Response to discussion on simple standard and transparent securitisations
Go back
a) Use of synthetic transfer to create an arbitrage or a short position;
b) Use of synthetic transfer to mitigate specific cases of impossible or very cumbersome true sale;
c) Use of synthetic transfer to organize a specific risk transfer, and where usually only a junior or mezzanine tranche is transferred to specialised institutions such as hedge funds.
Clearly, case a) has to be ruled out of SST as it is not linked to the real economy, and a simple criteria to avoid arbitrage is to impose on the originator using synthetic transfer to actually hold the underlying cash assets, and commit to hold these assets for so long as the synthetic transaction is in place.
Case b), for public transactions, should not be ruled out of SST as long as counterparty risk mitigation is properly in place. For example, having all the securitised exposures fully collateralised in cash in a segregated account, or having all the underlying assets pledged to the securitisation SPV should be adequate counterparty risk mitigants, without reducing the simplicity of the structure.
Case c) is a powerful risk transfer tool and is widely used by originating banks to reduce their risk exposure on portfolios of assets. In such transactions, originating banks keep full control of the securitized assets on their balance sheet, and retain some securitisation tranches, including in particular the senior tranche. The main regulatory issue here is the treatment of the retained securitisation tranches held by the originating banks. The lack of true sale is not detrimental at all, as they continue to hold and control the assets.
In short, we believe Synthetic Securitisations should not be excluded from the SST eligibility:
• In case c) hereabove, when the underlying assets are held on the balance sheet of the credit institution that originated the assets and holds residual securitisation positions.
• In case b) hereabove, when the synthetic transfer is accompanied by strong additional counterparty risk mitigation.
See Also Appendix 1.
More details will also been given in answer to question 8 regarding criterion 5(iii) which raises a particular concern.
However, the priority should be the EEA. It is preferable not to jeopardize the implementation of a SST framework in Europe which helps funding European SMEs and the economy, if an extension to other countries means it is more complex to define criteria which include European assets with low loss history and exclude assets classes which have had significant losses and if it would mean delaying the implementation of SST.
Therefore in priority, the framework should be implemented for EEA countries for transactions where :
i) The underlying assets are originated in these countries; or
ii) the acquisition process by the securitisation special purpose entity (SSPE) of the underlying asset are governed by the law of any of these countries; or
iii) The originator or intermediary or the sponsor is established in these countries.
In addition pools of pan European assets should not be excluded (key for trade receivables) as exposures to multiple European countries provides diversification which reduces credit risk.
It is also important to note that some decisions have to be taken by all investors without any distinction of seniority and require approval of each class separately. For example for decisions that affect the economics of the transaction such as the maturity, the interest rate or the principal amount, it would not be justified to allow only the most senior tranche to decide modifications without the approval of the mezzanine and junior tranches.
A timeframe of 1 month after issuance to obtain the main underlying transaction documents seems reasonable. We believe there could be a certification that the Offering Circular / Prospectus represents fairly the legal documents.
Regarding loan-by-loan level data, if the required information is on the final underlying portfolio, it is not possible to communicate it before the closing (and a fortiori before the pricing date), because the underlying assets of the portfolio may change until the closing date. However loan-by-loan level data on a prospective / provisional portfolio could be disclosed to potential investors.
Indeed granularity of debtors is not the only factor contributing to diversification. For pools of corporate loans, it can be also obtained with geographical diversification or industry diversification.
In addition, it is difficult to compare the credit risk of very granular portfolios (mainly retail borrowers) which are by essence measured with statistical approaches, to credit risk of non-granular pools (e.g. commercial real estate, infrastructure projects, aircrafts, corporate loans portfolios, etc.) for which a very detailed credit analysis is thoroughly performed at individual loan level.
Therefore granularity should be considered a prime factor determining credit risk only for specific consumer asset pools such as RMBS, consumer loans etc., which are concentrated in one country. For other asset pools such as corporates and SMEs, granularity is less relevant, and should not be applied.
Only about 40% of the SME securitisations issued in the past 2 years in Europe (rated by Fitch) would pass the top borrower concentration criterion. European trade receivables and lease transactions would also often have a top-borrower concentration above 1%. In the case of trade receivables, the short maturity and specific credit enhancement mechanisms reduce the risk, so that there has been no loss on European securitisation deals. The 1% threshold would be a quasi-insurmountable obstacle to structure efficient risk-sharing transactions on large corporates, and on special asset classes (e.g. infrastructure finance, trade finance, etc.).
We propose to apply a similar approach as proposed in the BCBS document “Revision to the securitisation framework” (published on 11th December 2014): For wholesale assets, pools are deemed granular when the effective number of loans N is equal or above 25 .
For trade receivables, the borrower concentration is a risk only on the part which is in excess of the risk fully funded by the originator. The concentration can be authorized depending on the size of the credit enhancement and the credit quality (rating) of the borrower.
Criterion 1: Regarding Asset Backed Commercial Paper (ABCP), which is currently explicitly excluded from the scope of simple, standard and transparent criteria, we think that EBA should make a clear distinction between the different kinds of ABCP issued in the market, and more precisely take into account that since the crisis, the ABCP market has significantly changed with the disappearance of SIVs and arbitrage ABCP conduits. Today multi-seller ABCP conduits represent 82% of the European ABCP market. Multi-seller ABCP conduits are useful in funding the real economy and are the principal way certain assets, like trade receivables, are securitised, predominantly to finance the working capital of corporates.
Moreover there are two aspects to be taken into account when looking at ABCP conduits: (i) the regulatory treatment for commercial paper holders, but also (ii) the regulatory treatment for the liquidity facility provided by the sponsor bank to the conduit. It is important to define a specific qualifying securitisation framework for the ABCP investors , by taking into account the quality of the liquidity facility provided by the sponsor (fully supporting liquidity facility) and the fact that the CPs are issued by a multi-seller ABCP conduit. Having a clear framework for qualifying ABCP would be really useful for the current MMF reform in Europe, which could consider ABCP securitisations qualified as SST to be eligible investments for MMF.
