The original made text of 2015/35 included the 'F factor' as F in the formula for LGD on exposures to heavily collateralised reinsurers in Article 192(2), but then referred in the subsequent body text to F’. At that time F was defined in Article 197(7) as 50% and F' as 90%, except (in both cases) where the conditions for 100% were present. Not only was this textually inconsistent as regards the treatment of exposures to heavily collateralised reinsurers, it produced an absurd result as the basic LGD assumption was 90% but recovery from collateral was limited to 50% (assuming that the test for 100% was not met). A correction was included by Article 2(5) of 2016/467, to change F to F’ in the formula. This had the logical effect of making the LGD and collateral assumptions internally consistent for heavily collateralised reinsurers (as they were and are still, for other types of counterparty in what is now Articles 192(2) to (3c)). F’ was also the factor initially used for derivatives in Article 192(3) – again, consistently with the basic LGD assumption of 90% loss for such instruments. The same value did service for two types of unrelated exposure, a fact that appears to have given rise to unintended consequences in 2019 when the approach to derivative exposures was revised to take account of EMIR. By way of amendments set out in Regulation 2019/708, Article 192(3) was replaced by a new 192(3), (3a) and (3b) dealing with three new categories of derivative, one with LGD basic assumption of 18%, one with 16%, and one of 90%, in each case tempered by collateral allowance of 50% times factors F’, F’’ and F’’’ respectively applying that to the value of the collateral. F’, reused in the new Article 192(3), was redefined in Article 197(7) as 18%, and F’’ and F’’’ were given the values of 16% and 90% respectively. Derivatives not falling within any of those three new categories were dealt with in new Article 192(3c), keeping the old Article 192(3) approach of 90%, with no 50% reduction in their F factor, but with a requirement to risk adjust the collateral. However the F factor corresponding to 90% was now F''' rather than F', so that was the factor included in Article 192(3c). These changes had been discussed in EIOPA’s Advice EIOPA-BoS-18/075 28 February 2018 and the amendments made broadly followed the Advice, and their purpose is described in the Recitals to 2019/708. What does not appear to have been considered, in either the Advice or Recitals, is the fact that changing the value of F’ from 90% to 18% also affected the treatment of exposures to heavily collateralised reinsurers in Article 192(2). The basic LGD assumption remained 90% but the F factor for collateral was now 18%, breaking the internal consistency of the LGD calculation for this type of counterparty alone. From the lack of mention of this provision in the EIOPA Advice and the Recitals of the amending Regulation, it seems highly probable that no change was intended to the effect of Article 192(2). If so, the consequential impact on the calculation for heavily collateralised reinsurers was unintentional and a drafting oversight. If no change was intended to the effect of that calculation, the factor F' for exposures to such entities should have been changed to F''' but this was not done. Finally, we can also draw support for the argument that Article 192(2) is currently incorrect from the fact that the simplified approach introduced as Article 112a (also by Regulation 2019/708) is more favourable to the cedant than the formula for exposures to heavily collateralised reinsurers in Article 192(2). According to Recital (18) of that amending Regulation, the simplification is based on an assumption that the reinsurer collateralises to more than 60% of its assets. One would therefore have expected the resultant calculation to be identical to the formula for such reinsurers in Article 192(2). Instead, it uses a factor of F, i.e. 50%. The simplified calculation is less prudent than the unsimplified, which should not be the case. It is also possible, though we cannot state this conclusively, that Article 112a is also in error and was intended to use an F Factor of 90% (F’’’) to reflect the assumption stated in the Recital.

This question has been rejected because the Q&As cannot be used to answer questions about the correctness of the legal text.