PrefBlog

I gave an example of tranching when looking at the Bear Stearns product a few days ago. Now I want to clear up some possible confusion regarding the practice.

Quality is Very Expensive

Let us say, just for the sake of an argument, that we have a pile of AA rated securities that we want to securitize. We have two choices:

• We can securitize them in one big bucket, rated AA, or
• We can divide it in two tranches. The first gets priority in distributions and will be rated AAA; the second will be junior and rated A

The second choice will be familiar to preferred share afficionados – it’s the same process that is used in split share corporations. The holders of the senior tranche (the preferred shares) don’t really care a lot about the underlying portfolio, as long as it’s reasonably good quality and there’s a lot of it! The junior tranche holders care a lot, because they’re taking the risk they’ll get paid less than expected in exchange for an expected reward of getting paid more.

It will always be more profitable to the underwriter to split the issue because bond investors pay up for quality. I might get 10 cents less than base price for the junior tranche, but I’ll get 25 cents more for the senior tranche.

This is even more important with the mortgages, because it’s not a 50-50 proposition … there is a better than even chance any particular mortgage in the underlying pool will behave exactly as expected. There is, shall we say, some debate over just what the default probability is, but let’s make up some numbers. The mortgages are all one-year term. I expect  10% of the underlying to disappear completely; interest is received equal to 20% of the original pool; therefore, my return on the overall pool will be 10%.

These numbers are obviously very exaggerated. If anybody wants to make up better numbers – or, even better, wants to dig through ratings reports to get actual numbers – be my guest and let me know what you come up with. Put it in the comments or eMail me and I’ll put it in the post.

OK, so I look at the market and I see that AAA instruments yield 5%, while A instruments yield 20%. So what I do, is I create two tranches. The first tranche is senior and comprises $80 of the original pool of $100. It gets an AAA rating because the pool as a whole can lose 20% of its original value and this tranche won’t even notice. That’s double the loss rate I expect.

On the senior issue, I pay only 5% interest, which comes to $4.

The second tranche, worth $20, takes the first loss, which is expected to be 10% of the original pool, or $10. If the end value of the second tranche is $24 at the end of the year, it will have yielded 20% on invested capital. So on this tranche, I pay $14 from my interest receipts and the expected value of the tranche is $20 (invested) – $10 (loss on pool) + $14 (interest) = $24.

Now I work out my expectations: I’m going to receive $20 interest from the underlying pool. I pay $5 interest to the first tranche and $14 to the second tranche. Hey, looky looky! There’s $1 left over! I’ll invent an IO (Interest Only) tranche to receive that interest and keep that tranche for myself!

Bingo! Financial alchemy!

Why is Quality Expensive?

The basic reason is segmentation, which will be familiar to readers of my paper on Portfolio Construction. There might be some investors who don’t have a portfolio big enough to diversify. Maybe they can buy only one bond; maybe they’re just a little bigger and are only buying 5 bonds to make a ladder. Five bonds is not a lot. If one goes bad, that’s 20% of the portfolio. Such players, too small to diversify away specific risk, should stick to higher quality instruments so that their portfolios have a greater chance of behaving as expected. So these players are logically restricted to higher quality instruments and shouldn’t invest in our junior tranche. They have to buy the senior tranche virtually irrespective of how much extra they could expect from the junior, because they are highly risk-averse.

Another reason for quality segmentation is fiduciary policies. Maybe the East Podunk Widows’ and Orphans’ Fund has a set policy: AAA only. They’ll go to jail if they buy our junior tranche.

There’s liquidity as well. The senior tranche is worth $80, the junior tranche $20. It’s not unreasonable to expect four times as much trading of the senior tranche in the secondary market. Market timers, for instance, will have a prediliction for the senior tranche because they’ll be able to trade out of it more cheaply if the market moves their way and they want to realize a capital gain.

All these factors conspire to ensure that interest rates on the lower grades of paper will (normally) be higher than they would need to be if default risk was the sole consideration. Therefore, an investor who (i) has no arbitrary limits, and (ii) can purchase a well diversified portfolio and (iii) has no intention of trading a lot can (normally!) achieve excess returns by moving down the quality scale.

This was exploited in the late ’80’s, when the junk bond market was invented.

