Guide

TED spread explained

Harbor Capital’s $90M short-term cash sleeve rolled weekly into AA-rated commercial paper and 3-month Treasury bills. In September 2019, the desk’s risk dashboard showed calm: the VIX sat below 15 and investment-grade credit spreads were tightening. Yet the gap between 3-month interbank funding (then USD LIBOR) and 3-month T-bill yields — the classic TED spread — widened from 28 to 74 basis points in ten sessions. CP backup-line draws cost 22 bp more than modeled; the team had been watching equity vol and bond OAS while ignoring the money-market credit premium banks were charging each other at the three-month tenor.

After that episode, Harbor rebuilt its front-end stress playbook around TED-style spreads alongside SOFR-OIS and the financial conditions index. When the 3-month unsecured-minus-T-bill gap exceeds its rolling 90th percentile, the desk cuts CP tenor to overnight repo and extends T-bill WAL by one week. Forecast error on weekly cash drag fell from 14 to 4 bp; CP program utilization during the March 2020 shock stayed inside covenant limits. This guide explains what the TED spread measures, how it was constructed pre- and post-LIBOR cessation, what drives widening and compression, crisis episodes from 2008 and 2020, the Harbor Capital refactor, a technique decision table versus credit spreads and overnight funding gauges alone, pitfalls, and a production checklist.

What the TED spread measures

The TED spread is the difference between the yield on 3-month interbank unsecured dollar funding and the yield on a 3-month U.S. Treasury bill. The name blends T-bill and ED (eurodollar deposits, the market where offshore dollar LIBOR was set).

Classic formula

Historically:

TED spread = 3-month USD LIBOR − 3-month T-bill yield

Both legs are annualized money-market rates at roughly the same maturity. LIBOR reflected what large banks charged each other for unsecured three-month dollar loans. T-bills are backed by the U.S. government — the closest thing to a risk-free short-term dollar instrument. The spread therefore isolates bank credit and liquidity premium at the three-month horizon, stripped of pure policy-rate level.

What widening means

When the TED spread widens, lenders demand more compensation to lend dollars to banks unsecured for three months relative to lending to the Treasury. That usually signals one or more of:

Elevated counterparty credit concern — fear that a large bank or cluster of banks may not repay.
Balance-sheet strain — dealers hoard liquidity ahead of quarter-end, regulatory reporting, or known stress events.
Collateral scarcity — when secured funding is hard to obtain, unsecured rates rise even if default risk is unchanged.
Flight to quality — T-bill yields fall (or rise less) as investors pile into government paper, mechanically widening the gap.

Narrow TED spreads (often 10–40 bp in calm periods) imply banks can fund freely at only a modest premium to government paper. Spikes above 100 bp historically coincided with systemic stress; the 2008 financial crisis peak approached 450 bp.

Construction, data sources and post-LIBOR proxies

LIBOR cessation (June 2023)

U.S. dollar LIBOR publication ceased for most tenors after June 30, 2023. The FRED series TEDRATE (daily TED spread) was discontinued. That does not erase the concept — it forces practitioners to rebuild the gauge with modern benchmarks.

Modern proxy stack

Post-LIBOR TED-style monitors typically use one of:

3-month Term SOFR − 3-month T-bill — closest structural replacement; Term SOFR is a forward-looking compounded rate published by CME and licensed for contracts. It is secured-rate based but includes a term premium; interpret widening carefully versus pure credit.
3-month SOFR-OIS (or SOFR vs OIS) at quarterly tenor — overlaps with overnight SOFR-OIS but at the money-market strip horizon banks actually roll CP programs.
Commercial-paper vs T-bill spreads — A2/P2 financial CP over matched-maturity bills directly prices issuer credit at the front end.
CDS and FRA-implied bank risk — senior bank CDS or FRA strips can proxy unsecured bank funding cost.

Do not mix tenors

The TED spread is specifically a 3-month construct. Comparing overnight SOFR-OIS to a 3-month T-bill spread produces false signals. Build each leg at matched maturity or use futures-implied 3-month rates from SOFR futures strips.

What widens and narrows the spread

Calendar and plumbing events

Quarter-end and year-end window dressing routinely widen front-end spreads as banks shrink balance sheets for regulatory ratios. Treasury General Account (TGA) rebuilds that drain bank reserves can lift unsecured rates even when the Fed holds policy steady. Large bill issuance that crowds out private credit also moves the T-bill leg.

Policy and facility backstops

Fed facilities blunt TED-style spikes: the Standing Repo Facility (SRF), discount window, and foreign swap lines inject reserves when secured funding seizes. The overnight reverse-repo (ON RRP) floor affects the front end but does not directly set 3-month unsecured rates — watch both balance-sheet plumbing and term funding markets.

Credit-cycle linkage

TED widening often leads corporate OAS widening by days or weeks because money markets are where banks fund first. When TED compresses while HY OAS stays wide, the stress may be issuer-specific rather than systemic — a useful disambiguation rule for desks.

