Austrian Economic Chamber, Division Bank and Insurance
In most cases a control relationship triggers ‘single risk’ and hence if financial distress occurs, the controlling entity will intervene to support the troubled subsidiary and vice versa. This being said, it might be the case in situations where another party has such control as described in guideline No. 13 c), such as management rights based on contractual conditions, clauses, etc. Such types of control do not necessarily have any impact on financial difficulties, etc.
Hence in our opinion, there are situations, in which a control relationship does not lead to a single risk, as e.g. fund structures or SPV financings with paid-up equity commitment or ring-fenced financings with specific restrictions (on distributions, intercompany loans etc.).
Another situation where interconnectedness through control does not necessarily lead to a “single risk”, which is not very likely but deserves to be mentioned, occurs when a parent holding controls several sub-holdings but explicitly states that it leaves to one or more of them a high degree of autonomy without exercising managing and controlling activities. In this situation, if it is also likely that possible financial difficulties would not generate a domino effect, then the autonomous sub-holding should constitute a separate group of connected clients.
A very typical feature of many SPV types (basically those established for undertaking business initiatives, like project finance and real estate) in practice is their bankruptcy-remote style set-up. Such ring-fencing design typically results in most SPVs being economically independent from other parties´ (such as the originator´s) economic fate even if accounting rules require the relationship to be deemed a controlling one.
We only would agree with the concept that the existence of a control relationship as it would always lead to a ‘single risk’ if control relationship does not mean the same probability of default. Otherwise all counterparties within the same scope of consolidation of the parent undertaking would have to have the same PD/rating under the IRB Approach. In our opinion this would not reflect the economic reality.
This concept is also elaborated by external rating agencies (please see below the quote from Standard & Poor’s RatingDirect paper: “Parent and Subsidiary Ratings”, August 29, 2000):
It is possible for a subsidiary to achieve a higher rating than that of a parent or a consolidated family if it is insulated.
As a result of insulation, parent companies may be prevented or restricted from accessing the resources of the subsidiary. The two strongest means of insulation are through either regulatory or legal barriers, which prevent excessive dividend upstreaming, intercompany loans, or any non-arm's-length" transactions.
Generally, the rating of a bankruptcy-remote, special-purpose vehicle (SPV) can exceed a parent rating or group rating by up to three notches. Thus, while insulation can protect the credit of a stronger entity, it cannot totally delink it from a weaker parent or weaker group rating. Legal or regulatory insulation in and of itself can be sufficient to warrant an entity ICR above the group's credit quality. However, other factors in combination can serve to enhance a rating.
Insulation factors include:
· Legal structure—an SPV with a cash waterfall distribution that limits disbursements to owners;
· National legal systems—for example, in certain countries corporations can only declare dividends up to a limited amount of earnings;
· Explicit regulatory restrictions or the threat of repercussions via tariff setting authority;
· Separate management and/or board of directors;
· Different countries with different legal jurisdictions;
· Tax disincentives discouraging repatriation of dividends/cash between countries.”"
The impact is positive; it permits more accurate counterparty credit risk underwriting and steering by not mixing cash-flows that should not be related.
For most SPV financings, this should already be covered by the legal due diligence; for some other cases, extra due diligence may need to be commissioned at significant cost.
As we are dealing with information contained in the official documentation (e.g. annual reports, local registries), which is collected from group clients, it would be sufficient to document the lack of single risk in a detailed and comprehensive manner. Hence, we estimated that the impact would be overall null or low in line with the current operating practices. It is expected that no further information/analyses will be required in addition to the documents already set used by mapping functions in risk management. Otherwise, the impact will be relevant.
No further clarification required.
In our view there are no other indicators of control which should be added to the list. Nevertheless, we find the list not clear enough in terms of relations between the situations described, i.e. ranking of criteria. Especially, it should be clarified how the cases when more than one situation is fulfilled between different natural or legal persons should be treated. For example, if for the same legal entity one counterparty holds the majority of voting rights whereas a different counterparty holds the majority of shares of capital. According to our understanding the majority of voting rights should be a decisive criterion for the control relationship.
According to our assessment this is not a significant problem as far as this information can be retrieved from official documentation (e.g. financial statements, public registries). Additional costs may arise in countries where centralised databases on control relationships among companies are not available.
Indicators for such an assessment should be provided out of the Know-Your-Customer Procedures at no relevant additional cost. If there are control relationships which cannot be derived out of the KYC documents (e.g. corporate registries extracts) the cost of establishing the control relationship would be higher.
The guidance is clear. We understand that institutions are granted full flexibility in choosing the approach - full approach, alternative approach and partial alternative - to be applied to central governments, on a case by case basis, in order to limit the operating burdens.
With respect to the examples outlined in the consultation paper, the group of connected clients constituted by the controlled entities (A, A1, A2, B, B1, B2) in the case illustrated in Example CG4 seems not to be explicitly required by the CRR and therefore in our opinion it should not be considered.
