Straight outta Antarctica, Felix Salmon calls me forth from a shallow grave, with his rhythmic chanting, "I Heart CPDO / I Heart CPDO". And he mocks, mocks, the Sacred Order of the Credit Cassandra. (Our motto: "We may go bankrupt first, but eventually you will too, and then we'll tell you that we told you so.") [1] All I can say to Felix Salmon is. You bastard. You bloody bastard.

Okay, then. Let us despatch with godspeed Felix's nefarious and very naughty heresies. Felix begins his abominations by conjuring the devil herself, that long-legged vixen of debt-bubble capitalism made flesh, Citibank. (Oh, temptation. Impure thoughts... Curvacious bubbles. Must. Not. Yield. Must. Not. Spread. 1929-1929-1929-1929.) Anyway, Felix trots out somebody from Citibank, with graphs showing many different ways that Bad Things^TM can happen in credit markets, while CPDOs still do fine.

Just to be mean, I'll note there is nothing much new in these graphs, that they very much resemble the graphs in ABN's early CPDO presentation, and that the re-presentation of this stuff was like a child putting his hands to his ears and shouting "I don't hear you! I don't hear you!" when we, the beneficient, were only trying to show dear Felix the light. (It is a rather dark and depressing light, we'll grant you. It's not quite enough to read by.)

If we must respond to substance, we'll go all jujitso and yield (and spread) where the opposition expects resistance. We concede the truth and wonkiness of the Citibank graphs. We concede that CPDOs are very, very clever synthetics, almost as nice as that synthetic heroin in the Eighties that could give you synthetic Parkinson's disease. If the world behaves even remotely like it has over the last 10 years during the next 10 years, the likelihood of any CPDO going bust is practically nil. Let us understand and extol the cleverness of CPDOs in non-mathematical terms [2]. They are, yay and truly, wonderous inventions:

1. If nothing bad happens, if credit spreads remain broadly unchanged, CPDO NAV ("net asset value") never falls, but slowly rises. Under this scenario, CPDOs straightforwardly yield full principal and coupon, even shedding (nearly) all credit risk fairly early in their lifespan by "cashing in" when the strategy has earned enough to pay its obligations.

2. If something bad happens every now and again, a spate of defaults, or some exogenous widening of credit spreads, then CPDO NAV takes a brief hit, and NAV drops. But CPDOs are designed to take on more leverage when this happens (until they hit a floor very far beneath where they start), and the increased spread combined with increased leverage accelerates the post-credit-event earnings of the CPDO! Thus, as the Citibank graphs show, CPDOs undo the damage of a sudden credit widening rather quickly. With CPDOs, yesterday's bad news is tomorrow's very good news. The extra leverage and yield hastens the coming of the glorious "cash-in" event, henceforth to be spake as "the rapture", when the CPDO converts itself from a complex, leveraged play into a bank account with a predetermined withdrawal schedule.

3. CPDOs are built with a lot of headroom — they earn highly leveraged credit spreads, but promise to yield only 100-200 bps more than LIBOR. They are intended to be conservative instruments, and they can take some hard-knocks: As Felix points out, while they enjoy "roll yield" from selling more protection than they buy every six months, they don't rely on it. They can take severe hits to NAV and bounce back. Under benign circumstances, they "cash-in" very early in their long (10-year-ish) lifespans. If there are less than benign circumstances early on, they can bend a lot but not break, and still have plenty of time to make up all their losses once the market normalizes itself again.

These are clever instruments. If I was the grunt at ABN-AMRO who designed the first one, I'd be damned proud of myself. If it's your job to design structured credits that earn maximum yield for minimum risk under any reasonable model based on recent-past credit-spread history, you Ms-CPDO-Inventor, deserve a big gold star (and a 50 million dollar bonus).

So why am I complaining? (Yes, I am complaining.)

I think that if CPDOs become popular enough, they will break. I really do. But how? First, let's understand some scenarios under which they could break:

Multiple, sequential spread-widenings

Please refer back to the Citibank graph in Felix's post that shows what would happen to a hypothetical CPDO under a credit-spread jump. Look at that worst-case curve, where spreads suddenly jump by 150 bps. It looks bad initially, but recovers quickly, for reasons we've already discussed. Now, imagine that spreads jump not all at once, but in three separate events, one every six months. The fall in NAV under this scenario will be worse than the fall under the single jump scenario. Why? After each individual fall, the CPDO will have levered its credit exposure higher, to make up for the previously lost NAV. Still, the CPDO might well survive and recover. But suppose that there are four, or five, or six such events. Even if nobody defaults, the CPDOs will break.

"Conundrum" defaults

Suppose that there are a substantial number of defaults over the lifetime of the CPDO, but credit spreads don't widen? In this case, the CPDO loses NAV as it has to make good on some of the credit protection it has written, must increase its leverage dramatically to recover (as it earns the same paltry yield on its exposure), and takes deeper losses when yet more defaults occur. A few such episodes would be enough to break the instrument.

