Response to consultation on changes to ITS on reporting
Go back
On the other hand, it leads to the modification of our current systems and methodologies to get a figure that has not been present in previous definitions and documents issued by the EBA.
As “Concentrated Position” AVA is concern, we do fell it will lead to some relevant “potential gains” based on the length of the exit strategy and the volatility of the underlying asset that may not be consistent with real exit prices.
Take as an example Block Prices for a big stock holding that may, in many cases, be sold using a corporative or private placement with a discount that, in most cases, will be lower than the VaR of the position on the average exit period.
Our AVA calculation will take into account this kind of floor, whereas the 10% confidence level approach based on the VaR of the position on the average exit period will lead to a positive result with no cap and based only on the Market Risk of the position
“Early Termination AVA” is another example of how this new metric is not easy to understand as a good tool to define the quality of the related AVA; as in most cases, clients asking for a non contractual Early Termination may support an implicit fee over the fair value.
As for “Operational Risk” & “Future Administrative Costs” some more concreted rules should be necessary to understand how to move the current definition to an “upside uncertainty”
Breaking the fair value figures to get to such identification, while possible, is not a common practice and will force us to implement new developments to segregate the information (based on external not internal trades) at portfolio level to produce the quarterly report requested.
The information needed, although already in the systems and used in risk and management reports is not ready to be used in AVAs calculation as the system developed did not include such requirement based on the existing RTS that allowed the calculation to be implemented using Company Level Net Exposures
As for Other Fair Value Adjustments (OFVA), we do have some doubts regarding the compliance of some of the AVA categories with IFRS13 (especially as Concentrated Positions OFVA is concerned)
We do see no problem to split adjustments on Model Risk as they are focused on models used for some assets classes and always on exotic products but found some difficulties already when Market Price Uncertainty (MPU) or Close Out Cost (COC) adjustments are to be split as we consider they must be calculated based on net positions and in many cases some trades or even portfolios net other portfolios exposures
This is especially true in the case of all derivatives position that, to some extent, suffer from Interest Rate MPU & COC AVAs based on the curves used for discounting.
Should the trade be denominated in a currency different to the one used to fund the trade, it may also suffer from MPU & CO AVAs derived from the Cross Currency Swaps Curves used to calibrate their discount curves.
What is more, having implemented a Core Approach calculation based on the valuation net exposure at company level, in many cases trades and portfolios net exposures between each other, making difficult the criteria to split such charges.
It is true that some AVA may have a clear portfolio to be assigned (Equity Correlation MPU being clearly 1.1.1.1.1.) but for others the assignment is much more complicated (Equity Vega MPU being caused by 1.1.1.1.1. & 1.1.1.1.2.).
Taking it to an extreme, as mentioned before, some AVAs are present in all portfolios (i.e. MPU on Cross Currency Swap basis present on all portfolios with trades denominated in a currency different of the one used to fund the trade)
Should we need to calculate all of then at a portfolio basis that would deeply increase the calculation effort making us redefine the actual procedures and system and will create unrealistic figures to be cancel using the “diversification” concept when they simply do not exit at Company Level.
Should a risk factor may not be clearly identified as part of an activity (we would really appreciate if EBA could provide which general factor do consider clearly assigned to a desk) we suggest splitting the AVAs based on the sensitivities to the risk factor to the different desk (should they be of the same sign) or to the one with the biggest sensitivity (in absolute value).
As mentioned in Q3 & Q4, Interest Rate MPU & CO is present in all derivatives (both vanilla and exotic). That is also the case of Cross Currency Swap Basis as they are used to calculate discount curves when the trade is denominated in a currency that is not the same used to finance it
In other cases, the product class is clear (may it be the case of Vega MPU & CO) but it is difficult to define whether it must be assigned to the Vanilla or Exotic desk; as both will show exposure to such risk …
We will really appreciate more clear instructions in order to be able to comply with the attachment of those risk to an specific portfolio without having to conduct the calculation at desk level (being it not the request as specifically mentioned in the annex) as it will mean a change in our system which implies more time and budget as it was neither initially planned nor executed based on the RTS definitions that did not include the need of such splitting.
Should it be needed, we will have to modify the inputs in the system calculating AVAs to identify which portfolios, based on their investment policy should be classified as Vanilla or Exotic.
Exposures should be aggregate (from the different portfolios in the different companies of the group) to get the net exposures with the granularity requested (using a criteria to assign Interest Rate and Cross Currency deriving from non-Rates activities to Vanilla Rates?) and then calculate the figures.
This will mean additional calculations in addition to the actual one that should be keep to get to the figure requested in row 10 and to obtain, by difference the figure to fill row 180
We understand that an adequate explanation on the products and models subject to Model Risk will go beyond an standard reporting and do think may be more properly address by further information to be send to the supervisory authority on request
We will appreciate more information on how to split out AVAs in assets categories and portfolios when a Net Exposure approach is used. May you confirm that no calculation should be conducted at portfolio level and AVA should be input to just one portfolio so no diversification figure must be reported?
We will appreciate more information on the calculation of Upside Uncertainty in the case of AVAs categories apart from Market Price Uncertainty, Close Out Costs & Model Risk.
