The spotlight is on credit rating agencies (CRAs) once again, this time triggered by the crisis at, and default by, Infrastructure Leasing & Financial Services (ILFS). In this, the ‘issuer pays’ business model of CRAs seems to have borne the brunt of criticism.
While this appears to be an intuitive response, a comparison of other possible CRA business models would indicate that the problems lie elsewhere.
The tendency to associate CRAs with financial failures dates right back to early 20th century. The most incisive of such inquiries followed the global financial crisis of 2008, conducted by financial regulators including in the US and India.
What is pertinent, however, is that, after detailed deliberations, the issuer pays model of the CRAs was left untouched by all the regulators. This was an implicit endorsement of the model’s relative superiority, or at the very least, of it being the lesser evil.
It is normal that credit ratings change over time. The primary concern surrounding ILFS’ ratings is about timeliness, abrupt and steep transitions, slow/inadequate change even after evident distress, and inadequate use of ‘outlooks’ – that handy rating prognosis tool.
That warrants questioning of the CRAs concerned in recent cases – CARE, ICRA and India Ratings – and also of others culpable in other instances.
To be sure, the inherent conflict of interest in the issuer pays model is undeniable, since the rated entity is paying for the rating. But it is instructive to consider some nuances.
Any CRA business model must facilitate two objectives: 1) ensure ratings are of high quality, and 2) ensure all market participants have access to such ratings at a reasonable cost. Additionally, the models must involve minimal or no conflict of interest. ‘Minimal’ is a considered usage here since conflicts cannot be entirely eliminated.
The issuer pays model
Under this, the issuer pays the initial and subsequent surveillance fees to CRAs, and the ratings are publicly and freely available. Market dynamics determine the fees and issuers are contract-bound to afford CRAs access to business and financial details, which facilitates better analysis. The cost to investors is virtually zero. For issuers, rating fees count among the smallest components of issue costs.
Two aspects need reflection regarding potential conflicts, whereby a CRA might systematically assign inflated ratings to satisfy issuers.
One, if a CRA consistently indulges in such a practice, it will be harakiri, for sooner than later, it will lose credibility, which is a critical success factor in the business.
Two, typically, CRAs collect upfront 100% of the non-refundable rating fee even before beginning the exercise. This eliminates payment risk and significantly reduces the incentive to assign favourable ratings. Even if an issuer delays or defaults on annual fees, the CRA is obligated to monitor and disclose rating changes. Therefore, despite the inherent, and potentially subversive, conflict in the model, these two aspects considerably mitigate the risk of malfeasance in practice.
The analytical competencies and standards of CRAs are a different matter altogether, but these are not frailties attributable to the issuer pays model. Three specific statistics lend support to this argument: One, out of ~35,000 ratings currently outstanding in India, nearly three-fourths are in the ‘non-investment grade’ category. Two, during fiscal 2018, Indian CRAs downgraded over 2,400 ratings. Both these suggest CRAs haven’t shied away from assigning lower ratings or downgrading those who pay for the assessment.
Third is the ‘ordinality’ expectation in ratings default performance (higher rated firms expected to have a lower rate of transition-to-default, than lower rated firms). This is a key yardstick of ratings quality. The four largest Indian CRAs have systematically demonstrated lower default rates for higher ratings. So while CRAs have erred in specific instances, this statistic reflects overall good quality and consistency.
The ‘investor pays’ model
The investor pays model has the advantage of precluding the issuer-CRA nexus. But it poses another, equally serious, conflict involving the investors themselves. Private placements of debt account for about 90% of issuances in India, and the secondary market is subdued. Therefore, if investors are the payees, they can influence CRAs to give lower-than-warranted ratings to help them negotiate higher coupon rates. After placement, too, investors can resist downgrades of the securities they hold, as it can trigger mark-to-market losses. The scope for this conflict is real.
Then there are other problematic issues. One, only those who pay for a rating can access it. Two, the cost of rating can be exorbitant since only a few investors may seek it, and the public would likely be the most deprived. Lastly, under investor pays, issuers may not always share full information with CRAs, which can, ab initio, jeopardise the quality of ratings. Therefore, on all counts, the investor pays model is decidedly inferior.
The government (or regulator) pays model
Conceptually, the ‘government or regulator pays’ model can eliminate bias in ratings because there is no pecuniary incentive for the CRA (public sector entities could be an exception). Here, the regulator can also mandate free dissemination of ratings to all. So far, so good.
But this model introduces other complexities. One, the choice of the CRA and the payment mechanism are extremely problematic, as it is difficult to conceive of either the government or the regulator exercising a choice regarding CRA quality beyond a minimum threshold, or paying market-determined/ negotiated fees for ratings.
On what basis would the regulator pick a CRA when the value proposition fundamentally differs from that of auditors? Two, if fees are not market-determined, CRAs would be hard-pressed to build capability and retain sharp analysts. Moreover, once business is assured, there may be little incentive for quality and excellence.
The most serious problem with this model is the ‘moral hazard’ wherein all ratings can be seen as regulator-endorsed. This is antithetical to efficient market practices, and can pose potentially disastrous consequences for the government/regulator in the event of poor rating calls.
Which means, operational and moral hazard issues overshadow the intuitive appeal and advantages of this model.
Conclusion
Sure, each model has its pros and cons, and conflict/moral hazard potential. It’s imperative, therefore, to pick the model that best meets the primary objectives of credit rating – high quality and open access – when offering the best opportunity to manage any conflict.
By those yardsticks, the issuer pays model recommends itself.
These arguments do not seek to absolve CRAs of poor judgement, lack of timeliness or the incompetence they have exhibited from time to time.
But the cause of serious concerns that typically surface during financial debacles (and therefore solutions) lie elsewhere.
Structurally and behaviourally, there is a need for a deeper understanding of credit rating process, related systemic issues, and a more enlightened approach to evaluation of CRAs. That would facilitate greater transparency, high quality ratings, and healthy competition.
The author is a Professor at IIM, Kozhikode
Disclosure: Venkataraman S worked for nearly two decades with a premier CRA in India and was also the founder CEO of Caribbean-based regional CRA. He writes this as an independent analyst.
