, you are utterly lost.

As the FCIC graphic below illustrates, the capital structure of any deal was overwhelmingly weighted toward tranches rated triple-A. Yet, as of year-end 2009, the vast majority of non-triple-A tranches, and most CDO triple-A tranches, got wiped out.

Subordination explains why GSEs’ purchases had a nominal impact on the expansion of subprime securitizations. Fannie and Freddie were very specific about the triple-A tranches they purchased, limiting acquisitions to the most senior triple-A tranches, as illustrated in an FCIC graphic of one subprime securitization, CMLTI NC-2006. As you can see, Fannie Mae bought the top $154.6 million slice, which was senior to $580 million other triple-A tranches, and senior to all the other lower-rated tranches.

The GSEs purchased tranches that got repaid before all the others, and consequently had average lives of less than one year, until the end of the boom. As the GSEs kept buying more and more private label single-family mortgage bonds, their year-end holdings never increased. At year-end 2004, Fannie and Freddie held about $233 billion in single-family bonds. Three years later, after buying another $433 billion in single-family bonds, the GSEs’ year-end holdings were slightly down, to about $223 billion, or 10% of $2.2 trillion total private label bonds outstanding.

Just as the most senior tranches can repay very quickly, the most subordinated tranches can get wiped out very quickly. Each tranche is measured by a net present value calculation, which is updated monthly. When you have a static mortgage pool in liquidation, which is 100% debt-financed, the tranches rated below double-A, those that are subordinate to about 90% of the debt, allow scant margin for error. In fact, the data showed that these subordinated tranches could not withstand a normal cyclical downturn.

Time is money. If you’ve ever done a net present value calculation, you know it doesn’t take much tweaking in the first few years’ cash flows to change a positive outcome into a negative outcome. The impact of subordination on the non-triple-A tranches, sliced twelve ways in CMLTI NC-2006, was to rapidly accelerate the timeframe in which they got wiped out.

Too Big To Fail means, among other things, too big to fail quickly. The financial crisis was caused, not by mortgage loans, which deteriorate gradually over an extended period, but by deeply subordinated mortgage securities, which got wiped out as soon as home prices started falling. (Notwithstanding delayed announcements of rating downgrades.)

You know who overlooked the dangers of subordination? Not the GSEs, but the companies who relied on the rating agencies, which pushed fanciful notions about risk diversification in CDOs. In spite of evidence to the contrary, the rating agencies asserted that they could accurately measure the safety of a single portfolio packed with deeply subordinated triple-B tranches of different private mortgage deals. Using their own flawed logic, the rating agencies proclaimed the dominant senior chunk of a triple-B RMBS portfolio to be super safe, ergo rated triple-A.

Ordinarily, the triple-B corporate bonds in a big portfolio will not all deteriorate in tandem, and they will not all deteriorate very rapidly. That’s because people, the professional managers who run corporations, have different ways of adapting to changing circumstances. Structured finance deals do the opposite; they are designed to operate like dumb machines. Also, outside of structured finance and possibly banks, there are no triple-B bonds subordinate to 90% of a capital structure with near-zero equity.

Executives who relied on triple-A ratings for CDOs — at AIG, Citigroup, Merrill Lynch, UBS, MBIA, AMBAC — acquired over $250 billion in risk exposure to portfolios packed with deeply subordinated mortgage bonds. Because their CDO risk exposures got hammered so quickly, six huge financial firms faced the specter of liquidity crises and insolvency. And the U.S. government, by announcing its decision to let Lehman abruptly collapse, signaled to everyone else that all bets are off. Panic ensued.

The severe subordinated risk concentrations among major banks and insurance companies had nothing to do with the GSEs, or with affordable housing goals. As you may remember, GSE mortgage securities are subject to corporate guarantees, so the uncertainty among different tranches concerns when they get paid, not if they get paid.

It all becomes obvious once you attempt to answer a simple question, “Where are the losses?”

Don’t forget accrual accounting: One other point. Wallison found Zandi’s numbers, most notably the GSEs’ realized loss rate of 3%, to be unfathomable, given the size of the $187 billion government “bailout.” Zandi reiterated that he was speaking about realized losses, on loans that were no longer on the books. Wallison insisted that the GSEs’ GAAP losses were in fact realized losses, despite massive reversals in 2013, when most of those unrealized losses proved to be illusory. It was neither the time nor the place to tutor Wallison on the basics of accrual accounting for financial institutions.