German Banking Industry Committee

Only a few investors can bear the considerable cost of the tougher demands introduced in the past few years on internal processes. Most important from our perspective are the regulatory impediments. For example, we fail to grasp why the securitisation market is to be supported whilst at the same time the regulatory demands on securitisations - seven years after the onset of the crisis in the subprime RMBS segment in the USA - are being significantly tightened:

- The capital requirements for securitisations with AAA ratings (e.g. Auto ABS) are to be more than doubled from 2018 according to the revised Securitisations Framework published in December 2014. The risk rating floor is to rise from 7% to 15%. The EBA discussion paper also proposes a 15% floor. Thus, no alleviation for the capital backing is apparent for SST securitisations". Compared with covered bonds this would mean a 50% higher capital requirement.

- Unless adjusted to successfully established market standards, non-impairment demands on securitised assets will have significant negative effects on the further development of the securitisation market, albeit without any reasonable benefit in the form of the apparently expected greater security.

- Further impediments arise from the fact that the revised Securitisations Framework published in December 2014 could force IRB banks to use the external rating-based approach due to their lack of information about the underlying assets or rating procedures approved by the banking supervisory authorities. This could be the case especially if the development of internal IRB processes for securitisation positions is not worthwhile for those banks. In that case, the risk weights would well-nigh triple, namely from 7% to 20%. This could lead to a reduction in the investor base.

- Should the current Commission's bank recovery and resolution proposals come into force in this form, negative effects on the securitisation markets could not be ruled out. For example, Amortising Swaps—which are extremely important for the marketability of ABS—could probably only be offered by trading entities. This will lead to a further narrowing of the market. The number of possible counterparties is already extremely limited.

- Liquidity is the driving factor for the attractiveness of ABS. Market makers play an important role in this. The Bank Recovery and Resolution Directive envisages restrictions which will have a negative impact on the number of possible market makers.

- Significantly higher capital requirements for securitisations in the trading book are still being discussed. This could lead to a sharp drop in ABS trading volumes, with a corresponding negative impact on ABS and their liquidity.

- The relative capital requirements are too high even for high-quality securitisations under Solvency II compared with other assets, especially covered bonds. This could make it uninteresting for insurance companies to invest in securitisations in future.

- The large number of vague legal terms (e.g. relating to non-impairment requirements) has left wishes of investors for LCR- and Solvency II-compliant ABS paper largely unheeded.

- The Commission's current proposals would be tantamount to prohibiting money market funds from investing in ABS in future, even if they are recognized as highly liquid under the delegated act to the LCR. This is already leading investors to move towards the sidelines.

Against this background there is great uncertainty as to how these provisions, already adopted but yet to come into force, will impact the demand for securitisations in future. It is to be feared that the tighter regulations will shrink the investor base, with a corresponding negative impact on the depth and breadth of the securitisation market."
Synthetic transactions should not be excluded from the framework for simple standard and transparent securitisations. This holds true in particular when the only reason why a securitisation does not meet the EBA’s criteria for simple standard and transparent securitisations is because it has been structured as a synthetic securitisation. The essential benefit of synthetic transactions for many originating banks is the transfer of credit risk (e.g. SME loans) to third parties, when true sale transactions (traditional securitisations) cannot be employed because bank customers do not want the bank to sell their loans (transfer clause limitations). Furthermore, synthetic securitisations release risk-weighted assets which enable banks to extend new loans to the real economy. Moreover, synthetic transactions are often the only way to manage the risk arising from certain off-balance sheet exposures, e.g. letter of credits or guarantees provided to the bank’s customers. This also applies to certain on-balance sheet exposures like until-further notice overdraft facilities. In other words, synthetic transactions are a very efficient way of supporting real economy SME transactions in that they enable banks to transfer the risks of various lending products and accommodate the concerns of bank’s clients such as data secrecy or the unfounded but widely spread worry about a sale of the relationship to third parties such as hedge funds. Synthetic transactions also support risk sharing in the financial system.

