Demystifying CVA, CECL, and ECL: Understanding accounting for expected credit loss

The concept of market value attributed to the counterparty credit risk has been developed in the early 2000s. It was applied to valuation and risk management of derivatives. Known as Credit Valuation Adjustment (CVA) and linked to the counterparty, it further evolved to include entity’s own credit-risk adjustments (DVA – “debit valuation adjustment”). Both CVA and DVA together with additional adjustments based on regulation requirements are often referred to as XVA.

Accounting of CVA for derivatives came in the context of IFRS13 and ASU-2011-04. Since the concept is also applicable to bonds and loans, CVA came to limelight after the financial crisis of 2008-2009. FASB and IASB developed regulations to address the issue with the until-then-applicable delayed recognition of credit-related losses. FASB introduced the Current Expected Credit Loss (CECL), and IASB refers to Expected Credit Loss (ECL).

Despite its wide application and usage, CVA, CECL, and ECL concepts may still cause confusion. This may be due to different definitions within FASB and IASB standards, and also due to application in risk management. Further, there are different methods of calculating CVA.

For completeness purposes, it should be mentioned that the concept of the internal-ratings-based approach to credit risk (IRB) has been introduced to help calculate the capital adequacy of banks under the Basel II agreement. The outcome of the IRB is the unexpected loss assessment; that is, however, outside of the scope of this article; we focus on clarification of CVA/ECL/CECL and their relation.

Accounting for expected credit loss under ASC 326 and IFRS 9

Under ASC 326, a corporation shall recognize a “lifetime” expected credit loss. This is called Current Expected Credit Loss (CECL) and stands for all the to-be-expected credit losses over the lifetime of the asset. The concept applies to all financial assets acquired by company that are measured at amortized cost. ASC 326 is effective for publicly listed entities as of the end of 2019.

Under IFRS9, a similar concept is introduced as ECL. A corporation shall recognize either a “12-month” expected credit loss, or a “lifetime” expected credit loss, depending on the credit quality of the issuer of the instrument, or a counterparty in the case of loan transactions. Credit quality falls into one of the 3 “Stages” (also referred to as Levels or Credit valuation hierarchies): Stage 1 corresponds to the business-as-usual when counterparty/issuer pays their obligations on time. Stage 2 denotes deterioration of the credit worthiness; for example, company obligations are paid more than 30 days past due. Stage 3 implies a significant increase in the credit risk; obligations are paid more than 90 days past due. For Stage 1 assets, a 12-month ECL is recognized, while for Stage 2 and 3 the lifetime ECL shall be recognized. This is applicable as of the beginning of 2018.

Calculation methods for CVA and Expected Loss (EL)

CVA represents the market valuation of a counterparty credit risk. It can be calculated using various methods ranging from the full Monte Carlo simulation that considers many risk factors to simpler methods based on Credit Default Swap curves (CDS) of a counterparty or issuer.

Monte Carlo simulation is computationally very demanding. It is used, for example, by banks when trading derivatives to adjust their price for the credit risk of the counterparty.

CDS-based method involves probabilities of default for individual tenors coupled with the recovery rate associated with a counterparty/issuer. This data is applied to the cashflows of transactions with the counterparty, that is, to the expected exposures and their value dates to calculate the expected loss.

(Older methods involved application of credit spreads on top of risk-neutral curves. Such an approach, however, excludes important considerations – non-linear probability of default, recovery rates, or effects of netting. The approach is not recommended any longer.)

In accounting, the expected loss (EL) shall be calculated generally as:

EL = PD x LGD x EE

With PD being the probability of default over the relevant period, LGD is the loss given default which depends on the recovery rate associated with the counterparty/issuer, and EE is the expected exposure.

Due to this broad definition of the EL calculation, approaches developed earlier for calculating CVA for risk management and pricing purposes can be reused. For example, the CDS-based approach can be applied to calculate CECL/ECL figures of amortized-cost assets. This approach is computationally fast and, therefore, allows for an efficient computation of CVA for a large portfolio of transactions in the end-of-day accounting jobs.

Taking the full CDS curve (all tenors) and applying it onto the cashflows of a transaction leads to the lifetime ECL. This should satisfy ASC 326 requirements.

Limiting the CDS curve to only the next 12 months tenors and applying that onto the transactional cashflows results in the 12-months ECL. This corresponds to the ECL value for counterparties/issuers of Stage 1 within IFRS9. For Stages 2/3 the full CDS curve is applied leading to the same result as CECL.

“EE” is the exposure to a counterparty. This can be a gross exposure. However, if the entity has netting agreement with a counterparty, exposures can be netted. Only the net total then enters the calculation.

Subscription cost challenges and alternative approaches

Subscription to a large set of CDS market data for all the applicable counterparties and issuers can be costly. As an alternative, entities may choose to develop proxy CDS curves based on a smaller set of market CDS curves that represent “classes” of counterparties/issuers. The proxy curves are then applied to counterparties according to, for example, their Credit valuation hierarchies/Stages.

Further, in context of IFRS9, the systems should allow for setup of a CDS curve for the next 12 months and bind them with Stage 1 counterparties/issuers. If the creditworthiness of a counterparty/issuer deteriorates to Stage 2 or 3, the systems should allow switching to the full CDS curve.

The preferred approach has to be approved by the entity’s auditors.

The market value component attributed to credit risk and its naming/usage in different contexts can be summarized in the following way:

CVA calculation is of a broader usage and can employ advanced numerical approaches. The computational version utilizing CDS curves can be more easily embedded into Treasury management systems. This, in turn, allows for a simple CVA/DVA calculation for derivatives, and computation of ECL/CECL of assets as required by the international accounting standards.

ION Treasury management systems implement the corresponding features as of their introduction by FASB and IASB. This allows ION customers to comply with the regulatory requirements and to use those calculations for obtaining CVA of traded derivatives.