"Fundamentals, not liquidity conditions, are behind MBS crash"

Hoisted from comments by a helpful anonymous reader was a revealing post yesterday at FT Alphaville that we somehow missed. It’s a important addition to the discussion over the stress in the US housing and mortgage-backed securities markets.

CreditSights, an award-winning provider of credit research, disputes the commonly held view that the sharp falloff in the mortgage debt market is due to a bit of fundamental deterioration compounded by considerable, and somewhat unwarranted, panic. According to Alphaville (sadly, the research is restricted; we’d love to read it ourselves), CreditSights believes that the current pricing of MBS is in line with fundamentals. The big reason for their harsh view is what they call the “severity loss ratio” which I assume is roughly equivalent to recovery net of expenses. They inspected a series of deals and found the loss ratio to be weighted average of 35%, which is truly horrid. Historically, recoveries on mortgage foreclosures are more like 70% (I believe that is pre-expenses), and it’s a certainty that levels like that were used in structuring these deals.

And in a related development, Fitch released a report that reaches the not-surprising conclusion: deterioration in US subprime RMBS will lead to rising risk of events of default for collateralized debt obligations.

From FT Alphaville:

But, said CreditSights, in a note to clients on Wednesday, current pricing levels reflect fundamentals, even for the most highly-rated debt. Mortgage securities across the board are overrated and overvalued:

The harsh truth about the outlook for the AAA tranches – necessary downgrades, if not defaults – should put the lie to the argument that current low prices in AAA RMBS tranches – let alone AAA tranches of mezzanine RMBS CDOs – are somehow the victim of poor liquidity conditions, and do not reflect the true fundamentals of the situation.

CreditSights publish the results of a survey they have conducted on “188 individual relatively large RMBS deals”. The outlook, by all accounts, is grim.

At root, CreditSights calculate a severity loss ratio for lenders on individual defaulting subprime mortgages based on mortgage market data collected over the past few weeks. The survey results indicate that such loss severity rates on mortgages are “painfully high”. They range from 24 per cent to 55 per cent – with a weighted average at 35 per cent. And they’re expected to rise. For second-lien mortgages – that is, second mortgages on a property, the loss severity rates average 94 per cent.

So how do those figures translate into the capital structure of structured mortgage-backed debt? Foreclosure rates are rising higher and higher – which means the number of occasions when the above loss severity ratios have to be applied are increasing.

And it doesn’t look like the blame can be pinned on any particular vintages of MBS. Here’s a graph of foreclosures on vintages since 2004:

According to CreditSights, that should “up-end the idea that the 2004 vintage was perhaps sufficiently seasoned and composed of loans that had enjoyed enough home price appreciation since 2000, to avoid any further erosion.”

As it is, foreclosure rates are hovering at around 13 per cent on 2005 and 2006 mortgage debt. But CreditSights say there is “no end in sight” when it comes to that figure rising.

Consider then the outlook for delinquancy rates – a measure of mortgage loans not yet in foreclosure, but in trouble:

Add the 7 per cent delinquency rate for the 2006 vintage to the 2006 foreclosure rate at 12.6 and it’s already close to 20 per cent.

How then does that translate into the world of structured finance, and those RMBS tranches?

To trigger a default on the most secure subprime RMBS debt – rated AAA, and structured with a typical 18 per cent attachment rate – foreclosure rates would have to reach the 30 per cent.

As can be seen from the results of CreditSights’ survey, that scenario is indeed becoming “less and less unthinkable”. Adding the foreclosure and delinquancy rates takes us close to 20 per cent. Both are set to increase. Then there’s those painfully low severity loss ratios. Add it all together and that AAA debt is far, far, far from safe.

And we haven’t even mentioned prime tranches lower down the structure.

Far from mispricing RMBS, CreditSights even go so far as to suggest that actually, the ABX indices (which list AAA RMBS debt at around 80 cents in the dollar) are throwing up some pretty appropriate figures.

This 20% default figure is no surprise; in fact it is likely to be low due to vagaries in loan classification. We featured this quote from IndyMac Bancorp CEO Michael Perry back in May:

On subprime loans, one of the things that I think people aren’t aware of is that the Mortgage Bankers Association basically classifies the lender as a prime lender or a subprime lender. So for example, they classify IndyMac and Countrywide as prime lenders, and they classify New Century or whoever as a subprime lender. And all of their servicing portfolio is considered prime or subprime for the MBA. Ok? And so when you see that delinquency number in the press of 13% subprime delinquencies, it’s hugely understated. It is absolutely hugely understated. And the prime delinquencies are overstated.The subprime delinquencies are more like 18, 20, 22% delinquencies and that’s where I think you’re going to see the problems.

