Primary tabs


Overall, we see this draft ITS to fit for the purpose of the underlying directive, but only consider it complete once the missing / misallocated features have been corrected. The ITS data template in its current version alone would have the opposite effect of bringing transparency to the market. Instead, the ITS template would require auxiliary data tapes to fulfil the buyers needs. This would contravene the idea of harmonized data templates.
a. We deem the ITS and in particular the template a very good foundation to a common and unified data template for NPL sales. We appreciate EBA’s approach as we at the same time also unify our own data structures to establish new standards in pricing. We have observed in recent years that sellers’ NPL data tapes have become more granular and with a higher quality, structure and relationship between the data tape contents. However, we also see a tightening of deadlines and that it depends very much on the single seller how these data tapes are structured and provided. Moreover, we have seen some very rough data tapes where we would have wished for more.
b. For the purpose of a proper valuation of the target portfolios, we usually exchange own lists regarding data requirements. These are of course only suggestions, as the seller currently is free to decide what to deliver in the final stage. In a comparison to the EBA ITS data template, we found that only around half of all our current requirements are fully covered by the ITS template. For a number of data fields on our requirements list, the EBA template is not detailed enough. Also, we found that the EBA template contains details on information that we deem not essential, but as good auxiliary information that can potentially help to understand specific cases.
c. Our main concerns with the data template in its current form are on the technical side. With the current version of the template, we would not be able to build a proper waterfall of anticipated cash to be received through the lien ranks of a real estate collateral that would be expected in a foreclosure act or collateral sale. Also, the legal status of the specific case is not properly reflected. This information is essential to realistically price a portfolio of NPLs. Not providing such information would lead to significantly less accurate pricing.
The issues seem to arise from the wish to reduce complexity by reducing the number of parameter fields compared to the previous version of the template. It also seems to arise from a different perspective that the constructors of the new template had. The first point seems to build a controversy between the underlying directive and the ITS template, as the intention should always be to reduce information asymmetry between the seller and the buyer side. Instead, the last reduction of fields also reduced details of crucial information for the buyers. As a result, the data template is lacking details of the (loan) to (real-estate collateral) interaction when it comes to a more than 1:1 relationship. Moreover, the legal status of only counterparty and collateral are concerned, but not of the loan itself.
The payment information of template 5 make it seem as if the NPLs subject to this directive were well predictable: The collection period after cut-off date (COD) is much better reflected than the period before. In fact, most NPLs for sale have fallen out of the predictable scheme and are sold for that reason. Thus, from an NPL buyer’s point of view, the perspective should be flipped and much more attention should be given to the recent collection history before the COD. This is one of the most essential information (especially for unsecured NPLs) that NPL buyers base their pricing on.
d. The choice of mandatory fields is well done for many parameters. Regarding the principle of proportionality, we strongly disagree with the threshold being at 25k EUR on single-loan level. Please also see our comment under question 11. below.
Template 1 provides a comprehensive overview of counterparties involved. This is more than the information usually provided but is helpful in forecasting.
a. We disagree with the usage of the proportionality threshold of 25k EUR and refer to our answers to questions 11 and 12.
b. From the data template, it is not clear what field 1,43 “Date of Obtaining Order for Possession” should refer to. Assuming that this refers to the transfer of material assets like a real estate collateral, the counterparty could potentially be a third party, as well as the selling institution itself. In the latter case, we understand it would refer to a REO (Real Estate Owned object), that would be described in the Collaterals template 4 and that would be for sale with the portfolio. We would foster that understanding very much as REOs can easily become the content of an NPL sale through a foreclosure during the period the portfolio is on the market. It should therefore not be excluded from sale. This needs to be described more clearly so that buyers will be able to realistically price the portfolio.
