Guide

Credit default swaps explained

Harbor Capital ran a $240M high-yield corporate sleeve through a March stress week: credit spreads on the Bloomberg U.S. Corporate High Yield Index widened 68 basis points while Treasury yields barely moved. Selling the entire HY book would have crystallized spread losses, triggered tax lots, and missed the carry investors still wanted. Instead, risk managers bought protection on CDX.NA.HY (the North American high-yield CDS index) at 412 bps running spread on $180M notional — roughly 75% of the sleeve's credit beta. When spreads peaked and retraced 41 bps over the next two weeks, the CDS leg gained about $7.4M while the cash bonds lost $5.1M on spread widening alone, netting a $2.3M cushion without unloading illiquid issues. That is what a credit default swap (CDS) is for: transfer credit exposure without necessarily selling the underlying bond.

A CDS is a bilateral contract where one party pays a periodic premium and receives a payout if a reference entity suffers a defined credit event (typically bankruptcy or failure to pay). It is the cleanest liquid hedge for portfolio managers, the reference rate for many structured products, and a thermometer for systemic stress — but it is not “insurance” in the retail sense: margin calls, basis risk, and documentation quirks can surprise first-time users. This guide covers buyer and seller mechanics, spread pricing as implied default probability, recovery and auction settlement, single-name vs index CDS, the CDS–cash basis, the Harbor Capital hedge refactor, a technique decision table against cash bonds and ETFs, pitfalls, and a production checklist alongside duration and spread analytics.

Contract mechanics: who pays whom

Standard corporate CDS (ISDA 2014 Credit Definitions) has two legs:

Protection buyer — pays a quarterly premium (the CDS spread, quoted in basis points per year on the notional face amount). If a credit event occurs, the buyer receives compensation.
Protection seller — collects the premium and takes on the credit risk. On default, the seller pays out.

Example: buyer pays 300 bps on $10M notional. Annual premium = 0.03 × $10M = $300,000, usually paid quarterly ($75,000). If the reference entity defaults, settlement follows the contract's terms (below).

Key terms on the trade confirmation:

Reference entity — the issuer whose credit is referenced (e.g., a specific corporation) or an index (CDX, iTraxx).
Notional — face amount the protection covers; need not match a single bond's par value.
Maturity — common tenors: 1Y, 3Y, 5Y, 10Y; must align with the hedge horizon.
Restructuring clause — which debt modifications count as credit events (important for sovereign and some IG names).
Currency — USD CDS on USD debt is standard; mismatch adds FX risk.

Unlike an option, CDS have no upfront premium in a “par spread” quote — but when spreads are very wide or very tight, markets quote upfront + running spread (e.g., pay 5% upfront and 100 bps running) to standardize present value to par.

Credit events and settlement

A credit event triggers the swap. For corporates, the usual set includes:

Bankruptcy — filing or effective insolvency.
Failure to pay — missed coupon or principal after grace period.
Restructuring — if the restructuring modifier applies and bondholders accept unfavorable terms.
Repudiation / moratorium — more common in sovereign CDS.

Settlement is almost always cash via an ISDA auction:

An auction determines the recovery rate on deliverable obligations (e.g., 25 cents per dollar of face).
Protection seller pays the buyer: Notional × (100% − Recovery%).
With $10M notional and 25% recovery, seller pays $7.5M; buyer's net economic position mimics owning a bond that just defaulted with that recovery.

Physical delivery (handing over bonds) exists in theory but is rare post-2009 reforms. For index CDS, the index rolls when constituents default; the defaulted name is removed and auctioned separately while the index continues on survivors.

Pricing: spread as implied default risk

The CDS spread is the market's price of credit risk. A simplified back-of-envelope links spread s, annual default probability p, and recovery R (fraction recovered, so loss given default is 1 − R):

s ≈ p × (1 − R) (ignoring discounting and timing).

If recovery is assumed 40% and spread is 300 bps, implied annual default probability is roughly 0.03 / 0.60 = 5%. Real models use hazard rates, survival curves, and risk-neutral discounting — but the intuition holds: wider spread = higher perceived default risk or lower expected recovery.

What moves CDS spreads

Issuer fundamentals — earnings, leverage, rating actions, litigation.
Sector and beta — energy names widen when oil crashes; financials widen on bank-run headlines.
Liquidity and positioning — dealer inventory, macro hedge demand (pension funds buying protection in recessions).
Rate environment — secondary to credit for IG, but refinancing walls matter for HY when Treasury yields spike.
Technical supply — new bond issuance can widen CDS if hedge demand outpaces cash buying.

CDS are quoted on a spread duration (CS01): the dollar change in mark-to-market for a 1 bp move in spread. A 5Y $10M position with CS01 of $4,200 gains ~$4,200 when spreads tighten 1 bp and loses the same when they widen — this is how Harbor sized the hedge ratio against the bond book's spread DV01.

Single-name vs index CDS

Single-name CDS

References one issuer. Best when you hold concentrated exposure to that name (e.g., $50M of one telecom's bonds) and want precise hedging. Downsides: illiquidity in smaller names, idiosyncratic basis to your exact bond (coupon, maturity, seniority).

