GSE Reform Advocates Ignore Risk Diversification

Mortgage lending is about risk diversification. As his latest blog demonstrates, few write about the subject with as much eloquence and precision as my friend, former Fannie CFO J. Timothy Howard.  Whereas most advocates of GSE reform and GSE risk sharing transactions seem to ignore the issue. Maybe that's because their proposals, when evaluated in terms of risk diversification, lose credibility.

Most reform proposals, and the GSE risk sharing deals, embrace the same premise upon which private label residential mortgage backed securitizations (PLS) were created. It's the belief that an investment banker knows how to diversify the risks in a static pool of mortgages in liquidation. Supposedly, he can structure the deal in a way that assures all investors will get their money back with interest. This premise has been discredited. The math doesn't work. Which is why a popular GSE reform proposal, to have the Federal government guarantee these types of PLS transactions, above a predetermined first loss amount assumed by private investors, is folly.  

Remember, home mortgage lending is the only major credit market that relies on asset appreciation. The last 15 years have demonstrated that, post closing, the rate of home price appreciation--positive or negative--is the biggest driver in loan recovery.

For PLS, market timing is everything; and there is no way to adequately diversify that risk, even when loans are carefully distributed among different zip codes.  Why? Because the outcomes of PLS transactions are driven by two different industry cycles: the cycle in mortgage originations and the cycle in home price appreciation. When Lewis Ranieri and others invented PLS the 1970s and 1980s, they were affected by only one industry cycle, the cycle in home prices. The addition of the second industry cycle, which commenced in the early 1990s, invalidated the premise of PLS.

It's Always An NPV of A Mortgage Pool In Liquidation: Let's recap the basics. An investment banker structures a PLS mortgage pool, from which he estimates future net interest income and future credit losses.  The net present value is the combination of net interest income (or excess spread) and credit losses. Every NPV calculation counts your chickens before they hatch, and in the early 1980s it seemed reasonable to estimate net interest income based on statistical data on prepayments.

Back then, you could examine historical data to extrapolate trends because the prepayment variables were demographic: borrowers who sold their homes and borrowers who lost them in foreclosure. Back then, almost nobody refinanced his home loan, because mortgage rates never went down. Prior to 1987, second lien financing was less attractive because credit card interest was also tax deductible.

But all that changed in the early 1990s, when Alan Greenspan temporarily slashed interest rates to boost the economy. In so doing, the Fed Chairman triggered the first in a succession of boom/bust refinancing cycles. Huge numbers of homeowners refinanced, and thereby damaged the NPVs of different PLS pools, which suffered shortfalls in interest income. Remember, every static mortgage pool goes through a death spiral; the volume of interest-generating loans shrinks every month because of scheduled amortization and borrower prepayments. If the pool shrinks at a faster than expected rate, the shortfall in interest income is permanent 

As with almost any NPV calculation, disruptions in the early years have the greatest impact. So newly-minted PLS transactions, with bigger mortgage pools, were more likely to suffer bigger shortfalls. And, unlike demographic data, you cannot extrapolate trends to predict when the Fed will slash interest rates.

Faster-than-expected prepayments represent an adverse selection within the shrunken mortgage pool. Borrowers with robust home equity tend to refinance when it's advantageous; whereas borrowers with declining home equity tend to stick around and are more likely to default.

No matter how carefully an investment banker assembles a mortgage pool, its credit risk diversification will continually change in unpredictable ways. And when the pool changes for the worse, nothing can be done about it. There are no do overs with these structured finance entities, which are designed to operate like dumb machines.

Lewis Ranieri identified the problem in his famous 1994 lecture at Northwestern. “We did not build the system to finance refinancing," he said. "We built the system to finance housing.” He concluded that, "We have damaged the basic structure of the new housing finance system." But no one grappled with the problem that he raised.

PLS vs. GSEs, or Dumb Machines vs. Human Beings: The difference between PLS and GSE securitizations is no less profound than the difference between dumb machines and human beings. Corporate guarantors, such as the GSEs, are run by people who adapt to changing circumstances. The GSEs, like every bank, manage and rebalance their risk diversification to avoid excessive concentrations in any one area.

In stark contrast to PLS, the GSEs, like banks, face limited harm from any refinancing boom. The GSEs replace prepaid mortgages with new ones. Earlier-than-expected prepayments on one GSE securitization won't impair the guarantor's ability to cover future credit losses.  And no one has ever diversified credit risk as broadly and as successfully as the GSEs 

The PLS structure seems inherently unstable when you look at the duration mismatch. These deals finance 30-year amortizing loans to finance houses with economic lives that exceed 30 years. But the average life of a PLS deal may be under five years.

A Duration Mismatch Between Housing Cycles And PLS:  Which brings us to the really big problem posed by housing cycles, which are very, very long. If you bought a house in Los Angeles in 1990 and sold it in 2000, the cumulative appreciation on your 10-year investment would be zero, according to Case-Shiller. Of course, if you sold your house at the end of 1995, you would have suffered a 25% loss. From 1996 to 2006. California experienced an unprecedented housing boom,

Rating agency models had an identical 10-year run. The method by which the rating agencies rated PLS deals, using models that weighed FICO scores very heavily, began in 1996, when Standard & Poor's rolled out its LEVELS 5.0 model. Fitch came out with a similar model soon thereafter. Moody's followed in early 2001. The defects in these models became obvious to everyone in 2006. The housing bubble had concealed a multitude of sins.

The rating agencies like to say they rate through the cycle, meaning that a credit rating — which is supposed to remain stable over the life of a bond — factors in the cyclical highs and lows experienced by a business. That concept works when you're rating companies, which have indefinite lives. That concept works if you are rating commercial real estate, which has a very long economic life. But that idea is an oxymoron when you're talking about PLS transactions, which go through fairly brief death spirals.

Remember, the simple investment thesis of The Big Short?  It was that the triple-B tranches of subprime PLS deals could not survive a soft landing in home prices. And since history showed that virtually every real estate boom was followed by an extended period of flat or declining prices, the investment seemed like a sure thing.

Hollowing Out, Then Abolishing, The GSEs In Favor Of PLS: Corker-Warner, Johnson-Crapo and their progeny proposals—including ”A More Promising Road Toward GSE Reform,” and “Toward a New Secondary Mortgage Market ”— seek to dismantle the GSE system, which continually manages credit risk on a global basis and attained the best loan performance anywhere. These reform advocates believe they can create an efficient market for privately-owned first-loss pieces of individual transactions. But during the bubble years, the subordinate tranches of PLS deals were never actively traded; almost all of them were stuffed into CDOs. Ten years after the rating agency models failed, the PLS market is comatose and the CDO market is dead. As Jim Parrott points out, GSE reform is synonymous with reviving the PLS market. 

GSE risk sharing proponents want the companies to do more and more deals that hollow out the companies' ability to manage global credit risk. Tim Howard, who reviewed some of these recent deals, says the GSEs pay exorbitant fees to Wall Street banks, which assume risks that statistically equal zero. Mr. Howard does not use the terms "sham transaction" or "gift to Wall Street." I do.

9 comments - GSE Reform Advocates Ignore Risk Diversification

Leave a reply