Furthermore, we suggest to add columns for the following data fields:
• National identification number (social security number) (non-mandatory)
• Date of death (non-mandatory)
• Address status (e.g. valid, unknown or moved abroad) (non-mandatory)
• Landline number (non-mandatory)
• Mobile number (non-mandatory)
• Email address (non-mandatory)
• Share of creditors bankruptcy claim (non-mandatory)
We agree with a large number of data fields of template 3. However, key fields as separate Principal and Interest amounts are missing. Instead, a combined outstanding nominal amount (field 3,12) is provided which impedes accurate pricing by buyers.
a. The separation of principal amount and collateral amounts at cut-off date (other cost positions as well as the total accrued interests (unpaid interests pre and past due, also including those that are requested in field 3,14 “Accrued Interests” that are not set due at COD to our understanding) is important for pricing and is in some legislations a key requirement for sending the cases to the legal stage. Debt sellers and buyers in countries like Denmark have recently faced the issue of improper interest calculation. A thorough due diligence to investigate this issue, however, requires the separate provision of principal, costs and interest amounts.
Furthermore, based on Art. 10 (2) (g) of the NPL Directive, it is mandatory to inform the borrower of the outstanding owed amount split into “capital, interests, fees or other permitted charges”. Since the seller will have to provide this information in any case, it should already be part of the data template.
b. In the various templates, reference to legal actions is only on counterparty and collateral (execution) level, not on the level of the loan itself. For the pricing of a loan, it is important if a legal title was obtained for the loan, what stage it is at and at what date and for which amount it was obtained. This information should be mandatory input for all loans. This loan-level information is key for pricing and supplements the legal information of counterparties and collaterals and – only combined – gives a full picture of the legal situation and collectability of a loan.
During the life of a loan, there could have been many forbearance measures and many more to come that are already contractually agreed. From our understanding, template 3 is only requesting input on the current forbearance measure.
We do not find this sufficient as there are forbearance situations like the French Banque de France’s (BDF) procedure where potentially (a) grace period/payment moratorium, (b) interest rate reduction, (d) rescheduled payments and (e) debt forgiveness may be agreed between the BDF and the debtor. These measures act combined, are all of importance and potentially fall apart if the debtor doesn´t comply with the requirements of the plan. From our view, these situations are not reflected by the options in the fields provided in template 3. We recommend extending this by separate columns in the data template and to also give start and end dates for all known current and future forbearance measures according to a forbearance plan.
c. Furthermore, we do not agree with using the principle of proportionality based on the loan amount, as is the case in many fields of template 3. It should be mandatory to provide the same information for smaller loans as for larger, especially if a forbearance was granted. To our understanding, this information should be available in the sellers’ systems. For instance, the amount of debt forgiveness (field 3,45) should be mandatory even for loans < 25k EUR, especially if it can be revoked. Also, the filling of joint counterparties on a loan, other balances (including other charges, commissions, fees, …), interest rate, charge-off date and portion of syndicated loan that is held by the institution should be provided (mandatorily) regardless of the loan size. All these are relevant for pricing.
Consolidated, we are missing data fields that we normally request for proper due diligence:
• Contract date (non-mandatory)
• Termination date (mandatory)
• Current Principal amount (mandatory)
• Current Interest amount (mandatory)
• Total payments since default (mandatory)
• Number of external agencies that have worked the loan (mandatory)
• Date agencies started working on the loan (non-mandatory)
• Date last agency stopped working on the loan (date when the case retuned from the last agency) (non-mandatory)
• Maximum percentage of discount that was offered (non-mandatory)
• Loan is time-barred Flag (Yes/no) (mandatory)
• Legal status of the loan (e.g. sent to court, judgement, execution) (mandatory)
• Date of legal actions. E.g. for Poland, dates of case sent to court, legal title received, date of enforcement clause received (enforceable document mandatory for sending the case to the bailiff), bailiff start and bailiff end date (mandatory)
• Date of last change in legal status (non-mandatory)
• Date after which legal collection is not possible, if there are no further payments or activities (mandatory)
• All levels of a forbearance act including dates of start and end as well as amounts where applicable (mandatory)
• Real estate collateral economic lien rank (if not covered in modified template 4, see comments to that template) (mandatory)
• Real estate collateral legal lien rank (if not covered in modified template 4, see comments to that template) (mandatory)
We agree with the template in many fields, but also have found some caveats that need to be considered and fixed before the final template will be sent to the EU commission.
a. Our understanding from the data glossary and the ITS documents is that each line of template 4 reflects a single real estate collateral or other guarantee.
