Algo Collateral Newsletter
Issue 7 - June 2002
Responding to the ISDA Margin Survey made easy
This newsletter will offer assistance specific to questions six, seven, eight, nine, ten and eighteen from the ISDA Margin Survey.
With the ISDA Margin Survey being a lengthy process, Algorithmics is offering assistance in the shape of various system queries to easily obtain the answers required for the survey. These queries will be provided for both Oracle and Sybase databases. Please note that the collateral manager may still need to define all reference and static data in a correct manner for results to be generated.
The ISDA survey covers the calendar year ending December 31, 2001. Clients are advised to run these same queries again at the end of 2002 to ensure accuracy in your results to next year's survey.
The ISDA Margin Survey will be discussed at our first Algo Collateral European users meeting and at our second Algo Collateral American users meeting in September.
This newsletter will offer assistance specific to questions six, seven, eight, nine, ten and eighteen from the ISDA Survey.
Question 6. Which OTC Derivatives products does your function support?
This question may be answered only if the client has chosen to load all collateralized and un-collateralized transactions into Algo Collateral (Sentry) during implementation. Do not leave all un-collateralized transactions in the missing bucket, because clients will be unable to identify new counterparty codes for collateralized clients. Some clients have assigned all transactions to collateral agreements, but have switched the status of the agreements that hold the un-collateralized transactions to a status other than active.
Products will appear as they appear in Tables/Products/Traded products.
Results will be displayed as in the following example:
| Product | Total number of trades where agrmt = active | Total number of trades where agrmt < or > active | Total number of trades | Total number of trades where agrmt = active/ Total number of trades |
| Cap | 1,100 | 400 | 1,500 | 73% |
| IRSW | 3,200 | 5,000 | 8,200 | 64% |
| FX | 200 | 400 | 600 | 33% |
| Repo | 250 | 700 | 950 | 26% |
The Collateral manager needs to interpret which products fall under the ISDA proposed classification. If uncollateralized transactions are not fed into the system, only the first and second column will be produced.
Algo Collateral (Sentry) does not currently take in potential future exposures and cannot calculate the percent of exposure figure.
Question 7. What other margined products does your function support?
Algorithmics is designing Algo Collateral (Sentry) to become a truly enterprise wide margining system. At the moment, however, Algo Collateral (Sentry) only supports OTC derivatives, overnight FX transactions, repo and securities lending trades. This question will also be addressed by the output of the query for question 6.
Question 8. How many OTC derivative collateral agreements does your firm manage?
Algo Collateral (Sentry) contains the following data elements: Agreement Type (set at the Master Agreement level), Agreement Status (set at the Collateral Agreement level; only where = Active), Posting Method (set at the Collateral Agreement level, Party Details), Governing Law (set at the Collateral Agreement level, Party Details) and Eligible Posting Party (set at the Collateral Agreement level, Party Details).
The script for this question provides the total number of active agreements by Agreement Type, and subsequently grouped according to Governing Law and Posting Method. The command "Eligible Posting Party" is used to divide that total number of agreements into the aforementioned categories.
- Unilateral In Your Favor = Agreements where Counterparty is the only Eligible Posting Party
- Unilateral Against You = Agreements where Principal is the only Eligible Posting Party
- Bilateral = Agreements where both the Principal and Counterparty are Eligible Posting Parties
This query does not distinguish agreements where one party can post collateral at a future date, based on a rating change for example. Rather, it is effectively unilateral as of the date of the survey. Such agreements are deemed to be bilateral for purposes of this query.
Results will be displayed as in the following example:
| Agmt Type | Governing Law | Posting Method | Unilateral Prin | Unilateral Cpty | Bilateral | Total |
| ISDA | UK | Transfer | X1 | Y1 | Z1 | X1+Y1+Z1 |
| ISDA | UK | Pledge | X2 | Y2 | Z2 | X2+Y2+Z2 |
| ISDA | US | Transfer | ||||
| ISDA | US | Pledge | ||||
| ISDA | Japan | Pledge | ||||
| DRV | German | Transfer | ||||
| AFB | France | Transfer | ||||
| Etc. | X(n) | Y(n) | Z(n) | X(n)+Y(n)+Z(n) |
Question 9. Total number of collateralized counterparties by type.
Algo Collateral (Sentry) contains the following data: Tables/Entities/Contacts. Legal Entity/General. Industry Group or Industry Code.
The drop-down table in both fields allows clients to use the ISDA classification. If clients have selected this field for every counterparty in their system, the output of query will look exactly as requested by the ISDA survey. Collateral manager will need to interpret the results in cases where output differs. Only counterparties where an active ISDA collateral agreement is in place will be counted.
Results will be displayed as in the following example:
| Industry code | Number of cpties |
| Bank/Broker | 35 |
| Hedge Fund | 2 |
| Industry group | Number of cpties |
| Bank/Broker | 35 |
| Hedge Fund | 2 |
Error trapping will list all entities for which no industry code or group has been selected.
Question 10. Total number of collateralized counterparties by country of incorporation.
Algo Collateral (Sentry) contains the following data: Tables/Entities/Contacts. Legal Entity/General. Domicile. Only counterparties where an active ISDA collateral agreement is in place will be counted.
Results will be displayed as in the following example:
| Domicile | Number of cpties |
| Belgium | 3 |
| Brazil | 2 |
| USA | 14 |
The collateral manager will need to interpret the results to meet ISDA's classification.
