Response to second consultation on RTS on estimation and identification of an economic downturn in IRB modelling
Go back
We are aware of the importance to improve and ensure the credibility of Risk Weighted Assets (RWA) reducing its variability, in particular regarding the downturn identification and LGD estimation, that can be particularly challenging.
Nonetheless, given the significant impact these RTS and Guidelines can have on the RWA, we would like to raise in the following some issues that in our opinion need further clarification or amendments.
In more detail, regarding the RTS to identify an economic downturn, we deem that the EBA list of economic factors which are relevant for the purpose of specifying the nature of an economic downturn for a considered type of exposure is complex to retrieve, especially on a 20-year look back horizon. Furthermore, in case of more than one downturn, we are concerned about the choice of identifying the most severe downturn period, and we suggest to use the average, that is in our opinion, more representative of a downturn realization.
With respect to the Guidelines, instead, we deem some further details are needed on many aspects, particularly those related to the recognition the observed impact of downturn and its interactions with other estimation purposes (ELBE, IFRS9, Stress Test), as well as those related to the interaction between downturn, model component, and attribution of the downturn effect at overall LGD level in the estimation approaches proposed. Finally we deem critical the adoption of the Reference Value Approach as a challenger for the estimation, especially where a solid estimation has been possible, and we rather propose to use it as an alternative to the adoption of a +20% add-on when the observed or estimated impact is not available.
Finally, we deem important to provide clarifications regarding what mentioned in Article 5 (Final Provisions) with respect to the entrance into force on the 31st of December 2019 of the RTS on nature, severity and duration of an economic downturn. Indeed, since all the IRB Repair Program - including also the EBA Guidelines on LGD downturn estimation - should be targeted by end -2020, we deem fundamental to ensure full alignment of the implementation of this complementary guidelines, given also of the strict interconnections to each other, and considering that the full package is aimed at reducing the unjustified variability of the LGD estimates.
With regard to the workability of the new approach (question 1) we have two major concerns. In particular:
- On the Severity of the economic downturn, as specified in Article 3 of the RTS, the choice of identifying the downturn period as the most severe one remains, in our opinion, critical. In case of more than one downturn period, we deem that all of them represent relevant empirical evidences that should be considered in calibrating the downturn estimate. Since Article 181(3) of the CRR related to severity, nature and duration, does not necessarily require to use the worst case scenario, we suggest to consider, instead of the most severe, the most representative scenario, that could be calculated as an average value of the observed economic factors within all the historically occurred downturn events. Furthermore, the Credit Factors and LGD (conditioned to the analysis of their significance with the dynamics of the loss rates as correctly stated also in Section 5 of the Guidelines) should be representative not necessarily of the “worst downturn period” ever registered, but of the downturn period if characterized by higher values then the long run.
- On the Nature of downturn, it would be beneficial to better define “what is a downturn event” taking into account the different characteristics of each economic factors considered (e.g. for GDP a downturn event is when a negative yearly variation occurs). Moreover, Article 2 reports a long list of economic factors aimed at identifying the nature of an economic downturn that are hardly retrievable and not always available on a 20 years time span. As shown in the table below on the EUROSTAT (official EU Statistical Bureau) statistics, a number of requested series are not available or do not cover the whole time series requested.
see Table 1 (List of economic factors) in the attached file
Furthermore, as clarified by the Table, the large variety of index sub-specifications within the same macro-economic factor may be a relevant and undesired source of risk estimate and RWA variability. As a matter of fact, even using the same macro-economic factor, different institutions may actually select different sub-specification of the same variable, thus maintaining the current heterogeneity that this RTS wants to reduce. In this regard, we suggest to provide a clear reference of the sources where the data should be collected (in order to allow all Banks to adopt the same sources fostering a proper harmonization) both for EU and not EU countries, as well as for different industries (for the customization and co-movement analysis required in paragraph 3).
Furthermore, in our view, two more points should be better addressed:
- we suggest to provide detailed indication on the prioritization of the macroeconomic indicators which should be taken in consideration for the downturn estimation. We deem that, in order to identify the downturn period, some indicators must be associated with other macroeconomic factors, e.g. a pick in a factor time series shall be relevant only if combined with picks observed on other factors (considered with a high priority). With reference to the Article 3 on the severity of downturn, we think it should be better clarified whether in paragraph 2 the possibility to start the 12-months period for the considered economic factor at any points in time means to adopt e.g. quarterly data of annual realization of the economic factor or to have a different reference date (e.g. March, June, etc.), keeping in any case a yearly frequency. This choice is relevant for the analysis of adjacent peaks/trough and the possibility to have longer than 12 months period of downturn;
- the adoption of absolute value (i.e. level) of the economic factors (as in the graphical examples reported in the explanatory section) are, in our view, not appropriate considering that, as proved in the economic literature, the factors considered in the analysis are typically integrated of first order with trend components. Given also the need to assess the significance of the economic downturn with respect to the loss rates, we suggest to use its variation, using as a support the analysis at absolute level.
Furthermore, the different time horizon of CFs realization (1-year compared to the LGD one, covering typical a multi-year fashion) should be taken into account when the downturn effect should be included into downturn EAD estimation.
Question 1: Do you have any concerns around the workability of the new approach (e.g. data availability issues, burden on the analysis, split between the definition of the economic downturn and its impact on the internal loss data)?
UniCredit Group supports EBA’s efforts to harmonize IRB framework and welcomes the opportunity of participating to the consultation on the European Banking Authority (EBA) Second Draft RTS on the Specification of the Nature, Severity and Duration of an Economic Downturn, as well as the Guidelines for the estimation of LGD appropriate for an economic downturn”.We are aware of the importance to improve and ensure the credibility of Risk Weighted Assets (RWA) reducing its variability, in particular regarding the downturn identification and LGD estimation, that can be particularly challenging.
