To the uninitiated, default appears to be a neat, compact and self-explanatory word and concept. It occurs when someone fails to repay, as promised, money that has been borrowed. Right? Well, you could not be more wrong. There is much variety in how the world defines and treats default.
And as the financial system expands and adopts a more prudential approach to lending, the different ways in which default is dealt with sometimes make the difference between, well not quite life and death, but likely between profit and loss!
In the practical world, some define default as failing to pay after a grace period. Grace periods vary between a few days and even several months. Some even do not worry about missed payment. They are concerned only about the ultimate loss of their money in case they fail to recover the full amount.
In the practical world of big money, institutions like banks also forsake dealing with default as a singular concept and prefer to look at the broader concept of non-performing loans or assets on a portfolio basis.
In all this, the simple fact that the obligation to make good every payment is lost sight of and the definition of default becomes as esoteric as the definition of beauty!
Why is this definition so important? Regulators the world over are adopting risk-based supervisory norms. For instance, the new Basle Capital Accord, which is a code that all bank regulators subscribe to, has requirements that depend centrally on how defaults are recognised.
This capital accord, popularly known as Basle II, requires banks to credit rate its assets and then allocate lower or higher capital, depending on whether each asset has a high or a low credit rating. This makes credit rating an important discipline to the entire financial community.
The above regulation treats credit rating as an absolute concept that is uniformly understood and determined. Unfortunately, it is not. In fact, nowhere is the division of opinion on what default is sharper than with the credit raters!
They differ on two dimensions. Ratings, according to one school of raters to which Crisil belongs, denote the probability of default, whereas the other school says it denotes the expected loss. To elaborate, the former school says its higher ratings indicate a lower probability that the debt instrument (bonds, loans, etc.) will default on its obligations.
The other school says the higher its ratings, the lower will be the loss that the investor in those instruments will suffer. The expected-loss methodology factors in the probability that the instrument will default (the probability of default) and then multiply it with loss-given default to arrive at the concept known as expected loss.
While the claim of this methodology is it gives the ultimate guidance to investors, it is to be used with great caution, because it has the effect of masking occurrence of default.
For instance, when an instrument fails to pay as per its promise, the former school of raters will automatically recognise that the instrument is in default by assigning the default rating, and the latter school of credit raters will assess the extent of possible credit loss, and the rating will be adjusted based on the severity of that expected loss.
The big difference is the world will not know that a default has occurred in the latter instance, while in the former, there is no such consequence. Also, the methodology used by the rating agency in determining the loss-given default might be at variance from the methods used by the various institutions and therefore yield different loss expectations.
Also, in a country like India, where recoveries of the collateral take a long time and are usually ineffective, using a uniform approach to loss-given default is fraught with many approximations.
Lastly, I believe it lets the rating agency off the hook by allowing it the luxury of not acknowledging default on an instrument it rates, by offering a sense of comfort on recovery. It is precisely for this reason that Crisil has introduced bank loan ratings, where the rating is on two dimensions - one indicates the probability of default and the other the efficacy of recovery.
So if and when a default occurs, the rating will indicate the defaulted status and then give a sense of what is likely to be credit loss, given that the default has occurred.
The other aspect in which this difference in the treatment of default is striking is the way default statistics are compiled. Default statistics are a good way to measure the efficacy of ratings by a particular agency.
For instance, in the former case, if the default statistics indicate that instruments enjoying a higher rating display a lower default rate, it is an indication that the rating agency has got its methodology right.
In the latter case, since there is neither an attempt to capture default, nor are the ratings supposed to alert investors to default, default studies will have a serious limitation in indicating rating effectiveness.
Perhaps a suitable study will be to indicate the loss incurred on various instruments rated by them over time and to establish that the higher the instrument is rated, the lower the loss incurred. How the data will be gathered, standardised, and verified by an external source for such a study to be considered objective remains to be seen.
Apart from the regulators, the markets are using ratings in their quantitative tools while investing in debt instruments. They also treat ratings as a monolithic input into their models. While there is considerable merit in doing so, it is important to recognise that there are some important fundamental differences in the approaches various rating agencies use.
As we have seen earlier, ratings, though sharing tools that appear similar, have fundamentally different meanings. It is important for users to adjust the models to match with the system used by rating agencies, so that consistent results come out.
Personally, I would prefer to use the digital approach to default. It is the safest method, and does not give anybody - the borrower, the lender, or indeed the credit rating agency any cover to hide behind statistics and voodoo magic to throw up false comfort.
When a payment is missed, it has to be recognised and all statistical analyses must capture this fundamental truth. Like in life, you are either dead or alive; can't argue about it. In the same way, default should be made digital and absolute.
Given the rapid progress in regulation and markets, variations in the basic definition and treatment of default have to be resolved expeditiously before widespread usage leads to conflicting and faulty results.
The writer is managing director and CEO, Crisil Ltd.