Regarding the liquidity facility provided by the sponsor bank to the ABCP conduits, it is important in terms of capital charge for the sponsor bank. Such liquidity facilities are generally high-quality senior securitisation tranches, and should benefit from favourable regulatory treatment, in order to keep attractive for banks this well-functioning tool to finance their clients. See also Appendix n° 2.
Criterion 2: The prohibition of cherry picking should not exclude the compliance with eligibility criteria which could increase or reduce the credit risk of the asset pool compared to the overall originator’s portfolio. In particular, there are cases where an originator may want to securitise more risky assets (e.g., LTVs above a threshold) in order to transfer risk, and thereby reduce risk of their portfolios.
The most important element is that the selection is transparent in the respect of the given eligibility criteria and that there is no cherry picking on individual loans, or only to the benefit of the investor (improvement of the asset, for example to improve the quality of the data in the securitized pool).
The securitisations for which the underlying asset pool is “cherry picked” or managed should not be excluded if it is clear in the eligibility criteria that the result of the management / cherry picking process is an improved pool of assets (e.g. replacement of an asset by a better rated asset, exclusion of loans with insufficient information).
Criterion 3: This criterion should not be applicable for synthetic securitisation where there is no assignment of assets and so the true sale is not an issue.See also answer to Q2 and appendix 1.
Criterion 4: Credit exposure to certain SPE should not be excluded from the scope of eligible assets: for example specialised lending transactions defined by the CRR in article 147 (8).
It is useful to specify that the required homogeneity in term of asset type does not imply that the exposure should be on a unique geographical entity (national or regional) or industry. The implementation of the “non-deteriorating underwriting standards” criteria should not imply a limitation of the originator’s ability to change its underwriting standards depending on market and economic conditions. This criterion seems also to be redundant with the retention criteria.
Also, SST should not exclude assets in multi European countries. Trade receivables for example can contain assets of medium size corporates who export in other European countries. There is not a particular risk that should prevent such trade receivables to be SST. Similarly, multi currency assets should not be excluded when the FX risk is managed.
Criterion 5: As already said in answer to Q3, the definition of default should be consistent with the definition of default of article 178 (1), particularly regarding national options (number of days past due for some asset classes). Managing two definitions of default could be burdensome and would not be consistent with the recommendations of the Basel Committee on the harmonization of the definitions as requested in its guidelines (“Principles for effective risk data aggregation and risk reporting”). In addition, late arrears or default should not be excluded from the underlying asset pools for SST if they are not funded by the notes, but they are fully funded by the originator. The definition of default should be consistent with market practices of the underlying asset in the given jurisdiction (e.g. trade receivable to government).
Comments on Criterion 5 iii): We have a particular concern, under Criterion 5 (iii), about the definition of ‘a credit-impaired borrower’ that is clearly problematic. French banks would face an important problem of feasibility on the latter.
A number of jurisdictions do not have the types of credit registers referred to here (e.g. in France), meaning that it would be impossible to check the three-year track record of credit difficulties referred to. Indeed, in France, the situation of a borrower could be known at the time of issuance, but his credit history is not necessarily available to the institutions. As an example, payment incidents are recorded in a database (called FICP and handled by the Banque de France), but they are removed as soon as the incident is cured, as required by French law.
The current proposed credit impairment requirement would exclude in France all consumer loan securitisations (including auto ABS) and RMBS deals, because it will be impossible for French issuers of securitisations to know with certainty if the pool underlying to their issuances does not include this type of exposures (at least it is our interpretation of this criterion iii). If this provision is maintained, it may create level-playing-field issues within the European Union. We urge the EBA to pay close attention to this matter.
We also believe a debt restructuring process should not be an exclusion criterion for non-retail borrowers: if the delinquencies have been cured after debt restructuring, there is no reason to exclude the borrower. For example in the case of commercial real estate, the borrower may have injected equity. Therefore debt restructuring should be a criterion applied to retail consumers only. Moreover, adverse credit history should only be considered on loans. For example, a borrower who is not paying telephone bills because of disputes could be registered as having an adverse credit history. Such borrower should not necessarily be excluded from securitisation pools if his loan-payment record is good.
Criterion 6: This criterion is useful for many types of loans in particular to detect frauds. However it cannot work for short term assets such as trade receivables. It is the same situation as for credit cards receivables which are mentioned by the EBA and for which EBA recommends that they should not be excluded on this basis. Short term assets like trade receivables are likely not to have had any individual payment before they are securitised. Trade receivables should not be excluded if there has not been a payment yet made by the borrower.
Criterion 8: Some securitisation transactions such as some auto loans, revolving credit cards or trade receivables transactions do not have any derivatives for hedging / mitigating interest rate and/or currency risks. In such transactions the interest rate and/or currency risks are covered with other means such as specific reserves, additional or specific credit enhancement / subordination or a level of excess spread high enough (i.e. up to 10%) for rating agencies being comfortable with such interest and/or currency risks hedging / mitigating. Such securitization transactions should not be excluded for the SST label.
Criterion 9: Some assets interest reference such as prime UK mortgages are based on cost of fund. This is similar for ABCP conduits which often reference funding cost of conduits. Interest rates based on cost of fund should be eligible to SST.
Criterion 10: Concerning the Criterion 10 (ii): failure to generate sufficient asset is not necessarily problematic if the bonds issued by the SSPE are repaid with the excess of cash received or if the situation is temporary or for a small amount. This should not necessarily trigger the end of the revolving period.
Credit Cards, Trade Receivables and short term assets (seasonality adjustments) should be removed from criterion 10 (ii). Additionally, when there is automatic adjustment of the liability size (e.g. trade receivables, credit cards), there should not be end of revolving period.
Criterion 11: The sequential amortisation payment should concern only the post-enforcement priority of payment or accelerated amortisation priority of payment.
i) It should be made precise that 11 ii) is for underlying loans assets, not for physical collateral of these assets. For example in the auto and car fleets securitisatons the cars are physical assets that can be sold on secondary market which is liquid.
Criterion 13: This criterion will exclude private transactions which do not have an independent trustee. See also answer to Question 5.
Criteria 15 : This criterion should be applicable only to public securitisation transactions. ABCPs are not subject to the Prospectus directive.
Criteria 17 : For ABCP, this criterion should be limited to the ABCP conduit documentation, and not to the documentation of the underlying transactions.
