Fair-value accounting for CVA, part two

Fair-value accounting for CVA

The treatment of credit value adjustment (CVA) under International Accounting Standard (IAS) 39 is extremely complex. In the last issue of Risk (Risk January 2011, pages 111–115), the boundary conditions for measuring the fair value for CVA were described – and a key conclusion was that IAS 39 requires the evaluation of CVA to be performed on a fair-value basis by using market (credit) spreads, if available. The practical implications of measuring CVA under IAS 39 are now considered.

Various methods are used by banks to assess and mitigate counterparty risk, but only a limited number are acceptable for the purposes of measuring CVA fair value under IAS 39. So, how should a bank go about building an integrated model for measuring counterparty risk that is consistent with current practices and also meets the requirements of IAS 39? This article considers the issue of consistency and model integration, and also tackles four other topics:

• Consistency of CVA evaluated on a fair-value basis with performance measurement.

• Consistency of fair-value calculations and counterparty risk management.

• Risk-adjusted pricing (CVA).

• Quantification of risk capital requirements.

Collateral and netting under IAS 32/IAS 39

Collateral agreements and netting rules help to reduce counterparty credit risk. Netting allows counterparties to aggregate and net the positive and negative market values of financial instruments. Generally, netting is defined by legal netting agreements, such as the International Swaps and Derivatives Association master agreement for over-the-counter derivatives. Despite the legal definitions of netting, one has to distinguish between different netting concepts: netting for credit risk management; netting under the Isda master agreement; and netting under International Financial Reporting Standards (IFRS).

In case of a default, the Isda master agreement describes the relevant netting procedures – but these may not be legally enforceable in every jurisdiction. However, this is a necessary condition for internal credit risk management by the bank. Both netting concepts rely on the default event, so netting represents a conditional right. But IAS 32.42 requires an unconditional right to net derivatives exposure.

IAS 32.42 states: “A financial asset and a financial liability shall be offset and the net amount presented in the statement of financial position when, and only when, an entity: (a) currently has a legally enforceable right to set off the recognised amounts; and (b) intends either to settle on a net basis, or to realise the asset and settle the liability simultaneously.”

Condition (a) is not satisfied by an enforceable Isda master agreement – according to KPMG, Insights into IFRS, seventh edition 2010/11 (5.6.490.50), there is no legally enforceable right in this case as it is only a conditional right upon default. For netting purposes, an unconditional right is required. Condition (b) is not satisfied for typical OTC derivatives. Consequently, the defined netting (offsetting) rules under IAS 32 do not coincide with the rules of the Isda master agreement.

Therefore, the following question arises from an IFRS perspective: what is the basis for calculating CVA in the presence of a netting agreement?

One important feature of the Isda master agreement is close-out netting, which affects the evaluation of CVA under IAS 39. Close-out netting means that if a derivatives user goes into bankruptcy, a counterparty has the right to collapse all its offsetting positions into a single net amount owed by one party to another. Furthermore, in jurisdictions where close-out netting is enforceable, all transactions under the Isda master agreement constitute a single agreement between the two counterparties, instead of separate contracts. The confirmation of a transaction serves as evidence of that deal, and each trade is incorporated in the Isda master agreement.

With respect to the CVA evaluation, IAS 39.AG73, IAS 39.AG76 and IAS 39.AG82 imply that if close-out netting is present and enforceable under the jurisdiction of the balance-sheet preparer, netting significantly reduces the counterparty credit risk position. In case of a default, any OTC derivatives contracted under the Isda master agreement can be aggregated into one claim against the defaulted party, so this agreement defines the netting set or net position. This aggregation is performed by adding market values across different derivatives products and subtracting collateral within the netting set. Therefore, the appropriate basis of proration for CVA under IAS 39 is represented by the netting set defined by the Isda master agreement, if legally enforceable.

Under these circumstances, IAS 32 netting rules do not apply for the determination of CVA under IAS 39 – they only apply for the presentation in the balance sheet. Therefore, CVA for the purpose of IAS 39 is based on a net balance-sheet amount that is different from the gross balance-sheet amount under IAS 32. That means there is no direct connection between the presentation of derivatives in the balance sheet (IAS 32), for which the Isda netting rules cannot be applied, and the derived CVA for the purpose of IAS 39, for which the Isda rules can be applied if legally enforceable. This impedes reconciliation between the inventory of derivatives for CVA in the trading book profit and loss (P&L) and the balance sheet.

