Response to discussion on the review of the NPL transaction data templates
Go back
For example, the proposed templates would not make purchase price calculations any easier. The following examples aim to illustrate this:
– Exposure to individual client: either only the exposure (borrower unit) or the individual client (borrower) can be modelled. However, there are a large number of credit exposures consisting of several borrowers that must be considered individually when it comes to the calculation and the subsequent processing. It must therefore be possible to link several borrowers to a single borrower unit in the templates.
– The same applies to the relationship between assets or collateral and the exposure/client. Either only the exposure (borrower unit) or the individual client (borrower) can be allocated. However, especially in the case of complex exposures, multiple allocation must be possible (e.g. if collateral in a borrower unit covers liability for several borrowers, which is often the case for property developer exposures).
– It is not possible to subdivide individual assets (e.g. different land charges per property, collective land charges or other types), although this is a mandatory (“critical”) requirement for measuring exposures.
– A 1:1 allocation between collateral and account (not client/exposure) is not sufficient, because there may be different declarations of purpose for each real property lien (additional collateral entity required) in the case of real estate.
– In addition, freely editable data fields should be possible so that buyers can ask for supplementary data for special cases (e.g. heritable building rights), while at the same time making this data visible to all.
We agree in general, but it has to be emphasised that for a loan sale and particularly for the legal transfer, the contract level is essential, as several loans are often granted to several borrowers in a loan agreement. Since the IT systems of many banks usually operate from the loan level, manual rework is often required to comprehensively mirror the contract level required for the legal assignment (e.g. mapping of the different utilisations of the different loans with, in some circumstances, different collateral and different (co-)borrowers to the respective loan agreements).
The originally non-critical data fields which have been reclassified as critical for all asset classes are too extensive. Most of these new data fields (e.g. Fixed Assets, Current Assets, etc.) are often not applicable to private individual borrowers. Even in the case of corporate borrowers, these data fields should not be critical. In most cases this data is publicly available and data fields should not be classified as critical only because it is more comfortable for an investor to value the loan assets without the need for themselves to check for current data. Therefore, in our view, the existing data fields are already too comprehensive and we should not add any additional data fields (it should be noted that there are no information asymmetries [as COM and EBA complain] for publicly available data).
For example, the following data fields are not necessary for purchase price calculation:
1,01; 1,03; 1,13; 1,14; 1,15; 1,16; 1,17; 1,21; 1,22; 1,26; 1,45; 1,49; 1,57; 1,58; 3,13; 3,15; 3,19; 3,23; 3,25; 3,27; 3,28; 3,30; 3,31; 3,32; 3,49; 3,50; 3,51; 3,52; 3,53; 3,54; 3,55; 3,57; 3,60; 4,12; 4,30; 4,35; 4,37; 4,39; 4,45; 4,46; 4,49; 4,54; 4,55; 4,64; 4,65; 4,68; 5,02; 5,03; 5,04; 5,13; 5,14; 5,15; 5,16; 5,17
This relates to the following data fields.
Annex I to the Discussion Paper
“Template 1: Counterparty” “1.13 Internal Credit Rating at Origination” and “1.18 Current Internal Credit Rating”;
(“Template 3: Financial instrument” “3.50 Internal Credit Rating at Origination” and “3.53 Current Internal Credit Rating”).
Annex III to the Discussion Paper
“3.023 Internal Credit Rating at Origination”and “3.028 Current Internal Credit Rating”.
Particularly in the case of NPLs, only outdated financial statement figures are usually available (especially in insolvency cases). A further complicating factor is that the financial statement amounts, if they are outdated, have often been determined to reflect the going concern approach, whereas the break-up values or the gone concern approach are relevant for determining the purchase price.
It shouldn't be the purpose of the templates to make life easier for buyers etc. at the cost of the seller (see also answer to Q3 and Q6).
See also Q1.
Exposure measurement for NPLs is based on valuations of underlying collateral. The contractual data referred to in the question is not relevant for this valuation. Additionally, under German law, loans that do not have a default status (e.g. due <90 days) and therefore cannot be terminated may not be sold without the borrower’s consent; in particular, no credit agreement data from such (non-defaulted/non-terminated) exposures may be communicated.