The Prospectus Directive already ensures that, at issuance, the investors have access to all the information that is necessary to make an informed investment decision. If the criteria 15 is already complied with, in our opinion, it should not be necessary to add the necessity to give to the (potential) investors access to the underlying documentation since (i) the information contained in the offering circulars already contains a summary of the main agreements concluded to set-up the securitisation and (ii) too much information may become counter-productive and become an impediment for the investors to assess correctly the risk.
The underlying documents of securitisation transactions usually include a confidentiality provision which prohibit the communication of the underlying legal documentation except in some limited cases. Therefore, such exceptions usually do not include the communication of all the underlying transaction documents to the investors.
In case the legal documentation does not include such a confidentiality provision, French law on professional secrecy does not allow the communication of all the underlying transaction documents to the investors without the prior agreement of the relevant parties to such contracts.
To allow such communication of all the underlying transaction documents to the investors, it would be necessary when drafting the legal documentation to include a specific provision in this respect or to get the prior agreement of all relevant parties (for the outstanding securitisation programs).
It would be preferable to determine the list of the agreements strictly necessary to allow the (potential) investors to assess the risk of the transaction. Certain agreements may contain sensitive information related to the “know-how” of the arrangers on the structuring and/or on the business, organization and strategy of the originators. Leave the possibility to have access to the underlying documentation should not become an instrument allowing “competitive intelligence”.
Criterion 18: Payment holidays are not necessarily related to delinquency and defaults of underlying debtors. Mortgages commonly allow payment holidays. Payment holidays should be removed from criterion 18. They are already included in “other asset performance remedies”.
A liability cash flow model should be made available to investors, but this could be achieved by a 3rd party supplier instructed by the originator at the issuance of the securities.
For ABCP, no Cash Flow model should be required, as timely payments are simply assured by the liquidity line.
Criterion 19: On current securitisation transactions, an external review of a sample of underlying assets is often performed, but this is not always the case. For example for non-granular pools, or for assets which are acquired on the secondary market and for pools which need to ramp up, external reviews are not necessary. The trustee (or a party on behalf of the SSPE) should ensure that the assets are in compliance with the eligibility criteria.
The external review on the underlying asset can be done prior to issuance, not at issuance. The audit may be performed on a provisional portfolio of assets substantially similar to those of the final securitised portfolio.
Criterion 20 : Historical default and loss performance data are used in structuring the securitisation, in particular by rating agencies. However detailed performance data are currently rarely available to investors and prospective investors prior to a securitisation. If historical data need to be public and available to investors, who could disclose such data ? (for instance commodity financing). This requirement would need to be consistent with the regulatory definition of default under Criterion 5 (ii) i.e. : definition of default for SST securitisation should be the same as per Article 178 of the CRR so that financial institutions will be able to provide this data, otherwise they will have to manage two definitions of default and two set of historical data which is not appropriate (see also answer to Q3).
Depending on the underlying asset type, the requirement on 5 years of historical data could be too difficult to respect and/or not applicable in all the cases / underlying assets. For Trade receivables, it should be 2 years.
Moreover this kind of disclosure requirement is already addressed in Article 8b of the CRA regulation.
Criterion 21: For operational reasons cut-off dates for loan-by-loan data can sometimes differ from those of investor reports. Cut-off dates of loan-by-loan data are often constrained by the originator IT and reporting process. Securitisation investor reports are aligned with the payment dates of the notes which can have a different frequency. The loan-by-loan data should be the latest available before an investor report.
A loan by loan level report is not pertinent for highly granular SME or retail pools that contain far more than 1000 borrowers (more than 100’000 borrowers on auto loans transactions). This information would be of no help for this kind of pool and should rather be aggregated.
Moreover this kind of disclosure requirement is already addressed in Article 8b of the CRA regulation. And ECB eligible ABS are required to publish loan-by-loan information on an ongoing basis once the transaction is live. See also answer to Q6.
Criterion 22: Debt restructuring, debt forgiveness, payment holidays etc. : we expect many issuers to have difficulties to provide that level of detail on a loan-by-loan basis. Payment holidays may be part of the initial contract and in this case they should not be reported as part of the delinquency performance data.
Finally this kind of disclosure requirement is already addressed in Article 8b of the CRA regulation.
Criterion A: It seems that Criterion A is already covered by criterion 4 (ii) : “originated in the ordinary course of the original lender’s business”… In addition the article 18 of the MCD (Directive 2014/17/EU) or article 8 of CCD (Directive 2008/48/EC) are relevant for loans granted to consumer debtors but would be not applicable to all other asset classes (e.g. assets originated by corporates).
Criterion B: See answer to Question 7.
For trade receivables, the borrower concentration is a risk only on the part which is in excess of the risk fully funded by the originator. The concentration can be authorized depending on the size of the credit enhancement and the credit quality (rating) of the borrower.
Criterion C i): We would agree for the EEA jurisdiction (but only with a view to the convergence with ECB eligibility criteria, as OECD countries would make more sense from a credit perspective). Also, this criterion should be applicable at origination date only, as the borrower can change their residence / address.
Criterion C ii):
1- Only the funded loans should meet the risk weight conditions, not the unfunded loans which may be part of the portfolio.
2- For residential mortgages, we recommend a 60% risk weight (Standardised approach) limit to take into account jurisdiction where high LTVs are the standard.
3- The risk weight limits should be adjusted with any change of the local regulatory risk weights.
The credit risk criteria and notably B (concentration) and C should not prevent the securitisation of State guaranteed loans (and especially of residential loans) from being compliant with SST criteria.
Also, assets complying with the simple, Standard and Transparent criteria (pillars I, II, III) should be considered as SST.
• When the credit risk criteria (A,B,C) are also satisfied, the securitisations should benefit from a more appropriate treatment for Capital requirements (lower floor, lower surcharge of securitised assets compared to non-securitised assets (Recommendation 4)).
• In addition, for assets complying with all SST criteria and Credit Risk Criteria except the RWA thresholds of Criterion Cii), the FBF members consider that the senior tranches of such securitisation should also benefit from a lower floor and lower capital surcharge, is these senior tranches have sufficient protection. Senior tranches with credit enhancement above a multiplier of 3 times the capital requirement of the underlying asset under the standardized Approach would provide sufficient protection .
