A Robust Regulatory Framework For Credit Rating Agencies

The rating actions by the Indian credit rating agencies (CRAs) have recently come in for a lot of flak on account of their failure, in a few significant cases, to downgrade the issues before an occurrence of default in those cases. While the public discussions have mostly focussed on somehow forcing the CRAs to perform better, there has not been much discussion on how to achieve the same. There are deeper issues in terms of some fundamental flaws in the way credit ratings are regulated across the globe.  

The key product of a CRA is a statement of prediction about an obligor’s ability to meet its obligation. In that sense, it may be compared to many similar businesses like equity research, medical drugs, beauty enhancement, or even astrology. The product in each of these businesses offers just a hope to the customer with different levels of likelihood. The regulatory structure in all such cases focusses mainly on protection of the customer from patently false promises. The regulations basically aim to provide appropriate disclosure to the customers on the process and the past performance of the sellers of the product or service. The customer can then independently decide on the choice of availing the product. 

The usage of credit ratings is currently driven mainly by directives from the Basel-3 based regulations rather than the need of customers. That is a primary cause of the repeated failures of the CRAs, worldwide. 

The Securities and Exchange Board of India (SEBI) has recently issued a circular (SEBI/ HO/ MIRSD/ DOS3/ CIR/ P/ 2019/ 70, dated June 13, 2019) titled “Guidelines for Enhanced Disclosures by Credit Rating Agencies”.  It is a welcome move to ask the CRAs to increase the level of disclosures in standardised formats. It is commendable that these disclosures will also be maintained by the CRAs in the public domain. SEBI should continue to encourage the CRAs to disclose the highest quality of information, within the acceptable systemic cost, on their processes and past performances.

There are a few issues in the circular issued by SEBI that may need some review.

1. Several terms have been used in the circular, but not defined. The terms “default”, “marginal default rate” and “cumulative default rate” need to be either defined in this circular or referenced to a definition followed in the circulars of any other regulatory agency. For example, the Reserve Bank of India (RBI) circular has already defined “default” and the same may be adopted. In the absence of standard definitions, these terms may be prone to individual definitions and thus the disclosed data on these not fully comparable across the CRAs. 
2. The benchmarks for the probability of default (PD) for rating categories of AAA, AA, and A have been stipulated at zero percent for a one year time horizon. 

This is conceptually flawed and goes against the fundamental principle of risk assessment. Only a risk free asset can have zero probability of default. Any risky asset cannot, by definition, have a zero probability of default. 

To illustrate the counterfactual, let us imagine a situation. Suppose, as of today, a 1 year government security (G-Sec) provides a yield of 7%. Also, suppose a 1 year bond, rated AAA, provides a yield of 7.2%. If the two instruments were to be exactly similar in all respects (like tenor and risk), other than the credit rating, the investors would consider them to be fully interchangeable. In case of a difference in their prices, they would sell the instrument with a lower yield and buy the one with a higher yield. In other words, they would sell the G-Sec and buy the AAA rated bond. This will lead to the prices of both instruments converging to a point where both are quoted at the same price and provide the same yield. This would be the equilibrium point. However, we will then be ending up with a situation where two different instruments with different credit ratings have the same price. This is therefore an anomalous situation and cannot happen in real life.  

Generalising the learning from the above anomalous situation, it may be said that two basic principles cannot be violated while rating the instruments:

1. No risky asset can have a default rate of zero even for a moment. The default rate for all risky assets, including AAA asset, will be positive, even if very small, for a one year period.
2. No two assets with different levels of credit rating can have the same level of PD. Thus, multiple loans or bonds rated as AAA, AA, and A cannot have the same PD for any common time horizon.

This anomaly of PD being zero for any rated instrument needs to be revised in the circular. 

1. The circular has also referred to the use of bond spreads of debt instruments as a factor while considering a material event. It is true that bond spread is often a good indicator of the credit worthiness of an issuer. However, this falls in the domain of the methodology for arriving at the PD and the credit rating of a debt instrument. A CRA should be totally free to devise its own methodology and choose the underlying factors for arriving at the rating. It is not desirable, or even required, for a regulator to stipulate any factor. 

CRAs need to keep working on their prediction models to improve their accuracy. They should be given full freedom to devise their methodologies and choose the markets in which they wish to operate. At the same time, they must be required to maintain the highest levels of disclosure on their non-proprietary processes and history of performances. This will allow the investors to make appropriate decisions without compromising on the ability of the CRAs to provide the best services. 

In sum, credit rating is a matter of opinion. Let the markets be empowered to accept or reject the opinions from various sources. Regulations should strive to encourage level playing field and market efficiency.  Hopefully, SEBI and other policy makers would soon put in place a robust framework toward this.