Response to consultation on Implementing Technical Standards on NPL transaction data templates
Go back
We acknowledge efforts made to streamline the templates compared to their initial 2017 version and thank the EBA for the several roundtables, hearings and consultations aiming to get a better understanding of industry’s viewpoints, needs and constraints. This kind of exchanges is also a good opportunity for us to understand EBA’s thinking, objectives and constraints (including legal mandate, overarching policy objectives) when designing the templates.
We thus welcome this consultation as an opportunity to provide new elements of context that will hopefully lead to targeted changes which can have a positive impact on NPL sales.
The key point we would like to make is that we are not asking for a reduction of mandatory data fields just as a way to minimize disclosure requirements: we are trying to achieve the best possible balance between templates standardization and NPL sales efficiency, bearing in mind that the ultimate objective, shared by the EBA, investors and sellers, is to make NPL disposals easier, with a reduced bid-ask spread.
Please find below elements of context that are essential to understand current NPL sales and that could usefully serve as a basis for modification of the NPL transactions templates.
1) First of all, we are already regular sellers, and we intend to even increase NPL disposals in the future. As such, we are strongly convinced that quality of information is key in order to sell in good conditions and to get good price. This is why we are truly interested in improving the NPL transaction templates under discussion, which means, providing the right information under the right format.
2) Second, depending on the practices used in different Countries, we may use two sets/templates in our NPL sales. The first one contains the information necessary to price the NPL portfolios. The second one is provided to the buyer before closing, and contains more information, notably data protected by the GDPR and necessary to the good management of sold NPLs. As currently drafted, the EBA NPL transaction templates mix the two types of templates, with information useful for valuation + personal information (+ regulatory & accounting information). In our view, the EBA NPL transaction templates, as their name indicates, should only contain information necessary to price correctly NPLs.
3) Third, all the information we have on NPLs is not provided under the same format. Schematically, there are three categories of information:
o “Core data fields” (~[40-50]TBC, depending of the type of NPLs) are typically available in a standardized way and are key inputs for NPL valuation. Thus, they can be considered as mandatory in the transaction templates.
o The format of the remaining information varies, typically according to the type (segment) of NPL, to EU Member States specificities, etc. This information can be provided in nearly all sales, but under various formats, and should not be required to be produced under one standard format in the transaction templates.
o In addition, sellers can work on “enriched data” i.e. we can produce additional information tailored to a buyer’s needs, if the sales process is sufficiently advanced and if we believe this could create additional value. Producing such information requires significant resources. This information, typically, cannot be provided for each NPL / NPL portfolio. Those enriched data can be very specific to one project. They may add valuable information for one transaction but may be irrelevant (and not available) for another kind of debt.
“Forcing” data fields into pre-defined, standard labels would be counterproductive: the content of the information would not exactly meet the label, and adapting banks’ IT systems to transform the data and make it fit the existing label would be unduly costly. Overall, the impact on NPL sales would be negative.
It would also, in our view, unduly transfer the “burden of proof” from buyers to sellers. While today, potential buyers analyze the data based on the definitions and explanations provided by the sellers, once the ITS enters into force, the responsibility to interpret ITS definitions will be on sellers, who will have to match the information they have with data fields that are almost similar but may not match exactly.
This is the reason why we have until now advocated for a reduction of mandatory fields. This does not mean we want to limit the information we would provide to sellers: in any case, we will provide all the relevant information we have on NPLs in order to get the best possible price, even in different ways than a Data Tape (for example, Q&A process on portfolios with limited number of files but higher individual exposure such as Corporate/ Large Corporate counterparts). However, we cannot provide all this information in a standard format. This is why, following the very useful conversation we had during the Public Hearing, we suggest the following two solutions:
o Solution 1 would be to reduce the number of mandatory fields to the “core data fields” (~40-50) identified via industry’s feedback on this consultation. Other information could be provided separately, under no pre-defined format/label.
o Solution 2 would be the keep the templates as is but to allow the use of “N/D” options: if we have the information under the format specified in the NPL template, we will provide it; if we have it but under a different format, we will provide it separately. At the very least, sellers should be allowed to use “N/D” and explain why they do so, beyond the very restrictive possibilities detailed in the draft ITS.
