ECSDA (European Central Securities Depositories Association)

Before considering the “practical impediments” for calculating capital requirements for custody risk, ECSDA believes that the EBA needs to address fundamental conceptual flaws in the draft RTS.

(a) “Custody risk” in the Level 1 CSDR refers to a risk faced by the CSD itself, not its participants

First of all, ECSDA believes that the draft RTS contain a misunderstanding of the notion of “custody risk” in the context of art.47 of the CSDR. Custody risk is not defined in the Level 1 CSDR, but it is important to remember that the notion was introduced in the CSDR in relation to the CSD’s investment policy, to be consistent with Principle 16 of the CPMI-IOSCO Principles for financial market infrastructures (PFMI) on “custody and investment risks” . Custody risk is not addressed in a separate article in the CSDR Level 1 text because both types of risks (custody and investment risk) are covered under the same PFMI and because both types of risks were meant to be assessed in relation to the CSD’s own assets.

Whereas clearing members post margins to the CCP, CSD participants do not post assets to secure their obligations towards the CSD, and CSDs typically do not invest their clients’ assets. Recital 11 is thus incorrect as far as most CSDs are concerned . CSDs are different from banks and CCPs in that they are primarily exposed to operational risk, with investment and custody risks being very marginal. Even for CSDs offering banking type services, investment and custody risks are typically very low compared to ordinary banks of the same size.

The custody risks faced by CSD participants (whether in relation to assets held directly in a CSD, via CSD links, or elsewhere in sub-custody) are covered by participants themselves. The risks to the CSD in relation to participants’ assets, on the other hand, are fully covered under operational and legal risks. Should anything happen to the securities held by a CSD on behalf of its participants, the threat to the CSD’s capital is indirect, either as a result of claims for compensation received from participants (legal risk) or operational incidents affecting assets held via links maintained by CSDs with other entities (operational risk). There is thus unnecessary duplication in art.5 of the draft RTS. Removing references to clients’ securities is indispensable to ensure that the RTS are workable and consistent with the Level 1 text and the PFMI.

(b) CSD links should not be considered separately for the purpose of calculating capital requirements

As a result of the excessively broad interpretation of the notion of “custody risk”, the EBA also suggests take into account CSD links separately for the calculation of capital requirements. ECSDA believes that such a proposal is not appropriate for two reasons:

First, CSDs typically do not invest client assets and do not face custody risk in relation to client assets, including assets held via CSD links. Custody risks are faced by the CSD’s clients, and the CSD itself is subject to legal or operational risk, in case the loss of assets via a link would result in a claim towards the CSD. The amount of CSD capital set aside for covering legal and operational risks already covers scenarios related to the loss or damage of client securities via a CSD link, and so the consideration of custody risk in relation to clients’ securities and CSD links creates unnecessary duplication.

Second, there is nothing in the Level 1 CSDR or in the PFMI which requires (or even suggests) that custody risks faced by CSD participants, including via CSD links, should be considered for the purpose of CSD capital requirements. For ECSDA, the mention of “custody risk” under art.47 of the CSDR should be understood in the context of the PFMI, and both the CSDR and the PFMI contain separate and specific provisions (article 48 and Principle 20 respectively) on the risk assessment related to CSD links. In fact, Principle 20 only refers to custody risk as an “additional risk” to be taken into account when CSDs use intermediaries to operate CSD links. Unlike links between CCPs, standard links between CSDs do not involve any credit exposures among the linked CSDs. This is why art.19(5) of the CSDR only requires an authorisation for interoperable links (such as the “bridge” between the two ICSDs) and allows CSDs to notify all other links, without having to request a prior authorisation from the competent authority. Removing the mention of CSD links in art.5 of the draft RTS would thus be fully consistent with the Level 1 CSDR, the PFMI and current supervisory practices (to our knowledge, national regulators do not currently assess CSD links outside of operational and legal risks’ considerations for the purpose of calculating CSD capital requirements).

