Guide

Interest rate caps and floors explained

Harbor Capital refinanced a $95M five-year term loan at SOFR plus 175 bp, resetting quarterly. Treasury’s covenant model showed that if three-month compounded SOFR climbed above 4.50%, interest coverage would breach bank limits within two years — a realistic stress given the yield curve was still inverted when the loan closed. The desk considered a receive-fixed interest rate swap to fix the coupon, but that locked in today’s forward curve and forfeited savings if SOFR fell. They bought a 5.25% cap on the loan notional instead: each quarter, if SOFR exceeded 5.25%, the cap counterparty paid the excess times notional times day-count fraction. Premium was $1.4M upfront (roughly 147 bp of notional). When SOFR peaked at 5.38% in one reset period, the cap paid $312K; when SOFR later dropped to 4.12%, Harbor paid only the floating loan coupon with no hedge drag. A plain swap would have cost more in opportunity terms during the low-rate window. The cap matched the actual risk: quarterly index resets on bank debt, not par swap-rate timing on pension liabilities.

An interest rate cap is a series of call options on a floating index — each period’s caplet pays when the index fixes above the strike. A floor is the mirror: put-like floorlets pay when the index fixes below the strike. Caps and floors are the standard toolkit for corporates, REITs, and project-finance borrowers with floating-rate debt, and for lenders who want to guarantee a minimum coupon on floating assets. This guide covers cap and floor mechanics, strike and notional schedules, collars and zero-cost structures, Black caplet valuation and volatility surfaces, the Harbor Capital loan hedge refactor, a technique decision table vs swaptions and plain swaps, pitfalls, and a production checklist.

What caps and floors are

A cap is a strip of caplets — one option per interest period on a floating leg. For each reset date t_i:

Notional — principal the cap references (often matches loan amortization schedule).
Strike (cap rate) — threshold above which the cap pays (e.g., 5.25% on SOFR).
Index — floating rate observed for the period (SOFR compounded in arrears, Term SOFR, Prime, EURIBOR, etc.).
Payment — max(0, Index − Strike) × Notional × Year fraction, paid at period end (in arrears) or start (in advance), per confirmation.
Tenor — total life of the cap (e.g., 5 years aligned with loan maturity).

A floor pays max(0, Strike − Index) × Notional × Year fraction. Lenders buy floors to protect against rates falling below a minimum on floating-rate assets; borrowers sell floors (or buy collars) to reduce cap premium.

Caps vs single caplets

Dealers quote whole caps as an upfront premium in basis points of notional. You can also trade individual caplets for one period — useful when only one reset matters (e.g., before a refinancing window). A cap’s value is the sum of caplet values across periods, each using the forward rate for that reset and the implied volatility for that expiry and tenor on the cap vol surface.

Collars, corridors, and zero-cost structures

Interest rate collar

A collar combines a purchased cap and a sold floor at a lower strike. The borrower caps upside borrowing cost and gives up benefit below the floor strike. Net premium can be near zero if strikes are chosen for balance (“zero-cost collar”). Harbor’s lenders sometimes require collars instead of naked caps so the bank retains some upside on the asset side.

Corridor

Buy a cap and buy a floor at different strikes — you are protected only if rates move outside the corridor. Less common on corporate loans; appears in structured notes and some project finance.

Amortizing and accreting notionals

Loan principal often amortizes; cap notionals should step down on the same schedule or the hedge is mis-sized in later years. Bullet loans use flat notional. Revolvers may use a notional equal to committed amount even if drawn partially — document the convention in the ISDA schedule.

Who uses caps and floors

Floating-rate borrowers — cap maximum all-in coupon while retaining floating benefit below the strike.
REITs and property — cap SOFR on floating mortgages; floors on floating-rate CMBS tranches they hold.
Project finance — caps on construction loans; lenders often require them in base-case models.
Banks and insurers — floors on floating assets; caps on liabilities in ALM books.
Leveraged loans — caps at covenant-implied ceilings; strike often tied to interest coverage math.

Caps hedge short-tenor index resets. They do not immunize duration on fixed-rate bond portfolios — that is swap and LDI territory. Mixing the two causes the classic “we bought a cap but duration still moved” complaint.

Valuation: Black caplet model and vol surfaces

Each caplet is priced like a call on the forward rate for that period using Black’s model (or Bachelier for negative-rate environments):

Forward rate — from the OIS-discounted curve for the index (SOFR forward for each accrual period).
Strike — cap rate from the confirmation.
Implied volatility — from the cap/floor vol surface quoted by dealers (expiry × tenor grid, e.g., 1Y into 5Y).
Discount factor — OIS to payment date.

