Deutsche Bank

We generally agree that the scope of the discussion paper matches the Consultation request. We would however urge that, although unrealised losses are not covered by the EBA’s mandate pur-suant to Article 80 (4) of the CRR, they should nevertheless be considered in the EBA’s technical advice. In particular, we request that the effect of hedging activities be considered in conjunction with primary positions, and only the net unrealized gain or loss be considered for prudential ad-justment. As portfolio hedged positions are not subject to the uncertainty concerns expressed in the Consultation, we believe this is logical. We believe it is inappropriate to analyse the prudential and economic impact of filters on unrealised gains without taking into account the treatment of corresponding unrealised losses.

Applying a prudential filter on gains contradicts the Basel 3 provisions as currently implemented in Art. 35 of the CRR which eliminate prudential filters for unrealised gains and losses. Even if the alternative to apply no filter, is not covered by Article 80 (4) of the CRR, it should nevertheless be considered in the analysis of policy options. This would ensure a consistent overall approach and take into account the issue of creating a level playing field.
We generally agree that the criteria raised need to be explored. However, we feel that some of the risks discussed are often diminished through prudent risk management, and that the magnitudue of unrealized gains does not necessarily offer any insight as to the durability of the gain.

For capital purposes, positions are subject to the standards for Prudent Valuation, as expressed in Articles 34 and 105 of the CRR. Through this process, valuation adjustments are made to ad-dress many of the concerns raised in this Consultation, including adjustments for:

• Market price uncertainty;
• Close out costs;
• Unearned Credit spreads;
• Model risk;
• Position concentration;
• Funding costs;
• Administrative costs;
• Early termination loss; and
• Operational risks.

Prudent valuation adjustments for these issues supplement accounting reserves already held, to ensure that the net value reflected for capital purposes represents the true position exit cost at a 90% degree of certainty, given normal market conditions, and that CT1 is adjusted to reflect this cost. Capital is held to cover market risk and credit risk, both in normal and stressed periods

Great care must be taken to avoid double counting these risks within the current Consultation; we feel the large majority of the concerns expressed are actually already prudently reflected in capital requirements, and thus need no further adjustment.

Regarding the criterion provided in the discussion paper “For banking book items, the risk that the unrealised gains will disappear is not covered by a capital requirement”:

We note also that for banking book positions market risk capital requirements for FX and com-modities risk exist, i.e. the risk that unrealised gains may not materialise is already covered by a Pillar 1 capital requirement. In addition, all other market risks (e.g. interest rate or credit spread risk) are covered in Pillar 2. Thus, these risks are already measured and do attract regulatory capital. We believe it appropriate, therefore for banking book positions to take unrealised gains and losses into account – without applying a prudential filter – when determining regulatory capi-tal.

We are concerned by the distorted overall picture that would be the result of an asymmetrical filter for unrealised gains only, while unrealised losses continue to reduce regulatory capital. Such an asymmetrical treatment overstates a bank’s risk positions and leads to a definition of own funds that no longer reflects the bank’s true economic situation. This would especially and disproportionately impact banks subject to IFRS accounting with a large portfolio of fair valued assets and liabilities. It should be taken into account that even under a prudent approach, the solvency requirements are not intended to reflect a worst case scenario in contrast to the large exposure regime.
As stated in our response to Question 2 and the cover letter, we oppose the insertion of an asymmetric filter, and are generally opposed to the use of any prudential filters, consistent with the Basel 3 guidance.

However, should the EBA decide there is a need to distinguish trading book and banking book positions, we would support this dichotomy for the purpose of policy distinction. If this is the case, we would urge the trading book to be exempted from prudential adjustment.. This approach is in line with solvency regulation and ensures that the different capital requirements that apply for the trading- and banking book (which do not depend on the respective accounting classification) are adequately taken into account.
Yes, while the regulatory and accounting treatments are often the same, the classification rules are not identical and therefore, under certain circumstances, instruments may require different treatment between accounting and regulatory classifications. DB does have regulatory banking book positions that are held at fair value through profit and loss, either as trading assets or finan-cial instruments elected under the Fair Value Option. Additionally, DB does have available for sale instruments in the regulatory trading book. Items elected under the Fair Value Option for IFRS accounting purposes are the only instruments in which DB has a one-time election at incep-tion to account for the instrument at fair value through profit or loss instead of at amortised cost. All other differences between accounting and regulatory classification are dictated by the rules.
We believe there is an important interaction between prudential regulatory requirements and ac-counting standards that must be factored into the regulators’ consideration before making any final decision on this topic. The current exposure drafts and IASB discussions lead us to expect that, IFRS 9 will narrow the definition of financial instruments that can be held at amortized cost, and therefore, increase in the classification of financial instruments at fair value through OCI (“FVOCI”) and fair value through profit and loss (“FVPL”). Additionally, under IFRS 9, many instruments that are held for the purpose of collecting cash flows may be classified at FVOCI or FVPL because of terms that result in cash flows that are not solely payments of principle and interest. It is possible that this rule could increase the number of instruments classified as banking book for regulatory purposes, but held at fair value through profit and loss or FVOCI for accounting purposes.

