The Swedish Bankers’ Association, the Swedish Savings Bank Association and the Association of Swedish Finance Houses
General remarks
The Swedish Bankers’ Association, the Swedish Savings Bank Association and the Association of Swedish Finance Houses (“the Associations”) welcome the possibility to express its views on the Draft Guidelines on loan origination and monitoring.
As general remarks, we note that:
• It is our understanding that the objective of the guidelines is to promote convergence and a level playing field among credit institutions across Europe (“institution” and “bank” are used interchangeably throughout this response), which as such is a commendable target. However, many of the requirements in the current version of the guidelines, seem to go beyond existing legal texts and the harmonisation/clarification mandate of the guidelines. Any changes to existing legislative frameworks should be subject to the co-decision process.
• Although the text states that the guidelines should be subject to the principle of proportionality, we note that this principle is not respected in practice in the wording of various parts of the proposed guidelines. The wording in the guidelines suggest that the same level of information requirements, decision process, monitoring activities etc. are expected, regardless of the size of the credit institution, the risk level, the type and size of clients and the relevant credit agreement. An appropriate application of the proportionality principle should be ensured throughout the guidelines, allowing for differences in these parameters.
• There are inconsistencies in the depth on which the proposed guidelines tackle the different aspects. Whereas some parts are more general/high level, others are far too prescriptive. The very detailed parts of the guidelines might, in some cases, hinder reasonable adaption to the actual circumstances and prerequisites for different segments and rather lead to a check list approach when applying the provisions, which may defy the overall objective of ensuring a prudent approach to credit risk taking, management and monitoring. We see a risk that too prescriptive guidelines could lead to diverging application in practice, where some institutions and national competent authorities would focus on achieving the overall objective while others might apply more of a check list approach, thus defying the objective of harmonisation.
• As a general rule, we think that the information requirements and processes:
- Should be only indicative examples, leaving room for the institutions to adapt it to the indicators relevant to their portfolios (which could be more, less, or simply different to the ones suggested)
- should be gathered if it does not represent an undue cost;
- should be adapted to the materiality of the portfolios and the local specificities
- should also be adapted to the differences in approach towards exposure class retail and corporate respectively
• The scope of the proposed guidelines is very broad and includes some areas which are already regulated elsewhere. As a general rule, we suggest that aspects which are already or are soon to be regulated in other legal acts, such as anti-money laundering and green lending provisions, are not repeated or supplemented in these guidelines. Where relevant, reference could instead be made to the existing legal provisions.
• Definitions should be coherent with existing legislation. In cases where there is an already existing definition in other acts of law, we suggest that reference is made to such existing definitions rather than repeating or modifying the definitions in these guidelines. In the proposed guidelines, there are several definitions which are not aligned with the applicable legislation, for example the definitions of risk (e.g. transition risk), definition of CRE and RRE, or other definitions in the area of IT or infrastructure. There are already established definitions of those concepts, for example in the “G20 Green Finance Synthesis Report”, in BCBS document 239, or Article 4 of CRR. New terms such as “professional borrowers” should be avoided for clarity. New definitions also give rise to the likelihood of inconsistency in reporting.
• The guidelines do not appropriately take into account the degree of digitalization in the banking industry. The most problematic element in this regard is the limitation on the use of advanced statistical collateral valuation. Full application of requirements will lead to reducing effectiveness and/or customer satisfaction, or even drive some entities out of certain business areas due to unreasonable costs, limiting the supply of credit to the economy.
• Another objective of the guidelines is to ‘ensure that newly originated loans are of high credit quality’. However, credit quality is a risk appetite theme and the guidelines shall not restrict portfolio diversity. Standardisation would not only limit banks’ capacity to finance innovation and entrepreneurship, but it would also raise questions of financial stability due to lack of diversification. We therefore suggest that this particular objective is deleted or rephrased as to “ensure that the credit origination processes and credit assessments for newly originated loans are of high quality”.