Forbearance data is only relevant if one assumes that a loan will be sold at an early problem stage. This is not normally the case (and in some cases not legally possible without the borrower’s consent), as institutions generally use forbearance measures to try to transfer the loan back into the normal loan portfolio.
A sale is only considered once the loan has been moved to the liquidation portfolio. In the course of liquidation, however, past forbearance measures are no longer of interest, as recovery of collateral is again the top priority.
See also the reference to Europe-wide legal harmonisation for question 28.
The First Auction Date should at least be included in the “non-critical data fields” category, since the time difference between the termination and the first foreclosure sale date can be used to draw conclusions about the possible processing time and the capacity utilisation of the local courts.
The data fields related to “Units” (Annex I to the Discussion Paper “Template 4: Collateral and Enforcement” “4.18 Number of Lettable Units”, 4.19 “Number of Units Vacant” and “4.20 Number of Units Occupied”) should be deleted, and instead the data fields related to square metres should be retained (Annex III to the Discussion Paper “11.022 Building Area (M2) Lettable” and “11.023 Building Area (M2) Occupied”), as this information is more relevant for the valuation.
To the extent that definitions already exist, for example for AnaCredit, BSI Regulation or Finrep or, going forward, IReF/Bird, reference should be made to them in all cases, e.g. 1.41, 3.04, 3.12, 3.44, etc.
1.13: No description should be required for sales within a group or an IPS using the same rating methodologies.
4:09: Please provide a more precise description of the quality categories.
4.25: Similar to 4.24, preference should be given to third-party valuations in the case of multiple valuations.
4.27: We do not understand whether the individual variants are mutually exclusive.
4.44: The volume definition in litres is so restrictive that it is almost impossible to calculate for certain machines. Suggestion: height times width times length. (see also question 16)
4.45, 4.46: We do not understand the restriction to non-property in these definitions. (see also question 17)
The descriptions are very helpful to decide what information is meant. However, while the descriptions may help sellers to fill in the "right" data, they may not fit materially with the data provided by different banks in different countries. That's why the descriptions should be in a separate file and not part of the handover to the buyer. Any ambiguity should be clarified in the individual case between seller and buyer.
With one exception, we agree with the confidentiality aspects: data field "4.01 Legal Owner of the Collateral” containing information about the legal owner of the property or non-property collateral should be deleted in order to avoid violations of existing data protection requirements.
In practice, purchase price discounts have been established that not only represent the lack of data quality for missing data points, but often also provide insight into the processing quality.
We do not agree that there should not be any no data options. It is advisable to avoid mandatory filling of data fields with data or mandatory specifications regarding the format of the data. This may help with automatic data validation, but it considerably complicates the practical usability of the templates. Simple and flexible manageability of the templates promotes usability, which is why there should be a "no data" option (e.g. in the case of a heritable building right).
The following points should be taken into account:
– Simplicity of the portfolios to be sold (e.g. exclusively residential mortgages with one collateral security or mortgages with less than two collateral securities). Such portfolios are already traded and priced as standardised products today – with a significantly smaller dataset. Expanding the dataset in the case of simple portfolios would lead to a lower purchase price due to the higher administrative expense.
– Number of exposures/loans included in a portfolio held for sale
– Total risk exposure amount of NPLs held for sale
– Expected purchase prices of the NPLs held for sale
– No application of the NPL templates in the case of sales of NPLs within an IPS network or group.
As mentioned above, the templates should generally be voluntary for all institutions. This is particularly essential for SNCIs.
Despite the revision and reduction of the “critical fields”, the specifications are still very much statistically driven. This is demonstrated by the conformity of many data points with values already known from AnaCredit or collected there.
Numerous data points are surveyed that are of a more statistical nature but do not affect the valuation or processing of exposures (see comments on questions 4, 12, 15, 16, 17 and 18 as well as the attached spreadsheet listing references to other reports).
The templates are strongly portfolio-driven. This hampers the frequent sales of individual exposures or very small portfolios within IPS structures and SNCIs, when case-by-case calculations form the basis for determining the purchase price.
