“Is that BBB really BBB?”​

In a recent Thomas Day posting about the recent bond sell-off triggering concerns about the maturing low IG debt at the median of the issuer distribution for US, European and other credit markets. Is it really still BBB?

Was it ever really BBB? asks Dr. Chris Donegan

CDO markets around the world failed in 2008. It was a perfect test of whether corporate debt ratings predict corporate bond default, including bonds backed by structured debt and repackaged as corporate debt.

After the GFC, government-licensed credit rating agencies (NRSROs) wanted to go on rating CDOs and CLOs but they didn’t want to be liable for results that didn’t work out. So although NRSROs had been publishing bond default studies since the 1980s “proving” their ordinal rankings work like measures of default, they backed off. You may or may not have noticed their disclaimer. But you still see them rating CDOs and CLOs.

Common sense tells us a measure is not necessarily a prediction. My son weighed 8 lbs. at birth. It was a flawed estimator of his future weight and height, and this could be known with 100% certainty. But that statistic together with a valid logistical study, enabled me to estimate his expected adult height. All measures human follow logistical patterns. They are not perfect estimators, but they are made more perfect via continuous updating.

Let me repeat the ingredients for good “forecasting.” A bounded measure based on transparent inputs, and frequent or continuous updating.

The same logic holds true for predicting loan performance. Loan default rates follow a logistical curve. Different subpopulations have different characteristic curves, which are are largely explained by loan quality impacted by underwriting standards and collection quality. Underwriting quality is tied to resilience under macro-stresses. Collection quality is tied to loan management experience under stress. Structured ratings are (or should be) produced in a structural model that reflects these performance characteristics for a specific capital structure backed by a specified collateral pool.

In contrast, corporate credit ratings express degrees of financial resilience, from “defaulted/insolvent” to highly resilient payment ability under conditions of severe constraint, based on the general experience of previously rated bonds. I commend to you looking up the meaning of different levels in the rating scale in words as a worthwhile exercise–I’ve used these definitions in expert work when NRSRO structured models failed.

But what you cannot do is look up the numerical definition that a corporate rating is calibrated to. Ordinal ratings don’t imply numbers.

If you were not aware that structured ratings require the use of numbers and calibration to a numerical scale, you couldn’t have contemplated calling NRSROs to account by demanding that this information be made public.

But if you are well and truly unhappy with the shortcomings of the NRSRO system, why not start demanding a better architecture, starting with a fixed point scale for calibrating structured ratings at origination. In speed, cost, accuracy and reliability–I guarantee this approach to revamping our credit ratings system is Occam’s razor.

We can never know if a corporate BBB is or ever was BBB. But we can know if a CLO tranche rated BBB is BBB at origination. And given that CLOs are backed by the common woman and man’s working capital, borrowed in their sincere desire to better their life and the lives of their clients, we should value their commitment to pay by the same benchmarks as the institutions that have many more financial advantages. This seems to me to be a standard of fairness every American should be able to get on board with. SME loans and CLOs should start life on a level playing field.

What happens after that? Stay tuned.