I spent the best part of an hour on the phone today with Steve Lobb, global head of structured credit marketing at ABN Amro in London, and a man who has as much claim as anybody to inventing the CPDO. I'm now much better informed about this instrument than I was in the past, and a few things make a lot more sense to me – especially this Citigroup report on CPDOs, which is the first best thing you should read on the subject if you really want to understand it.

Specifically, I think I understand where some of the confusion is coming from. First, there's Steve Waldman's idea of ratings arbitrage: that a CPDO AAA is meant to be in some way "statistically indistinguishable from an ordinary AAA".

I think what's happening here is that Steve is confusing issuer ratings with issue ratings. A CPDO is rated AAA, and all that means is that there's an extremely high (greater than 99.2%, or thereabouts) chance that all the coupons and the principal of that particular issue will be paid in full and on time. The amount of money in the structure can decline precipitously, to well below the amount of money that needs to be repaid at final maturity. But so long as the interest and principal payments are made, there's no default.

The case of a normal AAA issuer, on the other hand (France, or the World Bank, or ExxonMobil) is very different. Such issuers can issue at any time, at any maturity, and the market will never have any worries about their ability to repay. In other words, those issuers, could, if they wanted to, repay any of their outstanding bond issues at any time. CPDOs don't work like that at all: if you want to get your money back in full, you need to wait until maturity. It's the difference, if you will, between a zero-coupon Treasury bond, which can be sold at a risk-free yield at any time, and the zero-coupon Treasury-bond collateral on Brady bonds, which can only be sold when it matures.

To give you an idea what I'm talking about, here's a chart from the Citigoup research paper. CPDOs, like any credit product, are damaged by extreme spread widening, and benefit if spreads tighten. At the moment, spreads on investment-grade corporate debt are in the 40bp range, but what happens if they gap out in a year's time? (As Citigroup notes, that's when the structure is at its most vulnerable, because that's when it has the highest leverage.) Here's the results:

The CPDO has promised to pay out a certain coupon, and then par at maturity. The combination of spread widening and coupon payments is damaging to the value in the CPDO, which can drop significantly. But even if spreads gap out from 40bp to 190bp, the instrument still makes its money back by maturity. There's volatility along the way – and the CPDO certainly couldn't issue new triple-A debt at say year two, as a plain-vanilla triple-A issuer like France can. So a CPDO is not, and does not pretend to be, a proxy for other triple-A debt. But if you buy your CPDO, go to sleep, and then wake up in 10 years' time, you can be very confident that all your coupon and principal payments will have arrived in full and as scheduled.

There are other confusions, too, but I'll take just one more before leaving for the weekend, from John Dizard in the FT.

A significant part of the return on the CPDO comes from the effect of the "roll down" of credit risk. You start out owning five-year "protection", or credit exposure, then at the end of the period own four-and-a-half year protection. Because this should be six months less risky than the previous five-year exposures, the spreads should narrow, and the investor will have a tiny mark-to-market profit. That profit is levered up 15 to one by the structure, and that is where part of the 200 basis points over Libor return comes from.

Here's the graph from the Citi report – and remember that Citigroup has not structured any CPDOs, and does not have a dog in this race.

At the moment, S&P assumes that roll down will be 7%. If it's less than that, the CPDO suffers. But it doesn't break. In fact, there can be no roll down at all, and the CPDO still pays out in full. All that happens, just as in the case of the spread widening, is that the amount of leverage in the structure stays higher for longer.

All of the criticism of CPDOs misses the really clever thing about them: the fact that, contrary to what many commentators are saying, they're largely immune to corporate defaults. If a lot of investment-grade companies suddenly were to start defaulting, that would be very bad for the markets, and spreads would widen. But as spreads widen, so does the income on CPDO structures. So one of the only ways to break a CPDO is for there to be a lot of defaults along with spread tightening. Which is something I'm happy to bet will never happen. (Remember we're talking about investment-grade defaults here: think Enron. They happen, but they're not very common. And there would need to be a lot of them in a very short time, since CPDOs incorporate 250 different credits.)

A CPDO which starts out with 15x leverage is not intended to stay at 15x leverage for long. If things stay relatively normal, the CPDO will end up in risk-free securities with no leverage relatively quickly. On the other hand, if things go to pot, then the CPDO will have to keep that leverage for longer in order to make all of its scheduled payments. That's why it has a ten-year maturity: to be very safe on that front. The large leverage doesn't only increase the coupon rate; it also increases the amount of losses that the structure can recoup.

It's true that a triple-A security paying 200bp over Libor seems a bit unnatural; it's also true that with leverage often comes increased risk. But those are just lessons learned from experience. Maybe the experience of CPDOs will help to change those lessons.