Crisis episodes: 2008 and March 2020

Global Financial Crisis (2008–2009)

From mid-2007 through late 2008, TED blew out as interbank trust collapsed. LIBOR diverged sharply from policy expectations — the “LIBOR-OIS spread” became a parallel overnight stress gauge. Peak TED near 450 bp implied banks would not lend to each other except at punitive spreads. The episode validated TED as an early systemic alarm, though later research showed LIBOR submission manipulation could distort the series in quieter periods.

COVID liquidity shock (March 2020)

When pandemic lockdowns hit, TED-style gaps spiked intra-month as funds fled prime money-market funds and CP markets froze. The Fed’s emergency facilities (MMLF, CPFF) and aggressive QE compressed spreads within weeks. Harbor Capital’s post-2019 playbook — pre-committed repo lines and T-bill sleeves triggered by TED proxies — avoided forced asset sales that hit peers who waited for VIX to confirm stress.

Harbor Capital refactor (worked example)

Before 2019, Harbor sized weekly CP issuance off fed funds futures implied paths alone. The desk assumed unsecured front-end spreads were a fixed 15 bp over bills. After the September 2019 repo spike and the 2020 rehearsal, they implemented:

Daily proxy: 3-month Term SOFR minus 3-month bill yield (post-2023) with a rolling 252-day z-score.
Alert bands: yellow at +1.5σ, red at +2.5σ versus trailing year.
Red-band actions: cut CP allocation from 45% to 20% of sleeve; shift to tri-party repo and bills; extend backup-line draw tests.
Reporting: weekly liquidity memo includes TED proxy, SOFR-OIS, and IG OAS on one page — no single-indicator dashboards.

Outcomes (internal benchmarks, 2020–2025): CP rollover failure rate 0%; weekly cash-drag forecast MAE 4 bp vs 14 bp pre-refactor; red-band triggers fired twice (March 2020, March 2023 regional-bank stress) with average 36-hour lead time before HY OAS jumped 50 bp.

Technique decision table

Your goal	Prefer TED-style 3M spread	Prefer instead
Detect interbank / money-fund stress early	3M unsecured proxy minus T-bill	IG OAS alone (lags) or VIX (equity-centric)
Overnight funding squeeze only	Overnight SOFR-OIS	3M TED proxy (too slow to catch O/N repo breaks)
Corporate bond relative value	Credit OAS and CDS	TED (issuer-agnostic bank gauge)
Recession probability dashboard	Combine TED proxy with yield curve and FCI	Single-indicator recession calls
Post-LIBOR compliance / reporting	Term SOFR minus bill (document methodology)	Discontinued TEDRATE series
Treasury cash sleeve allocation	TED proxy + SOFR-OIS + TGA trend	Policy rate target only

Common pitfalls

Using discontinued TEDRATE — FRED stopped updating; rebuild proxies and document the mapping in audit trails.
Tenor mismatch — overnight OIS minus 3-month bills is not TED; match maturities or use futures strips.
Ignoring the T-bill leg — spread widening from bill yields collapsing (flight to quality) differs from LIBOR rising (credit fear); decompose both legs.
Confusing secured and unsecured — Term SOFR is secured-rate based; add CP or FRA legs when you need pure bank credit.
Quarter-end noise as crisis — apply seasonally adjusted bands or compare to prior quarter-ends.
Single-threshold alerts — 50 bp meant crisis in 2007; in abundant-reserve regimes structurally higher baselines persist; use rolling percentiles.
Neglecting facilities — Fed backstops cap tails; pair spread monitors with H.4.1 facility usage.

Production checklist

Define 3-month unsecured leg (Term SOFR, FRA, or CP index) in writing.
Match T-bill leg maturity to unsecured leg (ACT/360 conventions documented).
Publish daily time series internally; archive pre-2023 LIBOR-based history separately.
Set alert bands on rolling z-scores, not fixed 20th-century thresholds.
Dashboard shows TED proxy, overnight SOFR-OIS, IG OAS, and TGA on one screen.
Pre-commit red-band actions (CP cut, repo shift, line draws) before stress hits.
Quarter-end calendar overlay on alert system to reduce false positives.
Stress-test cash sleeve against 2008- and 2020-scale spread widening.
Train desk on decomposing spread moves into credit vs flight-to-quality legs.
Review proxy methodology annually as SOFR market structure evolves.
Log every red-band trigger with subsequent OAS and policy responses.
Include TED-style metrics in weekly liquidity memo to risk committee.

Key takeaways

The TED spread is 3-month unsecured bank dollar funding minus 3-month T-bills — a pure front-end credit and liquidity premium gauge.
LIBOR cessation ended TEDRATE; rebuild with Term SOFR, CP spreads, or matched FRA strips and never mix overnight with quarterly tenors.
Widening often leads corporate OAS and signals money-market stress before equity vol confirms — 2008 and March 2020 are the textbook cases.
Harbor Capital cut weekly cash forecast error from 14 to 4 bp by pairing TED proxies with SOFR-OIS and pre-committed red-band actions.
Decompose both legs, use rolling percentiles, and dashboard multiple gauges — no single spread captures all funding risk.