What Credit Ratings Attempt To Measure

Credit ratings appear as short letter codes attached to bonds and issuers. The question they answer is specific, and much of what they are assumed to cover falls outside it.

The question is repayment, not value

A rating expresses an opinion about the likelihood that a borrower will meet its payment obligations in full and on time.

It says nothing about whether the bond is attractively priced, how much its price will move or how easily it can be sold.

A highly rated bond can lose substantial value if interest rates rise, without any change in the assessment of whether it will repay.

The scale is relative and ordinal

Rating categories rank borrowers against each other rather than attaching a specific probability to each. The gaps between adjacent categories are not uniform.

The broad division between investment grade and the categories below it is the most consequential boundary, because many institutions face constraints referencing it.

Crossing that boundary can force selling by holders whose mandates prohibit lower-rated holdings, which produces price effects beyond what the change in assessment implies.

Ratings are opinions produced by a process

Agencies analyse financial statements, business position and, for governments, fiscal and institutional factors, then apply published criteria to reach a category.

Ratings are reviewed periodically and changed when the assessment changes, so they move in steps rather than continuously as market prices do.

Market yields therefore often move well before a rating changes, since prices incorporate new information immediately while the review process takes time.

The issuer usually pays for the rating

In most cases the borrower commissions and pays for its rating, which creates an incentive structure that has attracted considerable regulatory attention.

Rules governing agency conduct, disclosure and the use of ratings in regulation differ between jurisdictions and have changed substantially over time.

Some regimes have worked to reduce automatic reliance on ratings in rules and mandates, precisely because forced selling on downgrades amplifies market stress.

Structured products stretched the framework

Ratings were extended to pooled and tranched securities, where the assessment depends on assumptions about how the underlying loans behave together.

Those assumptions concern correlation between defaults, which is far harder to estimate than the finances of a single company.

The distinction between rating a single borrower and rating a structure built from many remains relevant to how much weight the letter code deserves.