Who pays for a credit rating, and why does the market care?
A rating is an opinion, and someone pays for it. Who pays turns out to shape the incentives around the opinion — a structural point that has been examined at length by regulators, academics and the agencies themselves. This lesson describes the mechanics factually and generically; it is not a comment on any particular firm.
Two business models
- Investor-pays (subscriber model). Investors subscribe to a research service. The agency's customer is the reader. This was the original model in the early twentieth century.
- Issuer-pays. The borrower commissions and pays for the rating on its own debt. The agency's paying customer is the entity being rated.
Issuer-pays became dominant for public corporate and structured debt from the 1970s onward. Two forces drove the switch. First, cheap copying made subscription research hard to keep exclusive — the rating leaked and became a public good. Second, issuers actively wanted ratings: without one, a bond is far harder to sell into a market where many buyers require ratings by mandate or regulation.
The conflicts that follow, as documented
Official post-crisis reviews and a substantial academic literature have identified a consistent set of structural tensions in the issuer-pays model:
- Rating shopping. An issuer may seek preliminary indications from several agencies and formally commission only the most favourable one — with the unfavourable views never published.
- Revenue concentration. In some segments, particularly structured finance, a small number of arrangers were repeat customers responsible for large shares of a business line's revenue, concentrating commercial pressure on a handful of relationships.
- Ancillary services. Selling advisory or related services alongside the rating can blur the line between assessing a structure and helping design one.
- Downgrade reluctance. Timely downgrades are commercially uncomfortable when the downgraded entity is the paying client, which creates pressure toward slower action.
- The regulatory licence problem. When rules, mandates and capital regimes hard-wire ratings into who may own what, the rating acquires value independent of its information content. That makes it something issuers must buy, whatever its accuracy.
What was done about it
Post-2008 reform in major jurisdictions converged on a similar toolkit:
- Formal registration and supervision of rating agencies.
- Separation of analytical staff from commercial and fee negotiations.
- Look-back reviews when an analyst leaves to join a rated entity or arranger.
- Public disclosure of methodologies, rating histories and default-and-transition performance statistics, so that accuracy is measurable after the fact.
- Provision for unsolicited ratings, which partially counters shopping.
- Deliberate removal of mechanical ratings reliance from parts of the regulatory framework, to shrink the regulatory licence.
None of this makes the conflict vanish. It makes it disclosed, supervised and measurable — which is a genuine improvement and a different thing from elimination.
The stance this argues for
Treat a rating as one input with a known incentive structure, alongside others: the market spread (which has its own biases, discussed in Spreads move together), the financial statements, and the actual bond documentation. When those inputs disagree, the disagreement is the information. Published default-and-transition studies also let you check, empirically, how well ratings have ordered risk over time — which is exactly the kind of verification the disclosure rules were designed to enable.
This is a description of incentives, not an accusation about any firm, and certainly not advice about which ratings to trust or which bonds to hold.
Try it now
Cross-check rather than accept:
- Six governments' ratings from the three agencies are below. Note where the agencies disagree by a notch or more, and rank the six by rating.
- What the market charges to insure the same six, as credit default swap spreads. Rank them by spread, cheapest protection first.
- Compare the two orderings. Where they diverge, write one sentence on what could explain it — a slow rating, a liquidity effect, or a market view not yet reflected in an opinion.