The Bubble Has Been Fully Restored: Goldman Is Selling A Synthetic CDO

Earlier this week, Bloomberg ran a story that carried the headline “Goldman Sachs Hawks CDOs Tainted By Credit Crisis Under New Name.” Here’s an excerpt:

The 2008 financial crisis gave a few credit products a bad reputation.


Like collateralized debt obligations, known as CDOs. Or credit-default swaps. But now, a marriage of the two terms (using leverage, of course) is making a comeback -- it’s just being called something else.


Goldman Sachs Group Inc. is joining other banks in peddling something they’re referring to as a ‘bespoke tranche opportunity’…


The deals are ‘attractive for credit-savvy investors in the post-QE credit picker’s market,’ according to a January U.S. credit derivatives outlook by Citigroup Inc.,


The transactions offer the potential for higher returns than buying a typical corporate bond, especially if an investor focuses on first-loss slices or uses borrowed money.

This is the latest installment in a series of articles that pop up every so often in the financial news media touting the resurgence of structured credit and, more specifically, synthetic CDOs. The interesting thing about the Bloomberg piece is that it references a Citi note which attempts to make a distinction between traditional synthetic CDO deals and so-called "exotics", such as the much maligned Leveraged Super Senior (LSS) structure, prevalent in pre-crisis Canada.

LSS deals -- which allowed protection sellers to lever up 10X on otherwise unattractive (from a premium perspective) super senior CDO tranches -- were at the heart of an absolutely epic meltdown in the market for third-party Canadian commercial paper back in 2007. This incident precipitated all types of mayhem with unsuspecting retail investors, played havoc with the country’s pension funds, and led several former Deutsche Bank employees to accuse the bank of masking more than $10 billion in paper losses during the financial crisis.

The laughable thing about LSS deals was that they were effectively non-recourse, meaning that the protection seller was allowed to sell protection on a notional amount that was multiples of the collateral posted, but in the event the market moved against the seller enough to chew through that collateral and a margin call was made, that seller could just say “to hell with it” and walk away from the deal. More simply, I, the seller, insure $100 million in debt, but only post $10 million up front. If there’s a credit market meltdown and my $10 million is no longer sufficient and you, the protection (insurance) buyer, call me looking for more money to compensate you for the elevated risk, I can politely tell you to piss off. The risk that I tell you to piss off is called “gap risk.”

Now if you’re the protection buyer in that scenario, you’re likely sitting on a mark-to-market loss and that’s something that some risk managers think shareholders should know about, hence the former Deutsche employees’ contention that in some cases, these losses were not properly accounted for, leading the bank to look healthier than it actually was during the crisis years. The other side of that argument says that it really doesn’t matter because if no actual defaults occurred in the senior tranches, then the losses were just paper losses so, ultimately, who cares? Of course that’s a kind of “hindsight is 20/20” argument that it’s very easy to make years later: “There were no defaults, so there was no need to report the M2M losses.” Much like: “There was smoke billowing out my neighbor’s window and he wasn’t answering the phone, but it turned out he just burned his dinner, therefore I shouldn’t have gone and checked on him.”

The point here is that supposedly, regular synthetic CDOs (and "regular" here just refers to the full-recourse nature of the IM/VM framework and isn't meant to indicate anything about whether the tranches are bespoke) are suitable instruments for, as Citi put it last month, "credit savvy investors," and it was only the unique character of the leverage employed in the LSS deals that led them to blow up. As synthetic credit appears to be making a comeback, it's worth revisiting that assumption.

To be fair to synthetic CDOs, Bloomberg does gloss over a distinction in terms of how “leverage” is being defined in these deals.

Here’s Citi on structural leverage:

“To understand this, consider an equity (0%-5%) tranche that trades at a price of 80 pts upfront with a running coupon of 500bp. In bond terms (fully funded), the investor pays 20 cents on the dollar for a bond and gets paid 500bp coupon till maturity, when par is returned if there are no defaults. The returns are therefore magnified 5x, but if there are defaults, the investor’s losses are limited to the 20 cents paid upfront, i.e., there is no recourse beyond the initial investment.”

So that’s the non-recourse structural leverage built into the equity tranche. So what about good ol’ financial leverage (that’s the fun kind)?

Here’s Citi again:

“In addition, it is possible to get financial (or swap) leverage on tranches through an initial/variation margin (IM/VM) framework – this is full recourse and can further help to magnify returns. In this case, the investor puts up collateral for a fraction of the original notional at trade initiation (initial margin), but the margin is adjusted upwards if the market moves against the investor (variation margin).”

For example, here are some IM assumptions Citi used for modeling purposes:


Note that Citi is quick to explicitly note that this leverage is not the same as the leverage that was employed with the LSS deals that imploded in Canada in 2007:

“...and super senior risk (with spreads in the single digits) was only attractive if the investor was able to apply financial leverage. This produced exotic structures such as the leveraged super seniors, which eventually resulted in the demise of the Canadian conduits.”

So why exactly were the LSS deals “exotic structures” (suggesting they are wildly different from other structures) whereas other synthetic CDO deals are seen as traditional and therefore more suitable investments when both involve the posting of collateral that is only a fraction of the notional amount insured? It comes down to whether the deals are non-recourse or full-recourse. In other words, with LSS deals, the investor can simply walk away in the event of a margin call and in traditional deals, the protection seller is obligated to post more collateral.

Here’s an excerpt from a Euromoney piece (which ironically discusses a deal Citi was pushing a few years back) which explains the distinction:

“Citi in its latest deal has introduced a feature that gives it full recourse to the investor, removing the option to back out. That means whatever the market does, Citi can still collect on its insurance, effectively swapping gap risk for counterparty risk.

So in sum, we should all feel better about these deals because now they’ll likely all be full-recourse and instead of gap-risk, the structures only create counterparty risk.

Of course this is all just semantics. Sure, all things equal it’s better to be able to sue your counterparty in the event they don’t meet a margin call as opposed to not being able to sue them, but in a crisis, what’s the difference between a protection seller walking away from a deal because they can and a protection seller walking away because they’ve blow up and can’t meet the margin calls?

At the end of the day, the idea that a “traditional” (whatever that means in the context of synthetic CDOs), full-recourse framework makes the use of leverage safer in synthetic CDOs seems to miss the fact that it was counterparty risk that blew up the system in the first place.