Tail-risk Credit Event

Suppose credit spreads jumped not by 150 bps over a single six-month period, but by 300 bps. Goodbye CPDO.

So, how likely are any of these events? That depends on your frame of reference. If you believe that the behavior of credit markets over the past decade is representative of how credit markets will behave in the future, then the odds of any of this happening are practically zero. If you believe that markets may behave in ways not captured by our experience of the recent past, than it's really a judgment call. My judgment is that the likelihood is significant. Why?

• You don't have to go back very far to find "tail-risk" credit events. Here's a bet: If you run a hypothetical CPDO through the markets of the late 1970s early 1980s, a combination of tail-risk credit events and sequential bad periods would kill a hypothetical CPDO. (I don't have time to do the work right now of testing this conjecture against the data. Doing so would require lots of judgment calls, since no liquid CDS market existed, but it is an exercise reasonable people could attempt, and the conclusions might be clear enough to overwhelm misgivings about the judgments.)

• The same conditions that created a "conundrum" whereby short rates rose while long rates did not could create a credit spread conundrum. There is preternatural (read "central-bank-and-petrodollar") liquidity in today's debt markets, and I would not be shocked to see a bunch of high profile defaults, followed by remarkably blasé spread widenings, and then all kinds of talk about how "the market" understands that the troubled firms were just "a few bad apples".

• If CPDOs grow very popular, if a whole lot of much money is invested in basically the same, publicly known, highly leveraged portfolio, then market participants will work to create the conditions that break the portfolio in order to profit from the carnage. Markets adopt new behaviors when too much money starts betting that they'll behave as they recently have. [More here.]

I'd like to have a punchier conclusion than this, but it's late, and I am, after all, a grumpy zombie. So. Yes, CPDOs are very clever. Yes, they're robust to most reasonable scenarios, where reasonable is defined by even the last twenty years of market history. No, I don't care. Yes, I think they might well break. And I'm sure they will be broken if the quantity of CPDO-invested funds grows suffiently large. And I don't like that, as a tax payer, I'm required to insure banks that take out 5-times as much CPDO debt as they are permitted to take on ordinary business debt.

Good night.

P.S. In response to a point of Felix's, I think I should say that yes, corporate-AAA debt is also not default free. But, at least corporate-AAA debt requires a default to break, and a default by a specific entity that one can evaluate independently of some credit agency's AAA seal of approval. If you think Pfizer is goin' down due to some liability issue, eff Fitch's and don't buy their debt. It's hard to evaluate a whole CDS portfolio, other than by stats like what percentage is BBB. CPDOs can be broken by broad market trends and manipulations while a specific, conservative, well-run business cannot be forced to default. If you choose your conservative, well-run business poorly, you might still be toast, but at least you have a choice. And AAA sovereign debt in a currency that the sovereign is allowed to print is an order of magnitude less likely to meaningfully default than any AAA corporates. (The potential for a late-payment due to some momentary budget standoff in the US congress does not count as a meaningful default, so long as the eventual recovery rate is 100%.)

P.P.S. Here are links to Felix's several posts that inspired my bleary rise from a blogospheric grave:

• The weekend's CPDO debate
• Big Pharma and CPDOs
• CPDOs for the weekend

Felix's posts are of course and as always excellent, as are many of the comments. You can be excellent and wrong though. (I usually am. But not here. Not now. I'm right, gosh-darn-double-dang-it!)

Update: I should note that this is a "for entertainment purposes only" kind of piece, really, it's after 5 am, and I was supposed to be doing something else. I mean it all — especially the part about Citibank being very sexy — but I've not really responded in a fair, measured, organized, or on-point way to anything anybody actually said. Sorry about that. Rereading the posts that inspired this, I do want to chime in on where Felix responds directly to a previous point of mine directly, regarding an arbitrage strategy of selling ordinary AAA debt and buying CPDO extra yield. If CPDO return-at-maturity is equivalent to a portfolio of other AAA debt held to maturity, an arbitrage strategy would still work. The point about the higher mark-to-market volatility and potential illiquidity of the CPDO is well-taken. But if CPDOs in the end always pull it out, a long-term investor able to bear MTM volatility should indeed sell diverse vanilla AAA and buy CPDO, with the intention of holding both to maturity, until CPDO and vanilla AAA spreads converge. (BTW — do banks have to mark their CPDO positions to market?) I also want to point out that Felix has actually spoken to the CPDO-sort-of-inventor-person, and contrary to the above, Ms-CPDO-Inventor seems to be a boy. I hope he did get the gold star and the 50M bonus. Even though, of course, it will be all his fault when Western Capitalism, nay, Civilzation Itself, crumbles beneath our sweatshop-shod feet.