How to Check a Credit Rating Agency's Default Study

A rating default study is the scorecard a credit rating agency has to publish on itself. It answers one question: of everything this agency rated AA three years ago, how much of it defaulted. SEBI standardised the method so the answer is comparable across agencies rather than a marketing document, and it requires publication every year with a decade of history kept online. This page covers where to find it, which table to read, and the traps in comparing two agencies.

Where to look

On the rating agency's own website, not on an exchange. SEBI requires the annual disclosures in the format specified at Annexure 27 of the master circular, made on a consolidated basis for all financial instruments the agency rates, and requires the historical data disclosed every year to be archived and kept available for the last 10 years. Transition rates follow the format at Annexure 28.

Two related disclosures sit next to the default study and are worth reading in the same session. The agency must publish standardised PD benchmarks for each rating category, for one, two and three year cumulative default rates, both short run and long run. And it must separately disclose sharp rating actions, including those on non-cooperative issuers, limited to listed or proposed-to-be-listed securities and other ratings required under SEBI regulations.

How the numbers are built

You cannot read the tables sensibly without the four definitions SEBI fixes, because each one closes off a way of flattering the result.

• Static pool. For default rates, the non-defaulted ratings outstanding at the beginning of any period. For transition rates, the ratings outstanding for each category at the start of a financial year, excluding ratings withdrawn or of non-cooperative issuers during the year.
• Default. Non-payment of interest or principal in full on the pre-agreed date, recognised at the first instance of delay.
• Default rate. The number of defaults among rated entities in the static pool, as a percentage of the total number of entities in that pool. It counts issuers, not rupees.
• Averaging. All averaging across static pools uses the weighted average method, weighted by the number of ratings in each static period.

The cumulative default rate is calculated issuer-wise using the Marginal Default Rate approach on monthly static pools. Withdrawals do not let a rating escape: for securities, a withdrawn rating stays in the computation until the cohort completes or the instrument matures, whichever is earlier, and debenture trustees must keep reporting delays or defaults to the rating agency for the lifetime of the instrument even after the rating is withdrawn.

Which table to read

Each agency publishes average one, two and three year cumulative default rates on a weighted average basis, for two different windows:

Then it publishes three versions of the CDR itself:

The gap between CDR (ii) and CDR (iii) is the size of the issuer not cooperating problem in that agency's book. If those two numbers diverge sharply, a large slice of the rated universe stopped supplying information, and the cleaner-looking number is cleaner because it dropped the riskiest names.

The transition matrix next to it

Transition studies are the stability half of the same question. Agencies publish their average one year rating transition rate over a five year period, calculated as the weighted average of transitions for each rating category across all static pools in that period, weighted by the number of issuers. Ratings downgraded to D are treated as default for the rest of the financial year; ratings upgraded from D count as a new rating in later static pools.

SEBI added the same non-cooperative correction here. Because excluding INC ratings from the static pool may not depict a true picture of stability, agencies must publish two additional transition matrices limited to listed or proposed-to-be-listed securities, alongside the standard one.

What to do with it

Three habits make the study useful rather than decorative.

• Compare like with like. Two agencies are comparable only on the same CDR version, the same tenor and the same window. A long run three year CDR against a short run one year CDR is not a comparison.
• Check the category against the PD benchmark. SEBI's benchmarks set an expectation for the top grades, with tolerance levels: zero one and two year default rates for AAA and zero three year with a 1% tolerance; zero one year for AA and zero two year with a 2% tolerance; zero one year for A with a 3% tolerance, all subject to unexpected legal events or mitigating circumstances (SEBI Master Circular for Credit Rating Agencies dated 11 July 2025, paragraph 26.2.6). A category running above its benchmark is the thing to ask about.
• Read it as a base rate, not a forecast. The study tells you what happened to a cohort. What is happening to one issuer now sits in its rating rationale, its outlook or watch marker, and the no default statement trail. Individual rating actions reach the market as exchange announcements, which is how rating revisions are tracked.

So checking a rating default study means going to the agency's website, picking the right CDR version for your universe, and reading the INC gap before the headline number.

Flock reports the filings themselves, each one dated and linked to its source. What a rating agency's track record means for your money is your call to make. Not investment advice.