Sorry about all this sub-prime stuff in a blog that usually sticks pretty close to its knitting, but this is a lot of fun!

Global Diversified Investment Grade Income Trust (TSX: DG.UN) (“Global DIGIT”) has announced:

that, considering the liquidity problems of MMAI-I Trust (“MMAI”) due to its inability to roll its maturing commercial paper in the present state of the Canadian asset-backed commercial paper market, it will not have sufficient financial resources to allow for the payment of the redemption price for the redemption of Units and consequently must announce the suspension, until further notice, of the August 31st, 2007 annual and quarterly redemptions of Units.

This is a fascinating investment scheme. What they have done, according to their financials is (with lots of rounding by me):

– taken $89-million of unitholders equity

– borrowed about $1,400-million, mostly as commercial paper

– put all this money into term deposits in banks, at the Bankers’ Acceptance rate plus a spread.

– written Credit Default Swaps against a portfolio of mainly mortgage backed securities and pledged the term deposits as security (note that writing a CDS gives you exposure and, hopefully, yield basically equivalent to what you wrote the CDS on)

The point? It’s discussed:

The Trust’s objective (save for any loss exceeding the first loss amount) is to provide a return on investment of 5.94% per annum to Unitholders up to September 7, 2009 and thereafter a floating distribution equal to the rate of bankers acceptance plus 2%.

Units cost $10.00 at issue in September 2004, of which sixty-five cents went to start-up costs. The provide a handy computation of their returns since inception, using a starting point of $9.35 net … they’ve made about 5.5% annualized.

It seems to me like a helluva complicated & (term-mismatch-) risky & very highly-leveraged way to go after BAs+200, but I should have said that three years ago if I wanted to be taken seriously.

And now they’re having a little difficulty rolling their CP, have suspended redemptions and I see that DG.UN is now quoted at $2.90-99 on the TSX, compared with a NAVPU of $9.17 reported as of 2007-6-30.

Y’know, the underlying doesn’t look all that terrible to me … having looked at it very, very briefly and with no intent of investing or recommending. The big question is how much of the lolly the CP holders are going to grab and, frankly, I’m not going to rip apart the prospectuses trying to find out. I’m not going to pay for any expensive legal opinions, either! But it does seem to me that this very, very distressed security that has enormous liquidity problems would be worth looking at … although, until we know more about what the CP guys are going to take, it’s a wild speculation and, what’s worse, blind.

Too bad the damn thing’s $1.4-billion … that’s getting into serious money that will be hard to finance in this environment … which, I imagine, is part of the problem. If it were smaller, it would be (I think) attractive to the hedge fund crowd. If any of my readers has $1.4-billion they want to put into a nice floating rate note with a term of nine years, call me and maybe we can help these guys out a little … after looking at this stuff a whole lot more closely, of course!

Hat tip : Financial Webring Forum.

Gee, the last one was so much fun I think I’ll do another! Let’s look at “Bear Stearns Asset Backed Securities Trust 2005-1”.

On August 24, S&P released the following:

Standard & Poor’s Ratings Services today lowered its ratings on seven classes from Bear Stearns Asset Backed Securities Trust’s series 2001-3, 2005-1, 2005-2 and 2005-3 transactions (see list).
     The lowered ratings reflect pool performance that has caused actual and projected credit support for the affected classes to decline considerably. All four transactions have experienced losses that have eroded overcollateralization (O/C) to levels that are significantly below their targets. Furthermore, delinquencies have escalated over the past six months. For series 2001-3, losses have, on average over the past six months, been approximately 2.10x monthly excess interest. For series 2005-1, losses have, on average over the past six months, been approximately 3.73x monthly excess interest. Severe delinquencies (90-plus days, foreclosures, and REOs) for the transactions, as a percentage of the current pool balances, range between
approximately 17.09% (series 2005-1) and 13.40% (series 2001-3). Realized losses for the transactions, as a percentage of the original pool balances, range between approximately 9.20% (series 2001-3) and 1.33% (series 2005-3).
     These performance trends have caused projected credit support for the transactions to fall well below the required levels. Standard & Poor’s will continue to closely monitor the performance of these transactions. If the transactions incur further losses and delinquencies continue to erode projected credit support, we will take further negative rating actions.
     Subordination, excess interest, and O/C provide credit support for the two transactions. The underlying collateral backing the certificates consists of both fixed- and adjustable-rate mortgage loans.

