For prudential and regulatory purposes, banks’ assets/liabilities are generally grouped into the trading book or the banking book. The distinction between trading and banking book is reflected in the different treatment of positions subject to market risk and those positions that are not subject to market risk.

The basis for the application of the proposed filters is the IFRS accounting classification (HFT and AFS at fair value whereas L&R and HTM at amortized cost) which differs from the classification for prudential purposes.

Both alternatives proposed in the discussion paper
• apply the filter based on the prudential categories of instruments
• apply the filter based on the accounting categories of instruments
necessitate a reclassification from one categorisation to the other.

We do not see big differences between the regulatory trading book and the IFRS category held for trading except for the different definition of trading derivatives and hedging derivatives (IFRS) versus banking book derivatives. In IFRS all non-hedging derivatives are part of the held-for-trading portfolio per definition.

Differences also occur from the accounting treatment at fair value through profit and loss, which are expected to further increase the traditional “banking book” financial assets under the upcoming IFRS 9, as well as from financial assets under the fair value option for several reasons (avoidance of bifurcation of embedded derivatives).
While for these instruments under IFRS a measurement at fair value through profit and loss is applicable they do not fulfil the criteria of “trading book” according to the regulatory definition.

There is a tendency to consider the balance sheet approach as less burdensome as in any case a reconciliation from balance sheet capital positions the regulatory ones will be requested under the disclosure regime.

We understand the arguments for both prudential and accounting classification approaches.
Distinction between trading and banking book would be reasonable because there is a capital requirement for the market risk of the trading book instruments reflecting the potential downside-risk. From this point of view just banking book instruments which are not subject to similar capital requirements might be considered in the need for potential filters for unrealized gains.

On the other hand, the approach using accounting classification ensures consistent reconciliation to financial statements and therefore an easy understandable bridge from accounting figures to prudential filters within the disclosure requirements.
As we think that application of prudential filter on unrealized gains is anyway not necessary before introduction of IFRS 9 we would propose to this question once the IFRS 9 requirements are already in place.

Yes, at least one of our member banks has a portfolio of hedged structured own issues which is treated at fair value through profit and loss to avoid bifurcation of embedded derivatives and ease hedge-accounting. This portfolio is part of the regulatory banking book.

In the near future there might be changes in the classification and measurement of financial instruments from the replacement of IAS 39 by IFRS 9. Based on the current wording of the standard it is expected that 30% of the loan portfolio of one of our bigger member banks will have to be measured at fair value through profit and loss, which means that the gap between a regulatory trading book and the accounting category “at fair value through profit and loss” will definitely widen under IFRS 9.

We agree on this proposal. We support the distinguishing between different categories when it comes to the question, how unrealised gains and losses should be treated in the own funds calculation, because:

• different instruments in different portfolios may be managed following different strategies. In a short term strategy the realisation of the currently unrealised gains may have to be judged differently as opposed to a longer term “holding” strategy.
• The possibility to actually realise unrealised gains is a question of market liquidity.
• Different instruments have different levels of volatility.

In principal the arguments are valid. Nevertheless we regard this approach as too prudent and prefer a portfolio approach. The potential downsides of an item by item approach are enormous and will lead to more volatility due to more realizations of unrealized gains.

A portfolio approach is consistent with the way banks manage their financial instruments (risk mitigation benefits achieved through diversification) and leads to lower volatility in own funds. Filtering out individual instruments could be in contrast to the overall portfolio management strategy.

If single instruments were filtered out, banks will be forced to realise unrealised gains on short term which could give rise to unintended effects in the financial markets and might even tighten a crisis situation.

If for example an institution sells an item at year end in order to avoid a regulatory capital impact due to the unrealized gains requirements and buys back the same instrument at the beginning of the following year the unrealized gains are realized and the regulatory capital contains 100% of the realized gain. However the risk combined with this instrument has not changed. Therefore an item by item approach would on the one hand lead to more volatility on the other hand to a very high impact on the regulatory capital.

Although the argument that gains of an instrument may disappear irrespectively of the movement of another instrument is understandable we would argue that unrealized losses may also disappear irrespectively of the movement of other instruments.

In case of obligatory item-by-item treatment full inclusion of positive valuation reserve would be justified.

An item by item approach would lead to unrealistic high capital impact for derivatives. For derivatives a portfolio approach is necessary. Also for debt instruments a portfolio approach is reasonable due to the fact that on the one hand unrealized gains will decrease towards the maturity but on the other hand unrealized losses might also decrease towards the maturity (leaving credit risk out of scope). For all other instruments we also plead for a portfolio approach.

For certain items (tangible assets), a materiality threshold could be defined.

