Credit Default Swaps Contracts

A credit default swap lets you buy or sell insurance on someone else's debt — and you don't need to own the debt to do it. That single feature makes CDS the most powerful (and most misunderstood) instrument in credit markets. The global CDS market hit $9.2 trillion notional outstanding at end-2024 (BIS/ISDA, 2024), but the real story is concentration: of 725 unique single-name reference entities traded in Q2 2024, only 26 names averaged 10+ trades per day (IOSCO, 2025). You're not trading in a deep, anonymous market. You're trading in a club.

The point is: CDS is straightforward in concept (pay a premium, get paid if the reference entity defaults) but treacherous in execution. The mechanics that determine whether you actually collect — auction dynamics, deliverable obligation disputes, cheapest-to-deliver games — are where the money is made or lost.

How CDS Actually Works (The Cash Flows)

Protection buyer pays a periodic premium (the "spread") in exchange for a contingent payout if the reference entity hits a credit event. Protection seller collects premiums and bears the loss.

Standard coupons are fixed:

• Investment grade: 100 bps (1%) annual, paid quarterly
• High yield: 500 bps (5%) annual, paid quarterly
• Payment dates: March 20, June 20, September 20, December 20 (IMM dates)

Because actual market spreads rarely match these fixed coupons, money changes hands upfront. The approximation:

Upfront payment = (Market spread - Fixed coupon) x Duration

If an IG name trades at 60 bps with ~4.5 year duration, the protection buyer receives upfront: (100 - 60) x 4.5 = 180 bps = 1.8% of notional — because the fixed coupon overpays relative to where the market prices the risk. If spreads are above the fixed coupon, the buyer pays the difference upfront.

Why this matters: Upfront economics determine P&L on day one. A protection buyer on a 350 bps HY name receives (500 - 350) x ~4 = 600 bps = 6% of notional at inception. That's real cash, not theoretical value — and it reverses if the spread compresses.

Credit Events (What Actually Triggers Payment)

Standard ISDA credit events:

• Bankruptcy: Legal insolvency proceedings filed
• Failure to pay: Missed payment beyond grace period (must result from actual credit deterioration since the 2019 NTCE amendment — designed to block manufactured defaults like the Hovnanian case, where a company deliberately triggered a technical default to benefit a CDS holder's position)
• Restructuring: Material modification of debt terms (excluded from most North American contracts; included as Mod-R in European contracts)
• Governmental intervention: Government imposes binding changes on obligations (added in 2014, designed after Eurozone bank bail-ins like the SNS Reaal nationalization)

The practical point: Whether your credit event "counts" is decided by the ISDA Determinations Committee — a panel of dealers and buy-side firms that votes on whether events qualify. This isn't automatic. The committee's governance was independently reviewed in 2024 and restructured in 2025, with a new Credit Derivatives Governance Committee overseeing operations (ISDA, 2025).

Settlement Mechanics (Where the Money Moves)

Since the 2009 Big Bang Protocol, virtually all CDS settle via auction-determined cash settlement. Physical settlement (delivering defaulted bonds for par) still technically exists but accounts for less than 1% of outcomes.

How the auction works:

1. ISDA Determinations Committee declares a credit event
2. Dealers submit initial bids and offers for the defaulted debt
3. Market participants submit physical settlement requests (net buy or sell)
4. A Dutch auction matches remaining open interest at a single clearing price (the "recovery rate")
5. Protection sellers pay: Notional x (100% - Recovery Rate)

Atos SE: What a Real Credit Event Looks Like (2024)

Atos entered accelerated safeguard proceedings in France in 2024. The DC classified this as a Bankruptcy Credit Event. The auction ran October 9, 2024.

Here's where cheapest-to-deliver matters. Atos had multiple deliverable obligations — unsecured bonds, a term loan, and a EUR 900 million revolving credit facility added to the deliverable list on August 30, 2024. The RCF traded far below the bonds. Protection buyers delivering the cheapest obligation pushed the final auction price to 3 cents on the dollar (down from an 8-cent initial midpoint).

If you sold protection on Atos at $10 million notional, your payout: $10M x (1 - 0.03) = $9.7 million loss.

What matters here: Recovery rate is not some abstract number from a Moody's table. It's determined by a live auction where deliverable obligation selection creates real economic consequences. The cheapest-to-deliver dynamic means protection sellers routinely pay more than the "expected" recovery would suggest.

