EBA’s consultation on economic downturn factors: Need of simplified approaches

The European Banking Authority (EBA) has recently launched a public consultation on its draft Regulatory Technical Standards (RTS) specifying the nature, severity and duration of an economic downturn, according to which institutions shall estimate the downturn of loss given default (LGD) and credit conversion factor (CCF), which are credit parameters that play an essential role in regulatory credit RWA estimation. These draft RTS are part of the EBA’s broader work on the review of the IRB approach aimed at reducing the unjustified variability in the outcomes of banks’ risk internal models, while preserving the risk sensitivity of capital requirements.

Prometeia as a leader in risk management solutions has given its contribution to the debate, acknowledging that enhancing harmonization of modelling and supervisory practices is essential in reducing unjustified variability of RWA across the system. 
However, as the EBA acknowledges in the Consultation Paper, this objective is somehow conflicting with ensuring necessary risk sensitivity of the overall AIRB framework. It is therefore required to seek an appropriate balance between different objectives. As the level of prescriptiveness of the proposed approach is significant and requires important investments both from institutions and supervisors, proportionality should also be taken into account, enabling simplified approaches for less material portfolios.

We support the opinion that institutions may adopt a target downturn solution leveraging on developed macroeconomic overlays. For instance, as far as LGD in-default modelling is concerned, it is straightforward to look at the downturn add-on, as required in the EBA/CP/2016/21, as the resulting of the macro/credit dependencies of LGD identified for Expected Loss best estimate (ELbe) calibration under stressed scenarios. Extending these rationales to LGD in-bonis would be a natural consequence. 

Moreover, banks are implementing IFRS9 requirements leveraging, in most cases, on AIRB risk parameters (PD, LGD and EAD) with appropriate adjustments in order to make them point-in-time and forward looking. In some instances, this is done leveraging on satellite models already developed for stress testing purposes. We believe that a more general multi-purpose understanding and modelling of macroeconomic overlays might be promoted for sake of consistency among different model usage. 

RTS do not exclude this possibility but might be more explicit in highlighting that such approaches are deemed as appropriate, while suggesting a more mechanical identification of observed downturn impact on individual model components and on overall LGDs and CFs as well as the identification of add-ons accordingly.

For these reasons we suggest allowing flexibility of practical application of model component approach principles, to be strengthened where appropriate in order to promote greater homogeneity, without pinning down operational steps.
 
We acknowledge the proposed model component approach gives a clearer regulatory view about what estimates are appropriate for an economic downturn. With regard to its feasibility, we highlight that the consistent application of the model component approach is particularly demanding in terms of data requirements. For this reason, its application is expected to be exposed to higher model risks, especially for low default portfolios. 

As the overall modellability of such portfolios is under scrutiny at international level, we suggest that over-reliance on human judgement and on the Memorandum of Cooperation definition shouldn’t be considered fully appropriate as it doesn’t guarantee an effective reduction of unjustified RWA variability.

Instead, for these portfolios the adoption of simplified approaches, leveraging on supervisory add-ons, should be considered a valuable option. As the proposed approach is significantly burdensome, it would be appropriate that such an option could be made available for immaterial portfolios as well.

Here is a summary of the most relevant remarks addressed in our feedback to EBA:

a) Principles related to the expected granularity of the downturn analysis and thereof adjustments might be strengthened as the model component definition refers only to LGDs and CFs distribution shape while exposure clusters with similar shape may show different level of correlation to identified economic factors;

b) As recovery processes may last for long, it is not straightforward to relate realisation of model components to a specific point of the economic factors time series; further factorisation might be considered appropriate in such cases instead of considering overall recoveries referred to time where most recoveries were realized;

c) As current economic conditions are still weak in some jurisdictions and the downturn impact on recoveries is still relevant due to time lags, then a full evaluation of the impact requires considering also the impact on incomplete workout and recoveries;

d) RTS and Guidelines do not stress the need for differentiating the downturn adjustments for LGD and LGD in-default, and, among the latter, based on time-on-book; this is critical for unsecured recoveries as for higher duration downturn impact tends to zero along with LGD near to 100%;

e) It might be made clearer that downturn adjustment to LGD in-default refers to the impact of an economic downturn to “residual recoveries” and not to overall workout recoveries;

f) The RTS sets what has to be intended for nature, duration and severity of downturn from the economic standpoint, while Article 6 and Amended GL would leave substantially unaddressed the way such conditions are reflected in downturnal parameters calibration, which is deemed to be a major source of unjustified variability;

g) Alternative approaches are found not fully economically grounded, while simpler approaches are deemed as appropriate for CCFs in general (considering directly the relation between economic factors and CCFs instead of factorizing CCFs in model components), for immaterial portfolios based on proportionality and for low default portfolios. 

• Read more under:
• BankingTech
• Data Protection