Moody’s view of sovereign ratings

By Felix Salmon
August 15, 2011
Adam Ozimek's interview with David Levey,  the former Director of Sovereign Ratings for Moody’s.

" data-share-img="" data-share="twitter,facebook,linkedin,reddit,google" data-share-count="true">

I’m very late to Adam Ozimek’s interview with David Levey,  the former Director of Sovereign Ratings for Moody’s. Levey makes a couple excellent points about sovereign ratings, starting with this one:

On methodology, things changed in the mid-2000s. Under pressure from regulators and issuers, the agencies were forced to “open the black box” and become much more explicit about their criteria, scorings, weights attached to various factors, etc. There was a tendency to move to a “scientistic”, quantitative, formulaic approach. I tended to resist that (being a great admirer of Hayek). I saw risk assessment as a multidisciplinary, highly qualitative, judgment process involving a varied weighting of factors.

I fear that this is exactly the reason why S&P started coming out with weirdly precise deficit-cutting targets — the “chop $4 trillion over ten years, or we’ll downgrade” phenomenon. Sovereign ratings are always more of an art than a science, and the pressure to become quantitative and formulaic has been deleterious at both major ratings agencies.

Levey also explains how ability and willingness to pay are inextricably intertwined:

Neither willingness nor ability can be defined independent of the other. “Willingness” depends on political calculations of the degree of sacrifice that would be required to make payment, which in turn depends on the financial resources available or easily raised. “Ability” depends on how much additional resources for debt payment the government can “squeeze out” through reductions in spending or increases in taxation — a political consideration. So the relation between them is -to use an old phrase-”dialectical” and the analysis is based on what Adam Smith called “political economy”. This may sound like “scholastic” nit-picking, but the point is vital for guiding the rating decision process.

It is reasonable, therefore, for S&P to keep one eye on debt ratios when it’s judging the US willingness to pay: the higher the amount of debt a country (or any debtor, for that matter) has, the lower that debtor’s willingness to pay becomes. If I only have $5,000 outstanding on the mortgage on my $300,000 home, I’m going to be extremely willing to pay that debt. If I have $500,000 outstanding, then, not so much. But as Levey says, these things aren’t scientific. And it’s silly to try to move from a general principle — the higher your debts, the lower your willingness to pay them — to something spuriously quantitative, like putting a dollar amount on the point at which your willingness to pay will fall.

Levey also says that I’m wrong about the difference between S&P and Moody’s. He’s much more of an expert on this subject than I am, so it’s worth taking his views seriously. But his stated reasons for disagreeing with me are pretty weak. He gives three:

First, S&P “notches” for subordinated debt, meaning that they are taking into account that a default on that debt is likely to have a greater severity than on senior debt.

No, it doesn’t mean that at all. Subordinated debt is more likely to default — that’s what subordinated means.* And because it’s more likely to default, it has a lower credit rating. You don’t need to look at severity in order to give a different credit rating to subordinated debt than to senior debt.

Second, market participants would find ratings almost impossible to use without comparability of meaning. So — in a sense — the markets more or less force equivalence of meaning on the agencies.

Essentially, Levey is saying here that S&P has to do what Moody’s does, since the markets think the two agencies are doing the same thing. I don’t buy it.

Third, if the meanings were that different, there would be a lot more “split ratings” (situations where the agencies rate differently) than there are.

This isn’t really true. The main case in which the difference in methodologies would result in a split rating is a case where there’s a significant risk of default, but where the likely loss given default is zero, and the likely recovery rate is 100 cents on the dollar. This is exactly what we’re looking at in the case of the US sovereign rating, and it applies to entities with US sovereign guarantees as well. But beyond that, it’s rare elsewhere. In general, if an issuer defaults, they’re not likely to pay off their debts in full. And as a result, S&P and Moody’s are likely to have similar ratings.

The ratings could diverge at the other end of the spectrum, too — Moody’s might have a lower rating than S&P, if the default probability were low but the loss given default was likely to be enormous. When does that happen? Mostly in the world of structured credit, where Moody’s had a vested interest in ignoring the fact that recovery values in the event of default were generally zero.

