Financial Times, Robert Pickel, chief executive of the International Swaps and Derivatives Association
In recent weeks, otherwise sophisticated commentators have made some basic yet potentially damaging errors about the amount of risk exposure held via credit derivatives. These assumptions, including those made by William Gross, Pimco’s chief executive, significantly overstate net settlement flows that would result on default of “underlying” entities. There is a danger this might create a climate of fear in financial markets in which companies, including Pimco, operate and which are already dealing with severe market volatility.
Critically, these errors ignore conventional market use of economically offsetting positions, which reduces the amounts at stake sharply. We seek here to illustrate some more down-to-earth assumptions; not to forecast the future. This is how the numbers really work:
First, the $50,000bn “notional” or nominal amount is just that; a nominal figure that references the “underlying” bonds and loans being protected by use of credit derivatives.
Focus on the net exposure of these transactions, many of which hedge or offset one another. A recent Fitch Ratings survey estimates net exposure at less than $1,000bn.
Factor in a probability of default of 2 per cent and a 25 per cent recovery rate and protection sellers would have to settle an aggregate $15bn of losses.
None of these amounts would be “lost” to the system; a credit derivative simply transfers a potential gain/loss from one party to another. Clearly, while $15bn is not trivial, it is a small fraction of aggregate write-offs to date on loans and securities; and less than a 10th of Mr Gross’s suggestion.
And our figures are conservative: we use a slightly higher ($50,000bn) figure for the total reference amounts; we round up the net exposure figure; we use a higher default rate than the 1.25 per cent used elsewhere; and our projected recovery rate is much lower than the 50 per cent used by Mr Gross.
Also, defaults in a downturn are usually considered to eliminate a much higher proportion of lower-rated companies, whereas surveys show that only about one-third of credit derivatives are written on sub-investment grade credit. So the impact of default would not necessarily be the same for credit derivatives as for credit markets as a whole. And such trading losses would be unlikely to hit simultaneously.
Even if default rates prove higher than those used here, settlement flows in credit derivatives would likely be much lower than in Mr Gross’s assumptions. This is not just theory. Settlement has occurred on this basis for several “credit events” where participants have organised net settlement through ISDA, with wide-ranging support, including from major bond funds.
Strip out the conservatism we build in here and those overstatements look extreme. And, while variations in each of these factors will have some impact on the final numbers, these are clearly most sensitive to the distinction between the aggregate reference amount of $50,000bn and that for net risk transfer of $1,000bn.
Meanwhile, industry and regulators remain focused on refinements to the critical discipline of counterparty credit risk management. Netting and collateral are important mitigants here, reducing so-called “presettlement” market exposures substantially.
Similarly, it is worth bearing in mind that not all defaults occur suddenly. A market participant that has written net protection will probably have an opportunity to manage its position in response to what is usually a gradual decline in creditworthiness by the reference entity. While this does not alter the net amount of protection written, it clearly reduces the financial impact on that individual participant of the entity’s default.
It is appropriate that a debate occur and that lessons be learnt from the way credit markets function. But we should recognise that privately negotiated credit derivatives markets have not only played a key role in allowing prudent and dynamic hedging of credit risk but have stayed open for business in spite of reduced liquidity in securities and interbank deposit markets.
The main point, however, remains a more basic one. Gross numbers are no basis on which to estimate the impact of the market in credit derivatives; net exposure is the place to start. [Source]
In recent weeks, otherwise sophisticated commentators have made some basic yet potentially damaging errors about the amount of risk exposure held via credit derivatives. These assumptions, including those made by William Gross, Pimco’s chief executive, significantly overstate net settlement flows that would result on default of “underlying” entities. There is a danger this might create a climate of fear in financial markets in which companies, including Pimco, operate and which are already dealing with severe market volatility.