Cap premium is sensitive to vega (vol level), theta (time decay), and the skew — OTM caps often trade at higher implied vol than ATM. During 2022–2023 hiking cycles, SOFR cap vol spiked; rolling caps before expiry required budgeting vol risk, not just rate direction.

Settlement and compounding

Post-LIBOR, USD caps on SOFR must specify compounded-in-arrears lookback and observation shift conventions matching the loan. A mismatch of even a few basis points between loan and hedge accrual creates unexplained P&L noise each quarter.

Harbor Capital refactor: cap on corporate term loan

The original plan was a five-year receive-fixed swap at 4.85% all-in. Problems:

Swap locked the full coupon even if SOFR fell — bad for a fund still in growth mode with variable cash flows.
CSA and bilateral credit lines for a new swap counterparty added operational overhead versus a cap from the lending bank’s derivatives desk.
Covenant stress was specifically “SOFR above X,” not “five-year swap rate above Y” — misaligned risk factor.

Refactor steps:

Map loan reset dates and amortization to cap schedule (20 quarterly periods).
Set strike at 5.25% — 75 bp above covenant stress SOFR of 4.50% for buffer.
Quote three dealers plus the lender’s desk; compare upfront vs running premium structures.
Elect hedge accounting (ASC 815) with effectiveness tests on critical terms match to loan index and notional.
Document quarterly settlement reconciliation: loan interest vs cap payment.

Outcome: effective all-in coupon capped at SOFR + 175 bp with SOFR floored effectively at 5.25% on the hedge leg only; average all-in over year one was 38 bp lower than the swap alternative because SOFR spent six months below 4.00%.

Technique decision table

Technique	Best when	Avoid when
Interest rate cap	Floating loan; want ceiling but keep downside benefit	Need fixed coupon certainty for budgeting or bond-like liabilities
Interest rate floor	Lender protecting floating asset yield	Borrower unless selling floor to fund cap (collar)
Collar	Premium budget near zero; acceptable floor give-up	Need protection below a hard minimum borrowing cost
Plain IRS (pay fixed)	Full coupon certainty; duration match to fixed-rate debt	High conviction rates will fall; want asymmetric upside
Swaption	Timing flexibility on entering a swap; pension derisking tranches	Risk is quarterly SOFR resets on bank loan, not swap rate level
No hedge	Low leverage; ample coverage headroom; cap vol too expensive	Covenant breach path is modeled inside 12–18 months

Common pitfalls

Index mismatch — cap on Term SOFR while loan pays compounded SOFR in arrears; terms must mirror the credit agreement.
Notional drift — amortizing loan with flat cap notional over-hedges early and under-hedges late.
Front-loaded premium pain — upfront cap premium is cash today; model IRR impact, not just accounting P&L.
Vol mean reversion — buying caps when vol is elevated (stress) is expensive; pre-purchase in calm markets if policy allows.
Floor surprise in collars — sold floor forces borrower to pay equivalent of higher coupon when rates plunge; read collar payoff diagram.
Wrong risk factor — cap does not fix swap spread or credit spread on loan; only the index leg.
Dealer markup opacity — request caplet breakdown and vol per period; single upfront number hides expensive tails.
Hedge accounting gaps — missing effectiveness documentation forces earnings volatility under ASC 815.

Production checklist

Extract loan reset dates, index, spread, compounding, and amortization from credit agreement.
Define hedge objective: maximum all-in rate, covenant line, or budget certainty.
Choose cap strike with stress test buffer above breach level.
Size notional schedule to loan principal path.
Request three quotes; compare upfront premium, running premium, and caplet vol grid.
Verify ISDA definitions (2006 or 2021) and SOFR conventions match loan.
Model cap payoff under parallel rate shocks and historical SOFR paths.
Document collar floor strike implications if zero-cost structure.
File hedge designation memos before trade execution.
Reconcile quarterly cap settlements to loan interest invoices.
Calendar cap maturity vs loan maturity; roll or re-strike before expiry.

Key takeaways

A cap is a strip of caplets that pay when a floating index fixes above the strike — the standard hedge for floating-rate borrowers who want a ceiling, not full fixation.
Harbor Capital capped SOFR on a $95M term loan at 5.25%, preserving savings when rates fell while protecting covenant headroom when SOFR spiked.
Cap premium comes from the cap vol surface via Black caplet pricing; SOFR compounding conventions must match the loan exactly.
Collars trade cap premium for a sold floor; zero-cost structures still have economic cost in forgone low-rate benefit.
Match instrument to risk: quarterly index resets need caps; pension swap-rate timing needs swaptions; full coupon certainty needs plain IRS.