Ardagh Packaging: When the Auction Itself Gets Contested (2025)

Ardagh Packaging triggered a Restructuring Credit Event in late 2025, affecting $3.5 billion gross notional across 1,100 outstanding contracts. An External Review Panel confirmed the restructuring in October 2025.

The twist: Arini Capital Management challenged whether senior unsecured notes should qualify as deliverable obligations — a dispute still live as of January 2026. This is not a theoretical risk. Deliverable obligation definitions are the highest-stakes legal question in CDS, and it plays out in real time with billions on the line.

The Index vs. Single-Name Split (Know What You're Trading)

Index CDS accounts for 94.7% of traded notional (Q2 2025) — single-name is just 5.3% of volume but 35.4% of trade count (many smaller, less liquid transactions). When people say they "traded CDS," they usually mean index (ISDA, 2025).

The test: If you're hedging a specific issuer with single-name CDS, can you exit that position in a stress scenario? The bid-ask on a single-name HY contract can blow out to 50+ bps during volatility. The April 2025 tariff shock pushed CDX NA HY spreads from ~300 bps to 376 bps in a single month; single-name markets were far worse.

Risks That Actually Bite

Wrong-way risk is the killer: your protection seller is most likely to default precisely when the reference entity defaults (because both are exposed to the same systemic stress). This is not theoretical — it's what happened across dealer networks in 2008. 83.1% of index CDS and 48.2% of single-name CDS now clear centrally (reducing but not eliminating this risk).

Basis risk between CDS spreads and cash bond spreads creates both opportunity and danger. The CDS-bond basis (CDS spread minus bond credit spread) should theoretically be zero, but funding costs, liquidity differences, and cheapest-to-deliver dynamics keep it persistently nonzero. When the basis goes negative (bond spread > CDS spread), arbitrageurs buy the bond and buy CDS protection — the basis trade. Research shows the residual basis strongly predicts excess returns: a portfolio formed on residual basis generates 1.79% abnormal return at 20-day horizon (BIS Working Paper 631).

Documentation risk is the sleeper. Whether a restructuring "counts" as a credit event, which obligations are deliverable, and whether the DC votes your way — these are not guaranteed outcomes. The Ardagh dispute proves it.

Counterparty Exposure Math (Work Through This)

CDS generates mark-to-market counterparty exposure as spreads move:

Setup: You bought $10M of 5-year protection at 150 bps. Spreads widen to 500 bps.

Your MTM gain: $10,000,000 x (5.00% - 1.50%) x ~4 years (duration) = $1,400,000

That $1.4 million is now your counterparty exposure to the protection seller. If the seller defaults with spreads at 500 bps, you lose both your hedge and your accumulated gain. This is why margining exists — and why $431 billion in initial margin and $1.0 trillion in variation margin sat across the derivatives market at year-end 2024 (ISDA Margin Survey, 2025).

Checklist: CDS Position Management

Essential (Do These First)

• Confirm reference entity legal name, RED code, and seniority
• Verify which credit events apply (no-R for North American; Mod-R for European)
• Calculate upfront payment economics before execution
• Know your deliverable obligation universe (it determines your worst-case recovery)

High-Impact (Ongoing Monitoring)

• Track spread movements against your entry level (daily for HY, weekly for IG)
• Monitor ISDA DC announcements for your reference entities
• Calculate counterparty exposure as spreads move; verify margin adequacy
• Review index roll dates (March and September) — series transitions affect liquidity

Advanced (For Active CDS Books)

• Decompose CDS-bond basis to identify relative value opportunities
• Monitor cheapest-to-deliver dynamics for your reference entities
• Track clearing rates and margin model changes (ISDA SIMM recalibrates semiannually)
• Stress test recovery assumptions using recent auction outcomes (Atos at 3 cents, not textbook 40%)

---

The bottom line: CDS is a precisely engineered tool for transferring credit risk — and the engineering matters. The difference between "I hedged my credit exposure" and "I actually get paid when the credit event happens" lives in the details: which events trigger, which obligations are deliverable, and what the auction produces. Textbook examples assume 40% recovery. Atos recovered 3 cents. The question isn't whether you have CDS protection; it's whether you've stress-tested the mechanics of how you'll actually collect.

Citation: ISDA. (2024). Key Trends in OTC Derivatives, H2 2024. IOSCO. (2025). Single-Name CDS Market Report. BIS Working Paper 631: CDS-Bond Basis.