No, the outlined approach does not appear to be sufficient for addressing all concerns and to take into account the different requirements of different loan types and asset/collateral classes. It cannot be ruled out that the mandatory use of the template increases barriers for market entry for new or smaller sellers of NPL, rather than boosting supply. Data quantity and quality are not the main concern and not the pivotal point for bringing more liquidity to the NPL market. We doubt that the high number of data fields as standard and an obligation to use the templates will have a positive effect on the NPL market. Currently, our experience is huge demand and liquidity on the buy side but still some resistance by banks due to the complexity and cost associated with transactions, even today. A mandatory NPL template would just further increase cost and regulatory complexity and therefore probably support the current rejection to sell NPLs in the market.
We consider the structure and coverage of the templates to be more appropriate in relation to portfolio sales. By contrast, the templates are less suitable for individual transactions, since in these large-volume cases a specific case-by-case approach leads to more appropriate results. To the extent that only individual exposures are sold, there will always be a case-by-case analysis, which can be both more intensive and less extensive than the information in the templates. It is therefore advisable to introduce different templates for the sale of portfolios and individual exposures.
The low use of NPL templates in the past is due primarily to two circumstances:
1. The data to be included in the templates is too extensive, not available at the institutions (or in some cases in their IT) and too complex to populate.
2. Standards for purchase price determination have become established in the market in which other techniques for data collection have prevailed, in particular the population of lean data templates, which were later supplemented or strengthened qualitatively.
If it was not possible to place or sell NPLs on the (secondary) market in recent years, this was not so much due to non-transparent data, but rather to the poor quality of the exposures and processing of some of the NPLs offered, and the excessively high purchase price expectations on the sell side. If NPLs held for sale make it impossible to realise hidden reserves or windfall profits because the sell-side purchase price expectations are too high, there is no market for them. This effect cannot be avoided even with (revised) templates.
In addition, (small) LSIs are unwilling or unable to tie up capacity and thus expenditure on reporting requirements because of their staff capacity (concentration on the market and thus on approximately 98.5% of the lending business). In this respect, the mandatory introduction of the NPL templates (in particular in conjunction with the introduction of mandatory fields for all data points to be reported) is more likely to lead to a situation where (small) LSIs back away from selling NPLs and instead (fully) adjust their value, which will additionally narrow their earnings potential. Equally, outsourcing to servicers will increase, counteracting the desired cleansing of bank balance sheets.
The acceptance and effectiveness of templates would only reach its full potential if aspects of the law going beyond the distressed debt area (foreclosure law, land register law, requirements for commercial and transparency registers) were to be harmonised across Europe (see also the question from the European Commission’s parallel consultation on which aspects of the law still need to/should be harmonised across Europe).
1. Do you agree with the proposed data structure and the relationship between templates? If not, please provide explanation.
The data structure and the relationship between the individual templates appear to be understandable at first glance. However, they have weaknesses because they contradict the practical realities.For example, the proposed templates would not make purchase price calculations any easier. The following examples aim to illustrate this:
– Exposure to individual client: either only the exposure (borrower unit) or the individual client (borrower) can be modelled. However, there are a large number of credit exposures consisting of several borrowers that must be considered individually when it comes to the calculation and the subsequent processing. It must therefore be possible to link several borrowers to a single borrower unit in the templates.
– The same applies to the relationship between assets or collateral and the exposure/client. Either only the exposure (borrower unit) or the individual client (borrower) can be allocated. However, especially in the case of complex exposures, multiple allocation must be possible (e.g. if collateral in a borrower unit covers liability for several borrowers, which is often the case for property developer exposures).
– It is not possible to subdivide individual assets (e.g. different land charges per property, collective land charges or other types), although this is a mandatory (“critical”) requirement for measuring exposures.
– A 1:1 allocation between collateral and account (not client/exposure) is not sufficient, because there may be different declarations of purpose for each real property lien (additional collateral entity required) in the case of real estate.
– In addition, freely editable data fields should be possible so that buyers can ask for supplementary data for special cases (e.g. heritable building rights), while at the same time making this data visible to all.
