Imagine this scene (and it should be fairly easy right now). A worried fund manager has just
lost a pile of money in collateralized debt obligations (CDOs) – so much money that his bonus,
his job, his department, and even his whole fund management business is at risk.
In an idle chat with an investment banker he sighs that "If only I could make a thumping great
profit before the year end – then everything would be okay."
Easy! And this is how it's done.
The USA's Federal mortgage financiers – known generally as Fannie Mae and Freddie Mac – habitually
slice up monthly mortgage payments, dividing them into the interest part and the repayment part
(the principal). They routinely chop up mortgages like this, all the way through to the final payment
from the home-buyer.
These slices can be recombined together to form custom-built packages of future cashflows, with
interest and capital repayment elements mixed to order for the investment banks. All these slices and
reconstituted packages are backed by Fannie and Freddie, so they're AAA-rated – the crucial
"investment-grade" credit rating needed to make them saleable to public and private pension funds,
perhaps including yours.
Imagine the investment bank buys $90 million of assorted interest-only payments, spread over 30 years.
It also buys $10m of principal – which pays no interest – due for repayment 30 years from now in 2037.
The bank then re-packages these two different mortgage investments in a very clever way. First, the
$10m principal pile is split down the middle. Half of it ($5m) is mixed into the $90m of interest payments.
Let's call this 'Package 1'. The other $5m of principal is left on its own and called 'Package 2'.
Both packages now have the same issuer, the same AAA credit rating, the same 30-year repayment date,
and an identical amount of principal ($5m). This becomes the face value of each package.
But Package 1 also has a 30-year string of interest payments, which makes it look very much like a
benchmark 30-year bond. Package 2, on the other hand, has no yield at all, and therefore it is worth
a lot less today than its face value.
Simple enough, and you can easily see which half you would value more highly. So here comes the clever bit.
The investment bank then folds both packages into a new trust, for which the investment bank itself acts as
trustee. It writes into the trust deed that "in the event of an early partial redemption, Package 1 must be sold
before Package 2."
Then almost everyone forgets about the details of the trust deed.
Our worried fund manager, the one with all those losses needing to be hidden away, now buys all of this new
trust instrument. He holds it for a month or two, before his investment banker telephones to suggest a sale of
half of the trust's face value.
Under the terms of the deed, all of the elements of Package 1 are now sold, including all its prospective interest
receipts, at a price equivalent to the benchmark 30-year bond which it so closely resembles. All of the low-value
Package 2, meanwhile, is retained within the trust – again, in keeping with the terms of the deed.
Even if the market has not moved during those two short months, this sale of Package 1 redeems 95% of the
fund manager's original investment, but it reduces the principal element in the trust by only 50% of its face
value. The benchmark 30-year bond – to which the actual sale price is very close – is then used as the valuation
basis for the remainder of the trust instrument, Package 2.
The worried fund manager is delighted! He has sold only half of his Trust Instrument and got nearly his whole
investment back – a big profit. Moreover, what's left is now priced according to the benchmark 30-year bond,
and it's allowed to float innocently up and down in line with the benchmark's publicized value.
For the next 30 years, Package 2 will sit on the books valued initially at perhaps 20 times what it would actually
fetch on the market, if sold. Of course, it won't be sold – and neither will it receive one shred of income for
between now and 2037. But the true benchmark bond will receive interest, and as the benchmark's interest
payments get paid each year, so the benchmark's resale value will slowly fall...and the value of Package 2
will slowly come into line.
When finally – in 2037 – the benchmark has no remaining income left to pay, the two will have the same value.
Package 2, which never paid any income, will be sold at the benchmark price against which it was routinely
over-valued these last 30 years, and no-one will be any the wiser!
Isn't that clever? Today's mega-loss has been converted, almost invisibly, into 30 years of miserable
under-performance, which eventually deprives the unwitting investor – the end-user who buys into this
scheme via the fund manager – of 90% of his money.
Think it can't be done? This was broadly the strategy implemented by US investment banks for their
Japanese customers in the aftermath of Tokyo's stock-market and real-estate crash in the early 1990s.
The Japanese have experienced horrible underperformance in just about all of their investments ever since.
The scam was exposed in F.I.A.S.C.O. by author Frank Partnoy, who was one of the investment bankers
involved. He is now a university law professor, and Partnoy broke the mould; he left his bank because he felt
ashamed – fully realizing that investors would be steadily relieved of most of their capital, while their bankers
and advisers got rich on un-merited bonuses, earned by hiding huge losses for decades to come.