For each immovable collateral, only the highest lien rank in favor of the seller is given as well as the cumulated amount owed to the seller and to others before the seller.
From the template however, it is not clear which lien ranks other than the highest noted above are in favor of the seller.
Moreover, it is not clear which loan belongs to which lien rank.
These two points are crucial information when it comes to calculating the waterfall of cash received from the lien ranks in case the real estate is being sold. This information is thus very important to be able to accurately price an NPL portfolio.
One way to overcome this issue is to allow for one line per lien rank in template 4. For clarity, template 4 should then also include one ID for the granular level (reflecting amongst others the lien ranks), which could be the already available field indexed 4,00 named “Protection identifier”, as well as a collateral group identifier that serves for having all protection elements of a collateral or guarantee object combined, for instance “Protection Group Identifier”.
All other fields than the lien ranks could then by means of template 2 be linked to the respective counterparties and loans without information loss.
b. It is not clear from the data glossary how to understand field 4,40 (Collateral Repossession Date) as only loans and advances should be subject of this data template.
This field, however, reflects property held by the seller. In a situation where the seller offers a Real Estate Owned object (REO), this field contains crucial information, and it should be clear from the description of the ITS that REOs are also included in the scope of the template.
If not accompanied by any other Flag, this field should also be mandatory. Also see comment to the data field indexed 1.43 (“Date of Obtaining Order for Possession”).
The loans associated with this collateral might have become effectively unsecured if the collateral is already repossessed and the workout strategy would be another than with existing collateral.
c. Template 4 provides valuation results for internal and external appraisals on real estate collaterals. In order to be able to make use of valuation results, the buyer urgently needs to be informed about the appraisal type – is it the appraisal for the market value, or is it meant to be for the liquidation value of the asset in a foreclosure act. The latter can vary significantly from the market value and thus the appraisal type should be mandatory information.
We further note that the template is missing further data fields for proper due diligence on the collaterals, which are
• Real estate collateral land registry identification number (mandatory)
• Real estate collateral cadastral municipality code (mandatory)
• Federal Region of real estate collateral (non-mandatory)
• Appraisal type (Market value or liquidation value) for appraisals in the data tape (mandatory)
• Real estate collateral operational Flag (Yes/No) (mandatory)
• Real estate collateral annual rental income (non-mandatory)
• Real estate collateral legal status (bankruptcy, enforcement, not in legal proceedings) (mandatory)
• Real estate collateral ongoing foreclosure start date (mandatory)
• Real estate collateral ongoing foreclosure stage (mandatory)
• Real estate collateral ongoing foreclosure file number (mandatory)
• Cash in court (mandatory)
• Real estate collateral economic lien rank (mandatory)
• Real estate collateral legal lien rank (mandatory)
We find the content of template 5 not appropriate for the sale of NPL cases as it does not allow for a proper due diligence on the payment history.
For the NPL portfolios we are pricing in many countries across Europe, there are no objectively ascertainable future cash recoveries that could fill the month 1-12 and years 2 and 3 after the cut-off date. Future recoveries significantly depend on the respective buyer’s workout measures (the only difference being post cut-off collection (PCC) arising in the period between COD and closing, which are known by the seller).
Thus, the emphasis of the template should be on the historical collections, at best from default date to cut-off date complemented by PCC. However, it should at least cover the last 24 months before cut-off date and all months after to the current date on a granular basis. This is a basic requirement for our pricing. An aggregation to years as done in the data template is much too rough, in our view, and should at least be broken down to monthly collections. Especially the payment dynamics in the previous 12 months give a good indication of the likelihood of a successful (re-)activation after take-over.
We also find the data fields concerning the external collection not sufficient. The data tape assumes that there is only one Debt Collection Agency (DCA) (as only one servicer can be entered in fields 5,04 and 5,05, Name and Registration Number of External Collection Agent). Instead, the case could have been handled by various DCAs in e.g. a rolling manner.
