Since a considerable share of SRT transactions is not public, it is very difficult to get a feeling of the actual size and distribution over asset classes etc. Nevertheless, based on our knowledge of the Dutch market, the data provided do not seem to be unrealistic in the context of the current shape of the European securitisation market in general.
Given the (inevitably) undisclosed nature of practices around SRT discussions with supervisors, there are not many sources outside the EBA questionnaire that could reveal these practices. We are however aware of the fact that some of our members are subject to far more detailed information requirements from their competent authorities than customary for SRT applications at other competent authorities. We are not against detailed information being shared, but the burden of information should be a function of the complexity of the transaction and not of the location of the competent authority.
Standardisation of securitisation processes is one of the key objectives of the Dutch Securitisation Association and against that background we welcome the proposals of EBA on the SRT assessment process. We do however note that the provision of informal early feedback by supervisors, as strongly desired by originators (as per section 70), unfortunately did not make it to your recommendations.
We support the proposal of several other market participants to require competent authorities to react within 30 days when notified, but would like to link the notification to (30 days before)
actually achieving SRT (f.i. by selling the relevant notes in a traditional securitisation) rather than the closing of the transaction.
Apart from the obvious ones (new CRR articles, new (homogeneous) types of collateral, additional inputs for new tests) we envisage no further template changes.
Again, we are strongly supporting standardisation. However, given the diverse nature of SRT transactions, a template should not be overly prescriptive. In that respect we refer to the ECB template of 24 March 2016, that seems to strike the right balance between detail and flexibility. Given the different characteristics of true sale and synthetic transactions, different templates for the two mechanisms would be appropriate.
The notification template should provide some kind of level playing field, where currently competent authorities in different jurisdictions have widely diverging requirements w.r.t. the information to be provided as part of the notification, as also indicated in our answer on Question 2.
Frequent monitoring is important but should not be overly burdensome and should not lead to re-assessment of the SRT status but just confirm that no material structural changes to the transaction have been made.
So we would suggest to limit the reporting to material/relevant deviations from the original application; any relevant quantitative tests will have to be met at inception of the transaction only and do not have to be monitored during the life of a transaction. The fact that losses materialise does not mean that that risk was never initially transferred and that SRT treatment should be revisited by recalculating the quantitative tests.
For our (dis)agreement with the assessments we refer to our answers on questions 8-16.
W.r.t. 3.2.7 Credit Events, which is not covered in the follow up questions, we and many others in the industry, note the discrepancies between CRR art. 178 as incorporated in your proposal, and the market standard of CRR art. 215/216.
To the best of our knowledge the list of structural features with SRT relevance is close to exhaustive. One missing element to be further elaborated on, is the maturity to be matched by the credit protection, especially for “until further notice” exposures (like overdrafts).
We agree with the proposal to include clearly specified contractual triggers as a condition for including (elements of / potential) pro-rata amortisation in SRT transactions and we also agree with the suggestion to have the trigger values reviewed by the competent authorities.
However, we would prefer to have also the actual definitions of triggers reviewed by the competent authorities rather than being determined by the 4 tests defined a as minimum in the EBA proposal. The 4 tests are not all equally valid or not always properly worded to
cover all kinds of potential SRT transactions (f.i. the granularity test is not always required, depending on the nature of the assets).
In this respect it could also be considered to reduce the number of tests, f.i. by combining trigger I and II by including non-matured defaults in the definition of cumulative losses in I.
Furthermore we would appreciate to receive confirmation that the proposal also allows for a choice of pro-rata or sequential amortization per tranche (again subject to review by the competent authorities) in order to meet specific preferences of investors in this respect.
We do understand the proposed safeguards against the possible scenario that an originator institution can use a time call to reassume credit risk that is still equal (or higher) than the risk transferred at the start of a transaction. However, in a situation where the credit risk is considerably lower than originally anticipated, it should be possible to call the transaction well before the end of the WAL of the securitised exposures.
Obviously, when exercising the call, compliance with regard to Implicit Support and Step-in Risk guidelines should be demonstrated to the competent authorities.
Also we would suggest that the exercise of the time call should follow the WAL of the securitised portfolio, in line with current practice, and not to add the replenishment period to the non-call period, which would seriously limit the flexibility of SRT transactions.
Furthermore we still fail to see why the regime for true sale transactions in this respect should be different from synthetic transactions, at least in a situation as described above (reduced credit risk, no implicit support).
We do agree with the proposed safeguards (not guaranteed or fixed, clearly defined also in terms of place in the waterfall).
In more general terms, we do question the way the proposal treats future/unearned excess spread, and in that respect we refer to our answer on Question 12 (below) on the use of excess spread in synthetic securitisation.
a. We do agree with a pre-determined calculation method for excess spread in synthetic transactions, but not with a pre-determined fixed amount or percentage.
b. We miss the rationale for not allowing the use-it-or lose-it mechanism, where according
to section 221 “the trapped allocation mechanism effectively counteracts SRT to a larger
extent than the use-it-or-lose-it allocation mechanism”.