Additional remarks on the specific subject of mandatory fields and NDs:
- To illustrate which data fields are necessary in our opinion, we have added one column to the Glossary Excel file provided for the consultation (see last column): this was meant to highlight which fields have to be removed, could be removed, or would raise issues. We have thus sorted the proposed data fields between “core data”, “optional fields” and “fields to remove”, to give EBA insight on FBF members’ view on the relevance of each data field.
It is worth noting that, as the discussions progressed, it became clearer and clearer that the assessment was highly correlated to the type of loan, asset class, loan features (secured/unsecured), etc. Hence the difficulty to land on a single template (as mentioned above), and also the importance of preserving optionality / allowing not to fulfill certain data fields in the template.
In this vein, it is paramount that, should the fields indicated as “to be removed” be finally kept, they definitely remain optional; otherwise, additional high hurdles would be put on the path of the selling of NPLs.
- There is no reason to block the sales of NPLs if the requested mandatory information are not fully (or easily) available at bank’s level as the bidders may take some assumptions / proxies on its sides (which, all other things being equal, will negatively impact the price). As mentioned by several parties during the public hearing, a large number of NPL sale transactions are done in Europe with datape having less than 20 information. For this reason, ESMA shall authorize banks/credit institutions having a minimum mandatory information (well below the 95/135 minimum fields indicated in the last draft of the transaction) to sell its portfolio of NPLs to prospective buyers in order to develop the secondary markets for NPLs in the EU even if they don't have the full mandatory information. As we can imagine, the price proposed by the buyers, as adjustment variable, will be negatively affected and so will, in a medium term, encourage all banks/credit institutions to improve over the time the quality / availability of their datatape (in particular by providing the optional information).
For instance: FINREP information requested in the NPL data tape may rise difficulties for banks as the reporting system will not be same or because these FINREP information are not easily available on a loan-by-loan basis.
In addition, we have the following comments:
• We reiterate that we are not in favor of a data hub, mentioned in the comments.
• It is suggested the cut-off date should be as close as possible from closing. For complex transactions, we believe that keeping a fixed cut-off date may be useful to secure the execution of the sale since it allows everybody to align on the same situation. Later updates can lead to new bargain based on the changes in the portfolio + additional work to update all valuations. We believe the cut-off date should be as close as the start of the tender (in line with the current market practice).
• A reasonable delay should be granted between the adoption of the ITS and the entry into force. We cannot adapt as long as the final details of the ITS are known, and adapting banks’ IT systems takes time.
More specifically on the data field “Total Balance” (3,13): this field seems to represent on balance exposure and therefore have the same meaning of Outstanding Nominal Amount (3,12), defined as the total outstanding net of write off, which for business opportunity we are not fine to share. On the other side, if we understood correctly, it seems we are missing the key data for every Investor, which is the Total GBV, or “Face Value”– which includes on-balance and off-balance exposure, before any write-off and represents the amount that can be claimed to the borrower. If by Total Balance we mean Total GBV, we are fine with this information to be included, but based on our understanding it seems it is not the case.
If the number of mandatory fields if left unchanged, however, we should be allowed to use an “N/A” option meaning “(similar) data collected in a different format”. If we have the information under the format specified in the NPL template, we will provide it; if we have it but under a different format, we will provide it separately.
As already mentioned, the sole use of ND4 for the mandatory fields would be a severe issue for banks/credit institutions leading to slow down significantly or stop any future NPL sales for a certain period of time (at least the time for IT developments).
