For the beginning, it is worth clarifying what are those Collateral Debt Obligations. I will cite from an instructive article of Salon.com. The author Andrew Leonard has his own gripping sub-blog on economic matters.

After a mortgage lender makes a loan to a homebuyer, that loan is packaged up with a bunch of other loans into a security -- a financial instrument that can be traded. Securities are rated by rating agencies according to the chances that the underlying assets will be defaulted upon. U.S. Treasury bonds, for example, get stellar AAA+ ratings because the U.S. government is considered likely to meet its obligations.

A security based on a pool of subprime mortgage loans would normally not deserve an AAA+ rating. Subprime, by definition, means "not so good." Subprime loans are made to people who can't put together a down payment or have bad credit, or can't prove they have a job. Subprime loans are risky!

Many investors -- particularly in pension funds and municipalities -- are prohibited from investing in securities that are not high-rated. Let the hedge funds and the investment banks play around with the risky BBB stuff, the "junk." The rest of us should be more prudent.

But investment bankers are clever fellows. In cahoots with the ratings agencies, they came up with a way to magically transform a low-rated security to a high-rated security.

[Enter] the collateralized debt obligation. The CDO takes a pool of risky mortgage loans and divides it into slices. (Wall Street calls these slices "tranches," but that seems to be a word that makes the brains of normal people freeze up, so we'll ignore it.) For simplicity's sake, let's say that a mortgage-backed security gets divided into two slices when it is transformed into a CDO -- a senior slice and a junior slice. Let's say that the senior slice gets rated AAA+ and the junior slice gets rated BBB-. But if anything goes wrong -- if the homeowners whose loans are part of this security start missing their payments -- the investors in the junior slice have to lose all of their money before the investors in the senior slice start feeling any pain. That's the beauty of the scheme. You take a bunch of bad loans and turn some of them into high-rated gold and some into lower-rated bronze. You sell the gold to the cautious and the bronze to the bold. If a few loans go kaput, the bronze investors suffer. If all the loans go kaput, everybody gets hurt. Unless there's a total financial meltdown, everyone is happily making money.

We keep hearing in the financial news about risk being "sliced and diced." Is that what you're talking about?

Yes. After the transformation, we now have an instrument that satisfies the desires of both conservative investors, who can just buy the AAA+ rated slice, and investors who have a taste for risk, who can buy the BBB- slice. It's a brilliant work of alchemy.

And very popular. CDOs tied to subprime mortgages became hot commodities, snapped up with gusto by traders all over the world -- even the riskiest, most likely to self-immolate, lowest-rated slices of those CDOs. Especially those slices.

Why was there high demand for the most wacky CDOs? An easy answer is that higher risk attracts higher reward - so greedy we are. The derivative market is broadly advertised as means to reduce risks. But who needs security in a booming market?! You better use financial instruments to leverage your investment tenfold!

Andrew Leonard makes an apt analogy between the derivative market and sports betting.

We often hear about the "real economy," that place where real people buy and sell real things, or go to work at real jobs where they make real stuff or deliver real services. Derivatives belong to what should be called -- but never is -- the unreal economy, a place where speculators make bets about what will happen in the real economy. Derivatives are vehicles for making such bets. If you think the borrowers whose loans are pooled together are going to make their payments, then buying a share in a group of such investments might be a good idea. That would be your bet.

A metaphor might be useful here. The real economy is like the Super Bowl. Real men on a real field push each other around and play with a real ball for a set period of time, and the team with the most points at the end wins. But while all this is going on, millions of outsiders who are not physically involved in the game bet on its outcome. Only they don't bet just on the outcome. They also bet on the spread -- how badly one team might beat the other. Or they can get more creative and bet on what the combined score of the teams might be, or which team's quarterback will be the first to be injured. There's absolutely no limit to the things that you can bet on, as long as you can find someone to take your bet.