Were there to be asymmetric capital filter for unrealised gains on instruments in the FVOCI cate-gory, this may potentially lead to behaviour where entities would classify such instruments in a FVPL category (which they may not have otherwise done absent the asymmetric filter), so that they can achieve a symmetrical capital treatment for the unrealised gains and losses whilst at the same time being relieved from an operational perspective of hedge accounting requirements un-der IFRS. Such outcomes would be inconsistent with their true business model for their instru-ments and as such not provide relevant financial reporting.

In any case, due to the interdependency between IFRS 9 that increases the scope of fair value classification and solvency regulation we request that the EBA wait for the final issuance of IFRS 9 and conduct an impact assessment before any decision regarding an asymmetric pruden-tial filter is taken.
We do not agree with an approach that distinguishes between different categories of instru-ments/items. The determining factor should be how the instruments are risk managed, rather than the form of the instrument.

Risk management for trading book position is in general performed on portfolio level, i.e. the trading book. There are circumstances where a position that qualifies as equity /own funds for regulatory purposes (e.g. a Tier 2 or Additional Tier 1 instrument) is used to hedge debt positions of a counterparty. Hence a split between interest and non-interest bearing instruments is not appropriate.

In our view a prudential filter for unrealised gains is not required for the trading and banking book. If a prudential filter is imposed, then it needs to be applied on portfolio level, i.e. the trading and banking book respectively.

For more detail see the answers to Questions 7 and 8 below.

NB - On p. 18, the discussion paper uses the term “equities”. It is not clear what exactly is meant by that (i.e. IFRS equity, equity investments in the sense of the RWA framework, own funds instruments etc.?).
We generally agree with the arguments supporting the portfolio basis. The application of a filter on an item-by-item basis leads to a distorted picture of the bank’s true economic situation.

Our financial instruments are managed on a portfolio basis and the regulatory treatment should be consistent with risk management. If a prudential filter is imposed, then it needs to be applied on portfolio level, i.e. the trading and banking book respectively, see also our answer to Question 6.
An item-by-item approach is likely to provide misplaced incentives (e.g. to sell and buy back, instead of encouraging diversification) and moreover leads to a distorted picture of the bank’s true economic situation because unrealised gains within a given portfolio (under the current proposal: within a certain category of financial instruments) are usually available to offset the corresponding unrealised losses. A portfolio approach would be in line with the current application of regulatory filters in most countries (including Germany) and the portfolio approach as permitted under IFRS 13. This also illustrates that the better arguments are the ones in favour of a portfolio approach.

We, therefore, strongly support a portfolio approach that distinguishes only between trading and banking positions.
No, the determining factor for the use of a portfolio basis should be how the instruments are risk managed, rather than the form of the instrument. The unrealized gains and losses calculated on a group of assets and liabilities should include all types of instruments risk managed on a portfolio basis. For example, unrealized gains and losses on a loan due to interest rate risk could be hedged with a derivative, or a debt security, or any other instrument with interest rate risk. The form of the instrument is not relevant in the risk management of the risk creating the unrealized gains/losses. We would reiterate our view, as expressed in response to Question 2 that many adjustments are already reflected in capital requirements through market risk, /credit risk and pru-dent valuation charges, and policy options should avoid double counting the same risks.
We envisage the impact to be material.
We do not agree with the alternatives presented. We believe that unrealised gains should be included without adjustment in Common Equity Tier 1 (i.e. the treatment envisaged by Basel 3, as currently implemented in Art. 35 of the CRR), inter alia, to ensure a level playing field.

Of the options presented in the discussion paper, we would be in favour of a partial inclusion in Common Equity Tier 1, subject to a haircut only, but not subject to an additional threshold. Inclu-sion in Common Equity Tier 1 would at least ensure a more level playing field between EU and not EU banks than would be the case if unrealised gains were completely derecognised in own funds (while outside the EU they would be fully included as a Common Equity Tier 1 item). Moreover, partial recognition in Common Equity Tier 1 would be the consistent approach since unrealised losses are also deducted from Common Equity Tier 1.