We do not agree on option 1 set out in Article 67 of the relevant discussion paper, because it is necessary to ensure:
• a global level playing field (EU/CRR vs. third countries)
• a best possible way of reconciliation and comparability of the accounting figures to regulatory (prudential) figures
• Avoidance of mismatch between the figures measured at fair value within the regulatory capital and those used for calculation of capital requirements
• Avoidance of any misunderstandings in analyzing and comparing accounting figures to regulatory figures

Full exclusion of unrealized gains from own funds is not appropriate from our point of view.
Additionally we want to point out that option 2 is also not in line with the above mentioned arguments, as the exclusion of unrealized gains anyway contradicts a global level playing field, contradicts the possibility to reconcile accounting to regulatory (prudent) figures and would therefore create a source of misunderstandings as well as misinterpretation.

As the competent local authorities could anyway decide to require full or partial exclusion of unrealized gains from the own funds during transitional period from 2015 (after fully exclusion of unrealized gains in 2014) to 2017 and afterwards consideration of unrealized gains will depend on accounting-development (IFRS 9). We therefore do not see the point why to call for such a treatment at this point of time.

We want to point out that consideration of unrealized gains within different layers of capital in case of fully consolidated entities may have an impact in the calculation of eligible minorities in case the entities are not owned by 100%.

We see no problem in case the unrealized gains are eligible within the CET 1 where the entity is not owned by 100%. In case the unrealized gains would be distributed to different levels of capital this would have an impact the parent as on the other hand tier 2 capital issued by the relevant subsidiary would lose eligibility. We want to ask EBA if this approach is appropriate as the issuance of tier 2 capital is a fundamental part of capital steering as well as capital planning. We believe that consideration of unrealized gains would lead to uncertainness on how to steer the capital of subsidiaries in case the issued amount of such instruments is not fully available on group-level.

We also believe that unrealized gains should be available as tier 1 at least as consideration within tier 2 is acceptable in respect of total own funds but not for reasons related to purposes of large exposure and/or leverage ratio.

With reference to the calculation of eligible and non-eligible minority interests we want to point out that any type of haircut or partial exclusion would have impact on the way on how to calculate the minority excess-capital. Therefore, in case haircut or partial exclusion of unrealized gains would be relevant for calculation of own funds we want to ask EBA to provide institutions with information on how the haircut or partial exclusion should be considered within the minority-calculation.

We also want to call attention that haircut of unrealized gains are already in the scope of the proposal on prudent valuation. For institutions using the standard approach for calculation of capital deduction as defined within the consultation paper on prudent valuation, 25% of unrealized gains shall be excluded from CET 1.

We also doubt that the point of time to discuss a possible haircut on unrealized gains is appropriate by now. As the transitional provisions for treatment on unrealized gains according to Article 468 CRR requires a haircut by at least 20% of all unrealized gains for the years 2015-2017 (100% in 2014) we do not see any reason for additional limitation of unrealized gains within this period.
A possible haircut or partial exclusion of unrealized gains from own funds should be required appropriately by considering the development of the respective accounting principles. With respect to institutions calculating and reporting their own funds based on IFRS the further development of measurement rules, in particular the development of IFRS 9 implementation should be taken into consideration.

To summarize our standpoint: Considering transitional provisions on unrealized gains and further development of IFRS 9 we doubt that a final definition on this issue is appropriate at this point of time.

Moreover, if a threshold for inclusion will be introduced it should be defined in relation to total CET1. There should be no threshold or haircut for inclusion in T2.

We think that equity and debt securities should be subject to the same policy options / treatment. Although equity prices could be more volatile than prices of debt instruments, this additional risk is anyway covered by higher risk weights for equity instruments in the banking book.

In particular we agree to the analysis pointed out in 88. The regulations for applying hedge accounting under IAS 39 are very strict and it is expected, that any ineffectiveness should be limited.

Unrealised gains and losses stemming from the effective part of a hedge should not be eliminated, as this would otherwise produce a mismatch and the criteria for hedge-accouting under IFRS are strict enough.
Under the current filter for cash-flow-hedges (Art 33 CRR) one needs to understand that the portion of the gains subject to this filter will be the “effective” one and that any additional ineffectiveness will not be covered by this filter.

With the new IFRS 13 and also derived from new market practice, a higher degree of ineffectiveness must be expected for the future also for hedges, that had in previous times been deemed “perfect hedges” as they fulfil the so-called critical terms match. However, aspects influencing derivatives valuation such as CVA/DVA as well as OIS-discounting (for collateralized derivatives) will automatically give rise to certain ineffectiveness.

One should also consider that under the new concept of hedge-accounting under IFRS 9 it is being discussed that time values for options and probably basis-risk-effects from cross currency swaps will also be booked into a kind of hedge-reserve (similar to the one for cash-flow hedges). The treatment of such reserves in relation to own funds will have to be addressed.

