Credit rating: what it measures and why it is not a guarantee

A credit rating is an agency's opinion on a borrower's relative ability to service its debt, expressed as an ordinal letter grade, not a calculation of the probability that a particular payment will arrive on time. A rating ranks: a bond with a higher grade should, on average, default less often than a bond with a lower grade. It does not promise that the former will never default at all. Between "less often on average across the group" and "will not happen to your issue" lies the entire difference between an assessment and a guarantee, and it is exactly this difference that readers most often lose sight of.

What a rating actually measures and what it leaves out

The subject of the assessment is credit risk, that is, the risk that financial obligations will not be met: a missed coupon, non-payment of principal, a restructuring on terms that are worse for the holder. Almost everything else that concerns the owner of a bond is left out. A rating does not say whether the price of the issue will fall, how wide the spread to government bonds is, whether you will be able to sell the bond without losing money on the bid-ask spread, or whether the current yield is adequate to the risk being taken. Market risk, interest rate risk and liquidity risk are separate things, and a high grade on the credit scale does not protect against any of them.

This leads to a practical conclusion: a rating is an input to the analysis, not its result. Comparing yields within one rating group makes sense; choosing an issue by the letter alone does not.

The national scale and why it is not compared directly

In Russia, ratings are assigned by agencies accredited by the Bank of Russia in accordance with Federal Law No. 222-FZ of 13.07.2015. The grades are assigned on the national rating scale, and this is fundamental: a grade on the national scale shows where a borrower stands relative to other Russian borrowers, not relative to issuers from other countries. The top grade of the national scale means "the best credit profile among Russian borrowers", not "the equivalent of the highest international category". The letters look alike, the content is different, and carrying a grade directly from one scale to another is a mistake.

The scales of different Russian agencies are not identical to one another either: methodologies differ, and one issuer may be rated at different grades by different agencies. The divergence is informative in itself — it shows where the methodologies disagree about the nature of the risk.

How the assessment is put together

An agency builds its assessment from several blocks. The business profile: the industry, the cyclicality of demand, market position, the concentration of revenue among customers. The financial profile: the debt burden, interest coverage, the maturity structure of repayments, the liquidity cushion and access to refinancing. The baseline assessment is then adjusted for support factors — ties to the state, the resources of the parent structure, systemic importance for the sector. The adjustment for support moves the final result up by a number of grades from the standalone level, and precisely this part of the rating is the most fragile: support is an assumption about the behaviour of a third party, not an obligation.

You can read for yourself the factual material on which the agencies base their conclusions: issuer disclosures are collected in the financial statements section, and corporate events and payments in the events calendar.

Outlook, review and migration

A rating comes with an outlook — an indication of the likely direction of the agency's next action. Reviews are tied to the scheduled cycle and to events: a major deal, a sharp rise in debt, a change of owner, a covenant breach. Hence the main limitation in terms of time: a rating reflects the assessment as of the date of the action, while the borrower's condition changes continuously. Migration between grades is the normal mechanics of the system, not a failure: in every period some issuers move down and some move up.

Why it is still not a guarantee

There are several reasons, and they work in different ways. A rating is ordinal: it cannot be converted into a "probability of non-payment" figure without additional default statistics, and those statistics are frequencies by group over past periods, not a forecast for your bond. A rating is conditional: it rests on a scenario, and a different one may materialise. A rating does not cover operational and legal risks that can stop a payment even when the balance sheet is healthy. And a rating does not cancel out correlation: in the contraction phase of the credit cycle, entire groups of borrowers deteriorate at once, because refinancing closes to all of them at the same time — more on this mechanism in the article on the credit cycle.

Where a rating has legal force

Beyond its informational role, a grade works as an admission ticket. Exchange listing levels, the range of assets permitted to institutional investors, collateral requirements and the terms of certain banking operations are tied to minimum rating levels set by the regulator's acts. That is why a downgrade sometimes hits an issuer not economically but mechanically: the bond drops out of the list of eligible securities, and holders bound by a mandate are obliged to sell it. The fall in price in this case is caused not by a change in solvency but by a change in the pool of buyers.

How to use it

Look at the grade together with the outlook and the date of the agency's latest action, not in isolation. Separate the standalone level from the uplift for support. Compare yields within a rating group, not between groups. Check for divergences between the grades that different agencies assign to one issuer. Start the analysis of an individual borrower with its instrument page — for example, {{instrument:SBER}} — and set it against the issues in the bonds section; the meanings of terms, including credit rating, are collected in the glossary, and rating actions are reflected in the news.

What this article does not contain: specific rating levels for individual issuers and default statistics by grade. These are data that change with every rating action, and their place is on the instrument pages and in the publications of the agencies themselves, not in an explanatory text.