Protection is a contract, not ownership of the bond
A swap, or CDS, transfers a specified credit exposure between two parties. The protection buyer pays for a contingent payment if a defined credit event occurs; the protection seller receives that compensation while assuming the contingent obligation. A single-name contract references one entity or obligation, whereas an index contract combines exposure to multiple reference names. New York Fed research describes both structures. [1]
The reference entity is not necessarily a party to the swap and ordinarily does not receive the buyer’s premium as new financing. That distinguishes a CDS transaction from buying a newly issued corporate bond. A swap can change who bears credit risk without changing how much cash the referenced company has raised.
This separation is useful for hedging and for taking a view on credit. It also creates room for confusion: the notional amount describes the scale used to calculate payments, not a sum automatically advanced to the reference entity or a measure of the holder’s immediate cash loss.
The trigger is defined before the event
A troubled borrower, a falling bond price and a contractual credit event are related but different things. Tradeweb’s publicly filed contract descriptions identify bankruptcy, failure to pay and restructuring among common credit-event categories, while describing settlement according to the contract’s terms. [2] The applicable definitions vary by contract, reference entity and documentation; restructuring is not automatically included in every CDS.
Analysis: a company’s bond could decline because investors expect weaker profitability, higher refinancing costs or a lower recovery in a future failure. Protection can gain market value as that risk is repriced without any event payment being due. Conversely, a transaction described colloquially as a default still needs to satisfy the relevant contractual conditions before the settlement obligation follows.
This makes documentation economically substantive. Seniority, reference obligations, maturity, currency and deliverable-obligation conditions help define which risk has actually been transferred. Two contracts carrying the same company name can differ in exposure and value. An issuer label alone is insufficient evidence that one instrument perfectly offsets another.
A running coupon and an upfront amount are different prices
Standardized CDS can combine a fixed running coupon with an upfront payment that reconciles the coupon with the market value of protection. The New York Fed describes this convention and its use in making contracts more standardized. [1] A quoted market spread therefore need not equal the actual annual coupon on a particular trade.
Consider a hypothetical $5 million notional contract with a 1% annual running coupon. Under a simplified quarter-year calculation, the buyer pays $12,500 per quarter. Actual accrual uses the specified day-count and dates. If comparable protection is now more expensive than that running coupon implies, a new buyer would generally make an additional upfront payment, all else equal. If the coupon is comparatively high, the upfront transfer can run the other way.
Analysis: the upfront amount is a present-value adjustment across the expected remaining premium and protection payments, not simply the spread difference multiplied by the calendar maturity. Default can end the premium stream, discounting matters and recovery assumptions affect valuation. Ignoring those elements can materially misstate a distressed contract’s price.
Worked example: the settlement price determines the payout
For an illustrative cash-settled contract, assume $5 million of notional and a valid credit event followed by an auction final price of 35% of par. The protection payment is $5 million × (1 − 0.35), or $3.25 million, before any applicable accrual or other contractual adjustment. The seller’s obligation is tied to the settlement measure, not necessarily to the precise price at which the buyer acquired its bond.
ICE’s historical credit-event auction primer explains why a common auction price can support cash settlement and reduce the need for every protection buyer to source deliverable bonds. It also describes the combination of auction-related bond trading with cash settlement to achieve an outcome comparable to physical settlement. The primer was updated in February 2010; each actual auction is governed by its own settlement terms. [3]
Suppose the buyer also holds $5 million face value of eligible debt worth $1.75 million at that same assumed 35% value. Adding the $3.25 million protection payment produces $5 million before premiums and other costs. That is a useful hedge illustration. It is not proof of a riskless return: purchase price, funding, settlement timing and any mismatch between the held debt and settlement obligations still matter.
The spread is not an observed default probability
Analysis: in a deliberately simple model with constant default intensity, fixed recovery and no risk premium, a fair annual spread is approximately default intensity multiplied by loss given default. At a 3% spread and 40% assumed recovery, that calculation gives 0.03 divided by 0.60, or 5% annual intensity. Under a constant-intensity assumption, the corresponding five-year cumulative probability is 1 − exp(−0.05 × 5), approximately 22.1%, not simply an exact 25%.