C iii): In cases such as the Dutch mortgage market, LTVs of 100% are standard. We would recommend to add some flexibility to the criterion such as limiting the loans with LTVs higher than 100% to 5 % of the portfolio, or to have a limit of 105% LTV at the time of inclusion.
This may be mitigated by introducing a modular approach with (i) simpler and fewer core criteria to define the SST securitisations, and with (ii) added specific criteria depending on the application (e.g. capital charge, LCR…) and the type of securitisation (with specific criteria for synthetic securitisation or for ABCP, for example).
For the use of the risk criteria (criteria A, B, C) for capital treatment of Securitisations which comply with the SST criteria (pillar I, II, III), we would also recommend to avoid threshold effects that are implicit in the EBA current proposal (all loans should have a RW below x%, granularity above 1%): These thresholds should not be limits which departs how capital is treated. Instead, the threshold effect would be limited by implementing a penalty for the fraction of the portfolio which does not comply (for example, penalty on the fraction above the granularity limit).
The French Banking Federation proposes that:
1- the simple, standard and transparent securitisations would be allowed a different hierarchy of approaches, with a re-calibrated Simplified Supervisory Formula Approach being above the external rating based approach:
o This would reduce reliance on external ratings.
o Additionally, being simple, standard and transparent, the ratio of capital requirements of the securitized capital structure compared to the capital requirements of the underlying asset pool should be much closer to neutrality. The revised Basel 3 framework for securitisation adds a surcharge for the securitized bonds which is very far from neutrality, and this is not consistent with the history of losses of the high quality securitisation that EBA has noted. Therefore the SSFA should be recalibrated for SST closer to capital neutrality assumptions.
2- SST securitisations should be subject to a lower floor (i.e. lower than the new Basel standard of 15%) for senior tranches. We believe a risk-weight floor of 7-10% would be acceptable for senior tranches of qualifying securitisations. Such a level would still be very conservative compared to the loss history of senior tranches of such securitisations in Europe.
To this end we encourage the EBA to contemplate the proposal of Duponcheele, Linden and Perraudin paper of November 2014 (“How to Revive the European Securitisation Market: a Proposal for a European SSFA”) of defining a specific SSFA for ‘qualifying’ securitisation positions: the ‘European SSFA’.
http://www.riskcontrollimited.com/public/How_to_Revive_the_European_Securitisation_Market.pdf
http://www.riskcontrollimited.com/public/Exec_Sum_How_to_Revive_the_European_Securitisation_Market.pdf
Moreover, regarding the leverage ratio, originators should benefit from a deduction (in the denominator) of the non-recourse funding received, without consideration to accounting re-consolidation due to limited risk transfer.
The application of Simple, Standard and Transparent Securitisation should result in lower risk weights and lower floor for the securitisations which comply with the set of criteria. It should not result in higher risk weights and higher floors from current proposals weights for the securitisations which are not deemed SST.
More generally, we agree that the securitisation market has unduly and excessively been impacted by the ratings given by agencies which have changed their methodologies and imposed country ceilings which are arbitrary instead of underlying detailed analysis of risk and recoveries. In this area, there is no difference between qualifying and non-qualifying securitisation.
Question 2: Should synthetic securitisations be excluded from the framework for simple standard and transparent securitisations? If not, under which conditions/criteria could they be considered simple standard and transparent?
The question of synthetic securitisations in relation with the SST label has to be analysed following three distinct angles:a) Use of synthetic transfer to create an arbitrage or a short position;
b) Use of synthetic transfer to mitigate specific cases of impossible or very cumbersome true sale;
c) Use of synthetic transfer to organize a specific risk transfer, and where usually only a junior or mezzanine tranche is transferred to specialised institutions such as hedge funds.
Clearly, case a) has to be ruled out of SST as it is not linked to the real economy, and a simple criteria to avoid arbitrage is to impose on the originator using synthetic transfer to actually hold the underlying cash assets, and commit to hold these assets for so long as the synthetic transaction is in place.
Case b), for public transactions, should not be ruled out of SST as long as counterparty risk mitigation is properly in place. For example, having all the securitised exposures fully collateralised in cash in a segregated account, or having all the underlying assets pledged to the securitisation SPV should be adequate counterparty risk mitigants, without reducing the simplicity of the structure.
Case c) is a powerful risk transfer tool and is widely used by originating banks to reduce their risk exposure on portfolios of assets. In such transactions, originating banks keep full control of the securitized assets on their balance sheet, and retain some securitisation tranches, including in particular the senior tranche. The main regulatory issue here is the treatment of the retained securitisation tranches held by the originating banks. The lack of true sale is not detrimental at all, as they continue to hold and control the assets.
In short, we believe Synthetic Securitisations should not be excluded from the SST eligibility:
• In case c) hereabove, when the underlying assets are held on the balance sheet of the credit institution that originated the assets and holds residual securitisation positions.
• In case b) hereabove, when the synthetic transfer is accompanied by strong additional counterparty risk mitigation.
See Also Appendix 1.
Question 3: Do you believe the default definition proposed under Criterion 5 (ii) above is appropriate? Would the default definition as per Article 178 of the CRR be more appropriate?
The default definition proposed under Criterion 5 (ii) proposes a 90 days past-due threshold. The default definition used in article 178 of the CRR refers to 90 days past-due, but leaves the possibility for competent authorities to replace the 90 days with 180 days for exposures secured by residential or SME commercial estate in the retail exposure class as well as exposures to public sector entities. In most securitisation transactions, the eligibility criteria exclude receivables with missed payments. However, we believe the default definition in article 178 of the CRR should be retained for bank originators: if the regulators in a given jurisdiction have decided to use 180 days past-due for certain asset classes, there is no reason why another definition should be used. The default criteria to define SST Securitisation should be consistent with the practice of the CRR. The case of the Public Sector is a good example. Assets such as trade receivables on EMU governments have a low credit risk, but they may not be paid in 90 days or 180 days, depending on the country, without being considered as a default. Moreover a parallel definition of default (one for securitisation, one for capital requirement purposes) would add some complexity in term of data management, history of default and recovery rate management. Also in the criterion 5, delinquencies and defaults which are clearly fully funded by the originator should not be excluded from the underlying exposures of SST.More details will also been given in answer to question 8 regarding criterion 5(iii) which raises a particular concern.
