|
 Rick's
Picks
An Exceptional Mind Envisions
a Disaster
Rick Ackerman
September 10, 2004
Excerpt from
the current Rick's Picks (website).
You can subscribe here.
I want to call your attention
to a well-reasoned and important think-piece from Howard Hill
that I ran earlier in the week. It was tacked onto the end of
Tuesday's comments but deserves to be featured more prominently
for several reasons. For one, Hill is the only person I know
with the knowledge and imagination to fathom why, in very specific
detail, the derivatives market is a disaster waiting to happen.
For two, his scenario is not just well-reasoned and disciplined,
but plausible; and for three, he is neither a scaremonger with
a book to sell nor a polemicist, just a guy with an extraordinary
mind that is not scared of darkness.
I do not desire merely to be on-record with this watershed essay;
I want to give it a spotlight and a drum roll, since it may well
be the most incisive think-piece we have on a topic whose details
have been refracted by a thousand lesser thinkers.
Hill's attempt at imagining how a derivatives disaster would
unfold began with a dialogue a while back between him and Henry
Liu. The venue was a K-Wave chat group. Liu, a blogger whose
thoughts and work are known to millions who surf the Web of ideas,
opened with some good questions, and Hill took it from there.
This is fairly heavy stuff, but if you want to cut to the chase
and find out exactly how it could all end, simply skip down to
the final four paragraphs.
How Derivatives Will Crash
"The battle against deflation is being waged in structured
finance through derivatives hedges. When the battle is finally
lost, counter party casualty will be massive," wrote Liu.
"How is that battle going? Are we winning? Is deflation
like the Soviet Army in WWII? Is Greenspan merely fighting a
holding action? Is there ANY chance he'll hold the line? When
might the line crumble? How can you fight against deflation using
derivatives? Are you talking about when the Plunge Protection
Team buys calls on the S&P500 to slow the deflation of the
stock market bubble?
Hill's response:
I'll take a shot at a start. The derivatives Henry mentions are
not stock index options, IMHO. They are all in the fixed-income
arena, first, and in commodities and currencies, second. While
I can't really do the topic justice in a short piece here, you
can look at them as being divided into two basic categories --
cash and "notional". Cash derivatives, like multi-tranche
securitizations, can be further divided into maturities, credit
"slices", and interest rate formulae.
Freddies Were First
Example 1: The first big public cash derivatives were in Freddie
Mac Series A CMO in 1983. A $1 billion pool of Freddie Mac MBS
(themselves a derivative for the real sticklers) were placed
in a trust and pledged to pay sequential bonds. We call this
an A,B,C sequential deal. Every month, interest from the MBS
was used to pay off interest on the bonds. Since the MBS would
pay for 30 years, but more than half of the principal was likely
to come in the first six or seven years (due to mortgage prepayments).
Class A would get all the principal from the MBS until it was
paid off, then Class B, then C. Under normal circumstances, Class
A would pay off in about 6 years, with an "average life"
just over three years. Class B might be fully paid in year 10
with average life just over 7 years. Class C would then get the
remaining 20 years of MBS principal payments, with an average
life of 12 or 13 years. Since the yield curve was steeply upward
sloping in 1983, a par bond with a 3-yr average life might have
had a 9% coupon, a par bond of 7 years average life might need
an 11% coupon, and a long bond, like the original MBS, and like
Class C, 12%. Freddie Mac would keep the extra 300 basis points
from the A bonds as long as they lasted, and the extra 100 basis
points from the B bond. Costs (underwriting, audit, Trustee,
legal, etc.) might eat up 1% of the whole balance, but the spread
income, or residual, still made the deal economically attractive
to Freddie Mac. A typical split of principal in and A,B,C deal
like this might be 30/45/25.
The next variation developed was the 'Z' or accrual bond. Each
month, the interest due on the Z class would be added to its
principal balance, and used to pay extra principal to the earlier
classes. In this case, we might call the structure A,B,C,Z. A
relatively small 12% bond at the back of the deal (about 5% of
the principal) would greatly accelerate the final payment date
of earlier bonds, especially bond C. Meanwhile, the Z bond would
grow to be that same 35% of the MBS pool balance by month 127
(1.01^127), so bonds A, B, and C would all be paid off in about
10 years. You could therefore make a 2 year, a 5 year, and a
10 year, which were all better benchmarks in the Treasuries,
anyway. There was also less uncertainty because the principal
on the early classes was not only paid by mortgage borrowers
paying off the principal on their loans. In an upward sloping
yield curve, the arbitrage was greater in this structure, i.e.,
the residual was worth more.
