Response to consultation on draft Guidelines on retail deposits subject to different outflows for purposes of liquidity reporting
Go back
Concerning the criterion that a “transactional account” shall be identified by historical data, in our view it should be deemed sufficient if a behavioural score exists for the respective client.
More specifically, in this context a more appropriate methodology seems to be the volume-based meth-odology discussed under Indent 6.
For assessing whether products are being actively managed, the following criteria might be used as indicators:
• The corresponding product category remains subsumed under new business acquisition;
• The product scores regular payment flows;
• Concerning the sales and / or process management, the product is part and embedded of banks’ activities.
Furthermore, as a special type of deposit guarantee, we suggest recognising also institutional protec-tion schemes. Institutional protection schemes are an important, additional factor which guarantees the security of customer deposits. Consequently, this should also be taken into account as an alterna-tive criterion with regard to outflows in the case of very high value of the deposits (> 500TEUR).
Concerning the thresholds for categorisation as a high or very high risk, we suggest that the EBA model its definition on the threshold defining retail exposure positions within the meaning of Art. 123(1) lit. c CRR, i.e. EUR1 million. This threshold should be applicable at least for a categorisation as very high exposure. Below this materiality threshold, a categorisation as a very high risk would be incompatible with the existing measurement mechanisms e.g. of retail holdings. Contrary to the privi-leged treatment when it comes to determining counterparty risk, we also have difficulties in compre-hending the stricter treatment of retail positions for liquidity purposes.
It is our view that the allocation of the risk factor and the allocation of a potentially higher outflow rate only apply to that part of the deposits which exceeds the forthcoming thresholds and we would ap-preciate a clarification to this effect. Otherwise this, too, would result in cliff effects (cf. also our gen-eral comments on this matter).
We also have concerns over the fact that rate decisions taken by the other banks in the peer banking group may potentially have a negative impact on the LCR ratio; this may, for instance, be the case if a bank does not emulate an interest rate reduction or if it only does so after a certain delay. In our view, this promotes a uniform behaviour within the respective peer group thus potentially further compounding emerging crisis scenarios during periods of stress. Hence, this criterion could even prove counterproductive with regard to the LCR’s key rationale - securing financial markets’ stability. Furthermore, it is not clear how supervisors are supposed to verify compliance or, moreover, non-compliance with this criterion. We hold the view that defining a peer group is a highly complex pro-cess. This complexity is compounded even further if this process has to take place in regular inter-vals, if market conditions are subject to fast change in retail banking and if the approach has to be agreed with national supervisors. Hence, it might be sufficient if banking supervisors were able to veto significant misconceptions or abuse.
First, compared to other business models, this approach would result in a wholesale discrimination against a certain business model in the absence of any valid statistical basis for this. Even during the last financial crisis, in the retail banking area (individual) internet banks were able to post considerable inflows; this is completely inconsistent with the EBA's assumption expressed at this juncture. Hence, with regard to internet banks, there are no behaviouristic signs indicating any particular risk aversion on the part of customers. In lieu of the latter, the level of the deposit guarantee scheme and the country where the bank is domiciled have a decisive influence on customer behaviour.
Furthermore, focussing the LCR on the classical distribution channels would fail to reflect the devel-opment which the banking market underwent in recent years. One of the hallmarks of recent devel-opments was that (alongside with the regular branch network) more and more clients adopted inter-net banking. This is also evidenced by a steadily rising number of electronic payment transactions.
Hence, there is no corroborating evidence for the theory that – in the eyes of the customer - an in-ternet bank (due to its “virtual nature”) was per se perceived as particularly prone to crises. Last but not least, to many clients, banks that are active in more than one region with branch network clusters primarily in larger cities may appear similarly virtual as internet-only banks. Based on the foregoing, banks which use the internet as their sole distribution channel should not be coerced into subjecting their retail deposits to higher outflow rates.
Above and beyond this, the EBA suggests that banks differentiate between deposits denominated in foreign currencies and deposits denominated in local currencies on the one hand as well as resident and non-resident deposits. For reasons related to the overall nomenclature, we have major concerns over an inclusion of these two criteria and perceive a need for further discussion; After all, customer behaviour and consequently higher outflows not only depend on a single stress scenario in the coun-try where the credit institution is domiciled. Quite on the contrary: Customer behaviour very much depends on the situation in the country of origin.
For instance, more likely than not, in the event of a revaluation of the foreign currency – i.e. if there was a “domestic currency stress” – there will be higher inflows. The same would be the case if there was an emerging crisis in the non-domestic banking market. Especially the most recent past has shown that this involves highly complex interactions. The generic definition of both criteria as risk factors does not reflect this complexity in an adequate manner. Given the proportion of such depos-its, for most banks the question arises anyway whether the meaningfulness of the LCR for the pur-poses of liquidity management would significantly increase if these factors were included.