• There is no evidence presented in the consultation paper that an organisation with detailed centralised steering is the best organisation to manage high credit quality for all institutions in all EU countries, having different local market conditions (some with other currencies than EUR), having different organisations currently (centralised or decentralised), different business models (holding financial assets or selling financial assets) etc. On the contrary, many other organisational types than detailed centralised steering, adapted to the geographical and individual conditions, have proven to be more successful in granting loans with high credit quality and monitoring the credit risk. Best practice from a systemic risk perspective is reached if the credit institutions themselves decide on the best organisational set-up, and competent authorities have the tools required to evaluate and put pressure, to ensure the institution actually has the best organisational set-up best suited for its specific conditions, in order to fulfil the regulatory requirements.
Question 1: what are the respondent views on the scope of application of the draft guidelines?
The guidelines should only apply to newly originated loans and credit facilities granted after the application date, and not to loans existing before that date. The regular credit review of a deal should not trigger any of the new requirements. Complying with the requirements regarding the collection of information is operationally unachievable for the stock of operations.
The draft guidelines take the approach of grouping credit assessment requirements according to the type of borrower (consumers vs professionals), whereas banks’ processes for corporate lending are typically based on industries / activities, size of the company or the exposure and the complexity in the customers operation. Banks need to be able to adapt the credit granting process to the approach taken for modelling credit risk, and the guidelines should allow for this.
The guidelines are in general appropriate in relation to significant-amount transactions to large corporates, which justify the additional costs connected with further detailed creditworthiness analysis and wider information collection required. On the other hand, some of the required information may not be available at all for consumers or small and medium enterprises. We would recommend applying the EBA requirements based on the customers’ and loans’ characteristics and/or a more granular differentiation based on exposure class. This would avoid implementation of unduly disproportionate requirements. This issue is particularly relevant in countries where the business environment is mainly characterized by SMEs. Consequently, the guidelines implementation could have negative effects on credit granting, especially for smaller companies if they are not properly calibrated to the business portfolio of banks.
We welcome the explicit statement that the guidelines should be applied in line with the proportionality principle but suggest that some sections of the guidelines are revised in order to allow for proportionate application in practice.
If adopted, the requirements in the guidelines significantly impact the credit granting and managing process, with expected increases in the work time required in the bank’s organisational procedures and in the costs for the clients. In particular, the greatest impacts will be on IT structure and staff training. Banks will need sufficient time to adapt their investment and operational structure to the new standards. The set deadline is not realistic in light of the proposed supervisory expectation level. Additionally, the proposals seem to us as going beyond the harmonisation scope of the guidelines.
Considering the complexity of implementing the EBA requirements as currently stated, it is fundamental to allow for a longer implementation period or alternatively for a phase-in period. In any case, the requirements should not apply before 31 December 2021.
Any requirements regarding sustainable or green lending set out here should not precede the further regulatory framework currently being created for ESG instruments. We would like to stress that any guideline requirements should be aligned with the Commission’s Action Plan and regulatory initiatives that decided not to include credit provision activities in the scope of the taxonomy & disclosure regulations. We would like to recall that the EBA was tasked in the CRR with a mandate to incorporate ESG factors in risk management via revised technical standards. Thus, these considerations will be already embedded into the loan origination process under the new technical standards.
We suggest that the proposed ESG risk related requirements (including climate risks) are deleted at this stage and that the matter is revisited at a later point in time, in line with the timeline set in CRR2. In any case, provisions regarding ESG risks should not be set effective before the beginning of July 2022. A substantial amount of work and resource spending is needed by banks in order to comply with the proposed ESG risk related requirements. That includes e.g. policy drafting, developing of various ESG risk assessment methodologies, defining processes, implementing the preceding into actual work streams and IT systems, testing the IT systems and improving them after tests, hiring of additional human resources and training of employees etc.
When the matter is revisited at a later stage, we propose that the any guidelines should be coherent with the practice to talk about “sustainable” rather than “green”. The first term is more encompassing and allows to refer to a wider breadth of initiatives that go beyond pure “green”. Moreover, it is important that the regulations give banks a possibility to have several different approaches to ESG risks (including climate risks) because of the multitude of differing situations faced by banks. Acting that way banks will be able to focus on material ESG risk cases/industries and manage the most relevant risks.