We agree in general, but it has to be emphasised that for a loan sale and particularly for the legal transfer, the contract level is essential, as several loans are often granted to several borrowers in a loan agreement. Since the IT systems of many banks usually operate from the loan level, manual rework is often required to comprehensively mirror the contract level required for the legal assignment (e.g. mapping of the different utilisations of the different loans with, in some circumstances, different collateral and different (co-)borrowers to the respective loan agreements).
2. Do you agree with the deletion of data categories ‘NPL portfolio’ and ‘Swap’? If not, please provide explanation.
We agree in general, since in NPL sales the relevant swaps will have crystallised to close-out exposures that do not need a separate data category.3. Do you think the suggested list of data fields capture all the relevant information on the counterparty needed for NPL valuation and financial due diligence? If not, please indicate which other data fields should be included and provide explanation for this.
No. The purchase price calculation for NPLs (especially workout cases) is generally based on a case-by-case calculation that is condensed at portfolio level.The originally non-critical data fields which have been reclassified as critical for all asset classes are too extensive. Most of these new data fields (e.g. Fixed Assets, Current Assets, etc.) are often not applicable to private individual borrowers. Even in the case of corporate borrowers, these data fields should not be critical. In most cases this data is publicly available and data fields should not be classified as critical only because it is more comfortable for an investor to value the loan assets without the need for themselves to check for current data. Therefore, in our view, the existing data fields are already too comprehensive and we should not add any additional data fields (it should be noted that there are no information asymmetries [as COM and EBA complain] for publicly available data).
4. Do you think any specific data fields should be excluded from the template? If yes, please specify the data fields and give explanation to your answer.
Any data field which is not vital or at least important for valuation should be considered as dispensable. This applies in particular to data whose purpose is mainly to ease a swift valuation of the NPL or to ease the takeover of servicing. Please consider that every data field envisaged in an official EBA template raises expectations on the buyer’s part that may not be fulfilled. This can have a negative impact on the price if such expectations did not exist before or at least did not exist in this form, as it may lead to the general impression of poor data quantity or quality.For example, the following data fields are not necessary for purchase price calculation:
1,01; 1,03; 1,13; 1,14; 1,15; 1,16; 1,17; 1,21; 1,22; 1,26; 1,45; 1,49; 1,57; 1,58; 3,13; 3,15; 3,19; 3,23; 3,25; 3,27; 3,28; 3,30; 3,31; 3,32; 3,49; 3,50; 3,51; 3,52; 3,53; 3,54; 3,55; 3,57; 3,60; 4,12; 4,30; 4,35; 4,37; 4,39; 4,45; 4,46; 4,49; 4,54; 4,55; 4,64; 4,65; 4,68; 5,02; 5,03; 5,04; 5,13; 5,14; 5,15; 5,16; 5,17
5. Do you agree that data fields on current external and internal credit scores and current external and internal credit scores at origination should be included in the template (for both private individual and corporate counterparties)?
No, because (i) banks are generally not willing to disclose internal scores for various reasons and (ii) due to contractual restrictions, banks are usually not entitled to forward or disclose external credit scores if they are not public. External ratings are usually public information and should be checked by the investor. In case there is a private external rating, disclosure restrictions might in any case prohibit disclosure. The data fields on internal credit scores should generally be deleted because of the critical appearance of a guarantee.This relates to the following data fields.
Annex I to the Discussion Paper
“Template 1: Counterparty” “1.13 Internal Credit Rating at Origination” and “1.18 Current Internal Credit Rating”;
(“Template 3: Financial instrument” “3.50 Internal Credit Rating at Origination” and “3.53 Current Internal Credit Rating”).
Annex III to the Discussion Paper
“3.023 Internal Credit Rating at Origination”and “3.028 Current Internal Credit Rating”.
6. Do you agree that data fields on corporate’s latest available financial statement amounts should be included in the template?