This is model arithmetic, not a forecast for any issuer. In market valuation, the inferred intensity is typically a risk-neutral quantity shaped by prices and assumptions. It need not equal the real-world frequency an analyst would estimate from business fundamentals. Changing assumed recovery changes the implied intensity even if the quoted spread stays fixed.
For example, retaining the 3% spread but assuming 20% recovery gives an approximate intensity of 3.75%. A lower recovery produces a larger loss per default, so the same premium can correspond to a lower modeled event frequency. A statement that a spread directly “means” one default probability without identifying recovery, horizon and model silently discards this ambiguity.
Counterparty risk returns through the protection promise
The CDS buyer depends on payment from the seller or the applicable clearing structure when the reference credit deteriorates. If the protection provider weakens during the same shock, the hedge’s economic value can be impaired precisely when it is most needed. Collateral reduces unsecured exposure but introduces cash and operational requirements.
Analysis: a protection seller can face margin demands well before a reference entity defaults. A rise in the value of protection can create a mark-to-market obligation even if the seller believes the issuer will ultimately repay every bond. The seller therefore needs to maintain the position as well as capital to absorb an eventual credit loss.
CFTC’s clearing-requirement materials describe determinations applying to specified classes of swaps rather than a universal rule that every CDS must clear. [4] Clearing can change the counterparty network and the handling of collateral; it does not make the reference credit incapable of failing or make margin funding costless. The relevant distinction is which risk the arrangement reduces and where the remaining obligations sit.
Bond-CDS basis exposes the limits of replication
Analysis: a bond and a CDS on the same issuer can price differently because the bond requires funding and carries its own , coupon and embedded-option characteristics. The CDS has its own contractual, collateral and counterparty features. The difference between appropriately compared is commonly discussed as the bond-CDS basis, but its exact measurement depends on the bond and valuation convention.
Imagine a bond offering 4% of credit spread while comparable protection costs 3%. The apparent 1% difference is not automatically a free return. Financing the bond, maintaining collateral, trading both legs and bearing any deliverability or maturity mismatch can absorb or exceed it. If funding becomes unavailable, a position with an attractive expected terminal payoff can still be forced to unwind at a loss.
An index hedge adds another layer. A portfolio concentrated in one industry or rating cohort may not move with the index used to hedge it. Even equal notionals do not establish equal sensitivity. The index can reduce broad credit exposure while leaving issuer-specific, sector and maturity differences intact.
Regulation and market evidence require precise scope
U.S. authority is divided. The SEC describes security-based swaps as including swaps based on a single security or loan, a narrow-based group or index, or certain issuer events. [5] Broad-based credit-index swaps generally sit in the CFTC’s swaps framework, while mixed products and particular activities can raise additional questions. Treating every CDS as exclusively CFTC-regulated would erase an important boundary.
The New York Fed’s 2019 staff report uses historical supervisory data to show why institutions’ simultaneous positions across single-name, index and other credit derivatives matter to interpreting activity. Its findings are historical research, not October 2026 market shares. [6] Gross notional alone cannot identify net economic direction, collateralized exposure or the amount likely to change hands after an event.
The broader lesson is that CDS make credit risk more separable and tradable, but not simpler in every dimension. A useful explanation connects the reference exposure, premium convention, event definition and settlement amount with the funding needed to carry the position. The quoted spread is an informative price, provided it is not mistaken for a complete probability forecast or a complete description of risk.
Sources
- Boyarchenko, Costello and Shachar, New York Fed Staff Report 879, revised August 2019; contract structures and fixed-coupon/upfront conventionsOfficial source · PDFBack to text: ↑1↑2
- Tradeweb SEF rule filing hosted by CFTC, February 2021; untranched credit-index contract descriptions, pages 86–90Official source · PDFBack to text: ↑
- ICE-hosted Markit/Creditex Credit Event Auction Primer, updated February 2010; settlement mechanism, not current auction termsSource · PDFBack to text: ↑
- CFTC, Clearing Requirement; rulemaking and determination resourcesOfficial sourceBack to text: ↑
- SEC, Security-Based Swap Data Repositories; statutory scope of SEC authorityFiling / reportBack to text: ↑
- New York Fed, The Long and Short of It: The Post-Crisis Corporate CDS Market, staff report landing page and historical research summaryOfficial sourceBack to text: ↑