Response to consultation on draft RTS on residual risk add-on
Go back
However, different institutions may model future realised volatility in different ways, either as an underlying or as a payoff.
• If future realised volatility is directly modelled as an underlying, then it is not captured by SBM and DRC and therefore qualifies for RRAO as an “exotic underlying” as per MAR23.3 FAQ1 .
• If future realised volatility is modelled as a payoff (for example, a variance swap on the S&P500 index may be modelled as an exotic payoff referencing the underlying equity, i.e. the S&P500 index) then in this case the instrument will be fully within the scope of delta, vega, curvature and default risk capital (DRC) for equity and general interest rates risk classes. Therefore it will not qualify for RRAO as an exotic underlying. However, such a pay-off on the S&P500 “cannot be written or perfectly replicated as a finite linear combination of vanilla options”, hence as per MAR23.4(1) it still qualifies for RRAO as an “Instrument bearing other residual risks”.
The Industry commented on the topic of RRAO for products involving future realised volatility in the response to the European Commission consultation on implementation of CRR3 , and we reiterate those recommendations below;
1. Ensure that only truly exotic underlyings are subject to the 1% RRAO charge, e.g. exclude future realized volatility from this category and thereby let volatility and variance derivatives be subject to the 0.1% RRAO charge.
2. For interest rate yield curves options: a reduction of RRAO charges to 0.01%, defining a risk sensitive notional, or an allowance to recognize positions that materially hedge the price risk of the exposure subject to RRAO.
3. Provide clarity as to whether long and short positions with same underlying risk can be netted.
Please also refer to our response to Q8.
We therefore consider that Asian options should not be in the scope of a residual risk add-on under Annex point 1 path dependent underlying options.
If the EBA does wish to make Asian options subject to RRAO, it is important to ensure that options which while technically Asian are clearly vanilla do not attract RRAO (as required by CRR325u(2)(b)(i)). Key examples include options on OIS and futures options where the underlying future is cash-settled based on the average spot price observed during the delivery month.
One way of making this exclusion explicit would be to add a new Article 3(f) either:
• explicitly excluding all Asians:
Risk of a path-dependent option where the only source of path-dependence is that the payoff is calculated based on an average price observed over a period of time.
OR
• explicitly excluding only vanilla Asians:
Risk of a path-dependent option where the only source of path-dependence is that the payoff is calculated based on an average price, and some linear products referencing the same average are listed or are eligible to central clearing.
The inclusion of Bermudan options in scope of residual risk under point 6 of the Annex is questionable. Bermudan options are similar to American options, differing only in that the exercise dates are at specified intervals for Bermudan options but on a daily basis for American options. We therefore recommend that Bermudan vanilla options should be out of scope of the RRAO capital charge.
Finally, please also see our response to Q8.
This would be consistent with the treatment of “plain-vanilla options on index instruments that meet the conditions in Article 325i(3) of the CRR”, as the above-mentioned IR-spread options have the same vanilla characteristics as index options: their payoff is straightforward, and their underlying, the IR spread, is treated as a single bundled underlying, analogously to an index, based on simple and observable information. The traditional instruments in this category are constant-maturity swap (CMS) interest rate spread options.
Furthermore, without this clarification, a strict interpretation of the current requirement leads to the majority of the RRAO capital charge being generated by simple hedging transactions. Penalising straightforward hedging transactions is contradictory to the very purpose of prudential rules, and also introduces an inconsistency within Art. 325u(2), which states in point (ii) that hedges included in the alternative correlation trading portfolio (ATCP) are not subject to RRAO, without a homogenous treatment for the other portfolios included in point (i).
For further details on the use and liquidity of CMS spread options, please refer to our response to Q8.
CMS spread options are simple and liquid products
The Industry acknowledges that in general spread options are in scope of RRAO due to the correlation risk, but argues that CMS spread options should be considered plain vanilla instruments and therefore be exempt from the RRAO charge. Indeed, these products have simple payoffs based on the spread between two maturity points on the same underlying yield curve, which is observable and liquid (see example box below).
It is important to note the great utility that these products provide for real money clients, as they are heavily traded by insurance companies and pension funds to hedge their yield-curve risk, in particular in Europe.
