Credit Derivatives and Credit-Linked Notes: CDS, Portfolio Products, and Funded Credit Protection
Book: Structured Finance and Insurance: The ART of Managing Capital and Risk Author: Christopher L. Culp Publisher: Wiley Finance, 2006 ISBN: 978-0-471-70631-1
Chapter 12 wraps up Part 2 on Traditional Risk Transfer by covering credit derivatives. These are the most insurance-like of all derivatives. They sit right at the boundary between the insurance world and the capital markets world, and understanding them is essential for everything in the structured finance chapters ahead.
Credit derivatives went from virtually unused in 1995 to an estimated $8.42 trillion in notional principal by year-end 2004. That growth was not an accident. Credit derivatives solve real problems that neither insurance nor traditional derivatives handle well on their own.
Credit Derivatives vs. Credit Insurance
The core distinction is the same one we have been tracking all book. Insurance requires indemnity and insurable interest. Derivatives are parametric. Credit derivatives are parametric contracts triggered by credit events on specific reference assets or names, but the buyer does not need to own the reference asset or suffer any actual loss.
In practice, credit derivatives work a lot like credit insurance. The protection buyer makes periodic payments to the seller. If a default occurs, the seller compensates the buyer. But because there is no insurable interest requirement, anyone can buy or sell credit protection on any reference name. This opens up the market to participants who want to express a view on credit risk without owning the underlying bonds.
Single-Name Credit Default Swaps
The CDS is the workhorse of credit derivatives. Despite the name “swap,” it functions more like an option. The protection buyer makes periodic spread payments to the seller. If a default occurs on an eligible obligation of the reference name, the seller pays off.
Settlement Methods
Cash-settled CDS: The seller pays par minus the current market price of the defaulted asset. The market price reflects expected recovery. This is like an indemnity insurance contract.
Digital CDS: The seller pays a fixed amount regardless of the actual loss. This is the derivatives version of a valued insurance contract. Used when the reference asset is so illiquid that agreeing on a post-default price would be difficult.
Physically settled CDS: The buyer delivers the defaulted bond to the seller and receives par value in cash. The seller then owns the bond and whatever recovery it produces.
If you own the reference asset and think you can beat the expected recovery, cash settlement is better because you keep the bond and the recovery right. Physical settlement is better if you want to wash your hands of the whole thing.
CDS Spread and Pricing
The CDS spread is economically equivalent to a credit insurance premium. It is also approximately equal to the bond’s yield spread over LIBOR (the financing rate). If a bond yields LIBOR + 50bp and can be financed at LIBOR, the CDS spread should be around 50bp. In practice, several factors create small pricing differences.
No Insurable Interest
This is the biggest practical difference from insurance. You can buy CDS protection on Ford without owning any Ford bonds. This means CDS trading volumes can far exceed the actual bonds outstanding. It also means that CDS markets serve a price discovery function for credit risk that the bond market alone could not provide.
Trigger Events
The ISDA credit derivatives documentation gives counterparties flexibility to define what constitutes a default. Options include: failure to pay, insolvency filing, downgrade, cross-default on other obligations, or repudiation. Restructuring as a trigger has been especially contentious. Protection buyers want it included because restructuring usually means weaker payment terms. Dealers push back because restructuring is hard to define precisely.
CDS vs. Bonds
A CDS is basically a financed bond position, but not exactly. A few differences:
- Par assumption: A CDS settles at par, but bonds may trade at a discount. The loss on an actual bond is market price minus post-default price, not par minus post-default price.
- Cheapest to deliver: Physically settled CDSs often specify multiple deliverable obligations. The seller will deliver the cheapest one, which creates an option that widens CDS spreads.
- Accrued interest: CDS spread payments stop at default. Bond accrued interest may be lost entirely.
- Liquidity: CDSs are generally less liquid than bonds, which tends to widen spreads.
CDS vs. Financial Guarantees
CDSs and guarantees are economically similar but differ in important ways:
- CDS triggers tend to be broader than insurance triggers. A CDS can be triggered by any publicly disclosed default. Insurance can be triggered by undisclosed defaults.
- When a CDS triggers, it terminates. Insurance stays active until the policy limit is reached. Reinstatement provisions may even allow extending coverage.
- Tax and accounting treatment differs. Guarantees get insurance treatment. CDSs must be marked to market. CDS spreads usually are not tax-deductible the way insurance premiums are.
- Unwinding a CDS is relatively easy with the buyer’s consent. Getting out of an insurance obligation is much harder. The best the seller can do is buy reinsurance.
Portfolio Credit Default Swaps
Portfolio products cover defaults across multiple names. This market grew 49% in 2003 to $754 billion.
Basket CDS
A basket CDS covers a portfolio of reference names. It could theoretically cover all defaults on all names, but that would be very expensive. More commonly, it covers a specific subset of defaults and then terminates.