Measuring CVA on a fair-value basis under IAS 39 raises an important question: how can consistency with respect to performance measurement and counterparty risk management be achieved, since netting sets are unlikely to coincide with trading desks? If performance measurement is conducted on a more granular basis (trading desk level), it should be remembered that CVA is a non-linear function that is not additive within a netting set. So, if the CVA is evaluated at trading desk level, the evaluation of CVA has to be performed for each trading desk separately, and additionally on an overall basis, in order to calculate P&L under IFRS.

This is also an important property of CVA under IAS 39 in connection with performance measurement and segment reporting (IFRS 8 operating segments), because it is sometimes necessary to allocate several trading desks into one segment and evaluate CVA together, irrespective of the definition of a netting set.

Consistency of CVA evaluated on a fair-value basis with performance measurement

Since IFRS requires the evaluation of CVA, it becomes necessary to recognise CVA in daily trading P&L to ensure consistency between internal and external performance measurement. For some banks, this is already common practice – but not for all. Consequently, IFRS leads to standardisation.

The classes of counterparts for a bank can broadly be subdivided into two subgroups: the interdealer OTC market and corporate business. In the interdealer market, OTC master agreements and collateral agreements are in place and specify margin periods, margin thresholds and minimum transfer amounts for collateral posting purposes in order to reduce the net risk position on OTC derivatives. Unfortunately, one cannot deduce that CVA for this class of counterparty is negligible, since it depends on collateral agreement specifics and credit spreads (ratings). From an IFRS perspective, CVA has to be evaluated and booked for all counterparties.

If CVA is not evaluated and charged to the trading business, it creates the problem of adverse selection – traders are encouraged to enter into business with lower credit quality counterparties to increase their P&L, but this increases counterparty credit risk. If CVA is evaluated, a price for counterparty risk can be assigned to each transaction. As such, business with lower credit quality counterparts attracts a higher CVA charge – and the problem of adverse selection is solved.

Unfortunately, this is only one part of the story, as one cannot assume CVA evaluated under IFRS using credit spreads can be passed on or charged to the counterparty. Subsequently, three situations have to be distinguished:

• There are market segments, such as the interdealer OTC market, for which CVA cannot be charged entirely to the counterparty because it is not considered in the market price quotation. In this case, it is an individual business decision to enter into the deal or not. Any departure from the full CVA charge is recognised as an expense (cost) in performance measurement if the deal is done.

• There are market segments, such as the corporate business, for which CVA can be charged entirely to the counterparty.
• Sometimes, banks evaluate CVA for IAS 39 purposes differently from the CVA calculated by the trading desk. As a result, the full amount calculated under IFRS is not necessarily passed on to the counterparty in full. Any differences in IFRS are therefore recognised in performance measurement.

Consequently, one has to distinguish between CVA imposed by IFRS (CVA (IFRS)) and CVA charged to an external counterparty (CVA (ex)) for analysing incentives created by CVA. Figure 1 summarises this (please click on the link to the pdf below to see all figures for these article).

If CVA (IFRS) and CVA (ex) both have the same sign, and assuming that CVA (IFRS) is greater than CVA (ex), then the business is encouraged to move to other counterparties. If the CVA (IFRS) is less than CVA (ex), then the trading desk is encouraged to do more business with these counterparties. So, if the externally charged CVA (ex) is below the IFRS CVA, trading with lower-quality credits is encouraged (depending on the market position of an institution).

Additionally, CVA evaluated on a fair-value basis not only creates incentives between counterparties, but it also creates incentives between banking products. Consider an example where the loan business also engages in derivatives business (for example, an interest rate swap) for one of its clients in order to exploit cross-selling opportunities. According to IFRS 8.5 and IFRS 8.11, the operating segment has to be presented and it is assumed the loan business represents a different segment than treasury/trading business (figure 2).

Typically, if the loan business enters into a derivative, then the market risks and revenues are transferred to the interest rate trading/treasury desk. Hence, the analysis of incentives becomes more complicated, as at least three components have to be compared: CVA (IFRS), CVA (ex) and expected loss charge (loan). Assuming components are positive, one has to compare:

CVA(ex) - CVA(IFRS) versus expected loss charge (loan)

Assuming CVA (ex) minus CVA (IFRS) is positive and lower than the expected loss charge (loan), then the loan is more favourable than the derivative. If CVA (ex) minus CVA (IFRS) is higher than the expected loss charge (loan), then the derivative is more favourable than the loan. Of course, this analysis also depends on the evaluation of expected loss charge (loan), if credit default swap spreads or Basel II probability of default (PD) figures are used. Additionally, one can also distinguish between internal expected loss charges and those passed to the borrower.