Question 4: Do you believe that, for the purposes of standardisation, there should be limits imposed on the type of jurisdiction (such as EEA only, EEA and non-EEA G10 countries, etc): i) the underlying assets are originated and/or ii) governing the acquisition process of the SSPE of the underlying assets is regulated and/or iii) where the originator or intermediary (if applicable) is established and/or iv) where the issuer/sponsor is established?
We believe that there should be as little as possible limit on the type of jurisdiction, especially if a proper pledge or charge on the assets can be effected. Therefore countries should not be limited to EEA, and should include countries where assets securitisation is sufficiently developed (US, Australia, Japan, for example).However, the priority should be the EEA. It is preferable not to jeopardize the implementation of a SST framework in Europe which helps funding European SMEs and the economy, if an extension to other countries means it is more complex to define criteria which include European assets with low loss history and exclude assets classes which have had significant losses and if it would mean delaying the implementation of SST.
Therefore in priority, the framework should be implemented for EEA countries for transactions where :
i) The underlying assets are originated in these countries; or
ii) the acquisition process by the securitisation special purpose entity (SSPE) of the underlying asset are governed by the law of any of these countries; or
iii) The originator or intermediary or the sponsor is established in these countries.
In addition pools of pan European assets should not be excluded (key for trade receivables) as exposures to multiple European countries provides diversification which reduces credit risk.
Question 5: Does the distribution of voting rights to the most senior tranches in the securitisation conflict with any national provision? Would this distribution deter investors in non-senior tranches and obstacle the structuring of transactions?
Distribution of voting rights to the most senior tranches is a good practice for SST Securitisation, however this criterion may be an issue for more junior tranches investors if it is required that all voting rights are allocated to the most senior classes. In the current market, securitisations are generally designed to allocate enhanced voting rights to the most senior tranches of credit risk, but certain decisions (e.g. identity of special servicers) are more appropriately allocated to junior tranches investors when they are likely to be more affected than the senior tranches. Removing the control of junior tranches investors over decisions most likely to affect their recovery would certainly lead to reduced demand for those junior tranches.It is also important to note that some decisions have to be taken by all investors without any distinction of seniority and require approval of each class separately. For example for decisions that affect the economics of the transaction such as the maturity, the interest rate or the principal amount, it would not be justified to allow only the most senior tranche to decide modifications without the approval of the mezzanine and junior tranches.
Question 6: Do you believe that, for the purposes of transparency, a specific timing of the disclosure of underlying transaction documentation should be required? Should this documentation be disclosed prior to issuance?
Securitisation contractual documentations could be made available for public disclosure. However, this is not current market standard: : the most important underlying documents can be obtained from the offices of the issuer, but only for public transactions. The timeframe for closing transactions is usually very short, and underlying transaction documents are prepared until the closing date/issuance date. There has not been any difficulty caused by this timing, as the Offering Circular / Prospectus (for public transactions) already provides a detailed description of the transaction and of the main underlying transaction documents. Therefore there is little value to provide all underlying transaction documentations prior to issuance.A timeframe of 1 month after issuance to obtain the main underlying transaction documents seems reasonable. We believe there could be a certification that the Offering Circular / Prospectus represents fairly the legal documents.
Regarding loan-by-loan level data, if the required information is on the final underlying portfolio, it is not possible to communicate it before the closing (and a fortiori before the pricing date), because the underlying assets of the portfolio may change until the closing date. However loan-by-loan level data on a prospective / provisional portfolio could be disclosed to potential investors.
Question 7: Do you agree that granularity is a relevant factor determining the credit risk of the underlying? Does the threshold value proposed under Criterion B pose an obstacle to the structuring of securitisation transactions in any specific asset class? Would another threshold value be more appropriate?
We agree that granularity would be a relevant factor in assessing the credit risk of the underlying portfolio for homogeneous retail pools; however we think that setting a fixed percentage maximum threshold to individual loan exposures is too limiting for the structuring of transactions in asset classes different from the most granular ones of retail assets (consumer or mortgages).Indeed granularity of debtors is not the only factor contributing to diversification. For pools of corporate loans, it can be also obtained with geographical diversification or industry diversification.
In addition, it is difficult to compare the credit risk of very granular portfolios (mainly retail borrowers) which are by essence measured with statistical approaches, to credit risk of non-granular pools (e.g. commercial real estate, infrastructure projects, aircrafts, corporate loans portfolios, etc.) for which a very detailed credit analysis is thoroughly performed at individual loan level.
Therefore granularity should be considered a prime factor determining credit risk only for specific consumer asset pools such as RMBS, consumer loans etc., which are concentrated in one country. For other asset pools such as corporates and SMEs, granularity is less relevant, and should not be applied.
Only about 40% of the SME securitisations issued in the past 2 years in Europe (rated by Fitch) would pass the top borrower concentration criterion. European trade receivables and lease transactions would also often have a top-borrower concentration above 1%. In the case of trade receivables, the short maturity and specific credit enhancement mechanisms reduce the risk, so that there has been no loss on European securitisation deals. The 1% threshold would be a quasi-insurmountable obstacle to structure efficient risk-sharing transactions on large corporates, and on special asset classes (e.g. infrastructure finance, trade finance, etc.).
We propose to apply a similar approach as proposed in the BCBS document “Revision to the securitisation framework” (published on 11th December 2014): For wholesale assets, pools are deemed granular when the effective number of loans N is equal or above 25 .
For trade receivables, the borrower concentration is a risk only on the part which is in excess of the risk fully funded by the originator. The concentration can be authorized depending on the size of the credit enhancement and the credit quality (rating) of the borrower.
Question 8: Do you agree with the proposed criteria defining simple standard and transparent securitisations? Do you agree with the proposed credit risk criteria? Should any other criteria be considered?
