The WSJ recently published an article highlighting the speculative nature of approximately 10 billion CDOs insured by AIG. They became an issue because the Maiden Lane III facility, which is buying up the collateral underlying the portfolio of Super Senior CDOs AIG insured, can't buy the collateral in this case. The insured parties don't own it – there is no insurable interest.
The WSJ article mentions “Abacus” and Goldman Sachs (GS). A search on the two terms brings up a Fitch downgrade announcement which is very informative:
Abacus 2005-CB1 is a synthetic collateralized debt obligation (CDO) that closed on Dec. 7, 2005 created to enter into credit default swaps with Goldman Sachs Capital Markets. Abacus 2005-CB1's synthetically referenced portfolio is comprised primarily of subprime RMBS bonds (72.7%), CMBS (20.5%) and other structured finance assets. Subprime RMBS bonds of the 2005 vintage account for approximately 48.7% of the portfolio.
For an definition and explanation, here is a link to Investopedia.
I have reviewed the material available on synthetic CDOs on the internet, and find that these transactions have been relatively common and were generally considered as a useful innovation compared to CDOs constructed with actual assets as collateral. It was more convenient to reference the collateral by a CDS than to actually locate and buy it. In point of fact, it appears that those who held certain assets found it easier or more convenient to offload the risk via synthetic CDOs than to sell the assets.
Of course, if a seller knew certain assets were no good, referencing them in a synthetic CDO was a nice way to place a bet. Synthetic CDOs to a large extent have become part of a huge zero sum game. Frequently no insurance is provided due to the lack of insurable interest. The WSJ is accurate in characterizing many of these transactions as speculative.
A review of the prospectuses for CDOs insured by Ambac (ABK) reveals that some of them permitted synthetic collateral, typically limiting it to a specified percentage of assets. Searching such documents for the word “synthetic” yields numerous instances, scattered throughout page after page of definitions and other contractual provisions. Ambac does not disclose what portion of the CDOs it insures contain synthetic collateral.
MBIA's (MBI) CDO disclosures specifically note certain CDOs as synthetic. According to Jay Brown's Comments on TARP addressed to Hank Paulson, MBIA insured approximately 100 billion of such CDOs. MBIA is not required to post collateral and as such will not be pressured as AIG was.
Brown's take on such transactions is informative: he now regards them as a zero-sum game, large enough to swamp the TARP program if it got involved, and as having no effect on the real economy. This is the stuff that Congress saw fit to protect from regulation. What I see is a mechanism whereby legitimate investors could be duped into providing insurance on risky assets, and frequently without the poor excuse of an insurable interest.
This all gets back to points I have been hammering on, to little avail, for over a year:
CDS are insurance and should be regulated as such
Regulation should include a requirement of insurable interest
Regulation should ensure that those who write Credit Default insurance are adequately capitalized.
Looking at AIG's involvement: what they did, at the time they did it, was widely regarded as a normal and acceptable business practice. In hindsight, it was not. This is the price we pay for innovation untrammeled by regulation.
Disclosure: Author holds a long position in AIG; reducing positions in ABK and MBI.