The recognition of own credit spreads adds additional complexity to such an analysis, making it hard to compare banks due to their different funding situations. But it reveals CVA evaluated on a fair-value basis creates incentives among banking products and doesn’t affect only trading business.

Consistency of fair-value calculation and counterparty risk management

A consistent framework between the CVA on a fair-value basis and counterparty risk management requires the integration of price risk. This integration results in adjustments for major components of counterparty risk management. It has become obvious the evaluation of CVA requires a lot of data and extensive capacity of computational power (IT systems). Figure 3 portrays the major components and processes for the evaluation of CVA.

This suggests a high degree of coincidence between counterparty risk management and financial accounting processes. We distinguish between three major components of counterparty credit risk management and portray how fair-value measurement under IFRS can affect these directly or indirectly (figure 4).

Exposure quantification and limitation  

With respect to this component, counterparty risk management faces the following issues:

• To prevent single-name concentration risks and wrong-way risk, counterparty exposure limits must be enforced. This includes the quantification of margin requirements, management of netting agreements, downgrade triggers and break clauses.
• Due to the nature of derivatives prices, future exposures cannot be known deterministically. Hence, the exposure requires stochastic modelling, and consequently limits must be defined with respect to some statistical metric, such as quantile exposure (for example, value-at-risk) or expected shortfall. Active and passive limit breaches have to be distinguished if limit compliance needs to be enforced. Passive limit breaches occur if the market moves unfavourably for the bank, without changes in the composition of the derivatives portfolio. Active limit breaches result from new trades with a higher CVA than permitted.

As the quantification of counterparty risk management requires the modelling of stochastic exposures, a higher accuracy of modelling (in comparison with conservative approximations) is necessary to meet the requirements for CVA under IFRS, and this includes the modelling using market rates (risk-neutral measure). As a consequence, the existing limit framework has to include limits for counterparty exposure and CVA as market risk. To achieve consistency and comparability, the effects of the application of risk-neutral measures, as well as counterparty and own credit risk (bilateral risk), have to be taken into account. For the sake of clarification, CVA under IFRS implies the measurement of price risk and not necessarily counterparty risk. Furthermore, to reduce the impact of CVA under IFRS, it may be necessary to optimise the use of margin agreements and netting agreements.

Risk-adjusted pricing (CVA)

Consider the following aspects of risk-adjusted pricing:

• The expected loss due to counterparty credit risk is a direct function of the credit quality of counterparties.
• Market participants generally take counterparty credit risk into account. Those that do not will be subject to adverse selection, as it is more attractive to trade with lower-rated counterparties and less attractive to deal with higher-rated counterparties.

This is directly affected by IFRS, as the application of the fair-value hierarchy implies the use of CDS spreads/own credit spreads and – only in case of non-quoted counterparties – the use of Basel II PD data. Therefore, IFRS introduces market values into risk-adjusted pricing.

Quantification of risk capital requirements 
The main aspects with respect to the quantification of risk capital requirements are:

• Sufficient capital buffers must be in place to protect against counterparty losses.
• Even under effective current exposure-based margining, tail risks will remain and capital needs to be set aside to protect against those.

IFRS indirectly affects risk capital issues. The main impact stems from the fact that most banks want to avoid implementing different methodologies for capital requirements and CVA under IFRS. Three approaches to quantifying capital requirements must be considered: expected positive exposure; add-ons based on expected exposures; and VAR-based approaches. In general, all approaches can be adjusted to meet CVA considerations.

Value-at-risk-based approaches are only applicable for short-term exposures, such as the residual exposure within the close-out period of collateral agreements. For longer-term exposures, an evolution of the portfolio exposure path should also incorporate the ageing of transactions and other trade life-cycle effects.

In figure 5, the boundary conditions outlined in the previous article and the approaches for quantifying capital requirements are shown. The main conclusion is that all approaches can be adjusted to achieve compliance with the fair-value measurement under IFRS – certainly all approaches represent approximations, which are common practice, as exact pricing methods are too expensive.