We will appreciate more information on the figure we are expected to report as “overhedges”, as we do not understand the example given. We do consider the fair value of a digital trade to be the one including the cost of hedge. That’s the price used both in our P&L calculation and to manage collateral, as no one would trade at the price given by a pure Montecarlo with no “triangle” as such position is unhedgeable.
We will appreciate more information on the definition, on template C 32.04 (040 “Relevant Risk”). Is it supposed to be the exposure (Delta, Vega, …) we do consider creates the concentrated position risk being too big to be hedge or close in less than 10 days?
As Model risk is concern, the information requested seems to extend to cover a quarterly report, as implies a lot of methodological information impossible to explain in just 60 characters, being, in our opinion information that should be requested under the continuous supervisory work of the competent authorities
Question 1. The EBA believes that an understanding of where the accounting fair value sits within the notional range of plausible values at an aggregate level is essential context for assessing that the downside of this range, and therefore the AVA, is appropriately reported. Do you agree with this statement? If not please explain your reasoning? [Annex 2, page 1]
Whereas such information may be helpful to understand the range of possible fair values and, to some extent, determine if the actual fair value used by the institution is conservative or aggressive, we consider that -in many cases- it may give poor information as both price distribution and parameter’s ranges estimated using an expert approach may not be symmetric; making –to our understanding- difficult to figure out how to measure the goodness of the AVA figure based on this new calculationOn the other hand, it leads to the modification of our current systems and methodologies to get a figure that has not been present in previous definitions and documents issued by the EBA.
Question 2: Would the ‘upside uncertainty’ measure defined above and used in column 120 be suitable as a definition of the upside uncertainty? If not please provide reasons and any alternative suggestions for how such an upside measure could be defined. [Annex 2, page 1]
While such measure seem, somehow, direct in terms of meaning and implementation in the case of “Market Price Uncertainty”, “Close Out” and, to some extent “Model Risk”, “Unearned Credit Spreads” and “Investment & Funding Cost“ (although it leads to the modification of our current systems and methodologies to get a figure that has not been present in previous definitions and documents issued by the EBA), for other AVA concepts it seems harder to understand and to implementAs “Concentrated Position” AVA is concern, we do fell it will lead to some relevant “potential gains” based on the length of the exit strategy and the volatility of the underlying asset that may not be consistent with real exit prices.
Take as an example Block Prices for a big stock holding that may, in many cases, be sold using a corporative or private placement with a discount that, in most cases, will be lower than the VaR of the position on the average exit period.
Our AVA calculation will take into account this kind of floor, whereas the 10% confidence level approach based on the VaR of the position on the average exit period will lead to a positive result with no cap and based only on the Market Risk of the position
“Early Termination AVA” is another example of how this new metric is not easy to understand as a good tool to define the quality of the related AVA; as in most cases, clients asking for a non contractual Early Termination may support an implicit fee over the fair value.
As for “Operational Risk” & “Future Administrative Costs” some more concreted rules should be necessary to understand how to move the current definition to an “upside uncertainty”
Question 3: Is the above approach to splitting out fair valued assets and liabilities and fair-value adjustments on the one hand between the different types of AVAs and on the other hand between asset classes and product categories practical to implement? If not please describe the practical obstacles. Please suggest any alternative approaches (particularly if an alternative approach has been found useful for internal reporting purposes). [Annex 2, page 5]
We do not understand where the Fair Value Assets & Liabilities are to be split out between the different types of AVA on the templates provided.Breaking the fair value figures to get to such identification, while possible, is not a common practice and will force us to implement new developments to segregate the information (based on external not internal trades) at portfolio level to produce the quarterly report requested.
The information needed, although already in the systems and used in risk and management reports is not ready to be used in AVAs calculation as the system developed did not include such requirement based on the existing RTS that allowed the calculation to be implemented using Company Level Net Exposures
As for Other Fair Value Adjustments (OFVA), we do have some doubts regarding the compliance of some of the AVA categories with IFRS13 (especially as Concentrated Positions OFVA is concerned)
We do see no problem to split adjustments on Model Risk as they are focused on models used for some assets classes and always on exotic products but found some difficulties already when Market Price Uncertainty (MPU) or Close Out Cost (COC) adjustments are to be split as we consider they must be calculated based on net positions and in many cases some trades or even portfolios net other portfolios exposures
Question 4: Is the above portfolio-based approach to splitting out AVAs and other attributes between ‘Exotic’ and ‘Vanilla’ practical to implement? If not please describe the practical obstacles. Please suggest any alternative approaches (particularly if an alternative approach has been found useful for internal reporting purposes). [Annex 2, page 12]
In many cases most of the positions will be suffering from AVAs based on more than one category.This is especially true in the case of all derivatives position that, to some extent, suffer from Interest Rate MPU & COC AVAs based on the curves used for discounting.
Should the trade be denominated in a currency different to the one used to fund the trade, it may also suffer from MPU & CO AVAs derived from the Cross Currency Swaps Curves used to calibrate their discount curves.
