July 28th, 2014
by Keith Jurow, Capital Preservation Real Estate Report
Non-Agency Mortgage-Backed Securities
Non-Agency residential mortgage-backed securities (RMBS) are securitized mortgages that are not guaranteed by Fannie Mae or Freddie Mac or insured by the FHA.
These non-guaranteed RMBS existed prior to the bubble years of 2005 - 2007, but the outstanding amount was relatively small. By early 2004, however, that number had climbed rapidly to $644 billion.
Then the speculative mania began to really heat up. Subprime lenders enlisted an army of 50,000 mortgage brokerage firms to hawk loans to just about anyone who was breathing and could sign their name.
With Wall Street frantically securitizing these loans and the three large rating Agencies giving their stamp-of-approval with what later turned out to be highly-questionable AAA ratings, underwriting standards had completely disappeared by mid-2006.
By the end of 2006, speculation reached unheard of proportions and underwriting standards had totally collapsed. Mortgages for $500,000 or more with no down payment were commonplace. Borrowers could get mortgages even when their total debt-to-income (DTI) ratio exceeded 50%. I have written about speculators who were able to buy ten homes or more with little to no down payment on any of them. Nearly the entire nation had become caught up in a frenzy that could only end in disaster.
As major housing markets headed over a cliff, the total amount of outstanding sub-prime, Alt A and other zany mortgages soared into the stratosphere. When mortgage lending finally peaked in July 2007, an incredible $2.3 trillion of non-guaranteed RMBS were outstanding. At the time, no one could really tell what a catastrophe had been created.
Even before the bankruptcy of Lehman Brothers in 2008, homeowners with non-guaranteed mortgages started defaulting in increasing numbers. Servicers responded by modifying mortgage terms to slow down the avalanche of defaults.
Amherst Securities Group (ASG) was the leading firm supplying comprehensive data on the mortgage market. Its spring 2010 report on The State of the US Residential Mortgage Market provided an excellent look at how bad the situation had gotten by the end of 2009.
Using Loan Performance's enormous database on non-Agency securitized mortgages, ASG reported that roughly 613,000 of these mortgages had already been modified by the end of 2009. That was 11% of all the outstanding non-Agency mortgages. Hopes were high that modifications might stop the bleeding.
Unfortunately, borrowers didn't cooperate. They began to re-default on their modified mortgages in huge numbers. ASG reported that by the end of 2009, 61% of borrowers with modified mortgages had re-defaulted within 12 months. What was worse, a third of them had defaulted within 90 days of the modification.
The rate of default for non-Agency mortgages was directly related to how severely the property was underwater. The more underwater properties had become, the higher were the rates of default.
For example, ASG reported that for loans originated in 2006, the annualized default rate for properties with a combined loan-to-value (LTV) ratio of 80% was only 4.7% at the end of 2009. The default rate jumped to 10% when the LTV was 110 - 120% and these homes had gone underwater. With properties severely underwater and the LTV at a sky-high 150 - 180%, default rates soared to 22%.
Regretfully, the rating agencies failed to see this connection until years after the collapse began.
Modifications of non-Agency securitized mortgages have been taking place for more than six years. Take a look at the growth.
You can see that the percentage of non-Agency securitized mortgages which have been modified has risen steadily since late 2008. That includes nearly half of all outstanding sub-prime mortgages.
Has the modification of millions of loans helped to slow down defaults? Take a good look at this very recent TCW graph.
You can see that the re-default rate for the earliest modifications was extremely high. That was because modifications were given to just about any delinquent borrower. Once the servicers became more selective, default rates for the most recent modifications dropped substantially.
Although I have used this graph several times, I now have some reservations about the default rates for 2010 and 2011 modifications. Let me explain why.
A mortgage report issued by ASG early in 2012 contained a similar graph showing re-default rates broken down by year of modification. It showed that for modifications done in 2011, 30% were in default within 12 months. For those completed in the first half of 2011, nearly 40% had re-defaulted within 18 months. Both of these percentages are considerably higher than in the TCW graph. I have a great deal of confidence in ASG and its data. I will let you decide whose numbers may be more accurate.
The important thing to keep in mind is that for the past five years, serious delinquency rates would have been considerably higher had these modifications not occurred. Obviously, we do not know how high the delinquency rate might have climbed. However, it is absolutely clear to me that the delinquency rate announced every month by the Mortgage Bankers Association and others is totally useless for describing the state of the mortgage market.
Downgrades of Non-Agency RMBS by S & P
After the sub-prime collapse began in early 2007, 75% of all sub-prime mortgage RMBS tranches rated by Standard & Poor's (S & P) had been either seriously downgraded from a AAA rating or had their rating withdrawn as of early 2013.
Over the last few years, S & P has continued to badly underestimate the expected default rates on sub-prime RMBS. In September 2012, S & P surprised investors by lowering the ratings of dozens of sub-prime RMBS issued between 2005 and 2007.
A month before the September 2012 downgrade announcement, S & P had carefully explained why it was revising their default projections for mortgages collateralizing non-Agency RMBS. It is worthwhile to carefully review how they explained the change.
