Sebi's new default probability rule to monitor credit rating agencies

Raters have to assign default probability to each rated debt paper

The Securities and Exchange Board of India (Sebi) has introduced a “probability of default” mechanism to keep credit rating agencies (CRAs) in check. The directive could reduce the number of companies that are in the top-rated category. 

According to the new framework, rating agencies have to assign the default probability to each rated debt instrument, and disclose its benchmark by December-end.  

The move assumes significance given that it helps reduce recurrence of major default cases such as the Infrastructure Leasing & Financial Services (IL&FS) crisis, which brought rating agencies under the government’s and regulator’s lens. 

IL&FS commercial papers were downgraded from ‘triple A’ to ‘D’ (or, default grade) in just 40 days.

“In order to enable investors to discern the performance of a CRA vis-a-vis a standardised probability of default (PD) benchmark scale, CRAs, in consultation with Sebi, shall prepare and disclose standardised and uniform PD benchmarks for each rating category on their website, for 1-year, 2-year and 3-year cumulative default rates, both for the short-run and long-run,” Sebi said on Thursday. 

The standardised and uniform PD benchmarks should be disclosed on the website of each CRA — for ratings of long-term and short-term instruments on a consolidated basis for all financial instruments rated by a CRA — by December 31, 2019. Further, the regulator tweaked norms on computation of default rates by rating agencies. It says rating agencies may re-index probability benchmark and tolerance level from time to time.

According to it, CRAs will calculate the cumulative default rate based on the issuer, by using Marginal Default Rate (MDR) for the last 10-year period. 

While PD benchmark for rating categories such as AAA, AA and A shall be subject to any unexpected legal event/mitigating circumstances impacting the default rates, with a certain permitted tolerance level. 

Short-run benchmarks may account for spikes on account of economic cycles or unforeseen events, and thus have a wider band. The same can be calculated on a confidence interval of 99.7 per cent over the weighted average of 1-3 year default rates of the last 10 years. For long-run benchmarks, the same shall be computed on a confidence interval of 95 per cent, as this irons out the economic cycles and makes the benchmark narrower. 

Besides, rating companies have been asked to treat sharp bond spread deviation as a material event. To further increase transparency, Sebi asked to disclose the rating sensitivities in the press release to understand factors that would have the potential to impact credit worthiness of the entity. 

“CRAs should have specific section of rating sensitivities, which shall explain the broad level of operating and financial performance levels that could trigger a rating change, upward and downward. Such factors shall be disclosed in the quantitative terms to the extent possible, discernible to the investors, and should not read like a general risk factor,” said Sebi. 

Sebi added that rating agencies should frame a uniform Standard Operating Procedure with respect to tracking and timely recognition of default. The same has to be disclosed on the website of each CRA.

Welcoming the move, CRISIL said that it has always supported enhanced standards, disclosures and transparency in the rating industry. “Towards that end, we welcome Sebi’s suggestion to revise the methodology of computing default rates and bring them in line with global best practices and increased disclosures for liquidity and rating sensitivity factors. We believe these will provide greater insights into the ratings of the companies and also help all stakeholders to evaluate performance of CRAs on the basis of a more robust methodology. The initiative to ask CRAs to put out the standard operating procedures for monitoring and recognising defaults is a welcome step which may remove the current inconsistencies across CRAs and hence reflect true default status.”