The betting economy is the unreal economy. All those sports bets, no matter how kooky, are financial exercises whose value and meaning are derived from what happens on the field. [There's] so much money involved in gambling that the temptation to fix the results becomes irresistible. Players and referees, for instance, can be bribed.

There is no secret - an easy way to make good betting profit is to make a wacky bet and bribe players or referees. (Coincidentally, there is ongoing high-profile scandal with an NBA ref involved in betting his own matches.) Or in the derivatives market, you can ridge mortgage originators or rating agencies, if you can.

Still, the extent of CDO demand looks remarkable beyond ridging tricks. Still quoting Leonard:

The incentive for everyone to behave [that crazy way with mortgages] came from Wall Street, where the demand for subprime CDOs simply couldn't be satisfied. Wall Street was begging the mortgage industry to reach out to the riskiest borrowers it could find, because it thought it had figured out a way to make any level of risk palatable.

It is time to get to Henry Liu's article mentioned forefront. It has the following section.

New Geometry of Debt Securitization

The mortgage sector before the age of securitization was shaped like a cylinder in which risk was evenly spread throughout the entire sector thus all mortgages share the aggregate cost of default.  This even spread of risk premium is viewed as market inefficiency. Securitization through collateralized debt obligations (CDO) permits the unbundling of generalized risk embedded in all debt instruments into tranches of escalating risk levels with compensatory higher returns and in the process squeezes additional value out of the same mortgage pool by maximizing risk/return efficiency.

The geometry of CDO scuritization transforms the cylinder shape of the mortgage sector to a pyramid shape with least risky tranches at the top and the more risky tranches with commensurate premiums towards the bottom, so that a greater aggregate risk premium can be squeezed out by the security packagers and investors as profit. This extra value when siphoned off repeatedly from the overall mortgage pool required an ever larger base of sub-prime mortgages in the new pyramid shape, thus increasing the systemic risk further. Sub-prime borrowers are no longer just low-income borrowers.  They include high-income borrowers whose incomes and collateral value do not provide sufficient reserve for sudden changes in market conditions. A sub-prime borrower is one who over-borrows beyond prudent standards. The extra risk premium value thus taken out of the mortgage sector contributes to the increase in liquidity to further feed the debt market, pushing the low credit standard of sub-prime lending further down.  As prime credit customers having already borrowed to their full credit limits, growth can only come from lowering credit standards, turning more prime borrowers into sub-prime borrowers.

This is the structural un-sustainability of CDO securitization, irrespective of the state of the economy since risk of default is shifted from the state of the market to the direction of the market. Any slight turn in market direction will set off a downward spiral crisis. The initial upward phase of this cycle is euphoric, but any addiction, but the pain will come as surely as the sun will set in the downward phase. Not many economists or regulators have yet focused on this structural defect of CDO securitization. The recent Congressional hearings on sub-prime mortgages completely missed this obvious structural flaw.

If I understand well, security of a CDO trench is based on security of "relative" CDO trenches of lower rating, or of lower objective quality. By "relative" trenches I mean those sharing subsets of bundled mortgages. Exponentially more trenches of lower quality are required to secure "higher" trenches - quite a pyramid scheme indeed. So much security on derivatives. The demand for lowest junk CDOs becomes less mysterious as well.

What does this interpretation mean for the long term prospects of world economies? An optimist may say: we merely have a pyramid scheme unraveling. Market valuations will be reduced by the amounts invested in the junk CDOs, but the markets will merrily rise again. Some hedge funds and other players will suffer dearly, just as LTCM was indirectly condemned in 1998 by the Russian government defaulting its Ponzi scheme. But all business will go on as usual.

A pessimist may say: we just have a pyramid scheme unraveling. The real crisis spiral is just starting to unwind. I am a pessimist. Next time I may put up a crash script -- there won't be much new to fellow Cassandras. Or is there more wisdom in keeping silence?

P.S. I did a Google News search fo "CDO pyramid" or "CDO Ponzi". I found 4+6 entries: A B C D 1 2 3 4 5 6.