Veryan Allen has another rant up at his Hedge Fund blog, reprising his refrain that anybody who loses most of their investors' money is likely not to be a hedge fund at all, 2-and-20 pay rates notwithstanding. Most of the rant is annoyingly unspecific, naming no names and not even hinting at the identities of the few good fund managers Allen claims to admire. But every so often he drifts out of his normal hand-waving mode and gets specific, as in this riff on triple-A ratings:

It was absurd to assign a measure originally designed for rock solid government and corporate debt to the untested (till now!) financial alchemy of CDOs, CLOs and CPDOs. It is applying a fundamental metric to model-based credit structuring using wildly optimistic assumptions of default and recovery rates and correlations of different borrowers. Collateral is "sound" only if someone else will buy it at prices you "assume". If product structurers want to rate shop for a sellable classification that is their freedom but investors should ignore them. The only things "investment grade" are those assets whose rewards outweigh the risks. How shortsighted to gain a few hundred basis points for a while but end up losing 100%.

There's some truth here, but also a lot of disingenuousness. Two rhetorical tricks are worth calling out in particular.

Firstly, it's true that CDOs, CLOs and CPDOs, in their present form, are untested – that's because all three are relatively new products, and any new product is, by definition, untested. On the other hand, it's absurd to say that all new products are, by definition, unsafe. And Allen's implication is actually broader still: that asset-backed securities in general are untested, and that triple-A credit ratings should be assigned only to rock-solid unsecured debt.

In truth, there's no reason to believe that asset-backed triple-A securities are more likely to default than unsecured triple-A securities. Given a long enough time horizon, all assets tend to zero, and history is littered with stories of cities, countries and companies which once had great wealth and power but which collapsed with astonishing rapidity. At the same time, asset-backed bonds in general have a long history which can compare reasonably favorably to unsecured bonds. Both have suffered defaults, of course, but the default rate on asset-backeds is not appreciably higher than the default rate on similarly-rated unsecured debt. (Is it true, at least, that asset-backed bonds are more likely to default during a general economic downturn than unsecured bonds are? I don't know. Maybe.)

And to check on whether the ratings agencies were "wildly optimistic," you have to look at their work in toto. Obviously, in hindsight, any instrument which defaults is likely to have suffered from overly optimistic assumptions. But there will always be instruments which default, so this kind of ex post criticism is generally unhelpful. What would be much more useful would be if Allen could point us to other, non mortgage-backed securities which he thinks are based on wildly optimistic assumptions: that sort of ex ante analysis can save people a lot of money and pain.

Secondly, "investment grade" does not mean "worth investing in" – as Allen knows full well. It's possible to make a lot of money investing in non-investment-grade ("junk") debt, especially if you pick credits which are likely to get upgraded to investment-grade status in the coming years. Similarly, a credit which starts off with a AA rating and slowly gets downgraded all the way to BBB is going to lose a lot of value, even though it's investment-grade the whole time.

And in any event, almost no one has ever ended up "losing 100%" of their money by investing in investment-grade debt, unless they employed leverage or they invested in the unsecured debt of a major fraud such as Enron. Even WorldCom bondholders made out pretty well in the end, considering.

The ratings agencies are not – and nor do they purport to be – a service telling investors which securities are going to turn out to be profitable investments. They rate only the probability of a default, not the probability of a drop in price. In 1998, credit spreads gapped out sharply across the board, and a lot of investors lost a lot of money. That's market risk, and it's not something the ratings agencies can or should be concerned about. In fact, over the past decades, it's very hard to find areas where the ratings agencies have been spectacularly wrong.

Eventually, of course, one such area will emerge, and it looks very much as though subprime is that area. I do think that the ratings agencies did get subprime wrong, although we're still a long way from finding out just how high up the ratings scale there are likely to be defaults. But I don't think that it's worth dismissing everything that the agencies do as worthless just on the basis of a bad outcome in subprime.