This is what S & P said:
"We have increased our roll rate assumptions for loans that are currently delinquent (meaning we expect more of these loans to ultimately default). We are also raising default rate estimates for 'reperforming loans' vis à vis current loans."
S & P was admitting that its previous assumptions on default rates for non-Agency RMBS tranches were much too optimistic. This applied to both delinquent loans and modified loans that had become current.
S & P explained that its new default estimates applied to roughly 80% of all the outstanding RMBS that they rate. This included subprime, Alt A, negative amortization and prime mortgages originated before 2009.
Then they elaborated on how it reached these conclusions. It had performed an "impact study" covering about 10% of all the tranches to which its new estimates would apply. The results of this study suggested that for the sampled tranches, "approximately 30% of ratings will be lowered by four or more notches."
What about the remaining 70% of the sampled tranches? The answer of S & P was kept very unclear:
"The majority of the ratings (approximately 68%) will remain within three notches of the current rating." Would most of these rating changes be upgrades or downgrades? They didn't say. All they stated about upgrades was that "Approximately 2% of ratings will be raised by four or more notches" and that 2/3 of the upgrades will involve "movements from CCC to CC."
My reading of this lack of clarity is that S & P was not prepared to admit that the vast majority of the rating changes coming would be downgrades. Because of my familiarity with how S & P determined its previous default estimates, I am quite confident that most of the rating changes in store for market participants would be downgrades.
What did they say about the likely change to all the tranches it currently rates?
"Higher ratings are more likely to be lowered than lower ratings."
In researching this article, I discovered the following table from a recent analysis by three Michigan State University scholars showing the total downgrades for all non-Agency tranches rated by S & P as of February 2013.
Let's take a careful look at this comprehensive table. Of the 121,584 non-Agency RMBS tranches rated by S & P, more than half were originally given the highest rating at the time the security was reviewed. These were the senior tranches of the RMBS which had the most protection in case of defaults by borrowers.
Notice what has happened to these senior tranches. Only 25% of those initially rated AAA have maintained that rating. One-third of all those originally rated AAA have had such high levels of default that they have had their rating withdrawn (WR). I suggest that you think long and hard about that.
There were still more than 23,000 tranches rated A or better as of February 2013. S & P seemed to be saying that these investment grade securities have the greatest risk of being downgraded. That would spell big trouble for anyone with a portfolio of these higher rated tranches.
Toward the end of its explanation, S & P went on to give specific percentages for what it calls the "base case default frequency assumptions." It stated unequivocally that for any mortgages in non-Agency tranches which were 90 or more days delinquent, it expected that 100% of them would end up in default. Not most of them. Not the vast majority of them. All of them!
In reading S & P's explanation while preparing this article, I was even more shocked than when I first read it a year ago. S & P seemed to be warning investors that thousands of downgrades were to be expected. Yet Wall Street and the media gave the initial 2012 announcement hardly any attention at all. I found it almost by accident while doing research on non-Agency RMBS. I have still not found a single article or report covering it. Amazing!
Back in 2013, I had had a phone conversation with an S & P spokesperson. From him, I learned something else very important. For all non-Agency loans where the loan-to-value ratio (LTV) was greater than 120%, they assumed that 100% of these mortgages would also eventually end up in default.
That conversation just amazed me. For more than four years, I have written about the bubble era mortgages and the collapse in underwriting standards. I have no doubt at all that the vast majority of non-Agency first mortgages originated in the major metros in 2006 and 2007 have LTVs of more than 120%. Just think what this means for the holders of these mortgages.
Take a look at S & P's own analysis of the connection between expected default rate and LTV ratio for 2005 sub-prime loans.
The chart ends at 120% LTV because, as I explained, S & P now assumes that at higher LTV ratios, 100% of the mortgages will default.
Are there certain metros where this will have the greatest impact? According to American Securitization Forum's January 2013 report, 45% of all outstanding non-Agency RMBS mortgages were originated in California, Florida, Nevada and Arizona. The housing markets in these states suffered the largest decline in value when the bubble collapsed. Hence the overwhelming majority of these properties have LTVs in excess of 120%.
Downgrades of RMBS Continue Unabated
Although downgrades of sub-prime tranches have been massive since the collapse, don't think for a minute that these downgrades are winding down.
The June 2014 Mortgage Market Monitor from TCW reported that 34% of all securitized subprime mortgages were either seriously delinquent, in default, in bankruptcy, or already repossessed by the servicing bank. It was only because 46% of all subprime mortgages have been modified that this serious delinquency percentage was not substantially higher.
In April 2013, S & P published a new "US RMBS Recovery Analytics" report. Applying its new rating criteria from the previous year which I just reviewed, the report revealed that of the 7,111 pre-2009 tranches rated AAA in 2012, a mere 1,119 remained at this level. Incredible! Nearly 85% of these tranches have been downgraded. One out of five was downgraded to BBB or lower. It gets worse. More than 10% of them lost their rating completely.
Downgrades by rating agencies continue this year. On May 14, Moody's reported that it had downgraded $216 million of RMBS tranches holding sub-prime mortgages originated in 2003. All eleven tranches had previously been downgraded either in 2011 or 2012.