We understand, from the definition of “ND2”, that the EBA implies that all data come from banks’ reporting systems. What is the definition of reporting system? Many data actually come from operational systems (=business systems), e.g. all data extracted during the month as opposed to some reporting system that only provides a picture at month-end. We would like to draw EBA’s attention on the fact that all mandatory reporting information is not available at any time during the month, but most often, at month-end, although NPL sales do not all occur at the end of month.
For instance: if we sell a portfolio made up of 80% of “old” NPL (not in scope of the ITS), does it mean we would have to provide: “extended” EBA templates for recent NPLs above the €25,000 threshold, “reduced” EBA templates for recent NPLs under the threshold, and another format for other NPLs? This seems very complicated and inconvenient.
Another example would be a portfolio largely made up of NPL below €25,000€ (which is typically the case in consumer finance): it would not make much sense to impose the use of NPL transaction templates if only a few NPLs exceed the threshold.
This is why, in our view, it would be reasonable (and consistent with real-life practices) to set a threshold at portfolio level where the template would be the same for all files even if some of them exceed the threshold. A reasonable “portfolio threshold” could be 20% of portfolio face value. This means that (i) we should be allowed not to use the transaction templates if the percentage of “in scope” NPLs is below 20% and (ii) we should be allowed to use the “reduced” transaction templates if the share of NPLs above the materiality threshold is below 20%.
We also have a remark on existing forward Flow agreements, where a buyer and a purchaser enter in an agreement where the data to be delivered, the price and the obligation to sell and buy have already been defined in a contract. For the avoidance of doubt, the ITS should not apply to the continuation of existing deals and would only make sense in the perspective of new deals.
Should this pragmatic approach (adapted to market practices and to economic reality) be disregarded, which we would highly regret, then we would propose a threshold of €100,000 / €75,000 to avoid an unnecessary and costly burden bringing very little value added.
It indeed raises suspicion against data production. We remind that sellers have the contractual obligation to provide accurate data. In addition, data accuracy and quality are essential in building our reputation as a seller. There is thus no need to impose a specific and formal process. Importantly also, this process would entail rigidity that could delay time-to-market.
1. Do the respondents agree that these draft ITS fits for the purpose of the underlying directive?
The global answer is “no” based on the described objective.We acknowledge efforts made to streamline the templates compared to their initial 2017 version and thank the EBA for the several roundtables, hearings and consultations aiming to get a better understanding of industry’s viewpoints, needs and constraints. This kind of exchanges is also a good opportunity for us to understand EBA’s thinking, objectives and constraints (including legal mandate, overarching policy objectives) when designing the templates.
We thus welcome this consultation as an opportunity to provide new elements of context that will hopefully lead to targeted changes which can have a positive impact on NPL sales.
The key point we would like to make is that we are not asking for a reduction of mandatory data fields just as a way to minimize disclosure requirements: we are trying to achieve the best possible balance between templates standardization and NPL sales efficiency, bearing in mind that the ultimate objective, shared by the EBA, investors and sellers, is to make NPL disposals easier, with a reduced bid-ask spread.
Please find below elements of context that are essential to understand current NPL sales and that could usefully serve as a basis for modification of the NPL transactions templates.
1) First of all, we are already regular sellers, and we intend to even increase NPL disposals in the future. As such, we are strongly convinced that quality of information is key in order to sell in good conditions and to get good price. This is why we are truly interested in improving the NPL transaction templates under discussion, which means, providing the right information under the right format.
2) Second, depending on the practices used in different Countries, we may use two sets/templates in our NPL sales. The first one contains the information necessary to price the NPL portfolios. The second one is provided to the buyer before closing, and contains more information, notably data protected by the GDPR and necessary to the good management of sold NPLs. As currently drafted, the EBA NPL transaction templates mix the two types of templates, with information useful for valuation + personal information (+ regulatory & accounting information). In our view, the EBA NPL transaction templates, as their name indicates, should only contain information necessary to price correctly NPLs.