(c) The proposed method for assessing custody risk is inappropriate

In addition to creating unnecessary duplication and introducing inconsistencies with the Level 1 CSDR and the PFMI, the proposed method for calculating custody risks under art.5 of the draft RTS is inadequate. It refers to the Capital Requirements Regulation (CRR) Standardised Approach for operational risk. When trying to apply the formula to their activities, CSDs face several problems. For example, it is unclear how CSDs' activities can be mapped to the business lines of a credit institution, as described in art.317 of the CRR. CSDs are not involved in the activities described (corporate finance, trading and sales, retail brokerage etc.) and thus they cannot select any of the listed business lines. So-called agency services" are perhaps the only services that are conceivable in a CSD context, but then the definition provided in the CRR is different from the central maintenance service as described in the CSDR. Indeed, the type of custody services provided by other financial institutions are different from the safekeeping of financial assets by financial market infrastructures like CSDs, and are thus not comparable from a risk perspective.

Furthermore, no matter what numbers CSDs try to calculate using the Standardised Approach, the results always seem to be already included in the calculations under the Basic Indicator Approach for operational, legal and custody risk. This further confirms the duplication and shows that art.5 of the draft RTS should not only be amended in relation to clients' securities, but should be deleted entirely.

ECSDA recommends that references to custody risk should be included in art.4 of the draft RTS instead, in line with the spirit of the Level 1 CSDR (see our proposed amendment under Question 2)."
ECSDA agrees with the EBA that:
- All EU-authorised CSDs with an identical risk profile should be subject to similar capital requirements, ensuring a level playing field;
- The level of CSD capital should be proportional to the risks arising from the activities of a CSD. In other words, we support a risk-based approach, as opposed to a fixed and arbitrary amount of capital;
- Some CSDs will need to increase their capital level as a result of the new standards. ECSDA is not advocating the status quo or ‘harmonisation by the bottom’. We support solid capital requirements for all CSDs, but these must be fair, simple, efficient, workable and truly proportionate.

Against this background, some of the draft standards proposed by the EBA do not seem appropriate. In particular, some of the methods put forward for calculating the level of capital requirements of a CSD will not result in an accurate reflection of a CSD's risk profile, imposing a disproportionate burden on many CSDs, most of which are small or medium sized enterprises with a particularly low risk profile.

Title I of the draft RTS should be redrafted to reflect CSDs' specificities and to allow for a much clearer differentiation between CSDs based on their risk profiles. In particular, the standards need to take into account the low risk nature of activities performed by non-bank CSDs, which, unlike CCPs and banks, are not exposed to material credit risk in relation to their core activities. Relying on CRR approaches designed for banks, as is being proposed by the EBA, is not always workable and could make it difficult for competent authorities to determine the appropriate capital level for the CSD(s) they supervise. The proposed methods often put the focus on risks which are marginal for CSDs (investment, credit and liquidity risks) instead of allowing for a transparent and consistent assessment of operational risks which are at the core of every CSD’s activities.

It is especially important to remember that the main risk faced by CSDs, operational risk, is mitigated mainly via the design and the procedures of the CSD’s securities settlement system, not via capital requirements. No matter how well capitalised a CSD is, this does not in itself guarantee that the system is available to participants, hence the focus of CSD overseers on business continuity and recovery planning.

In addition to these general comments, in order to be as helpful as possible, ECSDA would like to put forward the following alternative solutions for the most relevant articles under Title 1 of the draft RTS:

 Article 1: Overview of requirements regarding the capital of a CSD

ECSDA agrees with the drafting of art.1 of the draft RTS.

 Article 2: Definition of capital of a CSD

ECSDA supports the proposed definition of “capital” under art.2 of the draft RTS (p.18).

 Article 3: Level of capital requirements for a CSD

(a) The treatment of custody risks

ECSDA disagrees with art.3(1)(b) of the draft RTS. Point (b) is redundant and inconsistent, as explained in our response to Question 1. CSDR article 47 on capital requirements requires that a CSD “is adequately protected against operational, legal, custody, investment and business risks (…)”. This means that the capital of a CSD must be sufficient to mitigate the risks described under articles 43 (legal), 44 (business risk), 45 (operational) and 46 (investment and custody risk). It does not aim to include custody risks faced by CSD participants, and point (b) should thus be deleted altogether, especially since custody risks are already addressed under art.3(1)(a).