The RRAO capital charge is unwarranted for simple CMS spread options
The RRAO requires a blanket charge based on the notional amount and leads to excessive charges for CMS spread products. The severity of the impact for CMS spread options is a consequence of derivative notional convention not taking account of the difference in DV01 (and hence, the quantity of risk) across different interest rate products.
An example illustrates this point. For swaptions, the quantity of risk in a trade scales with the [notional*dv01] of the underlying. Currently, that dv01 ranges from approximately 1bp for a 1y tenor swaption to 23bp for a 30y tail swaption. For a CMS product, the DV01 is fixed at 1bp in all cases. This leads to a wide discrepancy between the notional of a swaption trade and the notional of CMS trade for a given level of risk. Market participants would see approximately the same quantum of risk in a 100 million 10y tenor swaption as in a 1 billion 10y-2y CMS option.
Furthermore, the RRAO penalises simple hedging transactions that are not strictly “back-to-back”, and hence discourages hedging, which is contrary to the very purpose of prudential rules. This is very impactful since it is standard market practice to hedge the market risk of non-vanilla CMS products, which are duly subject to RRAO, with plain vanilla CMS spread options. In addition, several vanilla spread options can be needed to hedge the risk of a long-term non-vanilla CMS product, each increasing the RRAO charge based on their gross notional.
The Industry hence recommends that it should be clarified that simple CMS spread options are vanilla instruments, and therefore are out of scope of the RRAO. Such an exemption avoids vastly inflated RRAO charges on a hedged portfolio vs. the much lower RRAO charge attracted by the unhedged position, as well as the negative unwarranted impacts for end-users.
Example
The below and in the attached document provides an example of a EUR 100m notional 10Y-2Y cap hedged with the most liquid standard market instrument across different counterparties, corresponding to business practice, and illustrates the disproportional capital charge, along with the market liquidity of CMS products.
Example trade and RRAO Impact:
Sell: 20Y EUR100m Notional Multi-look 10Y-2Y Cap (Client)
Buy: 20 x 1Y EUR100m Single-look 10Y-2Y Cap (Hedge)
• Step 1: Client purchases EUR100m Notional 10Y-2Y Cap with a strike price of 1% from the Dealer for 20 years.
• Step 2: Dealer hedges the client trade with the most liquid standard market instrument by buying 20 x 100m Notional of single look 10Y-2Y Caps.
The term structure illustrated in the attached documents shows the 2 trades and the effectiveness of the hedge.
The total gross notional of this near-flat position (see above tables) is EUR2.1bn RRAO charge = EUR2.1m at trade inception.
Using a 10% annual cost of capital, over a 20y maturity, and allowing for hedge position roll-off, gives a lifetime cost of capital (LCoC) of (1.1m* x 10% x 20) = EUR2.2m, or 220bps on the original client trade accommodating for hedge rollovers, which renders this activity uneconomic.
* While the initial RRAO charge is 2.1 m, the average through the lifecycle charge in this example is 1.1 m.
Liquidity
The table in the attached document illustrates broker quote counts for selected IR products. (Data from Jun‘19 to Feb’20 by month.)
This demonstrates that the Market liquidity of CMS spread options is in line with other ‘vanilla’ products.
Industry Quantitative Survey Results
The survey conducted by banks provided data on CMS spread options and demonstrates that this is a key concern as it attracts 57% of the total RRAO capital charge (based on a median estimate).
In addition to this, data was provided for structured desks within banks which heavily trade these products and are even more severely affected.
The residual risk add-on is meant by definition to capitalise secondary risks not captured otherwise, while SBM and DRC should capitalise the primary risks. It is therefore expected that the residual risk add-on be less material than the SBM and DRC charges. However, for CMS spread options, the RRAO dominates their capital charge, and furthermore for non-flow interest rate desks represents 87% of their RRAO capital charge (based on median estimates). This highlights the very unwarranted application and calibration of the RRAO capital charge for CMS spread options, which are simple and liquid products, and hence the high extent to which the current RRAO rules threaten the economic viability of respective businesses. The graph in the attached document illustrates the excessive RRAO contribution to the total SA-FRTB capital requirement for non-flow interest rate desks, as well as a comparison to the current capital levels under Basel 2.5.
It can be seen that the RRAO increases the capital level for the CMS non-flow desk by about 870% of Basel 2.5 desk level capital (based on a median estimate). Although some exposure to RRAO is to be expected the size of the impact clearly illustrates issues with the calibration of RRAO for CMS options. For those non-flow desks the contribution of CMS spread options is 87% of the total RRAO and 63% of the total SA desk charge (based on a median estimates).