3) Third, all the information we have on NPLs is not provided under the same format. Schematically, there are three categories of information:
o “Core data fields” (~[40-50]TBC, depending of the type of NPLs) are typically available in a standardized way and are key inputs for NPL valuation. Thus, they can be considered as mandatory in the transaction templates.
o The format of the remaining information varies, typically according to the type (segment) of NPL, to EU Member States specificities, etc. This information can be provided in nearly all sales, but under various formats, and should not be required to be produced under one standard format in the transaction templates.
o In addition, sellers can work on “enriched data” i.e. we can produce additional information tailored to a buyer’s needs, if the sales process is sufficiently advanced and if we believe this could create additional value. Producing such information requires significant resources. This information, typically, cannot be provided for each NPL / NPL portfolio. Those enriched data can be very specific to one project. They may add valuable information for one transaction but may be irrelevant (and not available) for another kind of debt.
“Forcing” data fields into pre-defined, standard labels would be counterproductive: the content of the information would not exactly meet the label, and adapting banks’ IT systems to transform the data and make it fit the existing label would be unduly costly. Overall, the impact on NPL sales would be negative.
It would also, in our view, unduly transfer the “burden of proof” from buyers to sellers. While today, potential buyers analyze the data based on the definitions and explanations provided by the sellers, once the ITS enters into force, the responsibility to interpret ITS definitions will be on sellers, who will have to match the information they have with data fields that are almost similar but may not match exactly.
This is the reason why we have until now advocated for a reduction of mandatory fields. This does not mean we want to limit the information we would provide to sellers: in any case, we will provide all the relevant information we have on NPLs in order to get the best possible price, even in different ways than a Data Tape (for example, Q&A process on portfolios with limited number of files but higher individual exposure such as Corporate/ Large Corporate counterparts). However, we cannot provide all this information in a standard format. This is why, following the very useful conversation we had during the Public Hearing, we suggest the following two solutions:
o Solution 1 would be to reduce the number of mandatory fields to the “core data fields” (~40-50) identified via industry’s feedback on this consultation. Other information could be provided separately, under no pre-defined format/label.
o Solution 2 would be the keep the templates as is but to allow the use of “N/D” options: if we have the information under the format specified in the NPL template, we will provide it; if we have it but under a different format, we will provide it separately. At the very least, sellers should be allowed to use “N/D” and explain why they do so, beyond the very restrictive possibilities detailed in the draft ITS.
Additional remarks on the specific subject of mandatory fields and NDs:
- To illustrate which data fields are necessary in our opinion, we have added one column to the Glossary Excel file provided for the consultation (see last column): this was meant to highlight which fields have to be removed, could be removed, or would raise issues. We have thus sorted the proposed data fields between “core data”, “optional fields” and “fields to remove”, to give EBA insight on FBF members’ view on the relevance of each data field.
It is worth noting that, as the discussions progressed, it became clearer and clearer that the assessment was highly correlated to the type of loan, asset class, loan features (secured/unsecured), etc. Hence the difficulty to land on a single template (as mentioned above), and also the importance of preserving optionality / allowing not to fulfill certain data fields in the template.
In this vein, it is paramount that, should the fields indicated as “to be removed” be finally kept, they definitely remain optional; otherwise, additional high hurdles would be put on the path of the selling of NPLs.
- There is no reason to block the sales of NPLs if the requested mandatory information are not fully (or easily) available at bank’s level as the bidders may take some assumptions / proxies on its sides (which, all other things being equal, will negatively impact the price). As mentioned by several parties during the public hearing, a large number of NPL sale transactions are done in Europe with datape having less than 20 information. For this reason, ESMA shall authorize banks/credit institutions having a minimum mandatory information (well below the 95/135 minimum fields indicated in the last draft of the transaction) to sell its portfolio of NPLs to prospective buyers in order to develop the secondary markets for NPLs in the EU even if they don't have the full mandatory information. As we can imagine, the price proposed by the buyers, as adjustment variable, will be negatively affected and so will, in a medium term, encourage all banks/credit institutions to improve over the time the quality / availability of their datatape (in particular by providing the optional information).