Please see below our analysis on some criterion (no comments were made to the criteria not mentioned in the list below):Criterion 1: Regarding Asset Backed Commercial Paper (ABCP), which is currently explicitly excluded from the scope of simple, standard and transparent criteria, we think that EBA should make a clear distinction between the different kinds of ABCP issued in the market, and more precisely take into account that since the crisis, the ABCP market has significantly changed with the disappearance of SIVs and arbitrage ABCP conduits. Today multi-seller ABCP conduits represent 82% of the European ABCP market. Multi-seller ABCP conduits are useful in funding the real economy and are the principal way certain assets, like trade receivables, are securitised, predominantly to finance the working capital of corporates.
Moreover there are two aspects to be taken into account when looking at ABCP conduits: (i) the regulatory treatment for commercial paper holders, but also (ii) the regulatory treatment for the liquidity facility provided by the sponsor bank to the conduit. It is important to define a specific qualifying securitisation framework for the ABCP investors , by taking into account the quality of the liquidity facility provided by the sponsor (fully supporting liquidity facility) and the fact that the CPs are issued by a multi-seller ABCP conduit. Having a clear framework for qualifying ABCP would be really useful for the current MMF reform in Europe, which could consider ABCP securitisations qualified as SST to be eligible investments for MMF.
Regarding the liquidity facility provided by the sponsor bank to the ABCP conduits, it is important in terms of capital charge for the sponsor bank. Such liquidity facilities are generally high-quality senior securitisation tranches, and should benefit from favourable regulatory treatment, in order to keep attractive for banks this well-functioning tool to finance their clients. See also Appendix n° 2.
Criterion 2: The prohibition of cherry picking should not exclude the compliance with eligibility criteria which could increase or reduce the credit risk of the asset pool compared to the overall originator’s portfolio. In particular, there are cases where an originator may want to securitise more risky assets (e.g., LTVs above a threshold) in order to transfer risk, and thereby reduce risk of their portfolios.
The most important element is that the selection is transparent in the respect of the given eligibility criteria and that there is no cherry picking on individual loans, or only to the benefit of the investor (improvement of the asset, for example to improve the quality of the data in the securitized pool).
The securitisations for which the underlying asset pool is “cherry picked” or managed should not be excluded if it is clear in the eligibility criteria that the result of the management / cherry picking process is an improved pool of assets (e.g. replacement of an asset by a better rated asset, exclusion of loans with insufficient information).
Criterion 3: This criterion should not be applicable for synthetic securitisation where there is no assignment of assets and so the true sale is not an issue.See also answer to Q2 and appendix 1.
Criterion 4: Credit exposure to certain SPE should not be excluded from the scope of eligible assets: for example specialised lending transactions defined by the CRR in article 147 (8).
It is useful to specify that the required homogeneity in term of asset type does not imply that the exposure should be on a unique geographical entity (national or regional) or industry. The implementation of the “non-deteriorating underwriting standards” criteria should not imply a limitation of the originator’s ability to change its underwriting standards depending on market and economic conditions. This criterion seems also to be redundant with the retention criteria.
Also, SST should not exclude assets in multi European countries. Trade receivables for example can contain assets of medium size corporates who export in other European countries. There is not a particular risk that should prevent such trade receivables to be SST. Similarly, multi currency assets should not be excluded when the FX risk is managed.
Criterion 5: As already said in answer to Q3, the definition of default should be consistent with the definition of default of article 178 (1), particularly regarding national options (number of days past due for some asset classes). Managing two definitions of default could be burdensome and would not be consistent with the recommendations of the Basel Committee on the harmonization of the definitions as requested in its guidelines (“Principles for effective risk data aggregation and risk reporting”). In addition, late arrears or default should not be excluded from the underlying asset pools for SST if they are not funded by the notes, but they are fully funded by the originator. The definition of default should be consistent with market practices of the underlying asset in the given jurisdiction (e.g. trade receivable to government).
Comments on Criterion 5 iii): We have a particular concern, under Criterion 5 (iii), about the definition of ‘a credit-impaired borrower’ that is clearly problematic. French banks would face an important problem of feasibility on the latter.
A number of jurisdictions do not have the types of credit registers referred to here (e.g. in France), meaning that it would be impossible to check the three-year track record of credit difficulties referred to. Indeed, in France, the situation of a borrower could be known at the time of issuance, but his credit history is not necessarily available to the institutions. As an example, payment incidents are recorded in a database (called FICP and handled by the Banque de France), but they are removed as soon as the incident is cured, as required by French law.
The current proposed credit impairment requirement would exclude in France all consumer loan securitisations (including auto ABS) and RMBS deals, because it will be impossible for French issuers of securitisations to know with certainty if the pool underlying to their issuances does not include this type of exposures (at least it is our interpretation of this criterion iii). If this provision is maintained, it may create level-playing-field issues within the European Union. We urge the EBA to pay close attention to this matter.
We also believe a debt restructuring process should not be an exclusion criterion for non-retail borrowers: if the delinquencies have been cured after debt restructuring, there is no reason to exclude the borrower. For example in the case of commercial real estate, the borrower may have injected equity. Therefore debt restructuring should be a criterion applied to retail consumers only. Moreover, adverse credit history should only be considered on loans. For example, a borrower who is not paying telephone bills because of disputes could be registered as having an adverse credit history. Such borrower should not necessarily be excluded from securitisation pools if his loan-payment record is good.
Criterion 6: This criterion is useful for many types of loans in particular to detect frauds. However it cannot work for short term assets such as trade receivables. It is the same situation as for credit cards receivables which are mentioned by the EBA and for which EBA recommends that they should not be excluded on this basis. Short term assets like trade receivables are likely not to have had any individual payment before they are securitised. Trade receivables should not be excluded if there has not been a payment yet made by the borrower.
Criterion 8: Some securitisation transactions such as some auto loans, revolving credit cards or trade receivables transactions do not have any derivatives for hedging / mitigating interest rate and/or currency risks. In such transactions the interest rate and/or currency risks are covered with other means such as specific reserves, additional or specific credit enhancement / subordination or a level of excess spread high enough (i.e. up to 10%) for rating agencies being comfortable with such interest and/or currency risks hedging / mitigating. Such securitization transactions should not be excluded for the SST label.
Criterion 9: Some assets interest reference such as prime UK mortgages are based on cost of fund. This is similar for ABCP conduits which often reference funding cost of conduits. Interest rates based on cost of fund should be eligible to SST.
