I guess we should talk about Europe and credit ratings. Now France isn’t AAA and Italy isn’t A and Portugal isn’t investment grade and here is something that someone at S&P actually said:

Our role is to give timely information to investors and if you give them timely information, if you give it to them in modest increments, then we think that they can make their own judgments about how they are going to allocate their portfolios.

Really! That could be S&P’s motto, “timely information, but in modest increments. Also not really that timely.”

If you’re into this sort of thing, though, the action is not in France so much as it is in the European Financial Stability Facility. The EFSF is basically, France and Germany and the other eurozone countries issue a bunch of debt*, put it into a blender, pulse until smooth, and then issue it as “EFSF debt.” The EFSF gets the money and uses it to prop up Greece, buy Italian bonds, etc. Because all the things are also all the other things, people saw this and were like:

1. Hey, that’s a CDO!
2. CDOs suck boo etc.

Here’s what the EFSF had to say about those claims:

No, EFSF is not a CDO. The essential difference between EFSF and a CDO is that EFSF debt has no tranche structure. There is no seniority and all investors have exactly the same rights. Secondly, EFSF bonds are covered by the guarantees from the euro area countries. However, a triple-AAA rating from all three leading credit rating agencies is not assigned lightly. EFSF has put into place additional credit enhancements through the use of a cash reserve and loan specific cash buffer which are immediately deducted from the loan made to a borrowing country in order to provide additional reassurance to investors. Consequently, all claims on the EFSF are 100% covered by AAA guarantors and cash.

That last sentence: not so much true any more. And, thus, the EFSF has been downgraded along with France to AA+. Maybe that’s okay. Maybe it’s not. Early days yet, with EFSF bonds trading off but still funding with no obvious catastrophe.

As a non-expert on S&P ratings process, the EFSF, CDOs, everything, I am a bit flummoxed by all of this. Part of the reason people like thinking of the EFSF as a CDO was that the intention was always to leverage the EFSF: to raise money by, essentially, borrowing multiple times against the delightfully safe AAA EFSF bonds. Now that dream is maybe dead. Felix Salmon writes:

On the one hand, it can restructure the EFSF so that it retains its triple-A credit rating. That would almost certainly involve shrinking the EFSF in size. Or, it can be sanguine about the EFSF downgrade and just let it happen. But that’s not a pleasant outcome either, given that everybody’s bright idea, when it comes to Europe’s sovereign bailouts, is to leverage the EFSF to some multiple of its present size. Leveraging a triple-A EFSF is hard enough; leveraging a double-A EFSF is pretty much impossible.

What flummoxes me is that the EFSF’s blender seems not to function the same way as everyone else’s blender. Again I’m not an expert in how structured debt gets blended and rated, but fortunately we have a great recent example outside of the sovereign space. Lisa Pollack at FT Alphaville has written a wonderful, technical, series of posts about a delightful regulatory arbitrage – briefly:

1. Banks have like BB-area loans that require lots of capital against them.
2. So they effectively crystallize a first-5% loss on those loans by “buying” “insurance” from hedge funds in which the bank gives the hedge fund 5% of the notional, the hedge fund posts it as collateral, and the hedge fund clips a fee without any real risk since their max loss is less than or equal to their “insurance premium.”
3. What this does for the bank is render the rest of the package mostly AAA: let’s say 85% AAA, 10% BBB, and 5% the unrated slop that the hedge fund is “insuring.”
4. The AAA and BBB things have lower capital requirements, meaning that instead of $100 of 1.5x-risk-weighted stuff you have like $85 of 0.2x-risk-weighted-stuff, $10 of 1.0x-weighted stuff, and $5 of, like, loss, which means you basically have $27 of RWAs (plus $5 losses) instead of $150, which at like 8% capital means like $2.16 of capital instead of $12, and if you assume there’ll be losses anyway so that $5 of “premium” isn’t a total cost, and your cost of capital is high enough, that could be a good deal.
5. The Fed and Basel think this is kind of bullshit. I think it’s kind of awesome but, yeah, also amusingly bullshit.

I partially made up the numbers above, but they partially come from this post going through a public example of this process, a Moody’s rating of a loan “synthetic securitization” (don’t worry about it being that thing, it’s the other thing, above – amusing pseudo-insurance) where the output was the 85/10/5 AAA/BBB/garbage split above and the input was:

In its base case Moody’s analyzed the performing portfolio to have a 4.8% expected default rate (corresponding to an average rating of Ba2 over 2.9 years of pool WAL and which includes a cycle stress of 30%) and a mean recovery rate of 43.8%. The average pairwise correlation is 5.3% and the weighted average mean used for simulating stochastic recovery rates is 43.8%.

So: Take $100 of BB-average loans. Put in blender. Out comes $85 of AAA paper, $10 of BBB, and $5 of unrated-but-so-bad-you’ve-already-taken-the-loss.

Now take the EFSF. Just for fun let’s calculate an average rating:

EFSF participations from here, ratings from here, weighted averaging by just awarding 1 point for AAA, 2 for AA+, 3 for AA, etc.

Now, of course, that synthetic CLO thing is totally different! Corporate loans! Amortizing! Probably less correlated! (I guess? Why?) Different agency, even, Moody’s vs. S&P. But still: $100 of AA/AA- (closer to AA) average paper put into the blender, with the credit protections that the EFSF mentions above (albeit a bit tarnished given that some of the AAA guarantors, ahem France, are no longer AAA), generates $100 of AA+ rated paper.

So how much AAA paper could it generate? Let’s guess that $85 is a … floor? No? How much credit enhancement (in the form of issuing a subordinated tranche) would you need to get back to the AAA rating for the rest of the EFSF? Like, $10? $5? Less?

I don’t know. Like I said, I’m flummoxed. I think part of the point of this is that it strikes me that it would be only a very little bit more difficult to leverage an AA+ EFSF than an AAA one. Like, just by math. But part of the point is – why does this synthetic CLO pool of BB-ish corporate loans end up looking so much better than the continental-savior pool of AA-ish sovereign debt? In one sense it’s straightforwardly explicable: the synthetic CLO has credit enhancements that the agencies like (tranching), the EFSF doesn’t, fin. In another sense it’s weird: while Moody’s quietly and cheerfully negotiated a mostly-AAA-rated structure for that synthetic CLO, S&P has been noisily whinging for weeks about how horrible and vulnerable the EFSF structure is.

You can if you like think in terms of incentives. Hey guess what: European politicians hate the hell out of the ratings agencies. And don’t pay them. Meanwhile, that synthetic CLO mentioned above? Barclays paid Moody’s for a rating even though the synthetic CLO notes that it issued were unrated junk: the rating was solely for Barclays’ risk-weighting purposes. Rating banks’ structured assets just for regulatory risk weighting seems like … well, sort of not the intent of recent rule changes, but also very much a good business. Better than sticking your nose into Europe, anyway.

Euro Rescue Fund Sells Bills ‘Smoothly’ After S&P Downgrade [Bloomberg]

S&P downgrades Europe [Felix Salmon]

Barclays visits the securitisation BISTRO [FTAV]

* Really guarantees but I find it easier to think of it as debt + blender.