What is more, having implemented a Core Approach calculation based on the valuation net exposure at company level, in many cases trades and portfolios net exposures between each other, making difficult the criteria to split such charges.
It is true that some AVA may have a clear portfolio to be assigned (Equity Correlation MPU being clearly 1.1.1.1.1.) but for others the assignment is much more complicated (Equity Vega MPU being caused by 1.1.1.1.1. & 1.1.1.1.2.).
Taking it to an extreme, as mentioned before, some AVAs are present in all portfolios (i.e. MPU on Cross Currency Swap basis present on all portfolios with trades denominated in a currency different of the one used to fund the trade)
Should we need to calculate all of then at a portfolio basis that would deeply increase the calculation effort making us redefine the actual procedures and system and will create unrealistic figures to be cancel using the “diversification” concept when they simply do not exit at Company Level.
Should a risk factor may not be clearly identified as part of an activity (we would really appreciate if EBA could provide which general factor do consider clearly assigned to a desk) we suggest splitting the AVAs based on the sensitivities to the risk factor to the different desk (should they be of the same sign) or to the one with the biggest sensitivity (in absolute value).
Question 5: Do you think such mismatches between the portfolio-level AVAs and the institution-level AVAs would be significant? Please give examples. [Annex 2, page 12]
Yes, as mentioned before, calculating at portfolio level would both increase the calculation effort and will create unrealistic figures to be canceled using the “diversification” concept when they simply do not exit at Company Level.Question 6: Where the difference is significant what additional practical difficulties would arise from calculating AVAs for each of the portfolio categories in rows 050-170? [Annex 2, page 13]
As we do calculate AVAs based on the Company Level Net Exposure we find not clear how to split some of the AVA (especially MPU and CO) as they are generated in several portfolios with exposures that in many cases net each otherAs mentioned in Q3 & Q4, Interest Rate MPU & CO is present in all derivatives (both vanilla and exotic). That is also the case of Cross Currency Swap Basis as they are used to calculate discount curves when the trade is denominated in a currency that is not the same used to finance it
In other cases, the product class is clear (may it be the case of Vega MPU & CO) but it is difficult to define whether it must be assigned to the Vanilla or Exotic desk; as both will show exposure to such risk …
We will really appreciate more clear instructions in order to be able to comply with the attachment of those risk to an specific portfolio without having to conduct the calculation at desk level (being it not the request as specifically mentioned in the annex) as it will mean a change in our system which implies more time and budget as it was neither initially planned nor executed based on the RTS definitions that did not include the need of such splitting.
Should it be needed, we will have to modify the inputs in the system calculating AVAs to identify which portfolios, based on their investment policy should be classified as Vanilla or Exotic.
Exposures should be aggregate (from the different portfolios in the different companies of the group) to get the net exposures with the granularity requested (using a criteria to assign Interest Rate and Cross Currency deriving from non-Rates activities to Vanilla Rates?) and then calculate the figures.
This will mean additional calculations in addition to the actual one that should be keep to get to the figure requested in row 10 and to obtain, by difference the figure to fill row 180
Question 7: What are stakeholders’ views on the ability to usefully summarise in a few key words the models and products concerned, as well as on the associated reporting burden or IT issues? [Annex 2, page 15]
We believe some problems may arise as of for today our reports seem to have really short fields.We understand that an adequate explanation on the products and models subject to Model Risk will go beyond an standard reporting and do think may be more properly address by further information to be send to the supervisory authority on request
Question 8: Do you find the proposed instructions on prudent valuation clear? Are there specific parts where definitions or instructions should be clarified?
We will appreciate more clear info on how to split out Future Administrative Costs and Early Termination AVAs between portfoliosWe will appreciate more information on how to split out AVAs in assets categories and portfolios when a Net Exposure approach is used. May you confirm that no calculation should be conducted at portfolio level and AVA should be input to just one portfolio so no diversification figure must be reported?
We will appreciate more information on the calculation of Upside Uncertainty in the case of AVAs categories apart from Market Price Uncertainty, Close Out Costs & Model Risk.
We will appreciate more information on the figure we are expected to report as “overhedges”, as we do not understand the example given. We do consider the fair value of a digital trade to be the one including the cost of hedge. That’s the price used both in our P&L calculation and to manage collateral, as no one would trade at the price given by a pure Montecarlo with no “triangle” as such position is unhedgeable.
We will appreciate more information on the definition, on template C 32.04 (040 “Relevant Risk”). Is it supposed to be the exposure (Delta, Vega, …) we do consider creates the concentrated position risk being too big to be hedge or close in less than 10 days?
Question 9: Do respondents have any comments on the structure and content of the proposed templates on prudent valuation?
Being an interesting figure to understand companies' business, we understand that the split of AVAs in portfolios and products implies a relevant burden in IT investment that was not budgeted when the AVAs projects were launched as they were not defined in the RTS published last year and just adopted by the European Commission.As Model risk is concern, the information requested seems to extend to cover a quarterly report, as implies a lot of methodological information impossible to explain in just 60 characters, being, in our opinion information that should be requested under the continuous supervisory work of the competent authorities