Critically, these errors ignore conventional market use of economically offsetting positions, which reduces the amounts at stake sharply. We seek here to illustrate some more down-to-earth assumptions; not to forecast the future. This is how the numbers really work:
First, the $50,000bn “notional” or nominal amount is just that; a nominal figure that references the “underlying” bonds and loans being protected by use of credit derivatives.
Focus on the net exposure of these transactions, many of which hedge or offset one another. A recent Fitch Ratings survey estimates net exposure at less than $1,000bn.
Factor in a probability of default of 2 per cent and a 25 per cent recovery rate and protection sellers would have to settle an aggregate $15bn of losses.
None of these amounts would be “lost” to the system; a credit derivative simply transfers a potential gain/loss from one party to another. Clearly, while $15bn is not trivial, it is a small fraction of aggregate write-offs to date on loans and securities; and less than a 10th of Mr Gross’s suggestion.
And our figures are conservative: we use a slightly higher ($50,000bn) figure for the total reference amounts; we round up the net exposure figure; we use a higher default rate than the 1.25 per cent used elsewhere; and our projected recovery rate is much lower than the 50 per cent used by Mr Gross.
Also, defaults in a downturn are usually considered to eliminate a much higher proportion of lower-rated companies, whereas surveys show that only about one-third of credit derivatives are written on sub-investment grade credit. So the impact of default would not necessarily be the same for credit derivatives as for credit markets as a whole. And such trading losses would be unlikely to hit simultaneously.
Even if default rates prove higher than those used here, settlement flows in credit derivatives would likely be much lower than in Mr Gross’s assumptions. This is not just theory. Settlement has occurred on this basis for several “credit events” where participants have organised net settlement through ISDA, with wide-ranging support, including from major bond funds.
Strip out the conservatism we build in here and those overstatements look extreme. And, while variations in each of these factors will have some impact on the final numbers, these are clearly most sensitive to the distinction between the aggregate reference amount of $50,000bn and that for net risk transfer of $1,000bn.
Meanwhile, industry and regulators remain focused on refinements to the critical discipline of counterparty credit risk management. Netting and collateral are important mitigants here, reducing so-called “presettlement” market exposures substantially.
Similarly, it is worth bearing in mind that not all defaults occur suddenly. A market participant that has written net protection will probably have an opportunity to manage its position in response to what is usually a gradual decline in creditworthiness by the reference entity. While this does not alter the net amount of protection written, it clearly reduces the financial impact on that individual participant of the entity’s default.
It is appropriate that a debate occur and that lessons be learnt from the way credit markets function. But we should recognise that privately negotiated credit derivatives markets have not only played a key role in allowing prudent and dynamic hedging of credit risk but have stayed open for business in spite of reduced liquidity in securities and interbank deposit markets.
The main point, however, remains a more basic one. Gross numbers are no basis on which to estimate the impact of the market in credit derivatives; net exposure is the place to start. [Source]
Mahalanobis - am 2008-01-29 16:22
cb (guest) meinte am 30. Jan, 18:18:
I agree with you, but note that you are not including counterparty risk. If the primaries are running around the globe looking for capital in response to losing a few billion on mortgages, I wouldn't want to be holding the bag if a few clients of a primary couldn't honor their CDS commitments.
stxx meinte am 1. Feb, 04:00:
Why do single banks write down 10s of billions while global net expected loss is 15bn? I have no opinion on where the flaw in this logic is but these could be some indications: - There is no clear distinction between CDS, CDOs, and traded tranches.
- Cash CDOs have no swap behind them and cannot be easliy offset. The net exposure is the CDO notional.
- The 50tn figure is based on CDS with ISDA docs which covers mainly investment grade and speculative grade corporates and sovereigns. Subprime or any other exposure would then not be counted.
- Hedging is an abused term in the credit world where it often does not mean "hedging" but something more that "we offset our BBB exposure in XYZ with a BBB exposure in ABC".
- A five year cumulative default probability of 2% can only be assumed for investment grade corporates according to publicly available cohort analysis.