It is thus important information how many DCAs have been working on the case, when the first one started and when the last one ended their actions, and if the case has already returned from that last DCA. A list of the last 2-4 DCAs on the case would help as well as the periods the cases stayed with the DCAs. The information of how many DCAs have worked for and until which periods of time on the NPLs is a key quality indicator for any buyer when pricing the portfolio.
A good way to provide payment information would be a list of payments containing amongst the loan ID a few more columns as
• Payment date (mandatory)
• Payment Amount (mandatory)
• Source of payment (e.g. liquidation of security, discounted pay-off, loan garnishment, direct debit or bank transfer, …) (mandatory)
All in a single-transaction manner.
In general, we agree with template 2. But one significant aspect that is not covered is the interaction between loans and real estate collaterals beyond their relationship (see also comment to template 4.)
This applies for situations where there is a many-to-many relationship between loans and collaterals. The data template in its current status does not give sufficient information on how much is recoverable from which collateral for which loan.
Template 2 does not allow to reflect this situation unless there was a split of lines in the collaterals tab and a clear assignment of lien ranks and collectable amounts to loans.
We agree with the structure of the data glossary. Regarding the content, see remarks made for questions 1 to 6.
The instructions are mostly adequate. For a better understanding, an example file would have been helpful for us and will be of great value for the banks in the future. We note that once the data template becomes mandatory to use, such an example file should be provided to all sellers to warrant for a proper filling of the fields.
We welcome the ‘no data option’ as it adds more clarity and insight into formerly empty data fields.
Yes, from our view the proposed set is sufficient and does not need further technical tools or technology.
We agree with the principle of proportionality when it concerns debtor and loan type. However, we disagree with the discrimination by loan size. In general, we suggest that any data field that is mandatory for one of the two amount categories should be made mandatory for both. See also answers to question 12.
Irrespective of the threshold, sellers shall nevertheless act on a best efforts basis and provide all fields contained in the template.
We disagree with the choice of a 25k EUR threshold, especially as it seems to apply on single-loan level.
In the ITS, it is stated that the carrying amount of the loans is the reference. Studying the data template irritates as the focus is a different one for each template therein. On counterparty, it makes most sense on debtor/group level, whereas at the loan and collaterals tab, it should be on loan and collateral level (The corresponding carrying amount of loans aggregated on counterparty and collateral level can become much higher).
We understand that this level is appropriate for the purpose of AnaCredit reporting, but the NPL portfolio sale has a different requirement. The average loan size of a banking portfolio is clearly below the 25k EUR threshold in many unsecured NPL sales, as most common products are overdrafts, credit cards and personal loans. Only the latter – and of course mortgage loans – are potentially in the amount range >25k EUR. Instead of using a proportionality threshold for the loan amounts, we would rather suggest a materiality threshold being at e.g., 500 EUR on debtor level.
However, with reference to the data being available in the banking system for the provision of such information that is flagged differently for cases < and > 25k EUR in the glossary, we do not see why the provision of e.g. the debtor’s age and postal code (Field indexed 1,07, “Date of Birth” and 1,09, “Postal code of Counterparty (private individual)”) should not be possible for all loan sizes.
Since we assume that sellers´ internal data governance mainly does not distinguish data requirements by loan size, they should be available for non-mandatory constellations.
As an exception to the rule, we deem the efforts to the sellers too high to provide corporate counterparty details for fields as P&L and balance sheet information for low loan amounts. Here we agree with the proposed threshold of 25k EUR. If the “Role of the counterparty” (field 2,01) is “tenant”, all originally mandatory information should be made non-mandatory.
Overall, we agree with the operational procedures and confidentiality. Data governance should, however, not be left for the seller to decide but should be accompanied by the supervisory level to ensure adequate usage.
Annex II - EOS Feedback - NPL data templates glossary with questions for consultation.xlsx
13 Questions Answered_EBA.pdf