It is true that a bank’s balance sheet positions are subject to constant fluctuations in value. How-ever, on the whole, most of these fluctuations cancel each other out, largely because positions hedge one another. A filtering out of all gains would ignore this equalising effect. And banks’ risks would be substantially overstated, triggering considerable and adverse procyclical effects.

We believe that no prudent adjustment should be considered. If, nonetheless, it is decided to apply a filter for unrealised gains, a haircut would be sufficient to address the prudential concerns regarding unrealised gains: it is not a realistic scenario to assume the unavailability of all unreal-ised gains to cover losses. An additional hard limit for the recognition of unrealised gains in own funds is not necessary. A hard limit for the recognition of unrealised gains in own funds, whether additionally or exclusively applied, would in any case disproportionately impact banks subject to IFRS accounting with a large portfolio of fair valued assets and liabilities. Moreover, it would ignore the fact that unrealised gains are usually available to offset the corresponding unrealised losses and should therefore qualify as own funds to a corresponding extent, possibly subject to a haircut, as unrealised losses decrease own funds irrespective of the overall amount of all unrealised gains/losses.
Deutsche Bank has no comment with regard to this question.
No, we do not agree with a threshold whether additionally or exclusively applied.
Financial instruments (debt and equity positions) are managed on a portfolio basis and if a filter is applied at all, then such a filter needs to be applied on a portfolio basis. Hence we consider a split between equity and debt instruments to be inappropriate.
We agree with the analysis of hedge accounting. We note, however, that there appears to be an asymmetry in treatment with the existing economic hedges where hedge accounting is not ap-plied. This may be due to the fact that the requirements for hedge accounting under IFRS are very specific. An economic hedge that is not eligible for hedge accounting is
nevertheless effective for purposes of risk management and should therefore be treated similarly for regulatory purposes. Also Hedge accounting may not always fully reflect the economic position of the entity.

Hedge ineffectiveness is very small due the way hedges are designated hedge accounting. The same applies to economic hedges that qualify as an effective hedge under the internal control processes of the institution, as assessed by the competent authorities. Moreover, please note that other provisions of the CRR in the area of own funds already take into account effective economic hedges (e.g. Art. 76 of the CRR). Accordingly it would therefore make sense to recognise economic hedges also in the area of potential filters for unrealised gains in-line with the proposed treatment for hedge accounting. This is amplified when the use of a prudential classification is proposed instead of a distinction based on accounting.
The level of ineffectiveness recognised may increase due to changes proposed in the forthcoming hedge accounting requirements, instead of failing the hedge accounting if the hedge effectiveness is outside of the 80—125% range, any ineffectiveness will be recognised in the P&L without failing hedge accounting.
We understand the logic of the proposal not to apply the filter to unrealised gains and losses where the fair value option election is made to eliminate or reduce accounting mismatch. Eco-nomically an entity may be in a similar position where the FVO election is done on “managed on FV basis” – as that usually means economic hedges and managing open risk to reduce such open risk, however no “credit” is given to these situations. In many situations the FVO election can be made on the basis of either accounting mismatch or managed on a FV basis, so it seems inconsistent that one type of FVO election is treated differently from the other. We strongly oppose any asymmetric filter to be applied both either fair value option or trading instruments under IFRS.
Deutsche Bank has no comment on this question.
We strongly agree with the EBA’s assessment indicated in paragraph 100 on page 28 of the dis-cussion paper that the introduction of a filter in the trading book is not appropriate. The EBA rightly points out why such a filter for trading book financial instruments is not required. The capital requirements for market risk already cover the volatility of the instruments. Moreover, trading book positions are held with the intention to sell and only short term. Any remaining concerns regarding trading book positions are better addressed by the RTS on Prudent Valuation and additional requirements in the Pillar 2 process. The concern that banks may reclassify financial instruments from the banking book to the trading book (paragraph 98 on p. 27) is unjustified since for a trading book classification, trading intent needs to be demonstrated.
We agree that there is a significant risk of double counting as the prudent valuation adjustments taken together with IFRS fair value reserves ensure that it is prudential to take unrealised gains and losses into account to determine regulatory capital. Hence in our view imposing a prudential filter on unrealised gains constitutes a double counting and should thus be avoided.
Please see our response to Question 2.
Please see our response to Question 19.
If at all, a prudential filter should be applied on portfolio and not item by item basis.
Deutsche Bank