We do acknowledge that the originator as well as the investor might be exposed to an additional counterparty risk in comparison to true sale transactions. Nevertheless the counterparty risk in true sale transactions might also arise from interest or FX derivatives needed for transforming the cash flow of the underlying pool. However, the counterparty risk from the originator’s perspective can be mitigated by the provision of equivalent cash collateral (SPV enters into guarantee or credit derivative with originating bank to ensure risk transfer and issues credit-linked notes to external investors). The cash collateral is either deposited with the originating bank or a third bank and pledged to the originator. On the other hand any counterparty risk from the investor’s perspective can be mitigated by setting a rating trigger for the deposit bank. If the deposit bank is downgraded below a predefined threshold, the cash deposit (arising for the CLN) has to be transferred to another eligible bank. Thus the investor only faces a jump to default" risk, if the bank becomes insolvent immediately without a preceding rating shift. It should also be borne in mind that synthetic transactions offer advantages over true sale transactions. Since the securitised assets will not be sold to the SPV there are no risks such as legal validity of the receivables, commingling risk, settlement risk and collection risk. This implies that the investor will not suffer any losses arising from such risks since they are not credit default risks. Moreover, if the originator bank defaults, the guarantee or credit default swap will be terminated and the investor gets back the provided cash (from purchased CLN) exceeding any occurred credit events in the underlying portfolio (in contrast to selling the securitised assets or awaiting any scheduled repayments in the portfolio). This is of particular interest for investors who want to buy a certain credit risk but not the actual underlyings (and potentially having to wait until all the assets are sold before getting their money back).

As mentioned above, synthetic transactions can be structured in a simple and transparent way. Often the transactions and associated documentation are less complex for both issuer and investor because no sale of assets is involved. By way of example there are after fewer parties involved in a synthetic transaction, and the complexity of criterion 12 (documentation specifying contractual obligations, duties and responsibilities of the trustee, servicer and other ancillary service providers) can be significantly reduced.

A synthetic transaction could be considered to be simple, standard and transparent under almost the same conditions/criteria proposed for a true sale securitisation. The following criteria need to be amended:

 criterion 1: it should be a "synthetic transaction" as defined in the CRR (as per Article 242(11));

 criterion 3: in the case of synthetic securitisation, the underlying assets have to be clearly defined and belong to one originator. A legal opinion should confirm that the credit protection arrangement is legally effective and enforceable in all relevant jurisdictions. The counterparty credit risk is mitigated by the way of collateralisation;

 criterion 12: in the case where servicer and originator are the same legal entity, the default or insolvency of the current servicer should lead to the termination of the transaction and fund repayment if the transaction is (partially) funded;

 criterion 15: the criterion should only apply to publicly offered or listed transactions with securities issuance. In the case of private transactions, the originator ensures that investors have access to all the information that is necessary for making an informed investment decision before the closing.

In summary, we believe that when synthetic securitisations of bank loans meet the spirit of the remaining EBA criteria, their being structured as a synthetic securitisation should not exclude their qualification as a "simple standard and transparent securitisation" and regulators should encourage the ability of banks to manage the risks associated with bank originated loans."
Criterion 5 proposes a default definition that does not correspond to the CRR's existing one. We are of the opinion that this definition cannot be implemented as it stands, because the criteria for default are very vague (e.g.: credit score indicating significant risk of default") and thus give scope for interpretations. This is an extremely critical point in our opinion. In order to achieve a uniform approach throughout Europe, additional explanations or EBA guidelines would thus be necessary. For that reason, we would suggest that the common practice be applied and any claims which are overdue, in default as per Art. 178 CRR or for which an individual IFRS value adjustment has been made be excluded.

In this context we would like to point out that it is unclear when a credit assessment by an ECAI or a credit scoring indicates a significant risk. In addition, it is not clear whether external credit assessments and internal credit scorings can be used alternatively or whether they have to be used in combination. The latter would entail very high additional costs.

Furthermore, it is unclear which ECAIs and which credit scorings could be used under which conditions. Many originators of high quality ABS still use the credit standardized approach and thus have no IRB approval by the supervisory authorities although they have scoring procedures and models in place which are validated annually, are comparable to those of IRB credit institutions and have a significantly better discriminatory power than the scoring and rating procedures of ECAIs. Although the assessments of ECAIs are used in the credit process, they are in many cases not used for the credit standardized approach, and hence the requirements of the CRR to use the assessments of ECAIs have not yet been implemented because the costs would be too high due to the fact that the external ratings or scorings would have to be applied consistently worldwide. In addition, it has to be noted that the EU is looking to reduce the dependency on external ratings. However, the new requirement would actually increase this dependency on external ratings.
To ensure comparability of "significant risk" across Europe, a PD threshold would have to be determined. The use of scoring and PD models implies model risks. The same problems exist with respect to model risks for ECAI scores.