A Moody's press release described the reason for the new downgrade as "a result of deteriorating performance and/or structural features resulting in higher expected losses for the bonds than previously anticipated." That sounds like S & P's warning from August 2012. Moody's warned that ratings in the US RMBS sector "remain exposed to the high level of macroeconomic uncertainty, and in particular the unemployment rate." Good luck figuring out what that means.
This downgrade of sub-prime tranches is very troubling. Loans issued in 2003 - even subprime mortgages - had much higher underwriting standards than those originated in 2005 and 2006. If performance of these loans is "deteriorating" as Moody's puts it, what does that suggest about the worst-of-the-worst mortgages originated in 2006 and early 2007?
Putting Rating Downgrades in Perspective
Since the vast majority of these formerly top-rated RMBS tranches have been either seriously downgraded or lost their credit rating entirely, how much credibility should be given to their current credit rating? As I see it, not much at all. The fact that so many continue to be downgraded well after the credit crisis was supposed to have ended supports my skepticism.
It seems quite reasonable to doubt the soundness of the current credit ratings. The important question is how does this impact an investor in non-Agency RMBS? Put differently, how can a prudent investment advisor of a client who owns these tranches either through shares of a mortgage REIT or a hedge fund determine what risks they pose to return of capital?
To help you, let's talk a close look at a mortgage REIT with huge holdings in non-Agency RMBS tranches. MFA Financial (MFA) has one of the largest non-Agency RMBS portfolios of all the mortgage REITs.
As stated in its first quarter 2014 10-Q report, MFA held non-Agency RMBS tranches with a market value of slightly more than $5 billion. The majority of loans in this portfolio were originated during the bubble years of 2005 - 2007.
Only one-half of these mortgages were amortizing loans. The rest were interest-only. One-third of the non-Agency mortgages were cash-out refinancing loans taken out by the borrower to tap the rising equity in their home. For the 2006 and 2007 loans taken out by owners with high FICO scores, the outstanding balance averaged more than $500,000. Because of these characteristics, I am almost certain that the vast majority of these properties are badly underwater.
Although the total non-Agency portfolio had an overall serious delinquency rate of 15.8%, it was more than 22% for 2006 and 2007 loans where the borrowers had FICO scores below 715. In the past twelve months, 30% of them were delinquent at some point. Remember this -- Had so many delinquent loans not been modified, the delinquency rate would have been far higher.
In the 10-Q report, there is no breakdown of the non-Agency portfolio by type of loan. So we have no idea what percentage of them are sub-prime, Alt A, Option ARM or prime loans.
The report only categorizes loans originated before 2006 as "2005 and Prior." You must keep in mind that the overwhelming majority of these were originated in the bubble year of 2005. How do we know? In its April 2013 Recovery Analytics report, S & P had explained that only 7% of all 2004 and earlier securitized sub-prime loans remained in RMBS tranches.
The percentage of loans still outstanding is not much higher for other 2004 and earlier non-Agency loans. Only 12% of the Alt A loans remain in RMBS pools, 10% of Option ARMS, and a mere 7% of sub-prime loans.
Why is this important? These 2004 and earlier loans were underwritten with higher standards than the bubble era loans. Even more important is the fact that they had substantially more equity in the property than the 2005 - 2007 loans. The loans remaining in MFA's non-Agency portfolio are there because the borrower's credit was too weak to refinance later or because the property was underwater.
Here is one final scary fact in the 10-Q report. The protection that is built into an RMBS to shield the higher-rated tranches from principal losses is known as "credit enhancement." Losses up to that amount go first to the lower, riskier tranches. In some RMBS, the credit enhancement is 40% or more. However, the average credit enhancement in MFA's portfolio is a mere 2%. Tranches that contain 61% of their non-Agency mortgages have no credit enhancement at all.
That is quite shocking. It means that these tranches are fully exposed to principal losses from defaults. Remember what I said earlier about S & P's latest assumption. They expect that 100% of non-Agency loans where the LTV is more than 120% will default. Cumulative losses in MFA's non-Agency portfolio could be well over $1 billion dollars before the collapse is finished.
There is little in MFA's 10-Q report which would indicate even to savvy investors that these potential losses were likely. Like most investors in mortgage REITs, they would probably focus on the dividend yield and little else. When these losses finally become evident, they would be caught totally unprepared.
One thing that is apparent to even the casual reader of the 10-Q report is that cash and equivalents on hand was cut in half from the previous quarter. Paying the $0.20 dividend in the second quarter will be difficult without additional financing.
My advice to owners of MFA shares is simple. Sell them while the market is liquid and the price is still close to its 52-week high.
The non-Agency RMBS market is still large - roughly $800 billion in thousands of tranches. Even after the sell-off of a year ago, prices on the secondary market are substantially higher than the lows of early 2009.
Wall Street and most investment advisors are quite euphoric about the long-term outlook for the non-guaranteed RMBS market. I have tried to show you compelling reasons why this optimism is completely misplaced.
The crash in housing and in the bubble-era mortgages which enabled the speculative madness is far from over. I strongly urge you to shed the euphoria and act in a prudent, defensive way on behalf of your clients.