Investigating the issue on a broader scale shows that the impact is still visible in firm-level capital charges as visualized in the chart in the attached document.
The graph in the attached document provides a visual illustration of the impact to Market Risk capital from Basel 2.5 to FRTB SA for those banks who provided data within the survey at 2.5x the current Basel 2.5 capital level (based on a median estimate).
In addition, (using a median estimate) the graph illustrates the proportion of the FRTB SA capital associated with RRAO as 15% and furthermore the proportion of the RRAO charge associated with CMS Spread Options as noted above is significant at 57% of the RRAO and 9% of the overall total FRTB SA charge.
Q1: Do you think that any of the elements constituting the conditions in Article 325u(2)(a) require additional clarification? If yes, which elements should be clarified?
MAR23.3 states that “Instruments with an exotic underlying are trading book instruments with an underlying exposure that is not within the scope of delta, vega or curvature risk treatment in any risk class under the sensitivities-based method or default risk capital (DRC) requirements in the standardised approach.”However, different institutions may model future realised volatility in different ways, either as an underlying or as a payoff.
• If future realised volatility is directly modelled as an underlying, then it is not captured by SBM and DRC and therefore qualifies for RRAO as an “exotic underlying” as per MAR23.3 FAQ1 .
• If future realised volatility is modelled as a payoff (for example, a variance swap on the S&P500 index may be modelled as an exotic payoff referencing the underlying equity, i.e. the S&P500 index) then in this case the instrument will be fully within the scope of delta, vega, curvature and default risk capital (DRC) for equity and general interest rates risk classes. Therefore it will not qualify for RRAO as an exotic underlying. However, such a pay-off on the S&P500 “cannot be written or perfectly replicated as a finite linear combination of vanilla options”, hence as per MAR23.4(1) it still qualifies for RRAO as an “Instrument bearing other residual risks”.
The Industry commented on the topic of RRAO for products involving future realised volatility in the response to the European Commission consultation on implementation of CRR3 , and we reiterate those recommendations below;
1. Ensure that only truly exotic underlyings are subject to the 1% RRAO charge, e.g. exclude future realized volatility from this category and thereby let volatility and variance derivatives be subject to the 0.1% RRAO charge.
2. For interest rate yield curves options: a reduction of RRAO charges to 0.01%, defining a risk sensitive notional, or an allowance to recognize positions that materially hedge the price risk of the exposure subject to RRAO.
3. Provide clarity as to whether long and short positions with same underlying risk can be netted.
Q2: Do you think that the list of exotic underlyings should be extended beyond the ones mentioned in the CRR mandate (i.e. longevity risk, weather, natural disasters and future realised volatility)? If yes, which other exotic underlyings should be included?
The Industry believes the list should not be extended.Q3: Do you think that any of the elements constituting the conditions in Article 325u(2)(b) require additional clarification? If yes, which elements should be clarified?
Features of risk-free rates replacing IBOR rates as part of the benchmark reforms should not result in options on these rates being subject to RRAO. For example, if such a rate is fixed in arrears, then a cap or floor on that rate should not be subject to RRAO as a path-dependent option.Please also refer to our response to Q8.
Q4: Do you think that the terminology used in the non-exhaustive list of instruments bearing residual risks is clear? If not, please provide your views, including rationale and alternative terminology that it would be preferable to use.
Please refer to our response to Q8.Q5: Do you think that the non-exhaustive list of instruments bearing residual risks should be extended? If yes, which other instruments should be included?
The Industry believes that the non-exhaustive list should be enhanced using defined criteria. These criteria should identify that instruments which can be replicated by a bounded number of vanilla instruments should not be included. An example of this are digital options since they can be exactly replicated by a linear combination of a spot position and two vanilla options. A simple example is attached to this response to illustrate this.Q6: Do you think that the non-exhaustive list of instruments bearing residual risks should be reduced? If yes, which instruments should be excluded?
The Industry recommends removal of Asian options from the scope of RRAO residual risk under point 1 of the Annex. There are no additional risks related to Asian options. This is despite the slightly more complex pricing than under a simple Black-Scholes method since it depends on the full term structure of implied volatility. However, there are no risk management difficulties stemming from this difference.We therefore consider that Asian options should not be in the scope of a residual risk add-on under Annex point 1 path dependent underlying options.