Switching from the restrictive concept of 'default contagion' (“substantial existence-threatening repayment difficulties”) to the less restrictive concept of 'repayment difficulties' is counterproductive and effectively prevents proper counterparty risk assessment and risk steering.
By removing the crucial qualification that economic dependence exists only in case of 'default contagion', entire value chains may or must be connected. This creates artificial groupings that cannot be managed with the tools of counterparty credit risk management any more (financial analysis, counterparty rating,…) and significantly deteriorates the quality of credit risk management in a bank.
The further clarification provided by para 24 & para 25 (page 16) does not remove this fundamental problem of the proposed modification to the current definition.
We generally doubt whether individual loan officers and relationship managers do have the soft information needed to identify connected clients according to the new guideline (para 34, page 23). To fulfill these extended requirements banks could be forced to take additional external data providers to closely monitor their business relationships of the clients.
Economic dependence is an extremely important criterion for determining single risks that, by its very nature, can and should only be established based on a thorough case-by-case assessment. Except for para k), any of the criteria listed in paragraph. 23 (page 35) may indicate economic dependence, but does not conclusively define it. The case-by-case assessment must be conducted by adequately skilled professionals familiar with the individual case. Defining any list of criteria may – and in actual practice – does lead to situations where Group of Connected Clients formation is done algorithmically by IT-processes or by back-office functions not familiar with the particular case. This creates groupings that may not constitute a single risk which undermines the very concept and usefulness of a group of connected clients.
If “repayment difficulties” are used as basis for the establishment of client groups instead of
“Substantial, existence threatening repayment difficulties” (as used so far), the consequence could be
(i) That more effort needs to be undertaken and a more detailed analysis would be required in order to establish such difficulties (i.e. the analysis whether a compensation for a loss could be found by other means). In many cases, the detailed insight in the business might not be given or is legally impossible; thus it would be hard to see whether e.g. cost cuts are possible to compensate for a loss.
(ii) That a more judgemental factor is introduced, as one analyst might easier come to another conclusion as another if a credit repayment could be made.
(iii) More interdependencies between clients will exist which leads to more groups of connected clients. This in turn leads to higher workload, as more companies will have to be included in proposals than before. Due to the limit restrictions loan volumes will decrease. An estimation of costs is not possible without reviewing the whole portfolio in detail in order to establish the groups to be built out of this change of guidance.
As already mentioned in our general comments, we find it absolutely necessary to clarify the terminology used in the draft guidelines, especially in relation to the default definition. In our understanding (i) ‘repayment difficulties’ do not equate to ‘default’ and (ii) ‘single risk’ does not equate to ‘the same probability of default’.
In general, the guidance is clear. The threshold of “at least 50%” is a helpful benchmark.
However, room for reasoning should be left for individual cases, if one item of the list applies, but there are reasons, why nevertheless no dependency exists. Example: a client‘s gross receipts from renting a property to a tenant exceeds 50% of the client’s gross receipts, but the property is so easily rentable that another tenant can be found.
Economic interdependencies should be established, in any case, considering the materiality of the impact in the levels of credit risk for the bank due to the repayment difficulties of the connected counterparty. The examples provided in the draft guidelines help to identify circumstances where economic interconnectedness may be established; however, the guidelines should clarify that these examples do not automatically lead to an economic interdependence and therefore, these clients may not be automatically considered as a group of connected clients.
In the assessments of situations where control may exist and grouping must be done, except for companies included in the consolidation for accounting purposes, it is indicated that the listed criteria should be considered in the assessment process. Regarding economic dependency it is however written in the proposed guidelines that a dependency shall be deemed to exist (deem") if the criteria are met. In such cases it becomes particularly important that the conditions actually are believed to lead to default or non-payment issues. For many of the criteria such a condition is described as a prerequisite («… so significant for the … is likely to default or experience financial difficulties…»). For the second criterion this condition is not included. These are criteria for a possible grouping where it may be more likely than not that this will be the consequence, but not necessarily. It will, inter alia, depend on the relevant customer's relative debt burden, cost structure, general adaptability, etc. Also for criteria c (income / expense), d (production), f (assets / liabilities), h (customer base), i (ownership structure) and j (relations to co-borrowers) this should be included as a condition.
The existence of such criteria should not automatically lead to the conclusion that a grouping must be done. The expert judgement of the bank, based on credit experience and knowledge of its customers, should necessarily play a role in the decision about the existence of interconnectedness through economic dependency. Consistently, the 50% threshold should not represent a mandatory threshold that triggers the existence of such connection, but rather a warning, prompting the bank to further analysing the clients.
It follows from the introductory comments in the consultation document that the grouping should not be based on risk concentrations as a result of industry or geography. Meanwhile, criterion 23 e) could be read as that such a common source of income can be a geographical region or a sector (what you produce / sell). Thus, it would be beneficial if the guidelines, in the introduction to chapter 6, indicate that the risk concentrations as a result of industry or geography matter shall not give rise to demands for grouping.