Index CDS (CDX, iTraxx)

Standardized baskets traded as one instrument:

CDX.NA.IG / CDX.NA.HY — North American investment grade and high yield.
iTraxx Europe — IG and crossover indices.
EM indices — sovereign and corporate EM baskets.

Indices roll every six months (Series N matures, Series N+1 on new constituents). Liquidity concentrates in on-the-run series; off-the-run widens. Harbor used on-the-run CDX.NA.HY because the sleeve was 120+ names — index hedge is cheaper to trade and more liquid than 120 single-name lines.

Index tranches (synthetic CDO slices) once dominated structured credit; post-crisis they are niche but matter for understanding systemic risk narratives.

CDS–cash basis: when the hedge is imperfect

Cash corporate bonds and CDS on the same issuer should price similar credit risk, but they diverge — the CDS–cash basis:

Funding differences — owning bonds ties up balance sheet; CDS is a derivative with margin.
Delivery optionality — cheapest-to-deliver bond at auction may not be the bond you own.
Coupon effect — old high-coupon bonds trade rich vs par; CDS references par notional.
Liquidity premia — stressed markets: bonds gap wider than CDS (negative basis) or CDS blows out faster (positive basis in panics).

Harbor's 75% notional hedge ratio acknowledged basis: their bond basket had slightly wider OAS than the index and longer average duration, so full 100% CDS notional would have over-hedged spread risk while leaving rate exposure intact.

Harbor Capital HY sleeve hedge (worked example)

Problem — $240M HY sleeve, spread duration ~4.2 years; macro desk flagged recession odds rising; PM wanted to keep yield but cap tail loss for a 30-day window.
Instrument — buy protection on CDX.NA.HY Series 42 5Y at 412 bps, $180M notional (75% of face).
Cost — annual premium 0.0412 × $180M ≈ $7.4M; ~31 days of carry cost ≈ $630K (plus initial margin at clearing).
Outcome — HY index spreads +68 bps then −41 bps; CDS MTM +$7.4M at peak, bonds −$5.1M spread component; net cushion $2.3M after premium. PM rolled off protection when spreads normalized.
Accounting note — CDS marked fair value through P&L; cash bonds also MTM; hedge effectiveness tested quarterly under ASC 815 if formal hedge accounting elected (Harbor used economic hedge only).

Alternative rejected: selling HY ETF shorts — basis to basket, dividend drag, and uptick rules made CDS cleaner for a temporary macro overlay.

Technique decision table

Approach	Best when	Trade-off
Buy CDS protection (index)	Diversified book, temporary macro hedge, liquid overlay	Basis to your bonds; ongoing premium; margin calls
Buy single-name CDS	Concentrated issuer risk, merger-arb or loan hold	Illiquidity; documentation nuance per name
Sell cash bonds	Permanent risk reduction, simple governance	Transaction costs, tax, lose carry; illiquid names gap
Short credit ETF	Retail-sized, quick beta hedge	Tracking error, fee drag, not name-specific
Treasury / rate futures	Duration risk dominates	Does not hedge spread widening in credit crises
Put options on equity	Issuer is public; equity–credit correlation high	Vol premium, strike timing; poor hedge for pure IG

Common pitfalls

Wrong reference obligation — senior unsecured CDS does not hedge subordinated loans; match the capital structure.
Ignoring restructuring clause — modified restructuring vs full restructuring changes payout on distressed exchanges.
100% notional hedge — basis and duration mismatch cause over-hedge P&L noise; size with CS01, not face value alone.
Off-the-run index — wide bid-ask and roll jumps when series changes; plan rolls ahead of liquidity cliffs.
Margin spiral — protection buyers gain MTM in crises but sellers face calls; cleared CDS (ICE Clear Credit) reduces counterparty risk but not margin volatility.
Default ≠ immediate payout — auction timing can take weeks; liquidity plan must cover the gap.
Confusing CDS with total return swaps — TRS transfer bond total return including coupons; CDS is pure credit event insurance.
Regulatory constraints — some funds cannot use derivatives or face notional limits; check prospectus before trading.

Production checklist

Map portfolio spread DV01 (or CS01) by issuer, sector, and rating bucket.
Choose single-name vs index based on concentration and liquidity.
Match restructuring clause and seniority to underlying obligations.
Size notional from CS01 hedge ratio, not naive 1:1 face value.
Model premium carry and worst-case margin under +100 bp spread shock.
Confirm ISDA/master agreement and CSA with counterparty or clearing member.
Document economic vs hedge accounting treatment with controllers.
Plan index roll schedule six months ahead for on-the-run liquidity.
Stress-test basis: bond OAS vs CDS spread divergence in 2008-style scenarios.
Reconcile daily: CDS MTM, bond spread P&L, net hedge effectiveness.

Key takeaways

CDS transfer credit risk without selling bonds — the protection buyer pays spread premium and receives par minus recovery on a credit event.
Spread ≈ implied default probability × loss given default — CS01 sizes how much MTM moves per basis point.
Index CDS (CDX, iTraxx) are the liquid macro hedge for diversified corporate books; single-name CDS fit concentrated exposure.
CDS–cash basis means hedges are rarely perfect — Harbor used 75% notional and spread duration math, not face-value matching.
Margin, documentation, and auction timing are operational risks — treat CDS as a derivatives program, not a fire-and-forget insurance policy.