Print Friendly, PDF & Email


  1. jck

    Unfortunately for the “award-winning provider of credit research”, loss severity is a pretty useless metric.Cumulative net loss is much better.
    I’ll give you a link to the worst 2004 vintage I could find with loss severity at 71.54%, that translates into 0.91% cumulative net loss.Ouch …now I am really worried.Btw, that pool is 85.8% paid off so it will hard for investors to lose their shirts.

  2. Anonymous

    Not sure about this cut and paste, but any comments?

    Legally, a CMO is termed a special purpose entity that is wholly separate from the institution(s) that create it) for the last two and a half years and it may have sold about $100 billion in CMO’s in that period, according to ABAlert.

    As such, synthetic derivatives are almost never consolidated on the books and therefore, these bad bets are impacting future value to a degree never imagined in any risk models!

  3. EEngineer

    Wow that’s one scary ugly plot. It looks like the pressure wave gradient in front of a shock wave. Hmm… Come to think of it, it might be just that… Sonic boom approaching!

  4. Anonymous

    That Katrina-like sonic boom is going to be the giant sucking sound of America being sold off:

    And right now Persian Gulf countries are putting their cash to work through newly minted sovereign wealth funds (SWFs), which operate much like state owned hedge funds or private equity groups.

    In the past six months alone, Middle Eastern SWFs have doled out serious cash for stakes in major international corporations. Just take a look:

    Nov. 27: Abu Dhabi pours $7.5 billion into ailing Citigroup Inc. (C), which recently lost its status as largest bank by market capitalization to Bank of America Corp. (BAC).
    Nov. 26: Dubai International Capital, a state-owned holding company, acquired an undisclosed stake in Japan’s electronics and media juggernaut Sony Corp. (SNE).
    Nov. 16: Abu Dhabi invested $622 million (an 8.1% stake) in California-based microchip-maker Advanced Micro Devices Inc. (AMD).
    Oct. 20: Dubai International Capital agreed to invest $1.26 billion in the initial public offering of hedge fund Och-Ziff Capital Management Group LLC (OZM).
    Aug. 22: Dubai World, another investment arm of the state, plunked down $5.1 billion for a 9.5% stake in MGM Mirage (MGM).
    Aug. 14: Istithmar, part of Dubai World, was cleared to buy Barneys New York Inc. for $942.3 million from Jones Apparel Group Inc. (JNY).
    May 21: General Electric (GE) sold its plastics division to Saudi Basic Industries Corp. – the country’s largest public company, though 70% owned by the government – for $11.6 billion.

  5. Yves Smith


    As I indicated, I don’t have access to the underlying research, and I always feel a bit exposed when I rely on third party comments (unless the analysis is pretty straightforward).

    Even if you don’t agree with the CreditSights methodology (or more cynically, perhaps they chose to publicize a metric that would lead to more dramatic-sounding findings) does that invalidate their logic?

  6. Anonymous

    Re: Unfortunately for the “award-winning provider of credit research”, loss severity is a pretty useless metric.Cumulative net loss is much better.
    I’ll give you a link to the worst 2004 vintage I could find with loss severity at 71.54%, that translates into 0.91% cumulative net loss.

    Keep digging for older issues mate and then go into other layers!


  7. Anonymous

    Oh, its the shape of the loss curve..ohhh, I guess that would explain the hyperbolic upward trend versus the non-model data, or the old model which is now not in the previous curve; however, if we go back 50 years and smooth out the past year of the trillions lost in bad bets, then in 10 years all is well!

    Re: Static pool data, if available, is compiled by
    taking a discrete period of originations of the originator, such as a
    financial quarter, and that pool’s performance is tracked on a monthly
    basis as the loans amortize, particularly focusing on loans which have
    been outstanding (seasoned) 18-24 months and have been substantially
    paid down. This allows a determination of the shape of the loss curve
    and project timing of losses to be made. The cumulative net loss on the
    less seasoned pools can then be extrapolated from the older pools.
    Static pool data is preferred over active pool data, which can mask
    losses during periods when the originator’s pool of loans is rapidly

  8. jck

    It does invalidate their logic, because the losses when they come[and they will come I am not disputing that] will come when most of the structure is paid off.
    The 2004 vintage are 85% to up 95% paid off.The early 2006 are 40% paid off.The timing of the losses is extremely important and while you get some pretty fat yields.
    None of the trancghes on the ABX has lost a cent in principal writedown so far and you get libor +35% on the cash bond while you wait for the day of reckoning.

  9. Anonymous

    Losses when they come due?

    You swap the word structure for bet, i.e, like these are mortgages with versus bad bets that are being called before the “fat yield” bait is dumped on to some idiot that thought the structure was backed by some theoretical AAA rating from some bozo paid off by a shill like you!

Comments are closed.