(b) The cumulative approach"

ECSDA agrees that a CSD's capital should be sufficient to ensure that a CSD can continue to operate as a going concern and to ensure that, if a restructuring or winding-down process is necessary, such process can take place in an orderly way to minimise market impact and preserve financial stability. However, we believe that the cumulative approach adopted by the EBA in relation to points (a) and (b) of CSDR article 47(1) is excessive and needs to be adjusted.

The problems with the proposed cumulative approach are as follows:
- There are overlaps between the approaches under points (a) and (b) of art.47(1), for example as regards business risk under point (a) and winding-down under point (b). A given event (e.g. a substantial and unexpected loss of income) considered for business risk purposes might or might not fall under winding down scenarios, depending on the severity of the problem and whether it poses a threat to the operation of a CSD as a going concern. The same scenario might thus be considered twice, although with different degrees.
- The proposed cumulative approach is not consistent with Principle 15, Key consideration 3 of the PFMI, which allows the same capital to be used both for ensuring that the infrastructure operates as a going concern and for ensuring an orderly recovery or wind down of critical operations. Whereas we acknowledge that the EMIR standards go beyond the PFMI, given the specific role of CCPs as centralisers of risk in financial markets, we do not believe that it is the intention of the EU legislator in the Level 1 CSDR to impose significantly stricter requirements in the case of CSDs.

Taking into account the need for a consistent treatment of custody risk and the need to avoid an excessively strict interpretation of the "cumulative approach" for business risk and winding-down scenarios, ECSDA recommends amending art.3(1) of the draft RTS as follows:

- Delete art.3(1)(b) "(b) the CSD’s capital requirements for custody risks, referred to in point (a) of Article 47(1) of Regulation (EU) No 909/2014, calculated in accordance with Article 5 where it has its securities or its clients’ securities under custody by another CSD or an intermediary within a CSD link;"
- Add the following phrase at the end of art.3(1)(e) ",to the extent that these capital requirements are not already covered under point (d) above"

As regards art.3(3) and (4) of the draft RTS, ECSDA understands that the wording has been copied from EMIR, but we wonder whether a notification at 110% is justified in the case of CSDs, given that a CSD's capital is typically very stable. Unlike CCPs, CSDs are not required to provide capital for managing a participant’s default and they are thus not exposed to systemic risk to such an extent that a minor loss of capital (e.g. 1%) would require urgent corrective actions and strict monitoring from a prudential perspective. This is especially true as such minor loss is very unlikely to lead to any disruption in CSD services. As a result, to avoid unnecessary notifications, ECSDA recommends either lowering the notification threshold or providing more flexibility as regards the frequency of notifications to the competent authority (to be determined by each competent authority), for example by replacing the phrase "at least weekly" in art.3 by "regularly".

 Article 4: Level of capital requirements for operational, legal and custody risks

The references to the CRR used in art.4 of the draft RTS are partly inappropriate when applied to non-bank CSD activities. Some adjustments are needed to allow for a proper reflection of CSDs' risk profile in their capital requirements.

(a) A re-calibration of the 15% ratio

ECSDA supports the use of the Basic Indicator Approach for assessing operational and legal risks, but believes that the 15% ratio used for banks should be recalibrated to 10% to reflect CSDs' lower risk profile. Indeed, many ECSDA members have not suffered any operational losses over the last 10 years, and, when there are any such losses, they tend to be extremely low.

Furthermore, it should be possible to adjust the ratio to take into account the effect of CSD-specific insurance arrangements covering operational and legal risks. In some markets (e.g. in the Nordics and the Baltic region), CSDs are obliged by law to subscribe special insurance arrangements against operational risks. Such arrangements protect the CSD’s capital up to a certain amount should the risks materialise, and they are thus taken into account by national supervisors when determining minimum capital requirements for the CSD. There are sometimes group-wide professional liability insurance policies protecting CSDs against worst case loss-causing events in the area of operational and legal risks. These arrangements are a vital component of those CSDs’ risk-management frameworks and should be recognised in the RTS. In some instances, the insurance cover is supplemented by a statutory participant guarantee scheme (see the response of individual ECSDA members to the EBA consultation for more details).
ECSDA believes that, when CSDs have specific insurance arrangements to cover business and other risks, competent authorities should be allowed to take these into account (as well as any existing guarantee scheme, if applicable) when calculating capital requirements under art.4 of the draft RTS. This could be done, for example, by applying a lower ratio or by allowing CSDs to apply art.323 of the CRR (as regards the impact of insurance in the Advanced Measurement Approach) in the context of the Basic Indicator Approach calculations so that capital requirements for operational risk could be reduced by up to 20%.

