Sir, you seem to overlook the fact that AAA rating should imply a very low expected loss. The loss magnitude and distribution of AAA rated securities historically has been very low. For example the average annual default rate for AAA rated debt by corporate issuers from 1990 to 2001 was 0.00% according to Fitch.

I never said they did. This is true but as deals season they delever (or used to anyway). The Bloomberg article misleadingly applies this ratings test to seasoned deals in the ABX. My example was an exaggeration but similiar cases appear in the BBerg analysis. They highlight LMBLT 05-WL2 3A4 as not deserving a AAA rating despite 62.5% CE and 32.8% 90+ DQ. I didn’t even mention the fact that they make the mistake of saying “investment grade” is the same thing as “AAA”.

“Second, market prices show that AA tranches are not likely to get paid and AAA tranches will likely lose some principal.

Third, when the market is discussing how many pennies it will get on the dollar on an investment, AAA rating is obviously meaningless.”

I agree with both of these. Since I read the article BBerg actually toned down the article quite a bit, the first version had many more errors than the current one.

Feels like 1984! Rewriting the news to suit. Graph gone. Article gone. Try it. It’s been “Updated”.

First, AAA’s on home equity deals did not have 90% credit support. 15-20% credit support was more typical until late 2007.

Second, market prices show that AA tranches are not likely to get paid and AAA tranches will likely lose some principal.

Third, when the market is discussing how many pennies it will get on the dollar on an investment, AAA rating is obviously meaningless.

The market stopped pricing bonds to their credit rating many months ago. The implication that reratings will cause a repricing to true value is incorrect, this repricing has been taking place since the summer. As Paul points out 07-1 AAA is already priced at 52, cash bonds are going to be marked to the same level. The importance of ratings isn’t mentioned until the end of the article, its effect on capital requirements. This is a real problem as it will cause forced selling and more capital strain but the discussion here is very weak.

The reporter clearly doesn’t understand ratings methodology, neither its current overstated form or a hypothetical correct application. Investment grade criteria are not and should not be credit support = 2 * troubled collateral. A simple example is a AAA bond with 90% credit support and 50% “troubled collateral”. That bond fails the quoted test but in reality it is as AAA as can be. To take a loss 100% of the collateral will need to default with a loss severity of more than 90%. Loss severities in the subprime wasteland are only running at 50%. That AAA bond is pretty attractive since it gets paid down with 10% prepays or 20% defaults, recoveries are applied just like principal payments to the top of the capital structure.

As for Paul’s question haircuts will probably be similiar to the discount window. 8% for AAA 10 yr duration private label mortgages