Fixed Income

High-Yield Bond

Also known as: Junk bond, Sub-investment-grade bond, HY

Bonds from borrowers the rating agencies doubt — priced somewhere between fixed income and equity, behaving like both.

Asset class
Fixed income (sub-investment grade)
Instrument type
Corporate bond, rated BB+ and below
Traded
OTC dealer market + ETFs; ~$1.5tn (US) outstanding
Typical users
Credit funds, insurers (limited), income ETFs
The price–yield seesaw applies — but for high yield, the yield axis is mostly credit spread, and the real risk is the issuer, not the rate.
y₀Price–yieldYield to maturityBond price
1 · SnapshotThe one idea to remember
Key intuition: an investment-grade bond's main enemy is interest rates; a high-yield bond's main enemy is the business itself. You are lending to companies that need things to go roughly right — and being paid roughly correctly, on average, across the cycle.
2 · BeginnerWhat is it, really?

A high-yield bond is a corporate bond rated below investment grade — BB+ or lower — issued by companies with real default risk: heavily indebted, cyclical, young, or recently fallen from grace. The market pays you visibly for that risk: coupons and yields several percentage points above safe bonds, historically 3–5% extra in calm times and 10%+ in panics.

The asset class was effectively invented in the 1980s, when Michael Milken's desk at Drexel Burnham turned an odd-lot backwater of "fallen angels" into a primary market: companies could now issue speculative-grade debt at scale, which financed the LBO boom, a generation of telecom and cable buildouts — and Drexel's own indictment. The market matured; the nickname "junk" stuck.

Behaviourally, high yield is a hybrid: it pays coupons like a bond but sells off with equities in recessions, because both price the same question — will this company survive? A high-yield fund is closer to a defensive equity position than to a government bond portfolio, whatever the word "bond" suggests.

3 · IntermediateHow it works in practice

The arithmetic of the spread

The spread must cover expected default losses, with the remainder as risk premium:

$$ s \;\approx\; \underbrace{p_d \times (1 - R)}_{\text{expected loss}} \;+\; \underbrace{\text{risk premium}}_{\text{what you actually earn}} $$

Long-run HY default rates average ~3–4% a year (peaking above 10% in recessions) with recoveries near 40%: expected loss ≈ 2%. Historical spreads averaged ~4–5% — the difference is why diversified high yield has beaten investment grade over most long windows, and the crises are why it hasn't felt like it.

Structural features that define the market

  • Callable, almost always: issuers can refinance after 3–5 years at declining premiums — your upside is capped near the call price, so bonds trade on yield-to-worst, not yield-to-maturity.
  • Covenants: HY indentures restrict extra debt, dividends and asset sales — protections IG bonds don't need and HY investors negotiate hard for (and have been losing gradually for a decade; "covenant-quality scores" track the erosion).
  • Ratings tiers: BB behaves like almost-IG; single-B is the market's core; CCC is equity in a bond costume, driving most of the index's defaults and most of its recovery-rally upside.
  • Fallen angels vs. original issue: bonds demoted from IG have historically outperformed same-rated original-issue HY — forced selling at the downgrade creates the entry price.
Worked example: a single-B bond at 7.5% vs. 4.5% Treasuries. Spread 300bp; expected loss at 3.5% default × 60% loss ≈ 210bp. You're netting ~90bp for recession risk — historically thin, and the market's way of saying "late cycle". At 800bp spread the same math nets ~600bp — the entry points feel terrible and pay best.
4 · AdvancedPricing & valuation

Yield-to-worst and negative convexity

With call schedules attached, the bond's value is a min over call scenarios:

$$ P = \min_{c \in \text{calls} \cup \{T\}} \mathrm{PV}(\text{cash flows to } c) \quad\Rightarrow\quad \text{price compresses toward } K_{\text{call}} \text{ in rallies} $$

Tightening spreads push prices into the call ceiling — upside convexity is sold to the issuer. The consequence: HY total returns are carry-dominated; capital-gain rallies are structurally capped while drawdowns are not. Effective duration is also spread-dependent — as spreads widen, expected calls vanish and duration extends, exactly when you least want it.

Distress mechanics

  • The distress boundary: bonds beyond ~1000bp trade on recovery, not yield — price becomes (probability-weighted) expected recovery, and the analysis shifts to capital structure: what's above you, what's the collateral, where does value break.
  • Liability management: the modern loosening of covenants enabled "creditor-on-creditor violence" — drop-down financings (J.Crew), uptiering (Serta) — where subsets of holders exchange into senior claims and strand the rest. Reading the docs is now alpha, not diligence.
  • Recovery cyclicality: recoveries fall exactly when defaults spike (same recession), so expected loss is convex in the cycle — Moody's data show recovery on senior unsecured dropping from ~45% in calm years to ~25% in crisis years.

The index-liquidity mismatch

HY ETFs offer minute-by-minute liquidity on bonds that trade by appointment. In March 2020 the big ETFs traded at 3–5% discounts to stale NAVs — functioning, as later analysis conceded, as the price-discovery vehicle rather than the anomaly. The Fed's corporate-credit facilities (SMCCF), which bought HY ETFs directly, ratified the structure: the wrapper is now systemically load-bearing.

The formulas above are standard textbook formulations, simplified for teaching. They explain the mechanism — they are not a valuation tool, and they will not reproduce a dealer’s price.

5 · Desk notesHow practitioners think about it
Practitioner note: judge the market by spread, not yield — a 7% yield means nothing until you know if it's 250 over or 500 over. And judge a bond by its docs and its place in the capital structure before its coupon: in high yield, the coupon is the advertisement; the indenture is the product.