Credit default swaps (CDS)

Definition

A credit default swap is insurance-like protection on a borrower’s default: the protection BUYER pays a running premium (the CDS spread, in basis points per year); the SELLER pays out if a defined CREDIT EVENT (bankruptcy, failure to pay, restructuring — per ISDA definitions) hits the reference entity. For non-participants, CDS matter as an INFORMATION instrument: the spread is a continuously-traded, default-focused price on any large borrower — often the fastest-moving public gauge of solvency concern (bank CDS in March 2023 moved before equity in several names — episode-banking-stress-2023), and the underlying of the credit indices (CDX, iTraxx) that price credit-market stress in real time.

How it works / structure

• The contract mechanics: standardized coupons with upfront exchanges, quarterly payments, ISDA-defined credit events adjudicated by Determinations Committees, settlement via AUCTION establishing recovery value (protection pays par minus recovery); post-2008 reforms moved index and much single-name volume to central clearing (the documented counterparty-risk fix).
• Reading the spread (engine-relevant): rough translation — spread ≈ annual default probability × (1 − recovery); at 40% assumed recovery, 300bp implies ~5% annual default probability (crude, regime-adequate); the TERM structure adds information — INVERTED single-name CDS curves (1-year above 5-year) are the documented imminent-distress signature.
• The basis and the indices: CDS vs cash-bond spread (the basis) reflects funding, liquidity, and deliverability — its blowouts are stress gauges in themselves (2008’s negative basis is documented); CDX IG/HY indices give the tradable market-level read that indicator-credit-spreads uses alongside cash OAS.
• The 2008 lessons carried: AIG’s uncollateralized protection selling (insurance-scale losses without insurance-scale reserves) and counterparty daisy chains motivated clearing reform (episode-gfc-2008); the “empty creditor” problem (protection holders preferring default) is the documented governance distortion.

When it applies

Solvency monitoring on levered names and banks (single-name CDS as the fast gauge — where quoted); credit-regime dashboards (CDX indices intraday); ext-credit-analysis cross-checks (CDS vs Z-score vs bond prices triangulate); sovereign-stress reads (episode-euro-crisis-2012 traded substantially in sovereign CDS); event anticipation (event-rating-actions — CDS reprices before agencies act, documented).

Risk profile & failure modes

• Thin single-name reality: outside indices and large names, single-name CDS liquidity has shrunk materially (documented post-crisis decline) — quoted spreads on smaller names can be stale indications, not tradable prices.
• Definitional edge cases: what counts as a credit event has produced documented controversies (restructuring definitions, “manufactured defaults” — narrowly-engineered triggers); the instrument’s precision is legal, not intuitive.
• Recovery assumption sensitivity: the spread-to-probability translation swings on assumed recovery — distress-regime recoveries vary widely (documented dispersion); treat implied probabilities as ranges.
• Information asymmetry: CDS markets are dealer/institutional — retail reads the prices without the flow context; sudden moves may be one hedger, not a verdict (size and persistence before conclusions).

Evidence & limits

ISDA documentation defines the contract machinery; clearing reform, auction settlement, and the 2008 record are documented regulatory history; the spread-probability arithmetic is standard. Real-time single-name data is paywalled for most retail users — the KB flags data-access limits where the gauge is recommended.

Falsifiable-thesis examples

Illustrations only, not signals:

• “Names whose 1y/5y CDS curve inverts underperform equity-wise over the following quarter (distress-signature check)” — falsified by the cohort return spread.
• “CDS spreads of downgrade candidates widen >30 days before the agency action in most cases (anticipation thesis)” — falsified by the event-time alignment study.

Cross-references

• The analysis frame: ext-credit-analysis; the market-level gauge: indicator-credit-spreads
• The instrument context: ext-bonds-rates; the rating seam: event-rating-actions
• The episode exhibits: episode-gfc-2008, episode-banking-stress-2023