Skip to main content

What Are You Paying Ratings Agencies Zero Dollars For, Anyway?

  • Author:
  • Updated:

Fitch released a report today saying "ohmygod banks Europe" and the market went down and maybe there's a causal link, whatever.

The report mostly takes notice of US banks' European exposures in general, and the mystery of net versus gross derivatives exposure in particular, in which one asks "if Bank A sells CDS on $100bn of Italian debt to Bank B, and buys CDS on $100bn of Italian debt from Bank C, then when things go pear-shaped is it on the hook for zero (because it has no 'net exposure') or $100bn (because Bank C goes belly-up) or somewhere in between (because of collateral, sub-1 correlations, etc.)?"

It's an important question: net exposures are manageable, gross exposures are terrifying, and there are legitimate questions about whether in a stress case the netting could break down. Various people who are smarter than me have tried to triangulate around parts of the answer using public data.

I don't know the answer and doubt I'll find out, though my gut is that netting should kind of sort of mostly work (I find Graph 5B of this, and the definition of "bilateral netting," oddly comforting). What troubles me today, though, is that Fitch has no clearer answer than I do. From their report:

Exposures to Major Markets: Gross exposures to large European countries and major banks are greater than exposures to the stressed markets. For instance, cross-border outstandings to France totaled $188bn for the top five U.S. banks (based on 2Q11 FFIEC and 10Q data). Of this total, $114bn was to French banks (25% of T1C). Aggregate outstandings to the UK were $225bn, of which $51bn (11% of T1C) was to UK banks.

Exposures Overstate Direct Risks?: A large portion of this exposure is likely secured (such as repo-related exposures), subject to margin/collateral agreements and/or hedged. Nevertheless, the exposure data show the susceptibility of banks to contagion risk and the interconnectivity of large global banks. ...

Disclosure is generally limited to aggregate claims outstanding to a country greater than or equal to 0.75% of a bank’s assets. Consequently, it is not possible to gather complete exposures for all countries. Total country exposures include deposits, central bank balances, securities, loans, participations, acceptances, fair value of derivatives, and reverse repos, among other items, but details of many categories are not disclosed. The long lag time before FFIEC quarterly country exposure data become available also hampers analysis.

In theory, one way we could all find out for sure is something like the following:
1. Call Jamie Dimon,
2. Have him send over JPMorgan's derivatives book at a reasonable level of granularity, showing what kinds of derivatives it's bought and sold from what counterparties,
3. Read it,
4. Put those exposures into some sort of plausible model, plug in some plausible loss assumptions, and go all Margin Call on it, and
5. Figure out that everything's cool or rush into The Mentalist's office and to freak out and sell everything.

That's the approach that the market took with Jefferies, and it worked! They disclosed CUSIPS and swore they'd never even heard of CDS and eventually made people believe that when they said "no net exposure" they really meant it. Good on them. But they're Jefferies. JPMorgan won't do that. And, even if JPMorgan opened its books, and it turned out they'd bought a lot of CDS from Goldman, we'd need to run the same exercise on Goldman to know if Goldman was going to blow up and bring down JPMorgan's protection. Turtles, etc.

But it doesn't matter because JPMorgan isn't talking. But some people might have more luck getting details on the banks' exposures. Which people? Regulators, yes. God, we hope so, anyway. Again, I'm an incurable optimist, but I feel reasonably good that people at the Fed and elsewhere are actually keeping an eye on bank interconnectedness and European exposure and have some sort of ideas on all of this that are informed by quite granular bank data that is only made public in boringaggregatedways.

Equity research analysts, no. For lots of reasons, one of which being that Jamie Dimon isn't allowed to tell research analysts things that he's not telling everyone everyone else. So when research analysts come to the defense of big banks' balance sheets, or criticize them for that matter, they're relying on more or less the same data you and I can see.

Here are some people to whom JPMorgan could give a full download: Fitch. And S&P. And Moody's.* All of whom rate JPM, and GS, and MS, and C, and BAC, and ... Soc Gen and BNP Paribas for that matter. And who are in the business of, y'know, inquiring into those banks' stability and likelihood of defaulting on their debts. Fitch actually explains in some detail how it rates banks; here's a sample:

As part of its analysis of a bank’s credit derivative exposure, Fitch takes into account both the amounts of protection sold and bought by a bank, the motivation for the use of these products, a breakdown of credit derivative activity by product, and the underlying ratings of counterparties and reference entities. Any credit events relating to credit derivatives experienced by a bank, as well as the financial reporting procedures relating to its credit derivatives book are also examined.

I'm sure I'm missing something here, but I don't really understand why Fitch's banking industry analysts, who certainly talk a good game about developing their ratings using detailed nonpublic information about counterparties and reference entities that they obtain in a collaborative process with the banks that they rate, released this report based entirely on public information. One potential conclusion is that they have lots of more granular detail from the banks, which makes them feel good about their (sort of wishy-washy and unsurprising) conclusions in this report, but they are constrained to discuss only public information in the report. That's not particularly how the report reads, and if it's the case there doesn't seem to be much point in releasing the report.

The alternative conclusion is that Fitch rates the big U.S. banks without detailed or timely information about their European exposures, information that they claim to focus on in their ratings process. If you think, as anyone who owned MF Global probably does - and, hell, as Fitch apparently does - that understanding European exposures is pretty, pretty, pretty important to understanding the credit quality of US banks, then that shouldn't make you feel great about their ratings.

Financials Drag Market Lower, Fitch Raises Doubts About Euro Hedge Effectiveness [WSJ MarketBeat]

Life’s a Fitch for Big U.S. Banks [Overheard]

U.S. Banks – European Exposure [Fitch]

* Because of Dodd-Frank, ratings agencies no longer have a special exemption from Reg FD disclosure requirements, but can still receive nonpublic information to inform their ratings decisions subject to a confidentiality agreement.