All that said, I suspect that what Levey’s saying here does have some truth to it: S&P is well aware of how its ratings are interpreted by the markets, and does little to disabuse the markets of that impression. I do think that the difference between S&P and Moody’s explains why S&P downgraded the US and Moody’s didn’t. But I don’t think the difference is all that big in their general day-to-day operations.

*Update: Greycap points out, in the comments, that really the meaning of the term “subordinated” is what happens to the debt in bankruptcy: that the claims are junior to other claims after the company has defaulted. This is true. But by the same token, everybody knows this, and lots of companies restructure or otherwise work out their debts “in the shadow of bankruptcy” — essentially trying to replicate the likely outcome of a bankruptcy case without all its myriad associated costs. It’s very easy, in that situation, for subordinated debt to suffer a haircut or become converted to equity while senior debt remains whole.

Update 2: Greycap seems to think that a default on subordinated debt while the senior debt remains whole “arguably has never existed in the history of the world and will never exist until the end of time”. Just off the top of my head, isn’t that exactly what we saw at Anglo Irish Bank?


We welcome comments that advance the story through relevant opinion, anecdotes, links and data. If you see a comment that you believe is irrelevant or inappropriate, you can flag it to our editors by using the report abuse links. Views expressed in the comments do not represent those of Reuters. For more information on our comment policy, see

I fear that this is exactly the reason why S&P started coming out with weirdly precise deficit-cutting targets — the “chop $4 trillion over ten years, or we’ll downgrade”
Yes, and I’m sure it that the weirdly large size (along with the fact that there is no economic-’theoretical’ reason whatsoever to assume that massive government spending cuts will create economic health rather than a “double dip”) has nothing to do with the fact that S&P is owned by McGraw-Hill, and that the CEO of that firm is also the head of the Business Roundtable..

Posted by Foppe | Report as abusive

“Subordinated debt is more likely to default — that’s what subordinated means.”

No it doesn’t. Subordinated means “subordinated”: lower in claims priority than other debt in the event the issuer defaults. So it has lower recovery, not lower default probability.

Posted by Greycap | Report as abusive

Now do you grasp Levey’s point? Say a company issues two bonds, one “senior” and one “subordinated.” The meaning of this is that in the event the company defaults, the claim of the first is senior to that of the second, and is paid first. The default probability of the two bonds is identical but the expected recovery in event of default is not. So if they have different ratings, the ratings can only be explained by recovery.

This case is extremely common, in contrast to your “main case”, which arguably has never existed in the history of the world and will never exist until the end of time.

Posted by Greycap | Report as abusive

David is a friend and former colleague. After the East Asian crisis and then Enron/WorldCom, investors wanted rating methodologies that were “transparent and replicable”. As a result, today Moody’s publishes somewhat prescriptive methodologies for all major sectors. While these are not binding on the rating committee, they provide clear signals about what matters and what doesn’t. This is good for investors and issuers who used to have to guess about criteria. It also introduces a greater degree of intellectual rigor, because it prevents dancing around the data. Do the methodologies produce rating outcomes that are consistent with expected loss? That question is answered by Moody’s default studies and rating performance metrics. S&P had clear sectoral criteria and specific criteria for the US. They followed those criteria. You may not agree, but at least they are not playing games.

Posted by nixonfan | Report as abusive

The view from the head of Dagong, the Chinese rating agency, in a short interview translated in the English version of Der Spiegel: d/0,1518,780502,00.html

Posted by aquacalc | Report as abusive

aquacalc, if you think Dagong is serious you should rush out and buy chinese bank debt and local government debt. China just spent 10% of GDP bailing them out. Let us know how that works out for you, maybe you’ll do as well as the geniuses who bought chinese companies that reversed onto US and Canadian exchanges.

Apparently everyone DOESN’T know the difference between sub debt and other debt. You and the “financial engineer” Mike Konzal apparently didnt know the difference with first and second mortgages. Also sov debt is completely different from corporate debt. For one thing, if the US gov defaults no one is going to be selling off the white house, as the debt is backed by the faith and credit of country. The resolution process is different too – ie no international chapter 11 or bankruptcy court.