ECSDA thus recommends the following amendments to art.4 of the draft RTS:
- Replace the phrase "…shall calculate its capital requirements for operational and legal risks in accordance with the provisions of the Basic Indicator Approach as referred to in Articles 315 and 316 of that Regulation" by " shall calculate its capital requirements for operational, and legal and custody risks in accordance with the provisions of the Basic Indicator Approach as referred to in Articles 315 and 316 of that Regulation, using a ten percent ratio. "
- Add a paragraph stating that "National competent authorities shall be allowed to apply a lower threshold in the case of CSDs having a comprehensive liability insurance policy in place to cover operational and other risks. "

 Article 5: Level of capital requirements for custody risks

As explained in our response to Question 1, we believe that art.5 of the draft RTS should be deleted and that custody risk should solely be addressed under art.4 of the draft RTS.

 Article 6: Level of capital requirements for investment risk

CSDs, unlike banks, do not typically seek to generate income by investing client assets; client fees are the primary source of income. Investment and custody risks are thus very marginal compared with operational risk. Besides, given the restrictive investment policy imposed on CSDs by article 46 of the CSDR, in particular the obligation for a CSD to invest its assets in highly liquid instruments, investment risk is typically extremely low for CSDs. The EBA should thus be aware that some aspects of the proposed method for calculating capital requirements in relation to investment risk will not apply in the case of non-bank CSDs (e.g. CSDs typically do not enter into derivatives transactions).

Furthermore, ECSDA wonders whether the consideration of tangible assets is adequate for CSDs. In particular, the fact that a CSD owns its office buildings (land and property used for the purpose of its own activities) should not result in a risk weight of 100% for assets as per art.134(1) of the CRR. Such property investments should not be considered in the context of investment risk (credit risk in particular) and the calculation of a CSD’s risk-weighted exposure amounts for credit risk should be limited only to CSD’s investment activities as described in art.46(3) of the CSDR (“A CSD shall invest its financial resources only in cash or in highly liquid financial instruments with minimal market and credit risk”).

 Article 7: Capital requirements for business risk and Annex II on business risk scenarios

Given the nature of CSDs' businesses, in particular the fact that CSDs operate on considerably simpler and lower risk business models than banks or CCPs, ECSDA believes that the 25% ratio for business risk is far too high. Using gross expenses as a reference is also inappropriate, as business risk should be primarily covered by net income (current or planned net income or EBIT). This would avoid unnecessarily high capital requirements, e.g. in case of a CSD that would remain profitable after the business risk scenarios described in Annex II have materialised. Equity coverage should only be required in case of net losses, hence there should be no 25% floor.

ECSDA thus suggests amending art.7(1) of the draft RTS by replacing the phrase "For the purposes of point (d) of Article 3(1), the capital requirements of a CSD for business risk shall be the higher of the following: (a) the sum of all estimates resulting from the application of paragraph 2; (b) 25% of the CSD’s annual gross operational expenses" by "For the purposes of point (d) of Article 3(1), the capital requirements of a CSD together with retained earnings and reserves for business risk shall be the sum of all estimates resulting from the application of paragraph 2".

As regards Annex II (p.68) on business risk scenarios, ECSDA is not convinced that predefined business risk scenarios are the most appropriate means to calculate capital requirements for business risk. We believe that a more flexible approach, similar to that in EMIR, allowing for example the use of reasonably foreseeable adverse scenarios relevant to the CSD’s business model, as approved by the competent authority, could be more efficient and proportionate.

