Although the NVB is a strong supporter of consistency in the application of regulations, we do not agree with the extension of the scope of connected clients. We understand the need of combining the exposures of connected clients in the context of the Large Exposure Regime (LER), which functions as a back-stop to prevent undesired single risk concentrations .For the other areas of the CRR we do not see the added value. It will increase the unwarranted risk weight variation (answers to Q3 and Q4), it might make the reporting less transparent (answer to Q5) at least when comparing institutions.
The proposal ignores the materiality threshold within the LER, which was carefully chosen to balance the goals of identifying excessive concentration risk and keeping the process manageable. Identifying excessive concentration risk will not benefit from the proposed change, while the operational burden will dramatically worsen at a high cost. Also, we fear that the reported numbers will become more volatile (see answer to Q5) which might confuse investors.
The CRR states in the corrigenda (‘53’) that the purpose of the connected clients principle is to ensure the proper identification of ‘excessive concentration’. We believe that this purpose is adequately addressed through the Large Exposure Regime. We strongly believe that connected clients that together do not form an excessive concentration should not lead to adjustments in the calculation of risk weighted assets or changes in reporting. The risk of excessive concentration is very remote within the retail assets class.
We lack to see any benefit, while we fear additional unwarranted risk weight variation to differences in balance sheet composition, as explained in the answer to question 3.
In the Netherlands there are legal limitations on the ability to use the information that is required to assess whether two or more retail clients should be considered to be connected. Two exposures to the same client – for example a mortgage loan for a residential property and a personal car loan – cannot be connected for the purpose of risk assessment, due to legal limitations. When the client would default on his car loan, the institution cannot use this information to re-assess the risk on the mortgage loan. Let alone combining personal data of two or more individuals that might be connected.
Should every retail client with a job be connected to his or her employer? Is an institution legally in its power to periodically (once every year) verify the status of the job contracts? Also, the two parents within the same family might well be economically connected. Should every retail client with a relationship be asked the question whether there was a significant change in their relation that might affected the economic connectedness?
The calculation of the Probability of Default for many (most) clients should be revised if the connected clients definition should be interpreted in line with the EBA proposal. These labor intensive processes should lead to clear prudential benefits, which we do not see.
Instead we propose to keep the intention of the current connected clients text in the CRR (article 123 and 147) close to the legal connectedness, or economic connectedness due to power of majority votes. We should keep in mind the main reason for the term ‘connected client’ which is – according to the CRR (‘53’) – the identification of excessive concentration risks.
footnote 1: corrigenda 53:
The monitoring and control of an institution's exposures should be an integral part of its supervision. Therefore, excessive concentration of exposures to a single client or group of connected clients may result in an unacceptable risk of loss. Such a situation can be considered prejudicial to the solvency of an institution.
No, we do not agree.
We fear that the proposal of applying the draft guidelines on connected clients to the development and application of the rating systems will have several negative side effects.
To start with, the pool of clients sitting in a model development set might changes more frequently compared to currently, which will not make the rating systems more stable in terms of performance. This unwarranted volatility in the model data sets will make it more difficult to build predictive and robust models.
Also it will lead to more unwarranted risk weight variation, illustrated by the following example: if two institutions have similar loans to the same SME. One of the institutions also has an exposure to a larger company being the distributor to the SME, which – in this case – leads to the situation in which the two are considered to be connected clients.
The first institution calculates the risks on the SME in isolation:
• Asset class SME (SME based correlation calculation)
• The PD is modeled based on the financials of the isolated SME
• Within the scope of the SME Supporting Factor
The other institutions calculates and reports the risk together with the larger distribution company:
• Asset class Corporate Other (Corporate based correlation calculation)
• The PD is modeled based on the financials of the SME and the impact of the connectedness?
• It falls outside the scope of the SME Supporting Factor (due to the combined exposures)
Although the risk of lending to the SME client is similar for the two institutions (besides the potential concentration risk, which is will be in scope within the LER), the two institutions will report different risk weights for this same SME client. We view this as a undesirable.
This example illustrates clearly that variability could arise as potential undesired side effect.
Furthermore, it is noteworthy that corporate exposures are reviewed annually; retail exposures are reviewed in a bucketing approach. This is accompanied by corresponding capacity/ staffing.
There are no particular additional issues that we see, next to the fact that due to connectedness the combined exposures will more often exceed the SME Supporting cap of EUR 1.5 million.
Again, in line with the example provided under question 3, institutions that lend to SME clients might end up with different levels of capitalization for the exact same risks, just because one institution also lends to another client that is economically connected to the SME client, which triggers an exposure outside the scope of the SME supporting factor limit.
Also it is less likely – compared to Large Corporate or FI clients with larger exposures – that two or more SME clients could lead to an excessive concentration risk.
In general, from an operational point of view the proposal seems to lead to an excessive reporting requirement that cannot be justified by the – in our view – non existing benefits. Before being able to report, the required processes, systems, policies and periodical checks and balances could be very substantial and for sure compete with scarce resources (people and money) that could be spent to processes that makes the institution more robust.
We believe that the calculation of the LCR and NSFR should remain bottom up and therefore not alter just because a client is connected to another client. The challenge however remains at the reporting side. Also we fear that the swings in the various ratios (LCR, NSFR) due to changes in connectedness per reporting bucket are hard to explain to investors. The swings might be incorrectly interpreted, which could harm the institution.