Another day, another derivative. This month's high-finance innovation is the CPDO, or "Constant Proportion Debt Obligation", and it is truly a wonder. In practical terms, a CPDO is nothing more or less than a synthetic bond. Investors pay money up front, receive coupon payments, and their principal is returned after a set period of time. Investors stand to lose if borrowers default or credit conditions deteriorate. But CPDOs work a secret miracle. These synthetic bonds are designed to score a "triple-A" grade from major bond rating organizations, while paying a spread of up to 2% more than "natural" AAA debt!

CPDOs appear to violate the core constraint of finance, the no arbitrage rule. If the ratings are accurate, selling short a portfolio of ordinary AAA debt and purchasing a portfolio of CPDOs would be a perfect arbitrage, earning risk-free profit for the arbitrageur with no net outlay of capital. Either the CPDO opportunity must be transient (because the number of issues that can be synthesized is limited, or because CPDO and AAA yields will soon converge), or the ratings must be wrong. Or else the wizards at ABN have invented an infinite free-money machine for well-placed arbitrageurs, the financial equivalent of a perpetual motion machine.

So is there a catch? And if so, what is it? Let's first understand how a CPDO works. Despite the complicated acronym, it's not rocket science.

A CPDO issuer accepts principal from investors, and commits up front to a coupon and principal repayment schedule. The issuer puts the money in a leveraged portfolio that includes high yield, risky debt (or credit derivatives), earning a yield higher than would be required to cover coupon payments to investors. In the most benign scenario, after a while, the CPDO portfolio earns enough extra money to trade in the risky debt for a risk-free portfolio of government bonds sufficient to cover the coupon and principal repayments promised to investors. Thus, the CPDO issuer has temporarily taken on credit risk to earn the promised excess spread, and then quickly locks in gains by putting investor assets into ordinary AAA bonds.

But what happens if something goes wrong? Suppose that while the CPDO holds its leveraged, risky portfolio, credit conditions deteriorate. Then the portfolio loses value, and the issuer's ability to meet the agreed-upon payment schedule becomes uncertain! Wouldn't this possibility translate into lower-than-perfect ratings by rating agencies? You might think so. But the CPDO-issuer makes a promise that offsets this risk. The CPDO-issuer promises that if the risky portfolio loses money, the CPDO will double-down, increasing the degree of leverage as required to make up for the loss and meet the structure's promised payment schedule to investors.

Well, that makes me feel better. You? Let's give the devil her due: This is a very model-tested approach. CPDO-issuers have carefully reviewed credit-spread history, and have come up with rebalancing-and-releveraging schemes that should nearly always manage to recoup losses. If there is no structural change in the bond markets, if the markets behave as models say they behave, then the likelihood that a CPDO will experience a sufficiently long sequence of adverse events to prevent the doubling-down strategy from recouping losses is very, very small, comparable to the probability of a default on an ordinary AAA-rated bond. And the independent bond rating agencies have double-checked this work. All of the bond industry's prevailing models support the view that a "perfect storm" of deteriorating credit conditions sufficient to tank a CPDO is no more likely than, say, France defaulting on its sovereign debt.

But what are the odds that some structural change in credit markets occurs, such that industry-standard models no longer hold? There is no good way to attach a probability to that event. Structural change in financial markets does happen, usually accompanied by what from the perspective of earlier models look like improbable "long-tail" events. But there is no "meta-model" that we can trust to estimate their likelihood of structural change. We are left with nothing but human judgment to decide whether the model-generated AAA rating of a CPDO issue is in fact as sure as a the same AAA on a traditional "risk-free" bond, given market conditions likely to prevail in the future, rather than conditions of the recent past on which the models were based.

This is good news. There are no perpetual motion machines, no huge gaps in the theory of finance. We can understand where the extra spread in a AAA-rated CPDO comes from: It is a model risk spread. CPDO-buyers will rationally price-in model risk, the risk that despite what Fitch or Moody's says, these complex, "gambler's ruin"-style instruments might not handle a changing credit environment as well as traditional AAA debt. Taking model risk into account, CPDOs ought to have a rating somewhat below ordinary super-high-quality bonds. But when bond-rating agencies generate their ratings from their models, model risk is left out of the equation. And that fact is the loophole these instruments are really designed to exploit.

Who is buying CPDOs? Where is the excitement? If it were true that these instruments were every bit as safe as government debt, but paid a higher yield, nearly every investor would want them in their portfolio. But investors understand model risk. While there is general demand, a specific sort of investor is particularly enthralled by CPDOs. From a Fitch report on these instruments:

[The evolution from earlier principal-insured products ("CPPIs") to CPDOs] ...is mainly driven by Basel II: under the revised international capital framework, bank investors are likely to need a rating on both principal and coupon for their credit investments. [1]

Banks always face a trade-off between safety and profitability — the more risks they take with depositors' money, the more profit they can earn. The Basel II regulatory framework requires banks either to hold less-risky portfolios, or to hold high levels of capital in reserve. Either approach exerts a drag on bank profitability (in the interest of depositor and taxpayer safety). Under Basel II, the safety of bank investment portfolios is judged in part by the ratings on the debt they hold. If a bank finds an instrument that offers an unusually high yield for its rating, that is an opportunity for the bank to increase its profitability without increasing reserves. If the rating of the offering understates its real risk, its availability effectively allows banks to circumvent the spirit of the Basel II reserve requirements.

Bank investors understand as well as non-bank investors that, due to model risk, CPDOs are not as safe as ordinary AAA bonds. But bank investors aren't looking for safety. They are looking for ways to marry the appearance of safety before regulators with opportunities to enhance profits by taking on risk. One risk not included in credit ratings is credit raters' model risk. The investment industry, constantly innovating to serve customers, has invented an instrument that exchanges credit risk (reflected in ratings) for model risk (excluded from ratings), allowing banks to have their risk and hide it too. If all goes well, banks earn more money. If all goes poorly, taxpayers cover depositor losses, while bank managers demur that they complied with regulatory requirements to the letter.

Truly, this is the golden age of finance!