Should the EBA however decide to maintain Annex II and to impose the use of business risk scenarios, we believe that the following changes are necessary:
(1) The arbitrary percentages attached to each type of event should be removed. For example, an unexpected reduction of income of 30% will not necessarily affect the capital of a CSD, and it should be possible for a CSD and its competent authority to assess the relevant fall in income that could require additional capital requirements, based on the CSD's own business model and risk profile;
(2) The reference to credit ratings should be removed in view of international and European initiatives calling for a progressive removal of references to ratings in regulation to reduce overreliance on them. Furthermore the EBA should be aware that several CSDs do not have credit ratings and do not issue debt. Annex II (a) will thus not be practically possible to apply for those CSDs. The CSDR does not require CSDs to obtain credit ratings, nor to issue debt.
(3) Points (e) and (f) of Annex II referring to long and short client balances should be removed as there are no such "balances" at CSDs and the notion is totally irrelevant for non-bank CSDs.
(4) Regarding contributions for pension plans under point (d), ECSDA recommends that the EBA should make it explicit that this requirement only applies to defined benefit schemes and not to defined contribution schemes.

ECSDA thus recommends either deleting Annex II entirely or amending it as follows:
- Add "when CSDs have credit ratings" at the beginning of point (a);
- Replace "(b) an unexpected increase of funding costs of 10%" by "(b) an unexpected substantial increase of funding costs";
- Replace "(c) an unexpected reduction of income of 30%" by "(c) an unexpected substantial reduction of income";
- Replace "(d) 10% higher than planned cash contributions for pension plans" by "(d) "higher than planned cash contributions for pension plans, in the case of defined-benefits schemes";
- Delete points (e) and (f).

 Article 8: Capital requirements for winding-down or restructuring and Annex I on winding-down or restructuring scenarios

As regards Annex I (p.67-68), ECSDA agrees with the proposed definitions of restructuring (situations where the CSD no longer meets capital requirements but is able to continue to provide core services) and "winding-down" (situations where the CSD is no longer in a position to continue to provide all core services required under the CSDR licence) contained in paragraphs 1 and 2 respectively.

ECSDA also welcomes the recognition, in paragraph 4, that "the scenarios shall be commensurate with the nature of the business of the CSD" and understands that CSDs with the simplest and lowest risk profiles, including a low level of interconnectedness with non-domestic markets, will be able to assess capital requirements based on a shorter restructuring/winding-down period than CSDs with a more complex risk profile.

As regards paragraphs 5, 6 and 7, ECSDA does not believe that the list of events proposed by the EBA is appropriate, especially for non-bank CSDs. For example, the "failure of significant counterparties" referred to in point (a) will typically not lead to a restructuring or winding-down of a CSD, since non-bank CSDs are not exposed to counterparty credit risk in relation to their participants, unlike CCPs. The same reasoning applies in case of a "severe outflow of liquidity". It will thus be impossible for most CSDs to include the listed scenarios in a meaningful way in their winding-down plan. Imposing that all CSDs include all scenarios, as suggested by the current wording of the Annex, does not make sense. As a result, ECSDA suggests either removing all references to “idiosyncratic events” and “systemic-wide events”, at least for non-bank CSDs, or allowing CSDs, together with their competent authority, to select and calibrate the relevant scenarios according to their own business models. This approach would be in line with the PFMI and EMIR.

Finally, CSDs should have the possibility to remove certain expenses from gross expenses, in particular:
- when these are not relevant in a winding-down situation, e.g. because they can be cancelled immediately from the moment the CSD enters into restructuring (e.g. bonuses, staff and commercial events, large projects);
- when these do not involve a cash outflow, such as depreciation and amortization expenses (see Principle 15 of the FMI, footnote 137 p.90).

ECSDA recommends either deleting paragraphs 5, 6 and 7 entirely or alternatively making the following amendments to Annex I:
- Delete point 5 entirely;
- In points 6 and 7, replace the phrase "shall include all the following scenarios" by "shall, where relevant, include the following scenarios".

 Comments on the draft impact assessment (p.69-78)

ECSDA welcomes the EBA initiative to undertake a survey among CSDs, via national competent authorities, to assess the impact of the proposed measures on the current level of capital of CSDs. Given that the calculation method being proposed by the EBA for capital requirements is totally unprecedented for non-bank CSDs, it is all the more important that the EBA carefully estimates the quantitative impact of the proposed RTS on current capital levels of CSDs before the RTS are finalised. The simulation exercise is also important to identify potential problems and unintended effects prior to the finalisation of the standards.