Nth-to-Default CDS
This pays off when the nth default occurs in the portfolio. A first-to-default CDS pays on the first default, regardless of which name it is, and then terminates. A second-to-default CDS ignores the first default and pays only on the second.
This is useful when you know the expected frequency of defaults but not which specific names will default. A bank expecting 10% of its 100 loans to default could set aside loan loss reserves for the first 10 defaults and then buy an 11th-to-default CDS. It protects against worse-than-expected loss rates without tying up capital for that 11th loan.
Senior and Subordinated Baskets
Basket CDSs can be layered like reinsurance. In a portfolio of 10 names at $1M each, the subordinated basket might cover the first 6 defaults and the senior basket covers defaults 7 through 10. You can add deductibles and limits on top of that, creating structures that look very much like the XOL reinsurance treaties from Chapter 9.
Correlation Risk
Here is where it gets interesting. Basket products are all about default correlation. If defaults are uncorrelated, more than 2-3 defaults in a year on investment-grade credits is very unlikely. An nth-to-default CDS for n above 3 would be nearly worthless. But if defaults are highly correlated (think an economic recession hitting an entire sector), the first-to-default and nth-to-default products are essentially the same thing.
Pricing and hedging basket CDSs requires modeling the correlation structure of the underlying portfolio. This turned out to be extremely difficult in practice, and getting it wrong had enormous consequences. But that is a story for later chapters.
Credit Indexes
In 2003-2004, tradable CDS indexes emerged. The iTraxx Europe covers 125 equally weighted European names. The Dow Jones CDX covers North America and emerging markets. These indexes include sector sub-indexes, high-volatility indexes, and crossover indexes for names at the edge of investment grade.
These are not just published benchmarks. They are tradable products. Their emergence explains a lot of the explosive growth in portfolio credit derivatives.
Other Credit Derivative Types
Asset Default Swaps (ADS)
A CDS where the reference entity is a securitized product rather than a corporate name. Volume has been low so far, but ADSs are used inside synthetic CDOs.
Equity Default Swaps (EDS)
Protection on a stock price. The trigger is a large decline (usually around 70%) in the reference stock. Since stocks do not “default” like bonds, the trigger is called an “equity event.” An EDS is essentially a deeply out-of-the-money put option, sometimes structured as a down-and-in barrier put.
Total Return Swaps (TRS)
The protection seller pays LIBOR plus a spread. The protection buyer pays all interest income on the reference portfolio plus any change in its market value. If the portfolio gains value, the buyer pays more. If it loses value, the seller compensates the buyer.
A TRS is broader than a CDS. It protects against not just defaults but any decline in value, including downgrades and adverse market moves. The notional amount can be larger than the portfolio’s face value, creating synthetic leverage. Or smaller, creating a partial synthetic sale.
Credit-Linked Notes (CLN)
A CLN is the funded version of a CDS. Instead of an unfunded promise to pay, the protection seller prepays the potential loss by buying a bond.
The structure: the entity seeking credit protection issues a note to investors. As long as no defaults occur on the reference portfolio, investors receive regular interest (base rate plus a credit spread equivalent to a CDS premium). If a default occurs, the issuer withholds interest and/or principal to cover the loss.
From the issuer’s perspective, a CLN equals a normal bond plus a CDS purchased from the bondholders. The par value of the note covers the maximum possible CDS payout. Investors get an above-market yield in exchange for bearing default risk on assets they may not even own.
The big advantage over an unfunded CDS: no counterparty credit risk for the protection buyer. The money is already there, locked in the bond. The disadvantage: CLNs are bonds that need to be marketed, possibly rated, and distributed to investors. This adds cost and complexity.
CLNs can incorporate any credit derivative structure discussed in this chapter. A first-to-default note is a bond plus a first-to-default basket CDS. The reference portfolio, trigger events, and payout structures can be customized almost without limit.
My Take
This chapter completes the toolkit for traditional risk transfer. We now have insurance (Chapters 8-10) and derivatives (Chapters 11-12) fully laid out. The recurring theme is the convergence between these two worlds. Credit derivatives are parametric but function like insurance. Insurance contracts have payoffs that look like options. The line between them is legal and regulatory, not economic.
The most important concept for what comes next is the funded vs. unfunded distinction. A CDS is unfunded (a promise). A CLN is funded (cash prepaid). This difference drives the entire architecture of structured finance in Part 3. When you turn credit protection into a tradable bond, you have created something that looks a lot like the securitized products we are about to encounter.
The correlation risk issue in basket CDSs is also worth flagging. Culp mentions it briefly here, but it becomes one of the most consequential modeling challenges in all of structured finance. Getting default correlation wrong was one of the central failures of the 2007-2008 financial crisis. But that is several chapters away.