No (see also answer to Q3), it should not be the purpose of the templates to make life easier for buyers etc. to the cost of the seller. Especially as the standalone amounts might be misleading given different accounting rules. It should be part of the buy side due diligence to analyse financial statements as a whole but not to rely on potentially outdated financial statement amounts given in a template.Particularly in the case of NPLs, only outdated financial statement figures are usually available (especially in insolvency cases). A further complicating factor is that the financial statement amounts, if they are outdated, have often been determined to reflect the going concern approach, whereas the break-up values or the gone concern approach are relevant for determining the purchase price.
7. Do you agree that data fields related to corporate counterparties’ assets and liabilities, market capitalisation should be included in the template?
No. As a general rule, outdated balance sheet amounts are not meaningful, especially as these were prepared on the basis of going concern accounting rules, whereas break-up values must be applied in the event of liquidation. In the case of restructuring exposures, the bank is required to review the restructuring plan presented by the borrower. Depending on the cause of the crisis, the focus may be on different aspects. This individuality cannot be reflected in a single data template. At most, there is an overlap in the case of short- and medium-term liquidity planning. Even in this case, however, the assumptions – which diverge from case to case – are decisive. It would make more sense to model the contents of restructuring concepts, if available (maturity extensions, planned debt waivers, other restructuring measures). However, these are normally qualitative values.It shouldn't be the purpose of the templates to make life easier for buyers etc. at the cost of the seller (see also answer to Q3 and Q6).
8. Do you agree with the proposed Template 2 of Annex I? If not, please provide explanation to your answer.
No. Collateral is only represented in a single dimension in the template. In order to be able to represent 1:n relationships from real life (e.g. several mortgages for a single property or one mortgage covering several properties, the representation of intermediate rights, subordinated rights, quotas, private assignments, etc.), supplementing with additional entities is necessary.See also Q1.
9. Do you agree with the inclusion of the data fields related to interest rates and other information as per contractual agreement for the valuation and financial due diligence of NPLs, especially when they are not more than 90 days past due? Please provide data field-specific explanation to your answer.
No, because this information is not crucial for NPL valuation and financial due diligence, which are likely to make up the majority of the exposures to be sold and should be the main purpose of the template.Exposure measurement for NPLs is based on valuations of underlying collateral. The contractual data referred to in the question is not relevant for this valuation. Additionally, under German law, loans that do not have a default status (e.g. due <90 days) and therefore cannot be terminated may not be sold without the borrower’s consent; in particular, no credit agreement data from such (non-defaulted/non-terminated) exposures may be communicated.
10. Do you agree with the inclusion of the data fields related to forbearance measures for the valuation and financial due diligence of NPLs?
No, at least not all of the given data fields as some of the data is not crucial for valuation and financial due diligence (e.g. field 10,009 Start Date of Forbearance).Forbearance data is only relevant if one assumes that a loan will be sold at an early problem stage. This is not normally the case (and in some cases not legally possible without the borrower’s consent), as institutions generally use forbearance measures to try to transfer the loan back into the normal loan portfolio.
A sale is only considered once the loan has been moved to the liquidation portfolio. In the course of liquidation, however, past forbearance measures are no longer of interest, as recovery of collateral is again the top priority.
11. Do you think the suggested list of data fields capture all relevant information on financial instrument needed for NPL valuation and financial due diligence? If not, please indicate which other data fields should be included and provide explanation for this.
No, in fact the list is too extensive while other data is missing (see question 3). But in some cases more than necessary and at least not crucial for valuation (e.g. fields 7020, 7042, 7055, 7056, 7074, 7075, 8005-8009).See also the reference to Europe-wide legal harmonisation for question 28.
12. Do you think any specific data fields should be excluded from the template? If yes, please specify the data fields and give explanation to your answer.
Any data fields that just sum up the amounts given in other data fields are dispensable (e.g. field 7066) as well as all fields whose purpose is to ease valuation or the takeover of servicing (e.g. field 7068). In general, it would certainly increase acceptance of the templates by the banks if the templates concentrate on data that is absolutely necessary for valuation. Current market practice shows that this is much less than provided for in the templates. In addition, a careful review of the categorisation of data fields as "critical" is still required. Data that may be important for valuation and can be a value driver may still not be a "conditio sine qua non", i.e. a buyer may still be able to value the portfolio on a loan-by-loan basis without the information. If this is the case, such data should be classified no higher than "important".13. Do you agree with the data fields related to lease? Please provide data field-specific explanation to your answer.