In order to ensure a proper impact assessment, we recommend that the EBA considers not only changes in the amount of capital to be maintained by CSDs, but also changes to the structure of that capital. Indeed, art.46(3) and (4) of the CSDR state that the amounts of capital which are not invested in cash or highly liquid financial instruments with minimal market and credit risk shall not be taken into account for the purposes of art.47(1). This means that some CSDs might have to overhaul their asset structure without necessarily having to raise their capital, and we believe that such changes in the share of cash and highly liquid instruments in total assets should be reflected in the impact assessment.

Although ECSDA was not able to carry out a comprehensive assessment, we anticipate that the draft RTS would result in an increase of minimum capital requirements for almost all CSDs, raising the bar far beyond current domestic requirements and the PFMI. That said, the impact of the RTS on the capital level of most CSDs should be limited by the fact that CSDs are often significantly overcapitalised on the basis of current minimum requirements. This should not prevent a few CSDs from having to raise a significant amount of additional capital as a result of the implementation of the standards, with possible repercussions on the level of CSD fees."
ECSDA would like to point out that the methodology for the calculation of the capital surcharge is based on theoretical and extreme scenarios that are not always realistic. In particular, the assumed 10% drop in market value of all collateral, in addition to already very conservative haircuts, does not seem to be an appropriate scenario and should be withdrawn.

Moreover, the calculation of the capital surcharge which reflects intraday credit exposures duplicates to a certain degree existing capital requirements for end of day exposures which are already taken into account under the CRR. The latter should therefore be subtracted from the calculation of the capital surcharge under art.9.
Intraday liquidity monitoring is a key treasury function. However, reporting of intraday liquidity positions requires the provision of adequate information by the CSD’s correspondent banks. CSDs are not in a position to impose such requirement on (all of) their cash correspondents.

It is very important that the RTS be aligned with the framework for intraday liquidity monitoring which is being developed by the BCBS (see BCBS guidelines 248). Given that the proposed requirements are mainly introduced to monitor the liquidity risks of correspondent banks, they should be governed by bank regulation rather than by the CSD Regulation. If for some reason the requirements are not introduced by the BCBS (or in the corresponding EU or national bank regulation), but are included in the CSD Regulation, they will only be applicable to CSDs, which will result in distortions.

As the provision of the necessary information requires a huge effort both on the part of the regulated entity as well as its correspondent bank, there is a risk that CSDs' correspondent banks will not incur the effort to set up the required infrastructure, if only CSDs have to comply. ECSDA therefore suggests covering intraday liquidity requirements in the regulatory framework applicable for banks, rather than in the CSD Regulation, or to align both to the largest possible extent.

Moreover, ECSDA would like to point out that the recordkeeping requirements for intraday and overnight risks included in art.14 of the draft RTS are technically impossible for CSDs to implement given that intraday exposures change constantly and in real-time, resulting in a huge amount of data which cannot possibly be kept over a 10-year period.

ECSDA thus recommends deleting this requirement. This will also ensure that all record-keeping rules are consistently covered under the technical standards under art.29 of the CSDR.
CSDs providing banking services are substantially different from CCPs, in terms of business as well as risk profile. They neither take principal risk in transactions, nor novate trades. Not all EMIR requirements are therefore appropriate for those CSDs. This is particularly true in relation to the collateral acceptance policy. The proposed requirements for acceptable collateral under art.19 of the draft RTS are too restrictive and would mean in practice that the category of “other collateral” would be irrelevant. Being able to accept a meaningful range of “other collateral” is nonetheless important for CSDs that provide banking services, in particular in relation to their international business.

In particular, ECSDA believes that art.19(1)(c) which requires an average time to maturity of less than 2 years for the portfolio of CSD banking service providers is inappropriate and does not adequately reflect the exposures incurred by the relevant (I)CSDs. Restricting collateral in such a way could have a serious negative impact on the market and cause stress on the limited set of highly liquid assets.
According to art.27 of the draft RTS, CSDs providing cash to pre-finance corporate actions should ensure that “irrevocable intraday guarantees from paying agents or issuer agents are established before lending the cash”. ECSDA is concerned about the proposed wording because it is not possible for CSDs to oblige paying or issuing agents to issue irrevocable intraday guarantees. Maintaining this requirement would put EU CSDs falling under the scope of the RTS at a competitive disadvantage compared to non-EU CSDs and risks making issuance in the EU less attractive for issuers, going against the objectives of the Capital Markets Union.
Soraya Belghazi