The Credit Spreads Manual

Research note. Every figure computed at render from two files we maintain: Moody's Baa−10Y spread (monthly, 1953–present) and daily HYG/LQD ETF closes (2011–present) — the same data behind the live credit spreads tracker. Methods in How we checked it; editorial standards in How we use AI.

What are credit spreads?

Lend to the US government and you get the risk-free rate; lend to a corporation and you demand extra yield for the chance it doesn't pay you back. That extra yield is the credit spread, and it is the bond market's continuously updated vote on how dangerous the world is. When lenders are confident, the cushion compresses; when they smell recession, it widens — sometimes violently, because a bond holder's upside is capped and their reflex under stress is to sell first. Equity investors argue about stories; credit investors are paid only if nothing goes wrong, which is why so many allocators treat spreads as the adult in the room.

Are spreads warning right now?

As of the July 2026 reading, the Baa spread is 1.59pp — the 35th percentile of the 1953+ record, below its 12-month average of 1.69pp. The faster gauge — the daily HYG/LQD ratio — closed at 0.7471 on August 6, 2026, a regime the tracker labels tight (23 days running). The same Baa series also votes in our Cycle Score's credit dimension and appears as unscored context on the Bubble Tracker — same file, one story.

The two instruments we watch — and why both

No single free series covers both depth and speed, so this Manual (and the tracker) uses two. The Baa−10Y spread is Moody's seasoned Baa corporate yield minus the 10-year Treasury — investment-grade credit, monthly since 1953, the longest clean history freely available. It is the yardstick for "how bad can it get": every episode in the table below is measured with it. The HYG/LQD ratio divides the price of the biggest high-yield ETF by the biggest investment-grade ETF, daily since 2011. It is not a spread in basis points — it is a relative price — but it moves the moment risk appetite changes, needs no publication lag, and is investable, which keeps it honest. Monthly for depth, daily for speed; we always label which one a number comes from.

How to read them — level, direction, and speed

Three readings, in rising order of urgency. Level: percentile against history — tight spreads (low percentiles) mean markets price near-zero default risk, which is comfort and fragility, since there is no cushion to absorb surprise. Direction: a sustained turn wider from tight levels has preceded or accompanied every major equity drawdown in our records — the regime changes on the daily ratio are the earliest clean tell we have. Speed: fast widening is forced selling — funds de-risking, dealers pulling back — and it feeds on itself the same way margin calls do. The blowouts in the next section were all speed events. What spreads do not give you is timing: they can sit tight for years (2004–07, 2017–19, 2024–26) while imbalances build.

Every famous blowout, computed from the data

Five episodes define the record. Peak spread and its month, computed from the series at render:

Note the asymmetry: 2008 stands 2.2pp above every other crisis of the past seventy years — credit didn't just reprice, it stopped functioning. And 2020's spike, the fastest widening on record, was also the shortest-lived: the Fed's corporate-bond backstop reversed it within months, a policy precedent that now hangs over every reading of this gauge.

The 2007 lesson — calm is the warning

The most useful thing this series ever taught is not in the spikes. Through 2004–07, with subprime lending at full boil, the Baa spread ground down to 1.56pp (March 2005) — near the tightest of that entire cycle, and not far from the all-time record of 0.29pp set February 1966. Credit wasn't warning anyone; credit was the complacency. The information arrived as a change: spreads began widening in the summer of 2007, months before equities understood, and never looked back. Read this gauge accordingly — a tight spread is not safety, it is the absence of cushion, and the moment the direction turns from a tight base is precisely when it deserves the most attention.

Where credit spreads will mislead you

• Tight is not safe. The 2007 section above is the evidence. Low spreads describe confidence at maximum, which is when surprise does the most damage.
• The monthly series lags. Baa−10Y is a monthly average published with a delay — by the time it confirms stress, the daily ratio moved weeks earlier. Use the monthly for depth and history, never for timing.
• The ETF ratio is prices, not spreads. HYG/LQD embeds duration differences and fund flows along with credit risk. It is a stress thermometer, not a measurement of the spread itself.
• The Fed changed the game in 2020. Once the corporate-bond backstop exists, every future blowout carries a policy put. Comparing post-2020 extremes to 2008 without that caveat overstates what today's spreads can tell you.
• Rate moves contaminate the spread. Baa−10Y can widen because Treasuries rallied in a flight to safety, not because corporate yields rose. Check both legs before narrating.

The last 12 monthly readings

Full series as CSV: baa-credit-spread.csv; daily ETF data: credit_data.json.

How we checked it

The long series is FRED's BAA10YM (Moody's seasoned Baa yield minus the 10-year constant maturity), monthly from 1953, with the 12-month average computed in our pipeline. Episode peaks are the highest monthly readings inside each stated window; the 2004–07 tight is the lowest reading in that window; percentiles rank the latest value against every month in the history. The daily ratio divides HYG by LQD closes from our maintained price database, with regimes assigned by the same rolling z-score logic the tracker documents. Everything recomputes at render, so this Manual always matches the live page.

Frequently asked questions