The scope of any ESG related requirements should also be applied in a way that respects the principle of proportionality. As an example, the requirements proposed in paragraph 49 are too burdensome and difficult to track given also the scope of the Non-Financial Reporting Directive and difficulties to gather data from SMEs, which constitute a large loan portfolio for banks. It is important that any requirements in this area also work in a situation where a bank originates large number of sustainable loans (e.g. hundreds or thousands) in a year, compared to the relatively small numbers at present. Without adequate data for monitoring, sustainable loans cannot be offered. Also, we would like to suggest that any disclosure requirements do not require banks to publicly disclose such information about their sustainable lending targets etc. that is commercially sensitive.
We acknowledge the fact that the EBA pays attention to technology-enabled innovations. Technological innovations give potential to increase market access, improved methods for measuring credit risks, convenience for the clients and can lower the costs for the clients. The financial technology is transforming the business models of financial service providers; this goes for new players in the financial market as well as for incumbent banks. Diversity in assessment methods and distribution channels may also contribute to a more efficient and resilient financial system. Any requirements or qualitative expectation on credit risk management process should be neutral regarding techniques and methods as long as they are proven prudent, hence all parts of the guidelines should be permeated with this approach. In our view this dimension is however not fully taken care of in the proposed guidelines – the perception when reading the criteria is that there is still a tilt towards an analogue process. So, the conclusion is that several parts of the guidelines are not future proof from a technology innovation perspective
The wording in the section suggests that the same level of requirements regarding assessment, monitoring activities, reports and control are expected, regardless of the size of the credit institution, the risk level, the type and size of clients and the relevant credit agreement. An appropriate application of the proportionality principle should be ensured in this section of the guidelines, allowing for differences in these parameters.
In several instances, the Guidelines state the criteria listed are to be applied on a “at least” basis which would seem to imply they are binding. However, the criteria listed for example in Annex 1 are reasonably not applicable in all situations. As such, the expression at least" is not appropriate and does not allow for a proportionate application of the guidelines."
There are requirements regarding governance in already existing legal acts, such as GL 11, some of which are duplicated in the proposed guidelines. We suggest that aspects which are already regulated in other legal acts are not repeated or supplemented in these guidelines, but that reference instead is made to the existing relevant legal provisions.
We see an obvious risk that the requirements on credit decision making may limit proven and well-functioning lending activity and well-reasoned decision-making structures. Defining the organizational control and monitoring structures, policies and procedures on conflicts of interest based on the detailed requirements that appear to be set in the guidelines seems extremely challenging and do not give enough room for bank individual processes and structures that is already proven prudent.
In more detail:
• limitations in credit decision making in terms of time and number should be removed. The number of delegated credit decision is not correlated to an increase in terms of risks undertaken by the bank (e.g. para 59 limitations on the number of delegated approvals)
• Paragraph 63, allowing individual approval authorities only for small and non-complex transactions could significantly increase the complexity of the lending process. This could decrease the level of efficiency of banks. We propose to eliminate point a) para 63 for all the banks that can ensure a credit process that includes an independent opinion released by the risk management function (that ensures the independency of the overall judgment, limiting the discretion of the delegated role).
The requirement set out in the Guidelines for the Credit risk management and internal controls framework paragraph 76 to provide an “independent risk opinion to the credit decision takers” (par 76c) and an “independent/second opinion to the creditworthiness assessment” (par. 76g) seem to require an ex-ante supervision of the risk management function within the credit process. This approach, implying an active and operational role performed by the risk control function during the lending phase (which in our view seems to some extent contradictive in relation to the provisions in GL 11), might be hardly applicable as:
• the prior involvement of the risk control function appears not fully coherent with the separation of responsibilities between the ex-ante first line of defence (lending functions) vs the ex-post second line of controls (risk management) and, ultimately, with the regulatory principle of segregation of duty;
• the need to have second opinion to the creditworthiness assessment might trigger process inefficiencies related to the duplication of activities and skills in charge of different functions, entailing inter alia also additional staff costs.
• It seems to restrict the possibilities of using different operational models or different governance structures and by that hampers the bank’s possibility to adapt its credit risk management process to the chosen governance structure.
Therefore, risk management should not be interpreted as a function to be uniquely performed by a specific risk office or the risk control function and “independence” as a notion should not be tied to the second line of defence. The banks should be free to apply any governance model as long they take care of the principles of independence, segregation of duties, duality and avoidance of conflict of interest. The same reasoning is also relevant for the sections 4.4 and 4.4.1 in the proposed guidelines. It is our view that these comprehensive principles and room for manoeuvre for the banks ought to be reflected and more explicitly presented in the content of the guidelines.