If the EBA does wish to make Asian options subject to RRAO, it is important to ensure that options which while technically Asian are clearly vanilla do not attract RRAO (as required by CRR325u(2)(b)(i)). Key examples include options on OIS and futures options where the underlying future is cash-settled based on the average spot price observed during the delivery month.
One way of making this exclusion explicit would be to add a new Article 3(f) either:
• explicitly excluding all Asians:
Risk of a path-dependent option where the only source of path-dependence is that the payoff is calculated based on an average price observed over a period of time.
OR
• explicitly excluding only vanilla Asians:
Risk of a path-dependent option where the only source of path-dependence is that the payoff is calculated based on an average price, and some linear products referencing the same average are listed or are eligible to central clearing.
The inclusion of Bermudan options in scope of residual risk under point 6 of the Annex is questionable. Bermudan options are similar to American options, differing only in that the exercise dates are at specified intervals for Bermudan options but on a daily basis for American options. We therefore recommend that Bermudan vanilla options should be out of scope of the RRAO capital charge.
Finally, please also see our response to Q8.
Q7: Do you agree with the proposed approach for the treatment of correlation risk? If not, please provide your views, including rationale motivating your preference for an alternative treatment.
The Industry recommends that it should be clarified that interest rate (IR) spread options, where the spread is between two maturity points on the same underlying yield curve, are not “multi-underlying options” and should not be subject to the RRAO capital charge.This would be consistent with the treatment of “plain-vanilla options on index instruments that meet the conditions in Article 325i(3) of the CRR”, as the above-mentioned IR-spread options have the same vanilla characteristics as index options: their payoff is straightforward, and their underlying, the IR spread, is treated as a single bundled underlying, analogously to an index, based on simple and observable information. The traditional instruments in this category are constant-maturity swap (CMS) interest rate spread options.
Furthermore, without this clarification, a strict interpretation of the current requirement leads to the majority of the RRAO capital charge being generated by simple hedging transactions. Penalising straightforward hedging transactions is contradictory to the very purpose of prudential rules, and also introduces an inconsistency within Art. 325u(2), which states in point (ii) that hedges included in the alternative correlation trading portfolio (ATCP) are not subject to RRAO, without a homogenous treatment for the other portfolios included in point (i).
For further details on the use and liquidity of CMS spread options, please refer to our response to Q8.
Q8: Do you think that there are other products, not currently covered in these RTS (e.g. CMS derivatives), which are potential candidates for being covered in one of the parts of these RTS? Please provide your views, including rationale motivating the needs for such inclusions.
The Industry would like to recommend that the EBA clarify that simple CMS spread options (including caps, floors and swaptions) are vanilla products and hence out of scope of RRAO. At the very least, simple hedging CMS spread options should be exempt from the RRAO charge.CMS spread options are simple and liquid products
The Industry acknowledges that in general spread options are in scope of RRAO due to the correlation risk, but argues that CMS spread options should be considered plain vanilla instruments and therefore be exempt from the RRAO charge. Indeed, these products have simple payoffs based on the spread between two maturity points on the same underlying yield curve, which is observable and liquid (see example box below).
It is important to note the great utility that these products provide for real money clients, as they are heavily traded by insurance companies and pension funds to hedge their yield-curve risk, in particular in Europe.
The RRAO capital charge is unwarranted for simple CMS spread options
The RRAO requires a blanket charge based on the notional amount and leads to excessive charges for CMS spread products. The severity of the impact for CMS spread options is a consequence of derivative notional convention not taking account of the difference in DV01 (and hence, the quantity of risk) across different interest rate products.
An example illustrates this point. For swaptions, the quantity of risk in a trade scales with the [notional*dv01] of the underlying. Currently, that dv01 ranges from approximately 1bp for a 1y tenor swaption to 23bp for a 30y tail swaption. For a CMS product, the DV01 is fixed at 1bp in all cases. This leads to a wide discrepancy between the notional of a swaption trade and the notional of CMS trade for a given level of risk. Market participants would see approximately the same quantum of risk in a 100 million 10y tenor swaption as in a 1 billion 10y-2y CMS option.