First, there is a host of factors which, as has been pointed out, are partly also correlated with each other. This leads to a clear increase in complexity meaning that also the costs for preparing and con-trolling this ratio will most likely see an increase. This is further compounded particularly due to the fact that, following the collation of the risk factors (which partly cannot be calculated), a further algo-rithm becomes necessary for deriving the respective outflow rates. This translates into the need for a multi-tier calculation method.
Last but not least, the most recent financial crisis has revealed that the level of deposit guarantees as well as the country in which the bank is domiciled are of decisive importance for customer behaviour. Whilst, in certain sub-areas, the parameters and criteria presented in the Consultation Paper may in-deed be extrapolated and thus, theoretically, available, during past real stress scenarios these pa-rameters did not have an effect that would justify their reflection in the LCR calculation.
Do you agree with this criterion for identifying a transactional account?
The EBA identifies a transactional account on the basis of the criterion that it is the client’s transac-tional account and that the customer’s ongoing payment transactions are handled through this ac-count. In our view, these requirements are but an inadequate reflection of the banking market’s evo-lution in recent years. Thus, nowadays clients generally tend to have several accounts which they use for handling their ongoing payment transactions; yet, they only tend to have one single payroll ac-count. Hence, accounts should be identified as a “transactional account” already in the presence of one of the two criteria proposed.Concerning the criterion that a “transactional account” shall be identified by historical data, in our view it should be deemed sufficient if a behavioural score exists for the respective client.
Regarding established relationships, how would you assess that the contractual relationship with the institution and the minimum number of products are active in the sense of being actively managed?
For identifying “established business relationships”, the EBA wants to propose the criterion that the customer shall have a minimum number of active products with the institution. There may, in our view, be a possible link between the number of products used and the stability of the client relation-ship. On the other hand, some analyses by banks have also concluded that the number of products is no yardstick of stability. Conversely, even a single-product user with a sufficiently long-standing ac-count may constitute a relationship stable enough to make a higher outflow of funds in a stress sce-nario unlikely. Extensive studies in some banks have found a very high level of stability after doing business for a period of only nine months. In this regard, the correlation of the two parameters is highly customer specific; hence, this correlation is not fit for purpose as a defining element. In the light of the technical implementation in terms of IT, on the whole, we view the use of two or more active accounts not as an appropriate criterion.More specifically, in this context a more appropriate methodology seems to be the volume-based meth-odology discussed under Indent 6.
For assessing whether products are being actively managed, the following criteria might be used as indicators:
• The corresponding product category remains subsumed under new business acquisition;
• The product scores regular payment flows;
• Concerning the sales and / or process management, the product is part and embedded of banks’ activities.
What is your view concerning the threshold proposed for high and very high value deposits? Please give your reasons.
The EBA proposes certain thresholds for high and very high value deposits. On principle, we share the view that there is a correlation between the total level of a customer’s deposits and its sensitivity with regard to the stress scenarios. Having said this, in our opinion the definition of the threshold for high value deposits is a source for concern. This is due to the fact that any potentially higher coverage of these deposits through deposit guarantee schemes remains unconsidered, even if these deposit guarantee schemes, though established on a voluntary basis, are fully recognised by national lawma-kers. However, especially the coverage of potential losses during crises has a far greater impact on the clients' sensitivity during stress scenarios than the absolute level of the depots. Hence, in our view it would be more meaningful to define high value deposits as such deposits which are not cov-ered by a deposit guarantee scheme (statutory or voluntary). The threshold for extremely high value deposits should then be a multiple of this threshold for high value exposures (for instance, it could be modelled on the EBA proposal, i.e. a five-fold multiple). Furthermore, we take the view that it should be possible to exclude deposits covered by an effective deposit protection scheme from checks for any further risk factors since there is no pressure to withdraw such deposits even in the event of a crisis.Furthermore, as a special type of deposit guarantee, we suggest recognising also institutional protec-tion schemes. Institutional protection schemes are an important, additional factor which guarantees the security of customer deposits. Consequently, this should also be taken into account as an alterna-tive criterion with regard to outflows in the case of very high value of the deposits (> 500TEUR).
Concerning the thresholds for categorisation as a high or very high risk, we suggest that the EBA model its definition on the threshold defining retail exposure positions within the meaning of Art. 123(1) lit. c CRR, i.e. EUR1 million. This threshold should be applicable at least for a categorisation as very high exposure. Below this materiality threshold, a categorisation as a very high risk would be incompatible with the existing measurement mechanisms e.g. of retail holdings. Contrary to the privi-leged treatment when it comes to determining counterparty risk, we also have difficulties in compre-hending the stricter treatment of retail positions for liquidity purposes.
It is our view that the allocation of the risk factor and the allocation of a potentially higher outflow rate only apply to that part of the deposits which exceeds the forthcoming thresholds and we would ap-preciate a clarification to this effect. Otherwise this, too, would result in cliff effects (cf. also our gen-eral comments on this matter).