Posted by Danny_Black | Report as abusive

“isn’t that exactly what we saw at Anglo Irish Bank?”

There have been many examples of 100% recovery in history. In fact, considering only nominal amounts, recoveries are sometimes over 100%. One way this can happen is that interest can be accrued on delayed settlement. Another (which applies more to loans than bonds) is that restructuring can result in more onerous covenants on the borrower; covenants that can trigger various payouts if breached. These covenants have positive economic value which boosts effective recovery.

(It is normal to report nominal recovered amounts; something to bear in mind when examining recovery statistics. The record I have seen is 100% recovery … after 20 years!)

But I digress. The issue is that while *realized* recoveries may be 100% or more, a rating is supposed to indicate *expectations* of recovery. An expectation of 100% corresponds to zero recovery risk. I challenge you to produce a historical example of a market expectation of 100% recovery. Especially since you are the guy who always says there is no such thing as a risk-free asset!

Finally, I can see that my comments were both hyperbolic and snarky; I apologize. But you did wind me up!

Posted by Greycap | Report as abusive

Without going into the question of likelihood, I think there’s another reason why a default on subordinated debt could have a more severe impact on a financial institution than a default on senior debt: if the debt is truly subordinated and therefore puts these “tier-2″ holders in a position similar to that of equity holders, regulators often allow tier-2 issues to count toward a financial institution’s capital requirements. So, if a senior default weren’t already apocalyptic enough, a t-2 default could mean that an institution also fails to meet its capital requirements. Not good!

Posted by johndelaney | Report as abusive

I think Lewey has summed it in not only the way it is, but has to be. Has anybody here ever tried to come up with her own credit ratings, for corporations, ETFs, countries ?
I did, and you end up with very similar ways. It is kind of generic. But in the end there always has to be a certain judgement, and not just some official numbers, especially if you dont trust them, but cannot say so in public. If you look at Greece, Portugal, Ireland, S&P was the more agressive in downgrading, what, from my perspective represents their more independent positioning. But ultimately none of them can stay too far away from CDS and interest spread numbers (especially in Euroland)without getting ridiculous. Just like federal bank rate announcements. Rating the US AAA or AA also depends on things you can not say in public, and whether you were horrified of the discussion or, with some knowledge of the US history, just enjoyed the show.

Posted by genauer | Report as abusive

FWIW – Each of Levey’s 3 reasons why Moody’s and S&P ratings effectively mean the same thing, despite the supposed difference in approaches, is bang on. Default on sub-debt but not on senior debt can happen, but it is the exception rather than the rule. In practice, few market participants can be bothered to differentiate between the ratings assigned by the agencies, and the historical default statistics published by the agencies confirms that they are close enough that the difference isn’t worth worrying about.

Posted by BoringCanadian | Report as abusive

You can’t default on sub debt and not on senior debt. What you can do is either negotiate hard with the sub holders on the basis they will get more with the new deal or if you are the government you can just ignore the rule of law or change it. AIB didn’t “default” on it’s sub debt, it “renegotiated” it. There is a difference.

Posted by Danny_Black | Report as abusive

@Greycap — I never mind snark. But Anglo Irish senior debt didn’t *recover* 100%, there was never an event of default in the first place. And yes, @Danny_Black, the sub debt did see an event of default. Ask the people who wrote CDS.

Posted by FelixSalmon | Report as abusive

The test of whether S&P considers their notching of sub-debt to be a PD or LGD issue would be in how they treated it in the default statistics. If they count the default against the rating of the senior debt (even though it remained current), that suggests that the senior debt reflects the PD and notching the sub-debt was due to lower expected recovery. On the other hand, if they count the default against the rating of the sub-debt, then that suggests that they believed they were rating the PD of the sub-debt itself. I don’t see enough disclosure in their 2010 report to be sure which approach they chose.

Posted by BoringCanadian | Report as abusive