Furthermore, the RRAO penalises simple hedging transactions that are not strictly “back-to-back”, and hence discourages hedging, which is contrary to the very purpose of prudential rules. This is very impactful since it is standard market practice to hedge the market risk of non-vanilla CMS products, which are duly subject to RRAO, with plain vanilla CMS spread options. In addition, several vanilla spread options can be needed to hedge the risk of a long-term non-vanilla CMS product, each increasing the RRAO charge based on their gross notional.
The Industry hence recommends that it should be clarified that simple CMS spread options are vanilla instruments, and therefore are out of scope of the RRAO. Such an exemption avoids vastly inflated RRAO charges on a hedged portfolio vs. the much lower RRAO charge attracted by the unhedged position, as well as the negative unwarranted impacts for end-users.
Example
The below and in the attached document provides an example of a EUR 100m notional 10Y-2Y cap hedged with the most liquid standard market instrument across different counterparties, corresponding to business practice, and illustrates the disproportional capital charge, along with the market liquidity of CMS products.
Example trade and RRAO Impact:
Sell: 20Y EUR100m Notional Multi-look 10Y-2Y Cap (Client)
Buy: 20 x 1Y EUR100m Single-look 10Y-2Y Cap (Hedge)
• Step 1: Client purchases EUR100m Notional 10Y-2Y Cap with a strike price of 1% from the Dealer for 20 years.
• Step 2: Dealer hedges the client trade with the most liquid standard market instrument by buying 20 x 100m Notional of single look 10Y-2Y Caps.
The term structure illustrated in the attached documents shows the 2 trades and the effectiveness of the hedge.
The total gross notional of this near-flat position (see above tables) is EUR2.1bn RRAO charge = EUR2.1m at trade inception.
Using a 10% annual cost of capital, over a 20y maturity, and allowing for hedge position roll-off, gives a lifetime cost of capital (LCoC) of (1.1m* x 10% x 20) = EUR2.2m, or 220bps on the original client trade accommodating for hedge rollovers, which renders this activity uneconomic.
* While the initial RRAO charge is 2.1 m, the average through the lifecycle charge in this example is 1.1 m.
Liquidity
The table in the attached document illustrates broker quote counts for selected IR products. (Data from Jun‘19 to Feb’20 by month.)
This demonstrates that the Market liquidity of CMS spread options is in line with other ‘vanilla’ products.
Industry Quantitative Survey Results
The survey conducted by banks provided data on CMS spread options and demonstrates that this is a key concern as it attracts 57% of the total RRAO capital charge (based on a median estimate).
In addition to this, data was provided for structured desks within banks which heavily trade these products and are even more severely affected.
The residual risk add-on is meant by definition to capitalise secondary risks not captured otherwise, while SBM and DRC should capitalise the primary risks. It is therefore expected that the residual risk add-on be less material than the SBM and DRC charges. However, for CMS spread options, the RRAO dominates their capital charge, and furthermore for non-flow interest rate desks represents 87% of their RRAO capital charge (based on median estimates). This highlights the very unwarranted application and calibration of the RRAO capital charge for CMS spread options, which are simple and liquid products, and hence the high extent to which the current RRAO rules threaten the economic viability of respective businesses. The graph in the attached document illustrates the excessive RRAO contribution to the total SA-FRTB capital requirement for non-flow interest rate desks, as well as a comparison to the current capital levels under Basel 2.5.
It can be seen that the RRAO increases the capital level for the CMS non-flow desk by about 870% of Basel 2.5 desk level capital (based on a median estimate). Although some exposure to RRAO is to be expected the size of the impact clearly illustrates issues with the calibration of RRAO for CMS options. For those non-flow desks the contribution of CMS spread options is 87% of the total RRAO and 63% of the total SA desk charge (based on a median estimates).
Investigating the issue on a broader scale shows that the impact is still visible in firm-level capital charges as visualized in the chart in the attached document.
The graph in the attached document provides a visual illustration of the impact to Market Risk capital from Basel 2.5 to FRTB SA for those banks who provided data within the survey at 2.5x the current Basel 2.5 capital level (based on a median estimate).
In addition, (using a median estimate) the graph illustrates the proportion of the FRTB SA capital associated with RRAO as 15% and furthermore the proportion of the RRAO charge associated with CMS Spread Options as noted above is significant at 57% of the RRAO and 9% of the overall total FRTB SA charge.
Upload files
EBA_RTS_RRAO_Final.pdf
(435.69 KB)