Conclusion
The analysis shows the derivation of CVA under IFRS requires the simultaneous consideration of various aspects of financial accounting. These various aspects cause a certain level of complexity, including mathematical modelling, processes and reporting, and show that financial accounting cannot be considered as an isolated approach. As bank management requires an integrated approach, the evaluated CVA under IFRS also affects its core business. From an implementation perspective, existing approaches that are utilised to meet capital requirements can be adjusted to reduce the cost of implementation and to facilitate the application of IAS 39.

Dirk Schubert is a partner in the financial services division at KPMG in Frankfurt. Email: [e-mail]

References:

• Artzner P., Delbaen, F., Default Risk Insurance and Incomplete Markets, Mathematical Finance 5 1995, pp. 187 - 195.
• Assefa, S., Bielecki, T.R., Crépey, S., Jeanblanc, M., CVA Computation for Counterparty Risk Assessment in Credit Portfolios, Working Paper 2009.
  
• Basel Committee on Banking Supervision, International Convergence of Capital Measurement and Capital Standards, June 2006.
  
• Belanger, A., Shreve, S.E., Wong, D., A General Framework for Pricing Credit Risk,
• Mathematical Finance, Vol. 14, No.3, July 2004, pp 317 - 350.
  
• Bielecki, T.R., Rutkowski, M., Credit Risk: Modeling, Valuation and Hedging, Ney York 2002.
  
• Björk, T., Arbitrage Theory in Continuous Time, Oxford 1998.
  
• Bin Li, Yi Tang, Quantitative Analysis, Derivatives Modeling, and Trading Strategies, 2007.
  
• Blanchett-Scalliet, C, Jeanblanc, M., Hazard Rate for Credit Risk and Hedging
• Defaultable Contingent Claims, Finance and Stochastics, 8. 2004, pp.145-159.
  
• Brigo D., Pallavicini, A., Counterparty Risk and Contingent CDS Valuation und correlation between interest rates and default, March 2006.
  
• Brigo D., Capponi, A., Bilateral counterparty risk valuation with stochastic dynamical models and application to credit default swaps, January 2009.
  
• Canabarro, E., Duffie, D., Measuring and Marking Counterparty Risk, Asset/Liability Management of Financial Institutions, Ch. 9, Euromoney Books 2003.
  
• Cesari, G., Aquilina, J., Charpillon, N., Zlatko,F., Lee, G. Manda, I., Modelling,
• Pricing, and Hedging Counterparty Credit Exposure - A Technical Guide, Berlin Heidelberg 2009.
  
• Düllmann, K.; Trapp, M., Systematic Risk in Recovery Rates - An Empirical Analysis of US Corporate Credit Exposures 2004.
  
• Duffie, D., Huang, M., Swap Rates and Credit Quality, Journal of Finance, Vol. 51 No. 3, July 1996.
  
• European Commission, Regulation of the European Parliament and of the Council on OTC Derivatives, Central Counterparties and Trade Repositories 2010.
  
• Gibson, M.S., Measuring Counterparty Credit Exposure to a Margined Counterparty, in Counterparty Credit Risk Modelling, Risk Management, Pricing and Regulation edited by M. Pykhtin, Risk Books 2005.
  
• Gregory, J., Counterparty Credit Risk, Wiley 2010.
  
• IASB, IAS 39: Financial Instruments recognition and measurement, 2010.
  
• IASB, Developing Common Fair Value Measurement and Disclosure Requirements in
• IFRSs and US GAAP, Comprehensive Project Summary July 2010.
  
• ISDA Master Agreement, 2002.
  
• Jarrow, R.A., Turnbull, S.M. Pricing Derivatives on financial securities subject to credit risk, Journal of Finance 50, pp 53 - 85.
  
• KPMG, Insights into IFRS, 7th Edition 2010/11
  
• Pengelly, M. Charge Changes, Risk August 2010.
  
• Pykhtin, M., Zhu, S., Measuring Counterparty Credit Risk, Bank of America 2006.
  
• Singh, M., Youssef, K., Price of Risk - Recent Evidence from Large Financials, IMF Working Paper August 2010.
• Sundaram, R., Equivalent Martingale Measures and Risk-Neutral Pricing: AnExpository Note, The Journal of Derivatives, July 1997.
  
• Whittall, C. , Dealing with funding, Risk July 2010.
  

 

Click on the link below to read the full version of this article.

Fair-value accounting part 2