RATINGS LOWERED
  
Bear Stearns Asset Backed Securities Trust
Residential mortgage-backed certificates
                               Rating
Series      Class       To              From
2001-3      M-2         BB              A
2001-3      B           B               BBB
2005-1      M-6         BB              BBB-
2005-1      M-7         CCC             BB
2005-2      M-7         B               BB
2005-3      M-6         B               BBB-
2005-3      M-7         CCC             BB

… which looks pretty horrific. But now let’s look at ALL of BSABST 2005-1:

 US$395 million asset-backed certificates, series 2005-1 
Class  Maturity Date   Rating  Rating Date    
 
A  Mar 25, 2035  AAA  Feb 24, 2005    
 
M-1  Mar 25, 2035  AA  Feb 24, 2005    
 
M-2  Mar 25, 2035  A  Feb 24, 2005    
 
M-3  Mar 25, 2035  A-  Feb 24, 2005    
 
M-4  Mar 25, 2035  BBB+  Feb 24, 2005    
 
M-5  Mar 25, 2035  BBB  Feb 24, 2005    
 
M-6  Mar 25, 2035  BB  Aug 24, 2007    
 
M-7  Mar 25, 2035  CCC  Aug 24, 2007    
 
R-I  Mar 25, 2035  NR  Feb 24, 2005    
 
R-II  Mar 25, 2035  NR  Feb 24, 2005    
 
B-IO  Mar 25, 2035  NR  Feb 24, 2005 
 

The SEC ID for this trust is 333-113636, for those who wish to see all the gory detail. Tranche sizes, from the prospectus on SEC / EDGAR are (this is SEC document 0000911420-05-000084.txt : 20050216) are:

Class A : $313,746,000 (pays LIBOR + 0.35% before optional termination / +0.70% afterwards)

Class M-1: $37,689,000 (+0.70% / +1.05%)

Class M-2: $18,549,000 (+1.40% / +2.10%)

Class M-3: $4,341,000 (+1.60% / +2.40%)

Class M-4: $3,946,000 (+2.20% / +3.30%)

Class M-5: $2,960,000 (+3.00% / +4.50%)

Class M-6: $4,538,000 (+3.50% / +5.25%)

Class M-7 was not offered in the prospectus. R-I and R-II are “residual interests in the real estate mortgage investment conduits established by the trust”, and were also not offered. B-IO gets all the Excess Spread. None of these last three classes had a stated principal value; I’m not going to tear apart the prospectus analyzing them because I don’t really care a lot how they work … I’m just after the principal values here!

The “optional termination” becomes effective “when the stated principal balance of the mortgage loans and any foreclosed real estate owned by the trust fund has declined to or below 10% of the stated principal balance of the mortgage loans as of the cut-off date”. At this point EMC Mortgage corporation could purchase all the assets.

At any rate, it should be clear that – while downgrades are always bad, and to be deplored by all right-thinking people – the downgrades that sounded so awful at the beginning of this post lose a lot of their ability to terrify when put into perspective.

Perspective is what’s needed when thinking about sub-prime … and I can’t provide it. I’m not a specialist, and I think a specialist would need a pretty good database to get it. What I really want is a transition study that has dollar figures attached, not just number of ratings. It would be nice, too, if interest rates could be attached to such a study … because, well, gee, the guys in the downgraded class M-6 were getting LIBOR + 350bp (and still are, since the issue is not in default)!

I’ll keep my eyes out, however, and whenever I see something interesting, I’ll post again.

Update 2007-09-18: I became involved in a discussion at Econbrowser in which this issue came up. In the course of the discussion I retrieved a bit more information from the prospectus, which I shall reproduce here:

The following table summarizes certain characteristics of the mortgage loans as of the cut-off date:

Number of mortgage loans……………………3,527
Aggregate principal balance…………..$394,649,130
Average principal balance………………..$111,894
Range of principal balance………$1,041 to $800,000
Range of mortgage rates……………0.00% to 16.50%
Weighted average mortgage rate……………..8.032%
Weighted average combined loan-to-value ratio……………………84.50%
Range of scheduled remaining terms to maturity……….9 months to 361 months