Currently, many high quality securitisations are not based on ECAI assessments or internal credit scoring but on the non-overdue status. Accordingly, all underlying exposures that are overdue at the time of the selection (pool-cut) or additionally in the case of replenishment after inclusion at any time after issuance are excluded from the selection. In future, these tried and tested processes which have ensured low losses of the underlying assets from securitisations even in times of crisis would have to be changed.

Furthermore, it is not clear under which conditions and for how long a credit history is deemed adverse or no longer adverse after a company has recovered or a private person is once again able to pay after a period of unemployment. Such a requirement would prevent in particular the recovery of SMEs after an economic downturn due to the increased financing costs even if the company has returned to a good credit standing in the meantime.

Moreover, companies, such as SMEs, which have recovered after an insolvency or debt rearrangement process, should not be excluded if they are no longer impaired under applicable accounting rules. Such accounting rules also require an assessment after a recovery as to whether the borrower is still credit-impaired. If this is the case, then such loans would have to be excluded from the securitisation of high quality ABS. According to the current proposal such borrowers would be excluded for three years regardless of their current creditworthiness, which would be detrimental to the recovery of such companies.

Finally, it will be difficult to implement the requirements because the required information is not stored in IT systems in a manner suitably structured for facilitating the exclusion of loans deemed credit-impaired under a new definition which deviates significantly from the credit-impairment definition under accounting rules. Thus, it is unclear how to exclude those borrowers and lessees which are no longer credit-impaired according to the accounting rules or in default and are serviced in the meantime in the normal course of business after their recovery. The information required for identifying such borrowers is often only available in an unstructured manner in credit agency reports. Highly granular ABS can comprise up to 90,000 contracts. It is virtually impossible to check all those contracts manually.

As a result, many existing high quality ABS structures are not catered for under the proposed non-impairment requirements, even though the existing eligibility criteria, which were mainly based on the non-overdue status, performed extremely well even under the stress conditions caused by the last crisis. Thus, we propose excluding all exposures which are in default as per Article 178 of Regulation (EU) 575/2013 ("CRR"), credit-impaired with objective evidence of impairment as per the relevant accounting standard or appendix A of IFRS 9, or are overdue.

To accommodate a forward-looking perspective and a prudent selection of the receivables to be securitised we propose the following:

In addition to the past-due status, originators should have to have an internal scoring or rating procedure subject to at least annual validation that comprises as a minimum a review of the discriminatory power and the calibration. An IRB approval should not be an obligatorily requirement. If such a procedure is not
available, the ECAI should be allowed if the ECAI can demonstrate that the minimum validation criteria are fulfilled and if the originator obtains at least an annual validation report on the discriminatory power and calibration.

Furthermore, we propose a requirement to the effect that internal or external scores have to be used to support the credit-decision process based on an acceptance policy which has to be the same for the securitised and non-securitised portfolio. Loans with a credit-score that is worse than the policy's cut-off score would have to be excluded from the securitisation even if approved manually because it would be assumed that all exposures which are worse than the cut-off score have a significant risk.

To avoid any adverse selection, the loans should be selected randomly from a target portfolio. This applies especially for those selected on a past-due status during the life of the contract. By combining the past-due status and the use of scoring and rating models it would be possible to have in place both an objective procedure with a prudent backstop function and a forward-looking perspective based on the acceptance policy both for the securitised and non-securitised loans without having to rely on PDs with their inherent model risks."
No, a restriction to Europe is not necessary if the calibration of capital requirements distinguishes between assets originating in Europe and beyond Europe. To simplify the due diligence, however, the securitisation of assets from different jurisdictions also within Europe should be avoided, because the lack of harmonised civil law and diverging case law in the individual jurisdictions would otherwise lead to excessive complexity.
We reject this, because this would disadvantage the owners of junior bonds. This could have extremely negative effects on the market attractiveness of junior bonds. Without the marketability of the junior bonds, the entire ABS transaction might be rendered unattractive to the market. This would stymie the goal of promoting an expansion of the securitisation market.

Furthermore, there are reservations on the part of the originators. It is widely perceived that there are reservations about borrowers and the securitisation of claims. These reservations are partly fuelled by the negative reporting on the detrimental treatment of customers whose claims have been securitised.

Currently, most ABS structures have been set up as auto-pilots" with defined contractual clauses on how to respond to a default or insolvency event. The granting of voting rights to an "identified person" could be interpreted by the general public to mean that the "identified person" or the trustees might also have discretionary rights permitting them to interfere in the established customer-bank relationship. This will surely fuel the already existing reservations e.g. among consumer protection bodies and parts of the general public and could be detrimental to the fostering of the securitisation market.

