European Savings and Retail Banking Group

Dear Sir/Madam,

Thank you for the opportunity to comment on the EBA consultation on the new Internal Model Approach (IMA) under the Fundamental Review of the Trading Book (FRTB). We welcome the proposals made. At some points we feel that the proposals are far more normative than the Basel text. This could lead to unnecessarily rigid rules. The discretion of supervisors would be too limited. Against this background, we would like to share with you the following reflections that we hope will be considered by the EBA.

In general, ESBG agrees to the argument brought forward by ISDA/IFF that a further narrowing defi-nition of “legally obliged” would reduce liquidity and usage of quotes for main instruments. Further-more, we think that the requirements for the RFET should be set in a way that they can be fulfilled not only for the most liquid risk factors. Especially for volatility surfaces a proof of the modellability based on the current proposal would be simple not possible or in the best case very time consuming and/or expensive dependent on the availability of the data providers. In the current proposal the requirements are set independent of the significance of the overall position in a certain risk factor. ESBG sees these strict requirements as an entry barrier for the implementation of an Internal Model, especially for smaller and medium size institutions.

---

ESBG thinks that for certain markets/instruments one firm side could be sufficient e.g. available firm bids for corporate bonds would prove existing market liquidity – which is the overall purpose of RFET.
Some of our members have not contracted specific vendor service we cannot provide an answer. No impact analysis was done so far in those cases.
All 3 restrictions (Intragroup, Volume, Bid/Ask) are overly (and unnecessarily) complex in our view. It should be possible to use all quotes provided that banks show that the quotes reflect normal market conditions.
All 3 restrictions (Intragroup, Volume, Bid/Ask) are overly (and unnecessarily) complex in our view. It should be possible to use all quotes provided that banks show that the quotes reflect normal market conditions.
Yes, potentially. The question must therefore also be discussed with the data providers. Since it can be assumed that a significant amount of information would be required of them, it must be ensured that the requirements can be fulfilled with reasonable efforts by them. Otherwise the information needed for the model would not be available or be very expensive.
We have no additional proposals.
This topic is of high relevance. Some of our members use parametric functions for all important volatility surfaces (Interest rates, Equity and FX volatilities) as e.g. SABR model, SVI model. However, although pricing is based on the parametric representation, these parameters are not risk factors. The risk factors are still the volatility quotations themselves (which are shifted in each scenario and after-wards new SABR parameters are derived in the scenario calculation).
We do not have a strong preference for either. In fact, we share the opinion of ISDA/IIF that both options have significant practical limitations.
If option 1 requires a historic recalibration it would not be possible from an operation point of view. Option 2 can be based on all input factors – including non-modellable – a parametric model is to de-liver calibration parameters as well as output risk factors. On this output level the exclusion of risk factors from non modellable buckets would take place. In cases where pricing functions are setup on model parameters it would require to establish new pricing functions (e.g. SVI). In cases where the pricing functions is setup to use the output risk factors we see open questions with respect to use of a non modellable basis in contrast to a full exclusion of the bucket.
We support the ISDA/IIF proposal.
Some of our members intend to rely on the services of data providers to be established and avoid building up own tracking routines of original maturities vs. actual maturities. Notwithstanding, these are in favor of any possibility which helps to improve the diminishing demand for bonds with remain-ing short-term maturities 0-1,5years (which were often highly liquid when issued at e.g. 5Y original maturity).
Since some of our members will not have sufficient own trades, they will need to rely on data service provider.
As soon as a liquid markets new reference rate is established, we do not expect additional problems. The important question is to ensure recognition of equivalence of the new benchmark: the industry is in favor of a legal act to ease the transition with clients.
We believe that the RFET is primarily a stand-alone test. Daily marked-to-market prices are available shortly following the close of trading. Vendor service data will most likely not be available during this time but only later with a considerable time delay.
Roberto Timpano
E