IRB approval level (Group vs Country)
-
We're in touch with the CRO of the Italian legal entity of a large European Banking Group
Their peculiarity is that:
- At country reporting level, Italy is reporting with standardized approach
- At group reporting level, Italy is reporting under IRB together with other countries for which the Group has IRB approval
Questions that have arisen
-
Has there been a similar situation elsewhere?
-
In the event the Italian subsidiary chooses to pursue local IRB approval, which of these option would be better considered from the Supervisor?
- Italian LE asking IRB approval of group models (the same ones already approved at group level) and without any specific calibration with country-only defaults
- Italian LE asking IRB approval of group models recalibrated with country-only defaults
- Italian LE rebuilding models from scratch
For options 2 and 3, there will be inconsistencies in the calculation of RWAs between the group and country level – has anyone seen this before ? Or should the country-level RWAs be used for group reporting?
-
-
Options 1 & 2 only really make sense if the models in question are for global portfolios (large corporates; banks; …)
-
If this is about mid-market firms or below, I suspect a local regulator would want to see local models in most cases (although this may depend on portfolio size)
-
If you wind up with different answers for local vs group I think that to meet local use test, you should use local model results in local reporting. For Group reporting, I think you have to use the Group approved models (and probably take a Pillar 2 underestimation risk charge as part of the ICAAP if the local models would have given higher numbers). In general I assume the Principle for reporting is that you should use the models that align with the level at which you are producing FinRep/CoRep/Pillar 3 report
-
UK PRA has a specific expectations around use of multi-country mid-market models and also has something called an “Overseas Models Approach” which seems to deal with the issue of using a model approved by another regulator in group – shout if you want me to dig out some of the materials on this)
Note, due to divergence between ECB and UK PRA views on what is acceptable, we have this week pitched to help German sub of a UK bank build their own models – the question of what number to use for what purpose will be a hot topic in that too
-
-
There are some specific existing instances of this where different regulatory models are used for the same portfolio – one country-specific for local LE capital requirements, and one “Group” for consolidated requirements e.g., ABC UK / ABC Group for UK mortgages
-
Has there been a similar situation elsewhere?
This is quite common for large groups, esp. where the group and local entity have different regulators, with quite different combos possible, e.g.,
- US CRE exposures of UK client are reported on SA for the US entity, but model is approved by the PRA and thus on Slotting for group reporting.
- Dutch client has ECB-approved Large Corp model that has Turkish large corps in it that are booked in Turkish entity and reported under IRB at group consolidated level, but Turkish entity is on SA locally
- Spanish bank has Turkish sub that’s developing IRB models for local supervisor approval so they can report locally on IRB but group can’t report that portfolio on IRB unless ECB approves the local model
In case the Italian LE would consider to pursue local IRB approval, which option would be better considered from the Supervisor:
1. Italian LE asking IRB approval of group models (the same ones already
approved at group level) and without any specific calibration with
country-only defaults
2. Italian LE asking IRB approval of group models recalibrated with
country-only defaults
3. Italian LE rebuilding models from scratchDepends on the portfolio. If it’s a global model like large corporates or PF, you would submit the group model and just make sure representativeness, other testing, impact assessment etc. are done at country level
If there is sufficient population to calibrate or build own model, I guess the reason to do so would be if there is any capital benefit from it? But then you would need to potentially carve out the local population from any global group model – otherwise it would go against the one-client one-rating principle
For options 2 and 3, there will be inconsistencies in the calculation of RWAs between the group and country level – has anyone seen this before ? Or should the country-level RWAs be used for group reporting?
Some discrepancy may be allowed e.g., at group level they would impose a country
cap but for local reporting, you might not need this. Dutch client was enforcing
consistency though and applying country caps in local reporting as well -
We’ve seen a variant of this issue in US/Canada, with the large Canadian Banks generally having IRB approval at the Group level (including for their main US loan books) and their US subsidiaries being on standardized
In this case, there is no incentive for the US entity to seek IRB approval - but US regulators do care about the risk rating systems from a bank supervision perspective, which has raised some of the same questions about whether group models are suitable. Those banks have generally taken a view aligned to a previous poster, i.e., using local models for middle market and below, and trying to align to group models for larger companies and FI's
On the last point, I'll say that we've seen US regulators challenge the support for those decisions heavily, but at least in some cases it seems to have stood up to that challenge. In one case where the bank's prior analysis and documentation didn't provide great support, they are being pushed to redevelop those as well, though they are trying to do so in a way that they can ultimately extend back to group as well, for a C&I model that was getting a bit long in the tooth anyway