If banks and their risk management models and methodologies have come into question after what has happened in the US credit market in recent months, so have the principles of banking supervision as argued by me a couple of weeks back.

Another pillar that has been shaken is the rating agencies. This is, of course, not the first time that rating agencies' judgements are being questioned: they had been criticised a decade back for their inability to foresee the balance of payment crises in south-east Asia.

As for the derivative instrument products, based on structuring and securitising mortgage loans to weak borrowers, which have resulted in major losses for the banking industry and a lack of confidence in credit markets, the involvement of rating agencies is much closer and deeper than as mere raters of plain vanilla instruments.

The reason is that, in structured product ratings, the role of the agencies extends to the structuring of the product itself. They are often as closely involved in the exercise as the originator in determining the criteria for the underlying pool; the amounts, risk ratings and coupons of the different tranches in collateralised debt obligations, in prescribing credit enhancement where needed, and so on.

Perhaps the role is more comparable to that of an investment banker advising a company on an IPO, as distinct from an external analyst of an issue. This greater involvement has two corollaries:

• More complex the structure, higher can be the fees of the rating company (would high fees affect judgement?);
• It also carries greater responsibilities.

As bond after bond in different CDO structures has been downgraded, rating companies are attracting increasing scrutiny from the authorities both in the US and EU. There does seem to be some merit in the criticism.

To quote two specific points, in one recent case, an instrument rated AAA in April 2007, was downgraded 14 notches at one go, just six months later. More generally, the historical default probabilities of identical ratings of different types of instruments, seem to differ widely.

In other words, the probability of default of a plain vanilla AAA bond is far lower than the AAA tranche of a CDO: clearly, while all AAAs are equal, some are more equal than the others! Surely there is a case for differentiating between the rating symbols of structured financial instruments from their plain vanilla counterparts.

The IMF's Global Financial Stability Report, September 2007, had this to say about the rating methodologies: "There is a need to examine the risk analysis of credit derivatives and structured products and the role of rating agencies. Ratings and ratings agencies will continue to be a fundamental component in the functioning of financial markets.

However, there is some concern about the rating methodology of complex products, particularly when securities, with very different structures, assumptions and liquidity characteristics, receive the same ratings. Ratings of complex structured products may have become too connected to facilitating origination. In periods of turbulence, the rapid downgrades then raise questions about the reliability of these ratings and their usefulness for the investors.

"We repeat the call from the April 2006 GFSR for a more differentiated scale of ratings for structured products. Investors also have an obligation and responsibility to understand the dynamics and liquidity risks associated with the products they buy they wrongly assumed that a low probability of default meant a low likelihood of losses from market movements. In the case of complex structured credit products, investors need to look behind the ratings they should not assume that the simple letter ratings provided by ratings agencies show equivalent risks as those for other asset classes. . .

"When purchasing complex products, investors will need to consider the associated liquidity aspects and include an appropriate liquidity risk 'premium' as part of the price. Financial institutions holding such securities as collateral will need to assign a 'haircut' that factors in liquidity characteristics."

But such issues apart, one point should not be lost sight of: most of the losses reported have arisen from mark-to-market valuations, rather than actual defaults in the portfolio.

The tyranny of market prices of risky products is that they are too influenced by market liquidity, as distinct from fundamentals, and liquidity is fickle and impossible to estimate or model.

Held to maturity, many of the now devalued and provisioned bonds may well be honoured: after all, the banks who invested in LTCM, the troubled hedge fund in 1998, ended up making profits on the positions which, on MTM basis, had landed LTCM in trouble.

Or, is this optimism the hangover of your columnist's long connection with the rating industry?