Fast Forward
[note -- this level of detail will take all night, so I apologize,
but I'll have to do the rest in fast-forward mode] The next variations
for dividing principal and interest were 1) setting a schedule
of principal payments that would get priority, concentrating
prepayment variance in about half the bonds. The priority payment
bonds were called Planned Amortization Class (PAC) bonds, and
the others were called "companions" 2) synthetically
creating premium and discounts - in its simplest form, a 9% MBS
could "feed" a 6% and a 12% bond that ran 50/50 parallel,
or it could support a 75/25 split of 8% and 12% bonds.3) using
the same ratios, a 9% MBS could support 3 LIBOR-based floaters
with a 12% cap, as long as there was 1 parallel bond whose coupon
floated at (minus) 3*LIBOR, and could go to zero. 4) "strips"
of interest-only and principal-only bonds could be created from
the stream of principal and interest of the MBS.
Finally, there was concentration of losses into Senior/subordinated
structures for collateral that had credit risk. With enough subordination,
you could issue AAA rated bonds from junk collateral. This technique
of Sr/Sub was even used to create cross-currency bonds by taking
all the currency translation risk in one half of the issue (Dollar/Yen).
10 Different Types of Bonds
Where the fun comes in is to create 10 different types of bonds
for 10 different types of investors. As the yield curve moves
around, and alternative investments price at different levels,
the bond structurer has to juggle everybody's desire to get the
best bond they can with the reality that the collateral price
is moving, too. This is what I used to do for a living.
Now, we move on to non-cash derivatives. Of course, having learned
to mix all the above techniques together, we couldn't help ourselves
-- we started to include interest rate swaps, caps, floors, currency
swaps, Treasury zero coupon bonds, insurance policies, etc. to
further "engineer" the cash flows for the buyers and
sellers.
The goal is to sell, at a fair price, the risk profile that each
investor needs. We'll go back to the simple A,B,C,Z structure
to see who the natural buyers were, and why it all worked to
lower borrowing costs for home buyers. When there were 9.5% mortgages
bundled into $1 million (original face amount) 9% GNMA pools
starting in 1972, hardly anyone had liabilities that required
monthly cash flows for 30 years, so the GNMAs traded with hundreds
of basis points more yield than similar average life Treasuries.
However, once we had a little better "window" of principal
payments defined, banks could buy the 2-year average life CMO
at Treasury + 100, and sell 2-year CDs at around Treasury rates.
Property-casualty insurance companies could buy 5-year CMOs at
T+120 instead of Treasuries or instead of BBB-rated corporates
that might default. Pension funds with an aging workforce might
buy the 10 year Class C, and Life insurance companies that won't
be paying out until 20 years from now could buy the Z-bond, and
lock in a 12% reinvestment rate. Arbitrage lowered the spread
of GNMAs to around T+120 once the CMO machine got going, and
homeowners started paying about 2% more than 30 year Treasuries
instead of 4% more.
GNMA Pool
Needless to say, anything with a statistically understandable
cash flow can be made into bonds this way. Just a note on statistics
-- if you buy one $1 million GNMA pool, it has maybe 10 or 20
loans in it. One or two homeowners selling their houses means
you get a lot of your principal back early, in huge chunks. When
we did $1 billion CMO deals, there were now four large classes
of bonds, which smoothed out the statistics (thousands of individual
mortgage loans), and gave investors bonds that had a good chance
of being traded in a secondary market.
Today, securitization is far and away the largest capital market,
with new-issue volume likely to top $2 Trillion this year. Only
commercial paper has bigger new-issue volume, but that's heating,
since it so short.
The other areas in the non-cash derivatives that I think Henry
was referring to would be the fast-growing credit derivatives
market (one counterparty pays the other if a credit event occurs)
and the Granddaddy of non-cash derivatives, the swaps market.
When we see Ford or GE come to market with long-dated bonds,
you can be virtually assured that investors are swapping them
back into LIBOR, since most fund at a LIBOR spread these days,
and even the issuers are likely doing the same.
In my opinion, the likely "first domino" in the chain
reaction of counterparty credit that blows the whole system is
going to be a Money Center bank that is crucial to the global
LIBOR-based funding system, or it will be an enormous quasi-bank
like GE... that is, unless we manage to continue to limp along
from one near-crisis to the next.
Good night all,
hh
Rick Ackerman
***
Information and commentary contained herein comes from sources
believed to be reliable, but this cannot be guaranteed. Past
performance should not be construed as an indicator of future
results, so let the buyer beware. Rick's Picks
does not provide investment advice to individuals, nor act as
an investment advisor, nor individually advocate the purchase
or sale of any security or investment. From time to time, its
editor may hold positions in issues referred to in this service,
and he may alter or augment them at any time. Investments recommended
herein should be made only after consulting with your investment
advisor, and only after reviewing the prospectus or financial
statements of the company. Rick's Picks reserves
the right to use e-mail endorsements and/or profit claims from
its subscribers for marketing purposes. All names will be kept
anonymous and only subscribers' initials will be used unless
express written permission has been granted to the contrary.
All Contents ©2004, Rick Ackerman. All Rights Reserved.
You can subscribe here.
_______________
321gold Inc
|