Criterion 10: Concerning the Criterion 10 (ii): failure to generate sufficient asset is not necessarily problematic if the bonds issued by the SSPE are repaid with the excess of cash received or if the situation is temporary or for a small amount. This should not necessarily trigger the end of the revolving period.
Credit Cards, Trade Receivables and short term assets (seasonality adjustments) should be removed from criterion 10 (ii). Additionally, when there is automatic adjustment of the liability size (e.g. trade receivables, credit cards), there should not be end of revolving period.
Criterion 11: The sequential amortisation payment should concern only the post-enforcement priority of payment or accelerated amortisation priority of payment.
i) It should be made precise that 11 ii) is for underlying loans assets, not for physical collateral of these assets. For example in the auto and car fleets securitisatons the cars are physical assets that can be sold on secondary market which is liquid.
Criterion 13: This criterion will exclude private transactions which do not have an independent trustee. See also answer to Question 5.
Criteria 15 : This criterion should be applicable only to public securitisation transactions. ABCPs are not subject to the Prospectus directive.
Criteria 17 : For ABCP, this criterion should be limited to the ABCP conduit documentation, and not to the documentation of the underlying transactions.
The Prospectus Directive already ensures that, at issuance, the investors have access to all the information that is necessary to make an informed investment decision. If the criteria 15 is already complied with, in our opinion, it should not be necessary to add the necessity to give to the (potential) investors access to the underlying documentation since (i) the information contained in the offering circulars already contains a summary of the main agreements concluded to set-up the securitisation and (ii) too much information may become counter-productive and become an impediment for the investors to assess correctly the risk.
The underlying documents of securitisation transactions usually include a confidentiality provision which prohibit the communication of the underlying legal documentation except in some limited cases. Therefore, such exceptions usually do not include the communication of all the underlying transaction documents to the investors.
In case the legal documentation does not include such a confidentiality provision, French law on professional secrecy does not allow the communication of all the underlying transaction documents to the investors without the prior agreement of the relevant parties to such contracts.
To allow such communication of all the underlying transaction documents to the investors, it would be necessary when drafting the legal documentation to include a specific provision in this respect or to get the prior agreement of all relevant parties (for the outstanding securitisation programs).
It would be preferable to determine the list of the agreements strictly necessary to allow the (potential) investors to assess the risk of the transaction. Certain agreements may contain sensitive information related to the “know-how” of the arrangers on the structuring and/or on the business, organization and strategy of the originators. Leave the possibility to have access to the underlying documentation should not become an instrument allowing “competitive intelligence”.
Criterion 18: Payment holidays are not necessarily related to delinquency and defaults of underlying debtors. Mortgages commonly allow payment holidays. Payment holidays should be removed from criterion 18. They are already included in “other asset performance remedies”.
A liability cash flow model should be made available to investors, but this could be achieved by a 3rd party supplier instructed by the originator at the issuance of the securities.
For ABCP, no Cash Flow model should be required, as timely payments are simply assured by the liquidity line.
Criterion 19: On current securitisation transactions, an external review of a sample of underlying assets is often performed, but this is not always the case. For example for non-granular pools, or for assets which are acquired on the secondary market and for pools which need to ramp up, external reviews are not necessary. The trustee (or a party on behalf of the SSPE) should ensure that the assets are in compliance with the eligibility criteria.
The external review on the underlying asset can be done prior to issuance, not at issuance. The audit may be performed on a provisional portfolio of assets substantially similar to those of the final securitised portfolio.
Criterion 20 : Historical default and loss performance data are used in structuring the securitisation, in particular by rating agencies. However detailed performance data are currently rarely available to investors and prospective investors prior to a securitisation. If historical data need to be public and available to investors, who could disclose such data ? (for instance commodity financing). This requirement would need to be consistent with the regulatory definition of default under Criterion 5 (ii) i.e. : definition of default for SST securitisation should be the same as per Article 178 of the CRR so that financial institutions will be able to provide this data, otherwise they will have to manage two definitions of default and two set of historical data which is not appropriate (see also answer to Q3).
Depending on the underlying asset type, the requirement on 5 years of historical data could be too difficult to respect and/or not applicable in all the cases / underlying assets. For Trade receivables, it should be 2 years.
Moreover this kind of disclosure requirement is already addressed in Article 8b of the CRA regulation.
Criterion 21: For operational reasons cut-off dates for loan-by-loan data can sometimes differ from those of investor reports. Cut-off dates of loan-by-loan data are often constrained by the originator IT and reporting process. Securitisation investor reports are aligned with the payment dates of the notes which can have a different frequency. The loan-by-loan data should be the latest available before an investor report.
A loan by loan level report is not pertinent for highly granular SME or retail pools that contain far more than 1000 borrowers (more than 100’000 borrowers on auto loans transactions). This information would be of no help for this kind of pool and should rather be aggregated.
Moreover this kind of disclosure requirement is already addressed in Article 8b of the CRA regulation. And ECB eligible ABS are required to publish loan-by-loan information on an ongoing basis once the transaction is live. See also answer to Q6.
Criterion 22: Debt restructuring, debt forgiveness, payment holidays etc. : we expect many issuers to have difficulties to provide that level of detail on a loan-by-loan basis. Payment holidays may be part of the initial contract and in this case they should not be reported as part of the delinquency performance data.
Finally this kind of disclosure requirement is already addressed in Article 8b of the CRA regulation.
Criterion A: It seems that Criterion A is already covered by criterion 4 (ii) : “originated in the ordinary course of the original lender’s business”… In addition the article 18 of the MCD (Directive 2014/17/EU) or article 8 of CCD (Directive 2008/48/EC) are relevant for loans granted to consumer debtors but would be not applicable to all other asset classes (e.g. assets originated by corporates).
Criterion B: See answer to Question 7.
For trade receivables, the borrower concentration is a risk only on the part which is in excess of the risk fully funded by the originator. The concentration can be authorized depending on the size of the credit enhancement and the credit quality (rating) of the borrower.
Criterion C i): We would agree for the EEA jurisdiction (but only with a view to the convergence with ECB eligibility criteria, as OECD countries would make more sense from a credit perspective). Also, this criterion should be applicable at origination date only, as the borrower can change their residence / address.