We suggest that this section of the guidelines is revised in order to better respect the principle of proportionality, as expressed in the scope of application section. It should be clarified that the list in Annex 2 is only examples of information that could be collected and verified, only if they are relevant for the type of client and product, according to the proportionality principle. The expression “at least” does not seem accurate as it implies that this information always has to be collected, which would not allow for the application of the proportionality principle.
For example, there are situations, e.g. for short term credits used to buy standard consumer goods (see the ruling by the Swedish Administrative Supreme Court ref. 5868-16), where it would be appropriate to base a credit worthiness assessment on information from a standard scoring model, in which case there is a limited need to collect further information about the client.
As another example, asking for the mandatory availability of business plans and projections from all clients is in clear contrast with the proportionality principle and the evidence that smaller (and therefore internally not structured) counterparties do not usually have managerial ability to develop such detailed documents. In such cases banks’ assessment should be allowed to rely on most recent historical performances and sufficient key budgeted figures (where available) with the aim to understand their future sustainability. Involvement of internal specialist functions for all transactions is in fact not sustainable.
The Associations would like to emphasize that any requirements regarding collection of information and questions to be posed to the customer regarding creditworthiness assessment of consumers need to be adapted to the digital world. It is for example in our view not always the best approach to ask the customer about an extensive amount of information, when it could be sufficient, given the circumstances in the specific situation, to use standardised estimations of costs for inter alia living expenses or when relevant or even more reliable information is publicly available, e.g. via reports and/or could be retrieved from independent credit reference agencies.
The supervisor’s expectations regarding the sensitivity analysis in paragraphs 101 (and 114 and 121) should be clarified. The requirements must be properly delimited by the proportionality principle, as for limited proposals and retail SME customers the requirements are not proportionate to the risk.
The Associations propose that the wording in paragraph 93 is amended as follows, in order to take account of the proportionality principle: [Please see the proposed revised wording in the uploaded file.]
We suggest that this section of the guidelines is revised in order to better respect the principle of proportionality, as expressed in the scope of application section. It should be clarified that differences in the process for assessment of the borrower’s creditworthiness could exist depending on e.g. whether the client is a consumer or a corporate, an SME or a listed company.
We do not agree with the statement on page 11 that the consumer protection aspects set out in the guidelines when dealing with the creditworthiness assessment of consumers should not be subject to the application of the principle of proportionality and that the same consumer protection framework should be applied regardless of the size and complexity of the institutions or of the loan. The principle of proportionality is relevant when applying the consumer protection requirements and it would lead to negative effects for the consumer if the same level of protection would always be required. By way of example, the Swedish Supreme Administrative Court has in a ruling (ref. 5868-16) confirmed that creditors assessment of what information is relevant for the assessment of creditworthiness can vary depending on the size of the credit and other circumstances surrounding credit agreement. In the ruling, reference is made to the CJEU ruling CA Consumer Finance SA, C-449/13 (EU:C:2014:2464, p. 36-37). In the situation assessed by the Supreme Administrative Court, the credit agreement was for a limited amount, used to buy standard consumer goods and a credit assessment mainly based on a scoring model was considered to be sufficient.
The Associations acknowledge the importance of protecting the consumers’ and other clients’ interests but would like to highlight the need to consider whether the type and extent of information to be gathered for the purpose of a credit assessment is proportionate, in relation to the purpose of assuring that the loan is suitable for the client. Some of the provisions in this section seem disproportionate in this aspect. In general, we consider credit granting criteria set out in Annex 1 too detailed and it should be clarified that the criteria listed in the annex is only examples of criteria that could be relevant, depending on the type of client and credit product.
Furthermore, some of the requirements, e.g. in paragraphs 112 b) and c) and 166, are excessively burdensome and impossible or difficult to fulfil. As a matter of fact, lenders have no data and cannot be responsible for assessing the quality of architects, engineers who take part in the property development. The certification of the costs associated with the development is not easy to obtain and it could be very expensive for the borrower. We propose to eliminate these requirements.