Generally, yes but we don't agree with the classification as "critical"The First Auction Date should at least be included in the “non-critical data fields” category, since the time difference between the termination and the first foreclosure sale date can be used to draw conclusions about the possible processing time and the capacity utilisation of the local courts.
14. Do you think the suggested list of data fields capture all relevant information on collateral needed for NPL valuation and financial due diligence? If not, please indicate which other data fields should be included and provide explanation for this.
No. The valuation of a receivable is based on break-up values or the future forecast (restructuring report). But in some cases more than necessary and at least not really crucial for valuation (see answer to Q15 below).15. Do you think any specific data fields should be excluded from the template? If yes, please specify the data fields and give explanation to your answer.
Yes, fields 16002, 16006, 16008, 16023, 16026. This information can be provided if the transaction is consummated and does not need to be provided in advance as soft data, but can also be handed over later with the relevant documents.The data fields related to “Units” (Annex I to the Discussion Paper “Template 4: Collateral and Enforcement” “4.18 Number of Lettable Units”, 4.19 “Number of Units Vacant” and “4.20 Number of Units Occupied”) should be deleted, and instead the data fields related to square metres should be retained (Annex III to the Discussion Paper “11.022 Building Area (M2) Lettable” and “11.023 Building Area (M2) Occupied”), as this information is more relevant for the valuation.
16. Do you agree with the data fields on the characteristics of non-property collateral? Please provide data field-specific explanation to your answer.
See answer to question Q14 and Q15.17. Do you agree with the data fields related to the enforcement of collateral? Please provide data field-specific explanation to your answer.
See answer to question Q14 and Q15.18. Do you agree with the proposed Template 5 of Annex I for NPL valuation and financial due diligence? Please provide data field-specific explanation to your answer.
We agree with the data fields in Template 5 (Annex I to the discussion paper). However, we believe that an observation period of 24 periods is adequate. Due to our experience it is not necessary for valuation purposes to provide historical data for the last 36 months (fields 5.10 to 5.17). It is much more relevant for valuation what happened in the last 12 to 18 months than before.19. Do you agree with description of data fields presented in data dictionary?
Yes, with qualifications:To the extent that definitions already exist, for example for AnaCredit, BSI Regulation or Finrep or, going forward, IReF/Bird, reference should be made to them in all cases, e.g. 1.41, 3.04, 3.12, 3.44, etc.
1.13: No description should be required for sales within a group or an IPS using the same rating methodologies.
4:09: Please provide a more precise description of the quality categories.
4.25: Similar to 4.24, preference should be given to third-party valuations in the case of multiple valuations.
4.27: We do not understand whether the individual variants are mutually exclusive.
4.44: The volume definition in litres is so restrictive that it is almost impossible to calculate for certain machines. Suggestion: height times width times length. (see also question 16)
4.45, 4.46: We do not understand the restriction to non-property in these definitions. (see also question 17)
The descriptions are very helpful to decide what information is meant. However, while the descriptions may help sellers to fill in the "right" data, they may not fit materially with the data provided by different banks in different countries. That's why the descriptions should be in a separate file and not part of the handover to the buyer. Any ambiguity should be clarified in the individual case between seller and buyer.
20. Do you agree with criticality (and non-criticality) of data fields presented in data dictionary? If not, please provide suggestions and explanations related to specific data fields.
As previously pointed out in detail in the introductory remarks (see 0.1) and in response to questions Q4 and Q12, the "critical" category should be much stricter in scope and only cover data without which valuation on a loan-by-loan basis is simply not possible.21. Do you agree with confidentiality aspects of data fields? If not, please provide explanation.
It has to be highlighted that apart from data which is protected by GDPR and banking secrecy one has to bare in mind that some data may be protected by separate confidentiality undertakings. This is in particular the case regarding scorings, ratings and valuations which are often protected against a disclosure.With one exception, we agree with the confidentiality aspects: data field "4.01 Legal Owner of the Collateral” containing information about the legal owner of the property or non-property collateral should be deleted in order to avoid violations of existing data protection requirements.