For instance: FINREP information requested in the NPL data tape may rise difficulties for banks as the reporting system will not be same or because these FINREP information are not easily available on a loan-by-loan basis.
In addition, we have the following comments:
• We reiterate that we are not in favor of a data hub, mentioned in the comments.
• It is suggested the cut-off date should be as close as possible from closing. For complex transactions, we believe that keeping a fixed cut-off date may be useful to secure the execution of the sale since it allows everybody to align on the same situation. Later updates can lead to new bargain based on the changes in the portfolio + additional work to update all valuations. We believe the cut-off date should be as close as the start of the tender (in line with the current market practice).
• A reasonable delay should be granted between the adoption of the ITS and the entry into force. We cannot adapt as long as the final details of the ITS are known, and adapting banks’ IT systems takes time.
2. What are the respondents’ views on the content of Template 1? Please provide any specific comment you may have on the data fields in the dedicated columns of the data glossary (Annex II to the draft ITS) added for your feedback.
Please refer to the Excel file for specific comments on the data fields.3. What are the respondents’ views on the content of Template 3? Please provide any specific comment you may have on the data fields in the dedicated columns of the data glossary (Annex II to the draft ITS) added for your feedback.
Please refer to the Excel file for specific comments on the data fields.More specifically on the data field “Total Balance” (3,13): this field seems to represent on balance exposure and therefore have the same meaning of Outstanding Nominal Amount (3,12), defined as the total outstanding net of write off, which for business opportunity we are not fine to share. On the other side, if we understood correctly, it seems we are missing the key data for every Investor, which is the Total GBV, or “Face Value”– which includes on-balance and off-balance exposure, before any write-off and represents the amount that can be claimed to the borrower. If by Total Balance we mean Total GBV, we are fine with this information to be included, but based on our understanding it seems it is not the case.
4. What are the respondents’ views on the content of Template 4? Please provide any specific comment you may have on the data fields in the dedicated columns of the data glossary (Annex II to the draft ITS) added for your feedback.
Please refer to the Excel file for specific comments on the data fields.5. What are the respondents’ views on the content of Template 5? Please provide any specific comment you may have on the data fields in the dedicated columns of the data glossary (Annex II to the draft ITS) added for your feedback?
Please refer to the Excel file for specific comments on the data fields.6. Do the respondents agree on the structure of Template 2 to represent the relationship across the templates? If not, do you have any other suggestion of structure?
We have no objection to the structure of template 2.7. Do the respondents agree on the structure and the content of the data glossary? Please provide any specific comment you may have on the data fields in the dedicated columns of the data glossary (Annex II to the draft ITS) added for your feedback.
NA8. What are the respondents’ views on the content of instructions?
NA9. Do the respondents agree on the use of the ‘No data options’ as set out in the instructions?
We believe that current instructions are OK if the number of mandatory fields is reduced to the “core data fields” (~40-50) identified via industry’s feedback on this consultation. Other information could be provided separately, under no pre-defined format/label.If the number of mandatory fields if left unchanged, however, we should be allowed to use an “N/A” option meaning “(similar) data collected in a different format”. If we have the information under the format specified in the NPL template, we will provide it; if we have it but under a different format, we will provide it separately.
As already mentioned, the sole use of ND4 for the mandatory fields would be a severe issue for banks/credit institutions leading to slow down significantly or stop any future NPL sales for a certain period of time (at least the time for IT developments).
We understand, from the definition of “ND2”, that the EBA implies that all data come from banks’ reporting systems. What is the definition of reporting system? Many data actually come from operational systems (=business systems), e.g. all data extracted during the month as opposed to some reporting system that only provides a picture at month-end. We would like to draw EBA’s attention on the fact that all mandatory reporting information is not available at any time during the month, but most often, at month-end, although NPL sales do not all occur at the end of month.