The requirements in paragraphs 144 to 146 must be properly delimited by the proportionality principle; otherwise each credit decision must be accompanied by very complex information, by multiple stress tests - idiosyncratic, general, combined.
In general, we consider section 5 to be too prescriptive and detailed, not allowing for a relevant level of proportionality.
The chapters 4 to 8 in the proposed guidelines contain many detailed requirements, describing exactly how routines, processes and IT-systems should be set up. For institutions in the member states fulfilling the currently applicable requirements and regulations regarding loan origination and monitoring credit risk, but not having exactly the same routines and processes as those prescribed in the proposed guidelines chapters 4 to 8, and not using the same classification requirements as proposed, there will be large costs and significant work required to update processes and IT-systems to fulfill the proposed detailed requirements. The Associations acknowledges a need for sufficient tools for the competent authorities to prevent institutions originating new loans with too high credit risk and substandard credit assessment, but for institutions with existing highly functional credit risk management and proven low credit risk and high credit quality when originating new loans, the proposed guidelines would entail very high implementation costs, but are unlikely to result in improved credit quality.. The competent authority in each EU member state already has the mandate to monitor and follow up the process of loan origination and monitoring according to the purpose of the applicable legal provisions and take action when necessary. In our view, it is important to allow the competent authorities to adapt their supervision and communication according to geographical differences within the EU; overly prescriptive guidelines would in this aspect be a constraint in the authorities’ possibility to perform their supervision effectively.
In credit risk management pricing should never be a factor to justify credit risk. An assessed credit risk is based upon the borrower’s ability to fulfil the loan agreement over time, pricing should not be a factor when granting new loans. Therefore, the Associations question whether this section should be included in the guidelines and would like to suggest that the whole of section 6 regarding pricing is deleted.
We suggest that this section of the guidelines is revised in order to better respect the principle of proportionality and existing well-functioning market practices. It is also important to ensure that this section of the guidelines is consistent with other union acts of law regarding valuation, e.g. the provisions in the Capital requirements Regulation (CRR) and the Mortgage Credit Directive, and do not impose stricter or different requirements.
The proposed definitions of Commercial Real Estate (CRE) and Residential Real Estate (RRE) are not in line with the definitions used in the CRR, which will lead to difficulties when applying the guidelines. We suggest that the same definitions as in the CRR are used and that dynamic references to those definitions in the regulation are used, in order to ensure consistency over time.
The proposed guidelines do not seem to allow for the use of advanced statistical models for valuation purposes at origination, even though these models produce reliable results. Disallowing its use would in our view not contribute to making banks’ standards more robust. Instead, it would result in additional costs, without direct benefits to the client and the bank. These extra costs would directly and/or indirectly be carried by the customer, decreasing the incentive to switch lender and therefore increase inertia in the loan market, which would be of further disadvantage for the customers. For asset classes where advanced statistical models have proven to be reliable, we advise EBA to allow a continuation of its use, i.e. we are in favor of Option 3c, as presented in the draft cost-benefit analysis/impact assessment. If the EBA does not consider this to be appropriate throughout the union, the use of advanced statistical models could be allowed on a member state level, at the discretion of the relevant national competent authority. Similar solutions, for when there are well-developed and long-established market practices on a member state level, are found for example in CRR articles 199(3), 199(4) and 199(6).
The proposed guidelines allow for the use of a statistical model for revaluation purposes given that the model fulfills a set of criteria (e.g. the model should account for individual characteristics of the property, it should be property-specific, valid and accurate) and given that CRR article 208(3) does not apply. There are however mechanical valuation methods, including mechanical valuation methods with some human interaction, that may not be considered to be statistical models from a strict mathematical perspective, but which fulfill the set of criteria in the proposed guideline and thus would perfectly serve as a revaluation method. The Associations believe that the idea behind the guidelines is to allow these mechanical methods for revaluation purposes. This should however be clarified. The Associations propose that the wordings “advanced statistical models” and “statistical models” in section 7 of the proposed guidelines are replaced with the phrasing used in CRR article 174; “advanced statistical models”, “other advanced mechanical methods” and “statistical models and other mechanical methods”, and thus aligned with the provisions in the CRR.
