Our latest white paper provides practical guidelines on estimation of Exposure at Default, CCR default capital charge and standardized CVA capital charge, based on the methods proposed in Basel 2 and 3, and compares the capital charge imposed under different methods (Current Exposure Method, Standardized Method, Internal Model Method, and the recently proposed Non-Internal Model Method) and risk weighting approaches (Standardized and Internal Risk Based), using the calculations performed for a portfolio of derivatives in PrevioRisk software.

The results of practical implementation show that Internal Model Method of EAD estimation produces about a quarter (27-28%) lower CCR capital charge than Standardized and NIMM methods. Particularly, IMM allows to recognize fully the effect of netting and margining, and saves capital for trades influenced by market factors with low volatility. In addition, the results confirmed that IRB risk weighting approach results in lower required capital estimates than the Standardized one, with capital relief from IRB implementation reaching 35-40%.

The full text of white paper is available at our Research Papers page. You can also download it by the link below:

When a bank does not have regulatory approval to use Advanced CVA capital charge, which allows to calculate regulatory capital on counterparty-by-counterparty basis, the bank must calculate capital on portfolio level using the following formula (see Basel III: A global regulatory framework for more resilient banks and banking systems para. 103).

 
As was mentioned above, this is formula for Standardized CVA capital charge calculated at portfolio level, which in the given form does not allow to define main drivers and components of CVA capital charge. For this purpose we use decomposition, based on EAD on customer or even trade level. As we will see, it is possible to manage portfolio by hedging separate trades. Here is the example of CVA capital decomposition for portfolio of 250 derivatives with total notional of 4.5 bln (all numbers are in USD mln).

Read more…