Do you agree with the criterion for considering a deposit to be rate driven?
We share the EBA’s assessment that – also in crises - there is a link between a customer’s focus on rates and this customer’s propensity to change. Also the criteria proposed appear plausible. However, generally speaking, we have concerns over banks’ capacity to identify peer banking groups and to compare different products within this peer banking group on an ongoing basis. In our view, it would be more productive if banks themselves were entitled to identify a product to be rate driven. Due to the fact that banks know their own funding costs, this information would be easier to establish; also, it would feature a greater degree of precision.We also have concerns over the fact that rate decisions taken by the other banks in the peer banking group may potentially have a negative impact on the LCR ratio; this may, for instance, be the case if a bank does not emulate an interest rate reduction or if it only does so after a certain delay. In our view, this promotes a uniform behaviour within the respective peer group thus potentially further compounding emerging crisis scenarios during periods of stress. Hence, this criterion could even prove counterproductive with regard to the LCR’s key rationale - securing financial markets’ stability. Furthermore, it is not clear how supervisors are supposed to verify compliance or, moreover, non-compliance with this criterion. We hold the view that defining a peer group is a highly complex pro-cess. This complexity is compounded even further if this process has to take place in regular inter-vals, if market conditions are subject to fast change in retail banking and if the approach has to be agreed with national supervisors. Hence, it might be sufficient if banking supervisors were able to veto significant misconceptions or abuse.
Do you agree with the criteria to identify this risk factor?
The explanatory text lists a bank which conducts its customer transactions exclusively over the inter-net as one example for a bank which, due to higher risk distribution channels, should define higher outflow rates for its retail deposits. We hold the view that this definition is erroneous on various counts.First, compared to other business models, this approach would result in a wholesale discrimination against a certain business model in the absence of any valid statistical basis for this. Even during the last financial crisis, in the retail banking area (individual) internet banks were able to post considerable inflows; this is completely inconsistent with the EBA's assumption expressed at this juncture. Hence, with regard to internet banks, there are no behaviouristic signs indicating any particular risk aversion on the part of customers. In lieu of the latter, the level of the deposit guarantee scheme and the country where the bank is domiciled have a decisive influence on customer behaviour.
Furthermore, focussing the LCR on the classical distribution channels would fail to reflect the devel-opment which the banking market underwent in recent years. One of the hallmarks of recent devel-opments was that (alongside with the regular branch network) more and more clients adopted inter-net banking. This is also evidenced by a steadily rising number of electronic payment transactions.
Hence, there is no corroborating evidence for the theory that – in the eyes of the customer - an in-ternet bank (due to its “virtual nature”) was per se perceived as particularly prone to crises. Last but not least, to many clients, banks that are active in more than one region with branch network clusters primarily in larger cities may appear similarly virtual as internet-only banks. Based on the foregoing, banks which use the internet as their sole distribution channel should not be coerced into subjecting their retail deposits to higher outflow rates.
Above and beyond this, the EBA suggests that banks differentiate between deposits denominated in foreign currencies and deposits denominated in local currencies on the one hand as well as resident and non-resident deposits. For reasons related to the overall nomenclature, we have major concerns over an inclusion of these two criteria and perceive a need for further discussion; After all, customer behaviour and consequently higher outflows not only depend on a single stress scenario in the coun-try where the credit institution is domiciled. Quite on the contrary: Customer behaviour very much depends on the situation in the country of origin.
For instance, more likely than not, in the event of a revaluation of the foreign currency – i.e. if there was a “domestic currency stress” – there will be higher inflows. The same would be the case if there was an emerging crisis in the non-domestic banking market. Especially the most recent past has shown that this involves highly complex interactions. The generic definition of both criteria as risk factors does not reflect this complexity in an adequate manner. Given the proportion of such depos-its, for most banks the question arises anyway whether the meaningfulness of the LCR for the pur-poses of liquidity management would significantly increase if these factors were included.
Do you agree with the above analysis of the cost and benefit impact of the proposals?
We explicitly disagree with the EBA’s view that the proposed further differentiation of the outflow rates can significantly increase the meaningfulness of the LCR ratio whilst incurring but moderate costs.First, there is a host of factors which, as has been pointed out, are partly also correlated with each other. This leads to a clear increase in complexity meaning that also the costs for preparing and con-trolling this ratio will most likely see an increase. This is further compounded particularly due to the fact that, following the collation of the risk factors (which partly cannot be calculated), a further algo-rithm becomes necessary for deriving the respective outflow rates. This translates into the need for a multi-tier calculation method.
Last but not least, the most recent financial crisis has revealed that the level of deposit guarantees as well as the country in which the bank is domiciled are of decisive importance for customer behaviour. Whilst, in certain sub-areas, the parameters and criteria presented in the Consultation Paper may in-deed be extrapolated and thus, theoretically, available, during past real stress scenarios these pa-rameters did not have an effect that would justify their reflection in the LCR calculation.