Nonetheless, given the significant impact these RTS and Guidelines can have on the RWA, we would like to raise in the following some issues that in our opinion need further clarification or amendments.
In more detail, regarding the RTS to identify an economic downturn, we deem that the EBA list of economic factors which are relevant for the purpose of specifying the nature of an economic downturn for a considered type of exposure is complex to retrieve, especially on a 20-year look back horizon. Furthermore, in case of more than one downturn, we are concerned about the choice of identifying the most severe downturn period, and we suggest to use the average, that is in our opinion, more representative of a downturn realization.
With respect to the Guidelines, instead, we deem some further details are needed on many aspects, particularly those related to the recognition the observed impact of downturn and its interactions with other estimation purposes (ELBE, IFRS9, Stress Test), as well as those related to the interaction between downturn, model component, and attribution of the downturn effect at overall LGD level in the estimation approaches proposed. Finally we deem critical the adoption of the Reference Value Approach as a challenger for the estimation, especially where a solid estimation has been possible, and we rather propose to use it as an alternative to the adoption of a +20% add-on when the observed or estimated impact is not available.
Finally, we deem important to provide clarifications regarding what mentioned in Article 5 (Final Provisions) with respect to the entrance into force on the 31st of December 2019 of the RTS on nature, severity and duration of an economic downturn. Indeed, since all the IRB Repair Program - including also the EBA Guidelines on LGD downturn estimation - should be targeted by end -2020, we deem fundamental to ensure full alignment of the implementation of this complementary guidelines, given also of the strict interconnections to each other, and considering that the full package is aimed at reducing the unjustified variability of the LGD estimates.
With regard to the workability of the new approach (question 1) we have two major concerns. In particular:
- On the Severity of the economic downturn, as specified in Article 3 of the RTS, the choice of identifying the downturn period as the most severe one remains, in our opinion, critical. In case of more than one downturn period, we deem that all of them represent relevant empirical evidences that should be considered in calibrating the downturn estimate. Since Article 181(3) of the CRR related to severity, nature and duration, does not necessarily require to use the worst case scenario, we suggest to consider, instead of the most severe, the most representative scenario, that could be calculated as an average value of the observed economic factors within all the historically occurred downturn events. Furthermore, the Credit Factors and LGD (conditioned to the analysis of their significance with the dynamics of the loss rates as correctly stated also in Section 5 of the Guidelines) should be representative not necessarily of the “worst downturn period” ever registered, but of the downturn period if characterized by higher values then the long run.
- On the Nature of downturn, it would be beneficial to better define “what is a downturn event” taking into account the different characteristics of each economic factors considered (e.g. for GDP a downturn event is when a negative yearly variation occurs). Moreover, Article 2 reports a long list of economic factors aimed at identifying the nature of an economic downturn that are hardly retrievable and not always available on a 20 years time span. As shown in the table below on the EUROSTAT (official EU Statistical Bureau) statistics, a number of requested series are not available or do not cover the whole time series requested.
see Table 1 (List of economic factors) in the attached file
Furthermore, as clarified by the Table, the large variety of index sub-specifications within the same macro-economic factor may be a relevant and undesired source of risk estimate and RWA variability. As a matter of fact, even using the same macro-economic factor, different institutions may actually select different sub-specification of the same variable, thus maintaining the current heterogeneity that this RTS wants to reduce. In this regard, we suggest to provide a clear reference of the sources where the data should be collected (in order to allow all Banks to adopt the same sources fostering a proper harmonization) both for EU and not EU countries, as well as for different industries (for the customization and co-movement analysis required in paragraph 3).
Furthermore, in our view, two more points should be better addressed:
- we suggest to provide detailed indication on the prioritization of the macroeconomic indicators which should be taken in consideration for the downturn estimation. We deem that, in order to identify the downturn period, some indicators must be associated with other macroeconomic factors, e.g. a pick in a factor time series shall be relevant only if combined with picks observed on other factors (considered with a high priority). With reference to the Article 3 on the severity of downturn, we think it should be better clarified whether in paragraph 2 the possibility to start the 12-months period for the considered economic factor at any points in time means to adopt e.g. quarterly data of annual realization of the economic factor or to have a different reference date (e.g. March, June, etc.), keeping in any case a yearly frequency. This choice is relevant for the analysis of adjacent peaks/trough and the possibility to have longer than 12 months period of downturn;
- the adoption of absolute value (i.e. level) of the economic factors (as in the graphical examples reported in the explanatory section) are, in our view, not appropriate considering that, as proved in the economic literature, the factors considered in the analysis are typically integrated of first order with trend components. Given also the need to assess the significance of the economic downturn with respect to the loss rates, we suggest to use its variation, using as a support the analysis at absolute level.
Question 2: Do you see any issues of applicability of this RTS for estimating conversion factors appropriate for an economic downturn identified in accordance with this RTS?
The same issues considered above are valid also in the estimation of the CFs. In addition, since no GL is provided for CFs downturn estimation that in period of downturn and in presence of strong EWS / credit monitoring processes (strongly required also by ECB Guidance to banks on Non Performing Loans), it is important to remark that the exposure and the credit limit can be shrunk by means of managerial actions, thus resulting in lower CFs.Furthermore, the different time horizon of CFs realization (1-year compared to the LGD one, covering typical a multi-year fashion) should be taken into account when the downturn effect should be included into downturn EAD estimation.