You simply cannot act as a money dealer when the money you are dealing is highly suspect. I am not writing about money in the true sense, such as any tangible form that falls under property laws of custody and bailment, but rather the wholesale “money” that is derived under the much looser and unconstrained terms of financial laws and more so convention. This was another of the great lessons of 2008 that did not stay learned.

Take you pick of narratives, as there was no shortage of representative samples from among the network of dealer behemoths. It isn’t coincidence that the names of those who actually failed (insolvent no matter the cold fusion of liabilities being injected) were those more active in these kinds of matters. For this particular story, I’ll use AIG because it represents perhaps best what works least. To begin with, AIG was supposed to be an insurance company rather than a money dealer.

What felled AIG was not losses; it is never losses. In August 2008, AIG released earnings that included $6.8 billion OTTI charges (other than temporary impairment) that were losses in the accounting sense of securities that were not sold but, as the name implies, weren’t likely to see prior values anytime soon. This was actually the Enron legacy of using market inputs for valuations. AIG had been a big seller of protection on LSS structures, perhaps the biggest, though we will never be sure. LSS meant “leveraged super senior” tranches of securitizations, which were, essentially, the leftovers after wholesaling out all the mezzanines. There was very little default risk in these pieces because the “thickness” of the structures above it was sufficient that it would take literal (not financial) catastrophe in order for defaults to penetrate that far (and, as fanciful as it sounds, this was actually correct; no senior or super senior piece that I am aware of ever felt a money loss from default; it was all pricing and liquidity).

The fact that there was such high demand for “hedging” on even LSS tells you all you need to know about math-as-money. The problem for writers of protection on CDS for LSS was that it wasn’t default that triggered collateral or reassignment of terms. Many (most?) were structured so that a ratings action (downgrade) or even spread levels would be the standard. If you owned protection on any particular prime MBS super senior (which meant you owned maybe 10% of the piece and leveraged the rest with commercial paper) and the ABX spread index started to blow out, that might trigger a collateral call on the counterparty that had written the CDS. That feature was meant as protection for both sides – that a collateral call step or two would mitigate any binary option, meaning that spreads rising wouldn’t just trigger an immediate and catastrophic unwinding.

Starting sometime in July 2008 but intensifying through August and into September that year, CDO spreads had initiated $6 billion in collateral calls just on AIG’s super senior protection positions (written). On September 12, 2008, three ratings agencies downgraded AIG based not just on those calls but also on their own modeled effects of where credit spreads were going (and thus the expected necessity of still more collateral posting). Two AIG subs were then shut out of commercial paper, forcing AIG the parent to step in with lending facilities.