What Credit Ratings Actually Measure and Who Pays For Them
A letter grade compresses an enormous amount of analysis into a symbol, and the business model behind it contains a conflict that has never been fully resolved.
The Scale
Credit ratings run from the highest grade down through progressively weaker categories, with a critical boundary between investment grade and speculative grade, the latter commonly called high yield or junk.
That boundary matters enormously because it is written into rules rather than merely into opinions. Many institutional mandates prohibit holding speculative grade debt, so a downgrade across the line can force selling by holders who have formed no view of their own. A single notch can trigger forced supply.
What the Grade Estimates
A rating is an opinion about the probability that a borrower fails to pay as promised, and in some frameworks about expected recovery if that happens. It is a statement about credit risk specifically.
It is not a statement about whether a bond is a good investment. A highly rated bond can be a poor purchase if its yield is too low for the risk taken, and a speculative grade bond can be an excellent purchase if the yield more than compensates. Ratings say nothing about price, and price is what determines return.
They also say nothing about interest rate risk. A thirty year government bond carries the highest possible credit rating and can lose a third of its value when yields rise, which is precisely what happened across 2022 and 2023.
A rating tells you how likely you are to be paid. It says nothing about whether the price you paid was sensible, which is the entire investment question.
The Issuer Pays Problem
The structural issue is who pays. For most of the industry's modern history, the entity being rated pays the agency to rate it.
The conflict is obvious. An agency that consistently issues harsh ratings may find issuers taking their business to a competitor. Since there are few major agencies and issuers can shop among them, competitive pressure runs toward leniency rather than accuracy.
This model developed for practical reasons. Under a subscriber pays arrangement, ratings leak quickly and non payers benefit anyway, which makes the product hard to sell. The issuer pays model solved the funding problem and created the incentive problem.
What Went Wrong in 2008
The clearest failure came in structured mortgage products. Agencies assigned the highest ratings to securities built from pools of mortgages, on models assuming that regional housing markets were largely uncorrelated and that a nationwide simultaneous decline was extremely unlikely.
That assumption failed. When correlations converged toward one, diversification within the pools vanished exactly when it was needed, and instruments rated as safe as government debt suffered severe losses.
The deeper problem was that these products were designed against the models. Issuers structured pools specifically to achieve target ratings at minimum cost, which means the rating was an input to the design rather than an independent assessment of it.
How to Use Them Properly
The sensible use is as a starting filter and a description of institutional constraints, not as a conclusion. Ratings compress genuine analysis and they are reasonably good at ranking relative default risk across ordinary corporate borrowers.
They are least reliable for novel structures, for securities designed around the criteria, and for borrowers whose risk is political rather than financial. Knowing where the tool is weak is more useful than knowing what the letters mean.
The Bottom Line
Ratings estimate default probability, ignore price, and are paid for by the issuer. Treat them as one input describing credit risk, and never as a judgment about whether something is worth owning.