Criterion C ii):
1- Only the funded loans should meet the risk weight conditions, not the unfunded loans which may be part of the portfolio.
2- For residential mortgages, we recommend a 60% risk weight (Standardised approach) limit to take into account jurisdiction where high LTVs are the standard.
3- The risk weight limits should be adjusted with any change of the local regulatory risk weights.
The credit risk criteria and notably B (concentration) and C should not prevent the securitisation of State guaranteed loans (and especially of residential loans) from being compliant with SST criteria.
Also, assets complying with the simple, Standard and Transparent criteria (pillars I, II, III) should be considered as SST.
• When the credit risk criteria (A,B,C) are also satisfied, the securitisations should benefit from a more appropriate treatment for Capital requirements (lower floor, lower surcharge of securitised assets compared to non-securitised assets (Recommendation 4)).
• In addition, for assets complying with all SST criteria and Credit Risk Criteria except the RWA thresholds of Criterion Cii), the FBF members consider that the senior tranches of such securitisation should also benefit from a lower floor and lower capital surcharge, is these senior tranches have sufficient protection. Senior tranches with credit enhancement above a multiplier of 3 times the capital requirement of the underlying asset under the standardized Approach would provide sufficient protection .
C iii): In cases such as the Dutch mortgage market, LTVs of 100% are standard. We would recommend to add some flexibility to the criterion such as limiting the loans with LTVs higher than 100% to 5 % of the portfolio, or to have a limit of 105% LTV at the time of inclusion.
Question 9: Do you envisage any potential adverse market consequences of introducing a qualifying securitisation framework for regulatory purposes?
Implementing a two-tier framework carries the risk that the market for non-qualifying securitisations would collapse, considering the proposed capital treatment for these assets. With the detailed criteria in the current proposal, the proportion of the non qualifying securitisation could potentially be huge.This may be mitigated by introducing a modular approach with (i) simpler and fewer core criteria to define the SST securitisations, and with (ii) added specific criteria depending on the application (e.g. capital charge, LCR…) and the type of securitisation (with specific criteria for synthetic securitisation or for ABCP, for example).
For the use of the risk criteria (criteria A, B, C) for capital treatment of Securitisations which comply with the SST criteria (pillar I, II, III), we would also recommend to avoid threshold effects that are implicit in the EBA current proposal (all loans should have a RW below x%, granularity above 1%): These thresholds should not be limits which departs how capital is treated. Instead, the threshold effect would be limited by implementing a penalty for the fraction of the portfolio which does not comply (for example, penalty on the fraction above the granularity limit).
Question 10: How should capital requirements reflect the partition between qualifying and non-qualifying?
Generally speaking, and since the consultative document bases a part of its analysis of SST on credit risk, we think that ‘qualifying’ securitisations’ should benefit from a better capital charge framework than the proposed Basel framework.The French Banking Federation proposes that:
1- the simple, standard and transparent securitisations would be allowed a different hierarchy of approaches, with a re-calibrated Simplified Supervisory Formula Approach being above the external rating based approach:
o This would reduce reliance on external ratings.
o Additionally, being simple, standard and transparent, the ratio of capital requirements of the securitized capital structure compared to the capital requirements of the underlying asset pool should be much closer to neutrality. The revised Basel 3 framework for securitisation adds a surcharge for the securitized bonds which is very far from neutrality, and this is not consistent with the history of losses of the high quality securitisation that EBA has noted. Therefore the SSFA should be recalibrated for SST closer to capital neutrality assumptions.
2- SST securitisations should be subject to a lower floor (i.e. lower than the new Basel standard of 15%) for senior tranches. We believe a risk-weight floor of 7-10% would be acceptable for senior tranches of qualifying securitisations. Such a level would still be very conservative compared to the loss history of senior tranches of such securitisations in Europe.
To this end we encourage the EBA to contemplate the proposal of Duponcheele, Linden and Perraudin paper of November 2014 (“How to Revive the European Securitisation Market: a Proposal for a European SSFA”) of defining a specific SSFA for ‘qualifying’ securitisation positions: the ‘European SSFA’.
http://www.riskcontrollimited.com/public/How_to_Revive_the_European_Securitisation_Market.pdf
http://www.riskcontrollimited.com/public/Exec_Sum_How_to_Revive_the_European_Securitisation_Market.pdf
Moreover, regarding the leverage ratio, originators should benefit from a deduction (in the denominator) of the non-recourse funding received, without consideration to accounting re-consolidation due to limited risk transfer.
Question 11: What is a reasonable calibration across tranches and credit quality steps for qualifying securitisations? Would re-allocating across tranches the overall capital applicable to a given transaction by reducing the requirement for the more junior tranche and increasing it for the more senior tranches other than the most senior tranche be a feasible solution?
Please see also the proposal of Duponcheele, Linden, and Perraudin paper of November 2014 (“How to Revive the European Securitisation Market: a Proposal for a European SSFA”).The application of Simple, Standard and Transparent Securitisation should result in lower risk weights and lower floor for the securitisations which comply with the set of criteria. It should not result in higher risk weights and higher floors from current proposals weights for the securitisations which are not deemed SST.
Question 12: Considering that rating ceilings affect securitisations from certain countries, how should the calibration of capital requirements on qualifying and non-qualifying securitisations be undertaken, while also addressing this issue?
Rating ceilings impact the ratings of the tranches, especially the more senior tranches, i.e. only when using the external rating based approach to capital (ERBA). We believe if a particular asset class is impacted by the general economic environment which also triggers the sovereign rating downgrade, there should be a downgrade of the asset without using a cap. The rating agencies should publish the uncapped ratings, because it would allow investors and regulators to distinguish between transactions structured at the level of the sovereign cap and those structured to AAA level but rated lower because of the sovereign cap. We think that the latter should attract a lower capital charge. We believe there should not be a difference between the securitised assets and the same asset class non-securitised.More generally, we agree that the securitisation market has unduly and excessively been impacted by the ratings given by agencies which have changed their methodologies and imposed country ceilings which are arbitrary instead of underlying detailed analysis of risk and recoveries. In this area, there is no difference between qualifying and non-qualifying securitisation.