22. Do you agree with excluding no data options for data fields? If not, please provide suggestions and explanations related to specific data fields.
Declaring all fields mandatory is not rational. It is already known from the old templates that more than one-third of the data points were either not available in the banks’ IT systems or not available at all. Due to the planned future mandatory use of the templates, it will act as a deterrent (especially for LSIs) to have to (manually) collect data that is not available. This will lead to a situation where institutions back away from the option to sell NPLs (manual post-collection effort > possible proceeds from correctly valued NPLs). To make matters worse, subsequent collection of certain data will require the cooperation of the borrowers.In practice, purchase price discounts have been established that not only represent the lack of data quality for missing data points, but often also provide insight into the processing quality.
We do not agree that there should not be any no data options. It is advisable to avoid mandatory filling of data fields with data or mandatory specifications regarding the format of the data. This may help with automatic data validation, but it considerably complicates the practical usability of the templates. Simple and flexible manageability of the templates promotes usability, which is why there should be a "no data" option (e.g. in the case of a heritable building right).
23. Please provide your views on how proportionality considerations regarding the size of the exposures or portfolios being sold should be incorporated in the implementation of NPL data templates.
Proportionality considerations are crucial and should take into account the different loan types (corporate, consumer etc.) and the loan sizes as well as the existence of collateral. It is of utmost relevance that portfolio transactions of unsecured smaller loans are not encumbered by formal requirements that increase the costs of such a transaction. Given the very limited recovery amounts, every cent spent for data collection, data preparation and data validation may be sunk costs.The following points should be taken into account:
– Simplicity of the portfolios to be sold (e.g. exclusively residential mortgages with one collateral security or mortgages with less than two collateral securities). Such portfolios are already traded and priced as standardised products today – with a significantly smaller dataset. Expanding the dataset in the case of simple portfolios would lead to a lower purchase price due to the higher administrative expense.
– Number of exposures/loans included in a portfolio held for sale
– Total risk exposure amount of NPLs held for sale
– Expected purchase prices of the NPLs held for sale
– No application of the NPL templates in the case of sales of NPLs within an IPS network or group.
As mentioned above, the templates should generally be voluntary for all institutions. This is particularly essential for SNCIs.
24. Should there be a threshold (e.g. in monetary terms) for the application of the proportionality principle? If yes, then how should this be defined?
There should not be a threshold for the application of the proportionality principle. This is because proportionality not only depends on the volume of the single loans but also on the price expected to be paid at the time of the trade. Furthermore, data collection, data preparation and data validation may be more challenging and more expensive as regards certain loan types and/or collateral. If using the template will actually boost liquidity and as a consequence result in better prices, the templates as such might become a market standard for achieving better pricing. In this case we suggest to leaving it up to each seller to balance the likely higher cost for data validation and the potentially better pricing, as only the seller can make the right decision with regard to a specific portfolio and market condition at the time of the sale. Therefore, we would highly recommend avoiding any mandatory use of the template.25. Do you agree that the proposed approach takes into account, in an adequate way, the proportionality principle? If not, which additional elements should be considered?
See comments on questions 23 and 24.Despite the revision and reduction of the “critical fields”, the specifications are still very much statistically driven. This is demonstrated by the conformity of many data points with values already known from AnaCredit or collected there.
Numerous data points are surveyed that are of a more statistical nature but do not affect the valuation or processing of exposures (see comments on questions 4, 12, 15, 16, 17 and 18 as well as the attached spreadsheet listing references to other reports).
The templates are strongly portfolio-driven. This hampers the frequent sales of individual exposures or very small portfolios within IPS structures and SNCIs, when case-by-case calculations form the basis for determining the purchase price.
