Subscribe to DSC Newsletter

Extracting information from structured credit markets

Bank of England Working Paper by Joseph Noss


(I generally do not do this & not usually to BoE work)

They seek the Holy Grail….. Accurately value credit risk in the nightmare which is securitised credit derivatives which caused this mess in the first place. I have been looking at this whole topic again and the more I look into
synthetic CDO’s and the securitisation of the credit risk portion of structured securities with property and mortgages as base collateral the more ridiculous and stupid this entire financial engineering was. Utter madness which is why the central banks are full of the things that they have had to pay for with real money…… I understand the problem that the central banks have and clearly when they had to intervene they got dumped with the worst of the worst of the garbage that was sitting on the rescued entities balance sheets….


This type of structured security effectively severed the link between borrower and lender, then added in to the inherent PD the worst possible catalysts which is market risk volatility, fraudulent ratings and then to add to it they involved the whole CDS fiasco into it by hedging risk both in terms of selling off credit risk as a security and then splitting it between equity tranches and then subordinate tranches in layers over the equity tranche flowing down from the senior tranches. However; even the senior tranches are not secure as these have critical default levels at which cash flows divert at trigger points from seniors back into subordinate tranches so even senior tranches are not secure. Then we add in whether or not a CDS applies to the original securitisation…..or not; which again affects both valuation and PD wherever you are sitting in the subordinate tranches.

My own view is that the credit risk dimension is not possible to value accurately by standard quantititative methodology measures whether by Gaussian Copula, Poissons, fat tail cash flow analysis or any other methodology you wish to apply as the products are too individually dependant on too many variables from the valuation of the underlying asset, the PD rate within this, the nature of the original contracts with or without cash flow triggers between tranches, with or without CDS, the current default rate, the burn level within the tranches at this time, and the PD of future defaults, whether the rating is worth anything, the state of the property market and employment within the areas where the properties are located, the likely recovery value and costs in the event of repo which again varies by product and contract such cash surpluses if there are any returning to the investment vehicle….etc etc. Therefore to establish a valid credit risk model on a quantitative basis is in my view an unachievable Nirvana and this cannot be solved by generic methodology. I do not consider given this range of variables and the degree of fraud in these products that any credit risk model is valid as a catch all solution and to establish one is foolhardy as it will be inherently flawed for the reasons I state. These products are worth between 0 and their market value the accuracy of which is entirely input data dependant.

I do believe however that it is possible to achieve a best possible Market Risk valuation of these products but not a reliable Credit Risk valuation other than by a detailed contract analysis on a product by product basis. They are in any event a hybrid as I have been saying as they are a credit risk based security which has a value based upon credit risk, and more importantly are traded and therefore also have a market risk value based upon their market valuation.

To accurately value market risk to the best level possible standards given the clear obfuscation of credit risk via securitisation I would adopt the following approach:

  • Take market pricing valuation data from a best in class data vendor such as Lewtan who have done a detailed contract analysis by security and therefore have a better than market stab at PD, CDS cover, and regional housing value data and the cash flow position within the contracts. This is therefore an
    analysed valuation and not a guess, an M2M or formulaic stereotypical catch all which by its nature is flawed as it makes too many assumptions.
  • Then run VaR’s on market at least a 4 year historic, variance co variance, stressed. VaR results analysed by a market risk analyst.

I would then on a product by product basis consider that it would be possible to establish a reasonable MARKET risk adjusted market valuation of the security on an individual basis which would be a best possible valuation given these circumstances which would be as accurate as is possible.

As for standardised credit risk valuations on the securities as generics; I consider this a fool’s errand and utterly unachievable with any degree of accuracy at all given the large number of vastly differing variables and the degree of fraud within the original products from fraudulent lending, fraudulent ratings, and the inherent complexities within the products based upon fallacious assumptions and financial engineering gone mad. We therefore have what we have. The mother and father of economic global crises.

- David R Bristow, Siag Risk Management, Madrid

Views: 58

Tags: REH, asymptotix


You need to be a member of AnalyticBridge to add comments!

Join AnalyticBridge

On Data Science Central

© 2021   TechTarget, Inc.   Powered by

Badges  |  Report an Issue  |  Privacy Policy  |  Terms of Service