We do not agree with the way the excess spread commitment is treated in the quantitative assessments of SRT.
In line with the CRR, we agree that realized/earned excess spread (one year EL in a synthetic transaction) should be regarded as a securitisation position. Future/unearned excess spread should however not be a securitisation position. The EBA proposal:
-conflicts with the treatment of future excess spread for loan portfolios on bank’s books.
-leads to double counting of risk and consequently excessive capital requirements.
As regards the “Commensurate test”, please see our comments on Questions 23 and 24.
a. We agree that future/unearned excess spread and future (back-loaded) losses are taken into account in the self-assessment. Additional regulatory capital penalties by making future/unearned excess spread subject to own funds requirements are undesirable (see our answer on Question 12) and not needed, and
b. Those regulatory capital penalties would seriously hurt the economics of SRT transactions and further reduce the number of such transactions
Being part of the self-assessment should be a sufficient safeguard.
Our comments (see answer on Question 12) whether (earned/unearned) excess spread qualifies as a securitisation position or not should be equally applicable to the full deduction option.
In our view, investors should at least have the option to demand this clause and it should not be banned, but we also welcome developments around the BRRD framework that ultimately may dis-apply the application of the clause.
Subject to our earlier comments on amortisation, excess spread and time calls, we agree with the principle of a self-assessment. Some further clarification (f.i. on how to develop scenario’s for back-ended losses) would be welcome. Also we would welcome some more guidance on the use of EU-wide stress test results where the maturity of the SRT transactions will usually differ from the maturity applied in the stress test.
No, we are not aware of any of those transactions.
The test is overly robust. We expect that most transactions will (have to) rely on the mezzanine test. Furthermore we refer to the comments of AFME and the EBF on the technical flaws of / inconsistencies between the tests. A combined thickness test for protection tranches may be a way to get around these issues.
In principle, we do agree with the first loss requirement for transactions subject to the mezzanine test. We would however prefer a requirement that allows more flexibility to sell some first loss also through the mezzanine tranche.
No, the current mezzanine test in combination with the first loss requirement (but subject to our comment on flexibility in our answer on Question 20) should be sufficient.
To the extent we expect an impact on tranche thickness in transactions of Dutch originators, it should be a thickening of the sold tranches to compensate for the higher risk weights of the retained tranches.
We have a number of issues with the Option 1 test:
-our earlier comments on earned versus lifetime excess spread do apply again;
-additionally we wonder whether point-in time (ratio 1) and lifetime (ratio 2) can be
compared in the way proposed;
-furthermore, we see an inconsistency in treating the lifetime EL as transferred in the
commensurateness test for transactions that are designed to meet the mezzanine test
(where the EL is retained as part of the first loss).
-and we wonder why (in the case studies of Annex 4) the UL of transferred positions is not
calculated with the Supervisory Formula (with in the denominator also including the UL of all
securitized positions rather than the underlying portfolio, to safeguard consistency).
-the one year excess spread only materializes after one year, so should not be part of the
point-in-time calculation of this test;
-UL should be further defined (as reg. UL);
-as regards the 50% threshold, we notice your statement (in section 231) that “the test is
generally more severe compared to the existing quantitative tests”. The 50% is justified in
section 225 by the fact that this is in line with the policy objective for significant and
commensurate. However in section 205, in the context of first loss tranche thickness, it is
stated that 80% of UL has to be transferred. We would prefer to see the results of a
quantitative impact study before commenting on the proper threshold.
No, we see no rationale for differentiating between asset classes.
As regards (non-)STS classification, we are strongly opposed to using the STS qualification for making any differentiation in regulatory treatment other than capital treatment.
The unjustified use of STS in the EMIR review has demonstrated the unfortunate consequences of this misconception.
No, given the undisclosed nature of the mainly private transactions in our jurisdiction, for competition reasons originators do not want this information to be disclosed through the DSA, but they may be able to provide this information directly to you.
We generally agree with the market assessment and the material aspects mentioned.
However the DSA members do not claim a deep knowledge of this market.
We would like to emphasize transparency and availability of loan level data as a pre-condition for a well functioning (NPL) securitisation market.
Subject to our comments on the EBA proposals on selected structural features, they should generally be valid for NPL securitisation transactions as well.
We do agree with the proposed way of implementing both methods in this specific case. It should also be extended to the use of non-refundable purchase price mechanisms in general (NPL or not).
On the adverse outcomes in case of the SEC-SA we subscribe to the detailed comments of AFME and EBF in their reactions on the proposals (not to be repeated here).
To the extent the “first loss tranche” as per the new CRR covers the non-refundable purchase price discount, the tests should work.
Apart from our comments on this test (Question 23), the non-refundable purchase price discount should be properly reflected.
Apart from our comments on this test (Question 24), the non-refundable purchase price discount should be properly reflected.