How to Read a Credit RatingSome background helps

A rating is an ordinal opinion about one narrow question, produced under a business model worth understanding. What it measures, what it deliberately excludes, and what to read instead.

What a rating is, precisely

• A rating is an opinion about relative creditworthiness — the likelihood of not being paid in full and on time, ranked against other rated obligations.
• It is ordinal, not cardinal. A single-A is judged more likely to pay than a triple-B. The rating itself is not a probability, though agencies publish historical default rates by category separately.
• It is about credit only. Not price, not volatility, not liquidity, not whether the yield compensates you. A highly rated bond can lose a third of its value from a rate move without the rating being wrong about anything.
• It is an opinion, and is legally characterised as one. That framing is not a technicality — it describes the product accurately.

Issuer rating versus issue rating

• Notching is where the information is. A subordinated instrument is rated below the issuer, a secured one above, and the size of the gap is the agency's view on recovery.
• The gap between issuer and issue rating tells you about loss given default, which is half of expected loss — the expected-loss calculator shows how much the recovery assumption moves the answer.
• Buying on the issuer rating when you hold a subordinated instrument is one of the more expensive routine errors, and it is exactly what happened around the 2023 AT1 write-down.

Outlooks, watches and reviews

• An outlook indicates the likely direction over a medium horizon — typically one to two years. It is a soft signal.
• A watch or review is a short-horizon signal tied to a specific event: an acquisition, a refinancing, a regulatory decision. It is a much stronger signal than an outlook.
• Watches resolve faster and more often in the indicated direction, which makes them the more actionable of the two.
• The investment-grade boundary matters far more than any other notch, because mandates are written around it. A downgrade across that line forces selling by holders who have no view — the fallen-angel mechanism in the other side of the trade.

The structural issues worth knowing

• The issuer usually pays for the rating. This has been the dominant model for decades and its incentive problems are widely documented and much debated. It does not make ratings useless; it means they are a product with a customer.
• Ratings are sticky by design. Agencies aim for through-the-cycle assessments rather than point-in-time ones, so ratings lag market prices — which is a feature for a long-term buyer and a limitation for anyone using them as a warning system.
• Structured finance ratings behaved differently from corporate ratings in 2007–08, and the difference was about model assumptions on correlation rather than about individual credits. See 2008 and securitisation.
• The agencies' own historical default studies are public, and are the honest way to convert a letter into a rough probability.

What to read alongside it

• The market's own view: the spread. A bond trading far wider than its rating band suggests means the market disagrees, and the market repriced first — the spread calculator puts them side by side.
• The CDS level, where one exists, is a cleaner credit-only signal than a cash spread, because it strips out funding and liquidity effects.
• The rationale document, not the letter. Agencies publish the reasoning, the key sensitivities and the specific triggers that would cause a change. Those triggers are the most useful part of the whole exercise.
• The company's own numbers. Leverage, coverage and the maturity schedule are computable from the annual report, and a simple screen like the Altman Z-score takes minutes.

What a rating deliberately excludes

• Whether the yield compensates you. That is a price question, and agencies do not answer it.
• Market and liquidity risk. Entirely outside scope.
• Suitability. Outside scope, in every jurisdiction.
• Timing. A rating says something about eventual repayment, not about what the price does next year.

The checklist

• Issuer or issue rating, and which do I actually hold?
• How many notches from the issuer rating, and what does that gap say about recovery?
• Outlook or watch, and what specific event resolves it?
• How far from the investment-grade boundary — is forced selling a plausible scenario?
• What are the stated downgrade triggers in the rationale?
• Does the spread agree with the rating, and if not, which one moved?