Rating Agency Criteria for Structured Finance

A corporate bond rating asks a fairly direct question: how likely is this company to fail to pay? A structured-finance rating asks something structurally different: given a pool of loans with some expected default behavior, and a specific tranche sitting at some specific level of subordination, how much loss would that tranche have to see before it stops getting paid in full — and how likely is a loss that large? The rating is really a rating of the structure, applied to one slice of it.

The building blocks agencies model

Rating agencies start with the pool's expected default frequency and expected loss severity given default, drawn from historical data on similar loans (same asset type, similar borrower credit scores, similar geography and vintage). They then apply a stress multiple scaled to the target rating — a tranche targeting a AAA rating might need to withstand losses several times the pool's expected base-case loss, while a tranche targeting single-A only needs to survive a milder multiple of that base case. The tranche is rated at the highest level whose stress-scenario loss it can absorb without missing a scheduled payment, given its subordination and any reserve account or overcollateralization built into the deal.

Worked example

A pool has an expected loss of 2.0% of face value. To rate a tranche AAA, the agency's methodology requires it to survive a stress scenario at roughly 6 times expected loss, or 12.0%. If that tranche has 13% subordination beneath it (mezzanine plus equity plus any reserve account), it clears the stress test and can be rated AAA. A different tranche in the same deal with only 4% subordination can't survive a 12% stress loss, so it gets rated at whatever lower stress multiple 4% subordination actually covers — perhaps single-A, where the required stress multiple is only around 3.5 times expected loss (7%).

Why the same pool produces different ratings for different deals

Two deals backed by nearly identical loan pools can carry different ratings on their senior tranches if one deal's sponsor built in thicker subordination than the other. The rating reflects the structure chosen by the deal's arranger as much as it reflects the pool — which is exactly why rating-agency criteria are published: arrangers use them to engineer subordination levels that hit a target rating at the lowest possible cost of credit enhancement.

What this means in practice

Reading a structured-finance rating without understanding the subordination and stress assumptions behind it tells you almost nothing about what could actually go wrong; the number is a compressed summary of a stress test, not an independent opinion about the borrowers.