Error trapping should list all entities for which domicile has been selected. The results of question 8 should equal the results of question 9 and 10.
Question 18. Collateral holdings and collateral delivered.
To achieve US$ equivalents, the exchange rates of ISDA's Explanatory Notes to the Margin Survey are used. If a rate is not available in explanatory notes, the client must provide the rate or apply 1 as per explanatory notes. For clients that apply database maintenance on collateral balance for less than 180 days, may be unable to represent the positions as of December 31st 2001. The collateral manager will be able to amend the date for which to create the positions.
To address the first query, we assume a set up where Cash is defined as a product in Tables/Products. All eligible cash must be defined under Cash in Tables/Eligible Collaterals. In Positions/Collateral, these movements would therefore have Product = Cash. The Issue would typically be blank for these positions and the Currency field determines cash type.
Results related to cash collateral will be displayed as in the following example:
| Cash | Prin Pos | USD Equivalent | Cpty Pos | USD Equivalent |
| USD | Notional | Notional * 1.000000 | Notional | Notional * 1.000000 |
| EUR | Notional | Notional * 0.886000 | Notional | Notional * 0.886000 |
| GBP | Notional | Notional * 1.451500 | Notional | Notional * 1.451500 |
| JPY | Notional | Notional * 0.007627 | Notional | Notional * 0.007627 |
| Other 1 | Notional | Notional | Notional | Notional |
Clients may choose various set ups for non-cash collateral; Product, Description and Issue determines whether a collateral asset should be classified as a Government Security, a Government Agency Security, a Supranational and so on. The query will roll-up product, issuer and currency, if they are all the same. Description will be ignored. Apply the market rate of this bond on December 2001. If no rate was available on that day, assume 100. The market value column required is to be confirmed by ISDA.
Results related to non-cash collateral will be displayed as in the following example:
| Product | Description | Issuer | Ccy | Prin Pos | Notional | Market value | $ equiv. |
| BE Govt | Belgium Government Bond | BEF | |||||
| DE Equity | German Share Options | EUR | |||||
| DE Govt | German Government Two Year Notes | EUR | |||||
| FR Equity | French Options | EUR | |||||
| US Agency | Federal Home Loan Bank Debt | USD | |||||
| US Corporate | Asset Backed Securities | USD | |||||
| US Equity | Convertible Bonds | USD | |||||
| US Govt | US Treasury Bills/Notes/Bonds | USD |
The output of the query will require some client interpretation to complete the survey. Going forward, Algo Collateral User Meetings will discuss asset classification. The results will be grouped by currency. Client must move all non-government securities to bottom of output. Products where ccy = USD, ccy = EUR or GBP, ccy = JPY, ccy = other currencies should be grouped in that order. The query will incorporate FX rates for all currencies. All others will be listed as the same amount as notional. The collateral manager can still apply his/her own rate.
All positions where currency or product = NULL will be highlighted. Collateral managers should then update these positions in their database and then run queries again. It should be easy to identify which collateral agreement had data entered incorrectly. If a position is found in an old European legal currency, the following exchange rates will be applied:
| Euro 1 = | |||
| ATS | 13.7603 | GRD | 340.750 |
| BEF | 40.3399 | LUF | 40.3399 |
| FIM | 5.94573 | PTE | 200.482 |
| FRF | 6.55957 | ESP | 166.386 |
| DEM | 1.95583 | NLG | 2.20371 |
| IEP | .787564 | XEU | 1.00000 |
| ITL | 1936.27 | ||
A similar query will be produced to list collateral delivered.
A distinction between Algo Collateral (Sentry) Version 4 and pre- Algo Collateral (Sentry) Version 4 must be made to identify what can be re-used and what is legally blocked. The latter contains the following data elements: Tables/Agreements/Collateral agreement/Party details/Governing Law and Posting Method. If these fields are set to New York Law and Pledge respectively, then the collateral position associated with this collateral agreement will be listed as legally blocked. All other collateral can be re-used, including the cash balances.
Version 4 of Algo Collateral (Sentry) and beyond allows flags to be set to determine whether collateral is re-usable or not. The following data elements are made available for this purpose: Tables/Entities/Legal Entity/Rehypothecation; Tables/Agreements/ Master Agreement/Rehypothecation; Tables/Agreements/Collateral Agreement/Rehypothecation; Tables/Agreements/Collateral Agreement/Eligible. Prin Eligible/Is Rehyp. Upon removing a flag, any collateral associated with this entity or agreement will be listed as legally blocked.
Results related to re-usable collateral will be displayed as in the following example:
| Re-use | $ Equivalent |
| Legally Blocked | |
| Allowed |
The result of this and the previous query should be the same as the sum of the cash collateral and non-cash collateral holdings queries.
The re-usable collateral can then be grouped according to which collateral meets calls, is the sum of all cash collateral delivered and all re-hypothecated collateral, and is re-used by the repo desk. The collateral manager must have the repo desk set up as a ' collateral agreement' and would need to have 're-hypothecated' collateral to the repo collateral agreement every time a collateral piece was used by the repo desk. This query must be tailored to represent how this agreement has been called for each client. Finally, all collateral put in "custody" will be listed as the difference between total collateral holdings and collateral used to make calls and collateral re-used by the repo desk.
Results related to the re-use of collateral will be displayed as in the following example:
| Re-use | $ Equivalent |
| Used to meet calls | |
| Repo | |
| Put in custody |
As a control check the sum of this query should be equal to the Allowed part of the previous query.