No, the outlined approach does not appear to be sufficient for addressing all concerns and to take into account the different requirements of different loan types and asset/collateral classes. It cannot be ruled out that the mandatory use of the template increases barriers for market entry for new or smaller sellers of NPL, rather than boosting supply. Data quantity and quality are not the main concern and not the pivotal point for bringing more liquidity to the NPL market. We doubt that the high number of data fields as standard and an obligation to use the templates will have a positive effect on the NPL market. Currently, our experience is huge demand and liquidity on the buy side but still some resistance by banks due to the complexity and cost associated with transactions, even today. A mandatory NPL template would just further increase cost and regulatory complexity and therefore probably support the current rejection to sell NPLs in the market.
26. Please provide your views on the asset classes covered and whether any specific data fields, other than already foreseen, should be included in the templates for ensure full coverage of certain asset classes.
No additional data appears to be necessary.27. In your view, is the structure and coverage of the templates adequate for both portfolio transactions and transactions where an individual exposure is traded? Please explain your answer.
Yes, if the comprehensive number of data fields will be reduced. Even if the specifications for single exposures may require additional information in a single exposure trade, it wouldn't make sense to reflect any such specialities in the templates. Trying to address any speciality would make them excessive, as we have seen at the beginning of the process. It is advisable to concentrate on the core data absolutely necessary for a trade.We consider the structure and coverage of the templates to be more appropriate in relation to portfolio sales. By contrast, the templates are less suitable for individual transactions, since in these large-volume cases a specific case-by-case approach leads to more appropriate results. To the extent that only individual exposures are sold, there will always be a case-by-case analysis, which can be both more intensive and less extensive than the information in the templates. It is therefore advisable to introduce different templates for the sale of portfolios and individual exposures.
28. Please add any additional comments, remarks or observations you may wish to include in your feedback to the discussion paper.
Standardisation is a good idea in principle, and a solid data basis can also have a positive influence on price. However, it is questionable whether this can be achieved with the NPL Data Templates in the present form. While the number of data fields has been reduced and/or reclassified, it is still too burdensome. During the previous discussions, the influence of third parties with interests other than those of the sellers and buyers had a significant influence on the templates. We believe that such templates can only create a widely accepted standard if they follow market interest only and not - at the same time - supervisory and statistical purposes, as well as purposes of transaction platforms that have identified the data and its analysis and provision as a source of revenue, and advisors who want to sell their services around data validation. This reference cannot be applied, and should thus be removed. The whole idea behind this project, i.e. to support the NPL market, could backfire if excessive templates become mandatory. This is true in particular if they push up the costs of NPL transactions or set new hurdles for potential sellers. Data collection, processing and validation always comes at a price, which may simply be too high for certain loan types.The low use of NPL templates in the past is due primarily to two circumstances:
1. The data to be included in the templates is too extensive, not available at the institutions (or in some cases in their IT) and too complex to populate.
2. Standards for purchase price determination have become established in the market in which other techniques for data collection have prevailed, in particular the population of lean data templates, which were later supplemented or strengthened qualitatively.
If it was not possible to place or sell NPLs on the (secondary) market in recent years, this was not so much due to non-transparent data, but rather to the poor quality of the exposures and processing of some of the NPLs offered, and the excessively high purchase price expectations on the sell side. If NPLs held for sale make it impossible to realise hidden reserves or windfall profits because the sell-side purchase price expectations are too high, there is no market for them. This effect cannot be avoided even with (revised) templates.
In addition, (small) LSIs are unwilling or unable to tie up capacity and thus expenditure on reporting requirements because of their staff capacity (concentration on the market and thus on approximately 98.5% of the lending business). In this respect, the mandatory introduction of the NPL templates (in particular in conjunction with the introduction of mandatory fields for all data points to be reported) is more likely to lead to a situation where (small) LSIs back away from selling NPLs and instead (fully) adjust their value, which will additionally narrow their earnings potential. Equally, outsourcing to servicers will increase, counteracting the desired cleansing of bank balance sheets.
The acceptance and effectiveness of templates would only reach its full potential if aspects of the law going beyond the distressed debt area (foreclosure law, land register law, requirements for commercial and transparency registers) were to be harmonised across Europe (see also the question from the European Commission’s parallel consultation on which aspects of the law still need to/should be harmonised across Europe).