Under a fair value hedge, usually the measurement of the hedging instrument and the basis-adjustment of the hedged item (or the line item of the hedged monetary amount in a portfolio-fair-value hedge) should perfectly offset each other. Thus no elimination would be required and only the ineffective parts (stemming from CVA/DVA, OIS or overhedging) would have to be eliminated.

No significant differences. EBA should just deal with additional OCI item which arises when options are used as hedging instruments and time value from the options is separated from the hedging instrument (IFRS 9 will introduce additional reserves in OCI).

According to one bank all financial instruments, which are held with trading intention, are classified in the trading book. Nevertheless there are also financial instruments with embedded derivatives or financial instruments, which are managed on a fair value basis, but not held with trading intention. These instruments are classified in the banking book. Another one of our member banks stated that they have financial liabilities in the fair value option in order to avoid the bifurcation of embedded derivatives. As the affected instruments are clearly long term and thus banking book instruments, the Austrian regulator has clearly declined their treatment in the trading book.

We support not to consider unrealized gains of all instruments in the FVO due to accounting mismatch, as these economic hedges should not result in a large amount of unrealized gains which are not matched with unrealized losses.

We fully agree with the raised issue that for trading book it is technically impossible to identify unrealised gains and losses, as the historic information needed for this purpose is currently not available in the systems. Additionally it is part of the definition of the trading book that it only relates to short term instruments or such instruments where the realisation of gains is intended in the short term. Additionally the trading book is covered by a special mechanism for own funds requirement calculation and additionally by its own strict concept of prudent valuation.
We do believe that the existing concept for the trading book is sufficient and the introduction of a new filter is neither adequate nor called for.

There would be a double effect when applying a prudential filter and the requirements on prudent valuations. Especially the simplified approach would lead to a double effect, because net unrealized gains are the basis for the additional valuation adjustment calculation.

We believe that the currently existing regulations for the trading book are sufficient.

Moreover, we see an additional double gearing effect not only for the trading book, if RWAs are calculated for banking-book financial assets on their (higher) IFRS book value, while the related unrealised gains / revaluation reserve may not be counted into own funds.

In general we would welcome if those two requirements could be combined.

We think that there is a risk of double effect for banking book items in case of applying both valuations.
Only the amount already adjusted as a result of applying the prudent valuation requirements should be deducted via prudential filter applied on unrealised gains in order to avoid a double effect (instead of additionally filtering out the higher amount before prudent valuation).

a) Prudent Valuation addresses the whole valuation of the assets/liabilities while the prudential filters discussed in this paper only for unrealized gains (notwithstanding the option to apply a prudential filter on a portfolio basis, which would allow the netting of unrealized gains and losses between instruments in that portfolio). The unrealized gains are also part of the whole valuation of assets and liabilities, therefore prudent valuation already addresses unrealized gains as well. We would argue for an “either-or-decision” between prudent valuation and treatment of unrealized gains. Especially in the case of trading book assets and liabilities we plead to follow only the requirements of the prudent valuation regime. If unrealized gains requirements are set for the banking book, we would recommend using the prudent valuation requirements only for trading book in order to avoid double-counting.

b) The interaction between the prudential filters and the prudent valuation requirements (under the simplified approach, the core approach or the fallback approach). Under the simplified approach and the fallback approach, unrealized gains are already accounted for on a net basis. Therefore if an institution uses the simplified approach in the prudent valuation regime this circumstance should be accounted for in the unrealized gains regime. Institutions with net unrealized losses receive a favorable treatment in both the prudent valuation regime and the unrealized gains regime which is not understandable. The combination between potential future losses and unrealized gains is not documented. The core approach is not directly affected by net unrealized gains; nevertheless the aim of the core approach is the same as the aim of the simplified approach.

c) The implication of the application of a filter on an item-by-item basis or on a portfolio basis when considered together with the prudent valuation adjustment. As already mentioned an item by item approach seems to be too prudent and should be avoided. A portfolio approach might deliver similar results as the simplified approach depending on the composition of the portfolios.

Based on the feedback our member banks provided us with, this topic does not seem to be relevant because properties are not measured at fair value. Nevertheless we think that hidden reserves in tangible assets do have an economic value that can be built upon in a crisis situation.

We want to point out that unrealized gains should not be excluded from the regulatory capital anyway. Therefore we are not in line with the argumentation to apply an additional filter on measurement on investment property and property, plant and equipment.
Furthermore we want to state that treatment on FV-measurement of investment property and property, plant and equipment should be revised once the IFRS 9 is in place.

A filter on an item-by-item basis should be related to a materiality threshold for all types of assets, financial and tangible.
Certain type of real estate property could be adequate for treatment within a portfolio, e.g. real estate of comparable functionality (residential real estate, commercial real estate, real estate project finance).