Criterion j ("The relationship between a debtor and his / her co-borrowers") appears in any case not as a sufficiently precise indication of a situation that necessarily describe economic dependence as this among other things will have to depend on the relative importance of the loan they are co-borrowers for compared to their total economic and financial situations.
Furthermore, along with the size of the economic relationship, other aspects should be taken into account for the purpose of identifying genuine economic dependency, such as the stability of the interconnection. For example, the bank could deem that the clients should be regarded as connected only after a permanence of the link for a predefined period of time. This would help in reducing volatility in the composition of the groups of connected clients (see also the answer to question 7) and would affect the magnitude of the single risk (i.e. if a counterparty’s repayment difficulties affect temporarily or permanently the likeliness of repayment of another counterparty, the risk for the bank is different).
The list provided in paragraph 23 is detailed and exhaustive. However, in our opinion it will be very difficult to identify and prove these dependencies. It will not be feasible to assess the threshold of ‘at least 50%’ with reference to particular customers operating in less transparent markets where this information is deemed to be confidential. Generally, in such cases it is possible to check whether the situations described in paragraph 23 take place only if it is publicly available information. We think that the cost of research on economic dependency relations would be disproportionate.
We therefore call for maintaining the wording of the CEBS Guidelines ("substantial, existence-threatening repayment difficulties"), as the expression "repayment difficulties" alone is too vague. Although the wording of point (b) of Article 4 (1)(39) of the CRR only says "repayment difficulties", it was always the common understanding, that the conditions to qualify as repayment difficulties in the meaning of the CRR are: the difficulties have to be substantial and permanent economic dependency."
We are not aware of other situations that should be added. In relation to the case with the same
guarantor, we would argue that neither the guarantor has repayment difficulties, if a guaranteed entity has problems nor has the guaranteed entity repayment difficulties, if the guarantor experiences financial trouble (as it is “only” a guarantee). If there are no other factors of interconnectedness between the guaranteed entities, there would be no group to our mind.
According to our experience, this is a quite exceptional case; as the likelihood of simultaneous claims under guarantees to unrelated counterparties seems to be fairly low, we would not treat such clients as connected. The situation described in the explanatory box on page 35 would be a too far reaching interpretation of the term “single risk”. We believe that these cases shall not be considered as connected. First, such a treatment would lead to a disadvantageous treatment of smaller Member States, where the availability of guarantors is limited and therefore a group of connected clients would be reached more easily than in larger countries. Further, it would be difficult for the institutions to obtain these information and to monitor these cases.
We think that there are no other situations which should be added to the list from paragraph 23. As stated above, the list is already detailed and exhaustive. In particular, we do not find the additional situation described as constituting economic dependency, since the guarantee only refers to secondary credit risk. Different to the point (a) from paragraph 23, the primary credit risk in the situation described is unrelated to each other and therefore does not constitute a single risk due to risk diversification
While the guidance is clear, since interconnectedness via control will cover most cases, we would suggest an approach where first the interconnectedness via control is established, and thereafter economic dependency is assessed only for those entities not already covered by control, and added then if necessary, instead of doing an all-encompassing economic
dependency verification for all entities. This should limit double efforts and costs which
would otherwise be high.
The implementation of the envisaged measures is likely to determine in many cases multiple counting of risk positions to clients that should be included in different groups of connected clients. We deem that double counting of exposures (due to entities being included in different groups of connected clients) is not desirable. In fact, this would lead to the double counting of the bank exposure to the client in the bank records aimed at supervisory reporting. Consequently (unless the bank creates a double-track management system detaching the credit risk management from supervisory evidences), the double counting would affect the monitoring of bank exposure to the client for risk management purposes, in terms of determination of credit limits and allocation of credit decisions. Potential consequences could also arise with respect to the identification of the actual parent company for the purpose of the attribution of the credit rating to the client.
The occurrences of this situation should therefore be limited to the cases explicitly addressed in the CRR (that seems not to be the case with the simultaneous presence of interconnectedness through control and interconnectedness through economic dependency with different entities addressed in Section 7 of the proposed Guidelines).
In any case, the EBA proposal is quite clear, even if its application is not straightforward, for the considerations explained above but also because the assessment of the economic dependency is not always easy. In fact, the concept of “Economic dependency” and its relationship with the concept of “control” raises several doubts. Please consider the following example:
The set-up of a dedicated communication channel between banks and EBA is of the utmost importance, not only for questions that might arise, but also to harmonise the interpretations between banks.
Moreover, the concept application is not straightforward as the assessment of the economic dependency is not always easy because of the high degree of subjectivity and the lack of relevant information; furthermore the reporting of groups’ exposures in the LE framework would be more ‘volatile’ due to the changes in the groups perimeter.