Should these reservations not lead to dropping the idea of distributing voting rights to an identified person, it should be ensured that these voting rights are restricted to insolvency and default events. Furthermore, the owners of the junior bonds should be included in such cases, in so far as mezzanine tranches immediately above the first loss position are not involved, so as not to jeopardise the market attractiveness of these bonds of crucial importance for the market attractiveness of the entire ABS transaction. Any impression that the "identified person" would have any discretionary or intervention rights during the course of the normal execution or implementation of an ABS transaction which would impact the customer-bank relationship and the particular situation of mutual trust must be avoided at all costs."
Question 6 asks if any documentation should be disclosed prior to issuance. We would like to ask that the exact understanding of the term prior to issuance" be clarified. It should not be overlooked that the originator relies on a certain degree of confidentiality before the deal announcement, and this might be endangered by a premature disclosure of significant deal information. Any extension of the period in which an ABS term transaction is public but not closed (i.e. closing of the books and the end of the book building) would burden the originator with more uncertainty regarding the placement of a transaction. Furthermore, we are of the opinion that the requirement is not relevant for bilateral private transactions."
We believe that the threshold is appropriate for retail transactions. However, the requirements with respect to the group of connected clients should be softened somewhat in a way that includes according to the best knowledge" of the originator. The reason is that there are in part thresholds in retail bulk business for identifying a group of connected clients in order to reduce the workload. Such a practice does not endanger the identification of single risks but could mean that small exposures might not be identified as belonging to a group of connected clients. Although it is very unlikely that the granularity threshold would be exceeded without the originator's knowledge, it cannot be fully ruled out that there might be a very few cases which exceed the granularity threshold. In order to avoid a securitisation not being eligible as "qualified" in such cases, the wording should be softened as proposed.

With respect to corporate customers the threshold could be too low. We propose a threshold of between 2% and 3%."
We agree with the principles, but see some need for further adaption and specification in order to avoid those securitisations which are already well established in the market and already fulfil the basic principle requirement of simple, transparent and standard securitisation being excluded from recognition as a SST securitisation". Otherwise, there might be negative repercussions for these already well-established securitisations. In addition, the wording should be adapted to the wording used in article 13 to the delegated act to the LCR determining the eligibility criteria for qualifying securitisations to benefit from extensive discussions during the deliberations on the development of the delegated act with stakeholders.
Criterion 4

Under Criterion 4, underlying assets should be "homogenous in terms of asset types". It should be clarified that the securitisation of corporate and retail exposures in one transaction is not excluded by this requirement.

Under Criterion 4, iv) the repayment of a securitisation position should not depend, i.e. either fully or in part, on the residual value of the collateral provided for the claim. This requirement could lead, especially for car securitisations, to these securitisations not qualifying as simple securitisations because the wording "in part" could have a detrimental impact on financing contracts with indirect residual value risks. When concluding contracts of this type, car dealers undertake to buy back the vehicle at a price determined in advance that reflects the final instalment and the estimated residual value and to pay the final instalment. Any shortfall is to be borne by the dealer. Hence the repayment depends in part on selling the vehicle. Therefore, it must be ensured for this criterion that only direct residual value risks are affected. The phrase "in part" should accordingly be dispensed with. The wording in the delegated act to the LCR constitutes, from our perspective, a suitable solution.

Criterion 9

Referring to "Commonly encountered market interest risks" in the automobile business loans and leases are subsidized by the car manufacturers or car dealers to promote the sale of the cars. It should be clarified that loans and leases subsidized, for instance, by the manufacturer and/or the car dealer are not excluded.

Criterion 13

Criterion 13 requires the involvement of an "identified person" who acts in the best interest of investors and facilitating timely resolution of conflicts. We kindly ask that it be clarified whether it is the EBA’s intention to establish a further party to act as arbitrator or if the trustee can play that role or if this is already covered by the capacity of trustee in traditional ABS term transactions.

See also our comments on question 5.

Criterion 17

"Access to all underlying transaction documents" should be limited to "all material and reasonable" underlying transaction documents.
Criterion 18

Processes and especially certain kinds of action are not always predetermined but characterised by a certain level of discretion that decision makers have when taking their decisions. Hence this could conflict with predetermined processes and actions. In addition, processes, definitions and actions can change during the lifetime of an ABS transaction. Thus, a description of the prescribed processes, definitions and actions could force the originator not to change the processes, an undesired outcome that should be avoided.