10. What are respondents’ views on whether the proposed set of templates, data glossary and instructions are enough to achieve the data standardisation in the NPL transactions on secondary markets, or there may be a need for some further technical specifications or tools to support digital processing or efficient processing or use of technology (e.g., by means of the EBA Data Point Model or XBRL taxonomy)?
In order to be consistent with our previous comments, we believe that NPL sales cover extremely diverse situations due to market, debtors, size, products, country, legal environment, etc. Therefore, the idea to get a standard template that would embrace all situations does not seem reasonable. We will always have specific information on a specific portfolio that will not be replicable for all transactions. It would be time consuming to check for each transaction a never-ending list of potential criteria coming from a universal and conceptual model. Hence our pragmatic recommendation to have mandatory “core” data fields and another set of important data that can be reported with more freedom as regards their format.11. What are the respondents' views on the approach to the proportionality, including differentiating mandatory data fields around the threshold? Please provide any specific comment you may have on the data fields in the dedicated columns of the data glossary (Annex II to the draft ITS) added for your feedback.
Our key recommendation on proportionality is that the final ITS adopt a “portfolio view”. This is because we usually do not sell single NPLs but rather portfolios of several (sometimes thousands) of NPLs. As currently worded, the ITS would put sellers in very difficult situations.For instance: if we sell a portfolio made up of 80% of “old” NPL (not in scope of the ITS), does it mean we would have to provide: “extended” EBA templates for recent NPLs above the €25,000 threshold, “reduced” EBA templates for recent NPLs under the threshold, and another format for other NPLs? This seems very complicated and inconvenient.
Another example would be a portfolio largely made up of NPL below €25,000€ (which is typically the case in consumer finance): it would not make much sense to impose the use of NPL transaction templates if only a few NPLs exceed the threshold.
This is why, in our view, it would be reasonable (and consistent with real-life practices) to set a threshold at portfolio level where the template would be the same for all files even if some of them exceed the threshold. A reasonable “portfolio threshold” could be 20% of portfolio face value. This means that (i) we should be allowed not to use the transaction templates if the percentage of “in scope” NPLs is below 20% and (ii) we should be allowed to use the “reduced” transaction templates if the share of NPLs above the materiality threshold is below 20%.
We also have a remark on existing forward Flow agreements, where a buyer and a purchaser enter in an agreement where the data to be delivered, the price and the obligation to sell and buy have already been defined in a contract. For the avoidance of doubt, the ITS should not apply to the continuation of existing deals and would only make sense in the perspective of new deals.
12. Do the respondents agree with the proposed calibration of 25 000 euros threshold in line with AnaCredit Regulation? If not, what alternative threshold should be introduced, and why?
Once again, we believe that the proposed threshold should be considered at portfolio level. For a portfolio largely made up of NPL below €25,000€ (which is typically the case in consumer finance): it would not make much sense to impose the use of the extended NPL transaction templates if only a few NPLs exceed the threshold.Should this pragmatic approach (adapted to market practices and to economic reality) be disregarded, which we would highly regret, then we would propose a threshold of €100,000 / €75,000 to avoid an unnecessary and costly burden bringing very little value added.
13. What are the respondents' views on the operational procedures, confidentiality and data governance requirements set out in the draft ITS?
We disagree with paragraph #17 and with the part of the proposed ITS relating to this paragraph: “the draft ITS requires credit institutions to establish internal process ensuring that the information being provided to the prospective buyers have been validated by staff independent from the staff involved in the sales process, and is subject to an appropriate managerial approval.”It indeed raises suspicion against data production. We remind that sellers have the contractual obligation to provide accurate data. In addition, data accuracy and quality are essential in building our reputation as a seller. There is thus no need to impose a specific and formal process. Importantly also, this process would entail rigidity that could delay time-to-market.