The new approaches considered are aligned to the CVA calculations under IFRS and the market risk framework under the Committees’ Fundamental Review of the Trading Book (FRTB). A revision of the current framework addresses three major issues:
- Ensure all important drivers of CVA risk and CVA hedges are covered in the Basel regulatory capital standard
- Align the capital standard with the fair value measurement of CVA employed under various accounting regimes
- Ensure consistency with the proposed revisions to the market risk framework under the Basel Committees’ Fundamental
- Review of the Trading Book
The consultative paper proposes two frameworks to accommodate different types of banks with respect to the ability to calculate CVA sensitivities:
- The “FRTB-CVA framework” consisting of the standardized approach (SA-CVA) and based on CVA sensitivities
- The “Basic CVA framework“ (BA-CVA), based on a formula similar to the current standardized method
To highlight the key differences of current and future calculation approaches for regulatory CVA risk capital charges, including the eligibility criteria for using the different approaches, this article focuses on SA-CVA and BA-CVA.
New Basic Approach (BA-CVA)
For banks not able or willing to provide sufficient CVA sensitivities, a new basic approach, which is closely related to the current standardized method, should be used. Improvements to the approach include an enhancement of the definition of eligible credit risk hedges.
New SA-CVA
To use the SA-CVA the following requirements must be fulfilled:
- The calculation of CVA sensitivities for given risk factors comply with general principles for the calculation of CVA
- A methodology for approximating the credit spreads of illiquid counterparties is applied
- There exists a dedicated CVA risk management function (and control unit).
Qualifying banks need to follow general principles to calculate regulatory CVA in line with the FRTB-CVA framework. There are two options for generating scenarios of discounted exposures: accounting-based CVA and IMM-based CVA.
One could, in principle, base CVA sensitivities on add-on approaches to exposure calculations, which would mean relying on MTM sensitivities only, although the consultation paper focuses on Monte Carlo simulation or equivalent methods that are able to calculate CVA as a proper hedging cost of counterparty credit risk (CCR).
Sample calculations
For the sample calculations we selected synthetic portfolios, including real market data, in order to provide an impression of potential CVA risk capital charges for end of June 2015.The sample portfolios consisted of interest rate and cross currency swaps (USD and EUR). For the sample calculations we considered the positions of a medium-sized bank with two different types of counterparties:
- Interbank portfolios with investment grade ratings.
- Corporate client portfolios with investment grade ratings.
Simulation approach
We implemented a framework that matches the definitions from the consultative paper and the corresponding QIS instructions. For sensitivity calculations, a two factor semi-analytic model was used. Sensitivities were based on 1 bps tenor shifts for IR and Credit Spread Delta and relative 1% shifts for FX Delta, as well as relative parallel 1% shifts for IR and FX volatilities to calculate Vega sensitivities.
Calculation Results
Current Basel III CVA risk capital charge
Calculated as a baseline scenario that defines the current capital charges for all banks without an available advanced approach. By applying the same rating for the financial and corporate counterparty there was no difference between CVA risk charges for the two portfolios. For the calculations, credit quality “3” was assumed with risk weight 1%. The EADs were calculated according to the Current Exposure Method (CEM) as described in article 274 CRR, the recognition of netting was applied according to article 298.
Future Basic approach for CVA risk capital charge
Calculations were based on EAD figures derived from SA-CCR, the new standardized approach effective January 1 20172. Considered by regulators as a more risk sensitive approach than CEM, SA-CCR recognizes netting and margin agreements in an enhanced way, and incorporates the IMM multiplier α to account for model inaccuracies. For interest rate swaps without CSAs the SA-CCR EAD is significantly higher. While CEM recognizes CSAs only for in-the-money trades, SA-CCR offers significant EAD reduction for both considered types of CSA trades, and also takes into account MPOR.
Results for the future basic approach show a significant increase in capital charges for all considered trades, compared to current Basel III results. It is evident that the new basic approach significantly increases the capital charge for both counterparties.
Future SA-CVA charge
The future SA-CVA capital charge is highly beneficial for collateralized trades as it is the result of calculations with real CVA sensitivities. For trades with no CSA, SA-CVA is generally higher than the current capital charge, whereas for trades with CSA I and CSA II this order is reversed.
Increasing MPOR from 0 to 20 days makes the trade riskier, and thus increases both CVA and SA-CVA for CSA II when compared with CSA I. MPOR also changes the distribution of sensitivities in credit buckets, which are dominant in SA-CVA calculations. So whilst the sensitivity of a credit parallel shift is always higher for CSA II, some credit bucket sensitivities for CSA II can be lower than those of CVA I which may lead to a smaller capital charge.
Conclusion
The most important result is the increase of the CVA risk capital charge under the new basic approach (BA-SVA) compared to the current standardized approach. The impact of the new CVA risk regulation framework on calculation methods and infrastructure of banks could be the turning point for many medium-sized institutes. This is due to many having only recently started calculating exposures and CVA within a Monte Carlo simulation based framework for IFRS 13 compliant accounting.
The SA-CVA method may be an attractive way to reduce capital charges for CVA risk – provided that banks are able to install an active CVA desk that is managing CVA and CVA risk. This would be the first time the Basel committee recognizes the simulation methods used for accounting for regulatory purposes.
Regular calculation of CVA sensitivities is not something associated with a Monte Carlo installation for month end IFRS reporting. Therefore, banks seeking to adopt the SA-CVA method will be interested in fast and accurate CVA sensitivity calculations. Research as well as technology solution providers are able to provide various approaches that support automatic differentiation methods, Malliavin type derivatives, alternative likelihood ratio methods or fast GPU implementations. Other methods for increasing efficiency may include more effective streaming algorithms and utilizing dependency graphs for analysis.
As a final point, in addition to the quantitative impact study (QIS) on the CVA risk capital charge finalized in September 2015, the current QIS on CVA risk Ends 13 May 2016.
References
- The internal model approach IMA-CVA that has been introduced in the consultative paper and both QIS has been eliminated later on. Elimination has been published in a consultation paper regarding credit risk RWA
- Basel Committee on Banking Supervision. The standardised approach for measuring counterparty credit risk exposures. March 2014.