The infamous Abacus transaction that cost Goldman Sachs $550 million might have been designed to fail, but it turns out that it actually performed better than its peers, according to a new study co-authored by BlackRock and Columbia Business School.The Abacus CDOs’ performance, “while undoubtedly bad, was actually better than average among all bonds that had been similarly packaged.”
Beyond this interesting insight, the the study nonetheless supports the assertion that collateralized debt obligations (CDOs) created from asset backed securities (ABS) were designed to decline in value:
Using a unique database published by the investment firm Pershing Square Capital Management, Faltin-Traeger and Mayer identified the underlying bonds in some 528 ABS CDOs issued between 2005 and 2007, and compared their performance to similar bonds that weren’t included in CDOs.
They found that the bonds in the CDOs performed a lot worse. Even if one holds observable characteristics such as initial ratings and yields constant, the bonds in the CDOs suffered ratings downgrades that were 50 per cent to 90 per cent more severe. As of June 2010, for example, bonds with initial triple-A ratings had been downgraded by an average 11.84 notches, compared to 5.99 for those not in CDOs. The bonds in the CDOs were also more likely to have been rated by all three major credit-rating firms.
Was the poor performance the result of intent on the part of the issuing banks? On this point, the authors note, “It would have been very hard to randomly choose securities with such poor ex-post performance.”
This is, on its face, damning.
But is also points to the fact that the broader assumption behind the creation of ABS CDOs was deeply flawed: namely, that you could take marginally credit worthy assets and structure them in such a way that risk would be more effectively distributed.
Instead, this study shows CDOs did little to smooth risk and may have, in fact, been a dumping ground for assets the banks preferred not to hold onto.