Thus, there should be a reference that these processes are governed and documented in internal process descriptions and instructions that will be provided for due diligence purposes at the request of interested potential investors. In addition, it should be indicated that whilst processes and definitions can change during the lifetime of the ABS transaction, the originator will undertake not to lower the existing standards. In addition, it is proposed that internal process descriptions and instructions will be provided on an ongoing basis upon request. Internal process instructions typically leave leeway for discretion, so that the discretion issue should be dealt with by reference to internal process instructions as well.

Referring to "provide investors with a liability cash flow model" before pricing and on an ongoing basis, investors should have their own cash flow model. However, originators should provide investors with all the relevant data necessary for building a cash flow model.

Criterion 20

Criterion 20 requires access to data on historical default and loss performance. We agree that such information is available for typical bank originated loans. These data might not be available, however, for corporate customer receivables (e.g. German trade receivables) since the customer’s infrastructure does not usually provide such information. In addition, it should not be overlooked that the loss data for certain asset classes (e.g. German trade receivables) indicate that losses are very low (see also True Sale International 2010, 2014). Furthermore, silent assignment and transfer of the receivables is mandatory for almost all corporates (needless to say only if allowed under the applicable law). Although the data can be anonymised, the level of confidentiality a corporate client intends to maintain would be undermined and the relative value of a securitisation in comparison to loan lending would decline further.

Criterion B

Identifying the group of connected clients can be very challenging in the retail bulk business and can entail high operational costs with little benefit from a risk management perspective. In many cases there are thresholds in place in the retail bulk business to reduce the workload and so strike the right balance between operating costs and the need for identifying single risks in order to ensure proper risk identification and to comply with requirements. Thus, so as to ensure that the identification of the group of connected clients can be based on the existing processes of the originator who has to the identify group of connected clients, the words "to the best knowledge of the originator" should be added.

Criterion C

Criterion C of the credit risk criteria requires that the underlying exposures have to meet certain criteria in terms of their potential CRR credit risk framework risk weight. From our perspective this criterion contradicts the intention of the other EBA criteria on simple standard and transparent securitisation. Irrespective of the potential RWA consumption of the underlying assets, securitisation transactions can be set up as proper securitisation, where other risk than the pure credit risk related to the underlying exposures are properly mitigated. This applies in particular to underlying loans that meet criterion 4, which requires that the loans are consistently originated in the ordinary course of the original lender’s business. Furthermore, we wonder whether the requirement demands that the securitised assets were on a bank’s balance sheet. If so, corporate customer receivables included in a securitisation transaction in which the bank acts solely as a sponsor cannot meet the criterion. It should be clarified that the requirement is met, when the assets would potentially (assuming the Credit Risk Standardised Approach is applied to such assets) assigned to the listed risk weights."
No, in so far as the qualification as SST securitisation" is not coupled with the seniority of the ABS tranche. For this approach can help to avoid sharply rising capital requirements (cliff effects) for junior bonds ranking immediately below senior bonds."
According to the EBA observations the qualifying securitisations had hardly any losses. Therefore there is no need to increase the capital requirements for qualifying securitisations. Based on the proposed minimum risk weight of 15% (see BCBS revision to the securitisation framework) and actual minimum risk weight of 7%, the partition between qualifying and non-qualifying would be 0.5. This partition is to be understood as a minimum and should be adjusted after the holistic review of the regulatory framework for securitisations and other investment products.

The capital requirements should be based on historical data for qualifying and non-qualifying securitisations. This should include segmentation by European and non-European securitisations. As the EBA's empirical findings show, the historical SST securitisation and non-SST securitisation losses differ significantly for the same external rating.

For the IRB approach, in the external rating-based approach and the standard approach as per the Securitisation Framework the parameterisation should be based on historical data such that the capital requirements for SST securitisations are reduced.

The hierarchy in the Securitisation Framework should be retained because—despite all the reservations—the external rating-based approach has the highest risk sensitivity compared with rating agencies under the IRB approach. Furthermore, rating agencies have learned from their past mistakes and significantly improved their rating methodology.

Concerning the standard approach, historical data should be used to determine conservative average parameters for SST and non-SST securitisations and used as the basis for capital requirements.
Re-allocation across tranches has already been proposed by the BCBS. The use of a partition on the non-qualifying securitisation risk weight seems to be a feasible and simple solution.
We would propose, in the case of securitisations meeting the demands on SST securitisations, the capital requirements be determined before the sovereign-ratings cap for countries within the banking union, if this can be justified from the historical data.
Olaf Instinsky