It is important that regulators move in a direction that supports smarter and more automated valuation methods (AVM) in situations where the real estate markets are well established, especially for residential real estate. The regulatory burden and costs would increase considerably if banks were required to employ certified valuers in a well-established residential real estate market, where this is in fact not needed. The use of eligible AVM methods (not indexation) and registered data is increasing, creating more reliable data. It is important that AVM methods are put on the same footing as manual valuations (RRE) and considered as independent. In some member states these methods are to be considered ever so reliable, as the data they stem from often is extensive and accurate. Since these methods are based on numbers alone rather than subjective opinions from a valuer, one could argue that they in some cases are more accurate than the latter. The data is supposed back-tested by the AVM vendors and/or the bank in the same operation.
The most important aspects of property valuation are to do the valuation independently and with a high knowledge of the local real estate market. It is also important that the valuer has a suitable education and skill to perform a reliable and prudent valuation. In our opinion, the EBA guidelines should state and support that long-term competence and experience within the institution in valuing properties is of value. We suggest that the institution’s personnel working with valuation should have appropriate training (which could be done internally within the institution). Internal valuers should be appointed officers with a long experience of the real estate market, and there could be a requirement for internal approval and supervision. An appointed bank officer, using professional valuation methods in line with good market practice, should be considered acceptable to perform valuation of RRE.
The Associations find the wording “… at the point of origination” in paragraph 191 unfortunate. A recent valuation made some time, e.g. six months, before the point of origination would be sufficient, provided that the institution makes sure the valuation is still reasonable and relevant or reviewed and supplemented. The Associations hence propose that paragraph 191 is amended as follows: [Please see the proposed revised wording in the uploaded file.]
In order to allow for the use of advanced statistical models on a member state level, when this is appropriate, the Associations propose that a new paragraph 194 a) with the following wording is added:
“Institutions may derogate from paragraph 194 and use advanced statistical models for valuation of immovable property being used as security for loans to consumers, for origination purposes, if
- the property is situated within the territory of a Member State, and
- the competent authority in that Member State has approved the use of such models.”
The Associations moreover propose that paragraph 199 is amended as follows: [Please see the proposed revised wording in the uploaded file.]
The requirements proposed in section 200 are too prescriptive and there is a risk that they would lead to a check list approach, instead of ensuring a prudent valuation and documentation process, thus defying the objective of the guidelines.
The requirements in paragraphs 207 to 213 would overhaul the current monitoring applied to collaterals subject to revaluation and the frequency of the update. Many banks have just modified their evaluation processes on the basis of the recent NPEs guidance. Any new changes would require high IT disburses and longer time for their implementation than what proposed in the guidelines. For example, performing full appraisals for revaluation purposes as set out in paragraph 213 instead of the current desktop ones, would significantly increase the appraisals’ annual cost, and delivery time could be delayed. Additionally, mainly in case of NPE, the debtor/asset owner wouldn’t permit an internal visit of the Real Estate asset. Also, the proposed parameters in paragraph 208 to be used to structure the frequencies of monitoring are not necessarily the best. Market volatility and risk of deterioration regarding industry, technical infrastructure and location as well as respective market price developments are deemed more suitable. The institutions do have enough experience and market knowledge to judge on the best parameter reflecting the risk structure of their portfolio. Hence, parameter for determining different monitoring frequencies should not be predetermined by the EBA.
It is not always appropriate or even possible to require rotation of valuers, as proposed in paragraph 214. There is already a requirement for appraiser rotation for non-performing loans via EBA NPL Guideline. The processes of banks have just been updated to accommodate for this new rule. The expansion to all exposures would cause yet more changes just as processes have been updated. The existing requirement for only NPLs is assessed as appropriate, while we propose to remove clause 214 which is an expansion of existing rules.
Overall, the ongoing monitoring proposed in the guidelines appears overly complex. This framework represents a burden that is not justified in relation to the average size of the banks' portfolio loans and does not allow for the institution to determine its credit risk appetite. The monitoring activity shouldn’t lead to undue additional reporting or disproportionate increase of the administrative obligations for banks.
The requirements regarding stress testing in the monitoring process should to be framed by the proportionality principle. Otherwise, using a transaction-by- transaction approach, there is the risk of burdensome procedures, information and reporting requirements.