Determination of proxy spreads for the calculation of CVA risk under the advanced method

1. Does Article383 (1) CRR in connection with regulation 526/2014 (RTS) in principle allow to waive one or more of the attributes of rating, industry and region when determining a proxy credit spread for the advanced method for the determination of own funds requirements for credit valuation adjustment (CVA) risk? 2. What is the minimum granularity (number of categories) that a credit spread proxy must reflect to be appropriate with respect to the attributes of rating, industry and region of the counterparty?

Banks may have contracted financial instruments bearing CVA risk with counterparties for which no market observable Credit Default Swap (CDS) spread is available. In this case, banks should use a spread that is appropriate having regard to the rating, industry and region of the counterparty (proxy spread) in accordance with Article 383(1) CRR. Thereby, the availability of liquid market data depends on the granularity of the proxy spread segmentation: the higher the required granularity the lower the number of liquid CDS spreads available in each of the segments to determine a proxy. For example, to satisfy the granularity envisaged by Article 1 of the RTS banks (in total 72 different categories (buckets) by combining six categories for rating, four categories for region and three categories for industry) would have to base their models on a small number of spreads or in extreme cases on one or two spreads only. The necessity to adapt the proxy spread modelling to the liquidity is based on a significant decline in CDS trading volumes in 2015, as dealers withdrew from CDS market making. 1. May banks in these cases be allowed to disregard one or two of the attributes of rating, industry and region in order to allow for a larger number of liquid spreads to be combined to a more reliable proxy credit? 2. The minimum number of categories that have to be discriminated for each of the attributes has to be defined. For example, is it sufficient in case of low market data availability to reduce the granularity within the rating attribute to a distinction between investment grade and non-investment grade or do banks have to reflect always the required credit quality steps according to the RTS?