Part 1 of this blog looked at the significance of credit risk associated with exchanges, custody and prime brokerage services. Part 2 explores counterparty risk in a traditional financial institutional setting and provides insights on how to extend the traditional credit risk framework to the cryptoasset industry.
Counterparty risk is of upmost importance for banks, as it is more difficult to hedge. This was underlined during the 2008 GFC, during which time most of the losses were triggered by counterparty default and Credit Value Adjustment (CVA) volatility.
The quantification of counterparty risk can be divided into three types, which are defined by their risk metric and purpose:
- Potential Future Exposure (PFE) for setting trading limits
- Bilateral CVA (and other Valuation Adjustments) for adjusting trades prices
- Risk-weighted Asset (RWA) and CVA Value-at-Risk (VaR) for Basel required capital charges covering default and CVA risks
For cryptoasset OTC markets we expect counterparty risk to be a major consideration affecting the estimation of trade profitability.
PFE and Credit Limits
For internal credit limits, banks traditionally use PFE, which is similar to market risk VaR in that it measures the one-time tail of exposure of a portfolio against a counterparty. It can be defined, more precisely, as a potential credit exposure over a specified period of time calculated at some level of confidence, e.g. one-year horizon 99% tail. PFE is often calculated for various time horizons and then the max value of the time profile is used for the credit exposure limits. In terms of calculations, one can use either historical or Monte Carlo simulation techniques. In the latter case, market factors should be calibrated to historical data with ‘real-world’ volatility measures. Banks often calculate two credit exposure limits for each trade based on collateralised and uncollateralised PFE approaches.
For banks or other financial institutions trading cryptoasset products, setting limits based on historical data presents a significant challenge as there is insufficient data for cryptoassets. The high volatility profile of cryptoasset price time series also makes parameterisation and calibration of risk models more art than science, depending on the choice of historical period. An alternative approach is ‘add-on calculation’, which is current exposure plus some factor dependent on the type and maturity of the trade. However, this approach usually leads to more punitive exposure limits.
Bilateral CVA and Adjustments to Trades Prices
PFE does not include counterparties’ probability of default and loss given default (LGD) and takes into account only one value, either fixed horizon or max for a given time, but not the whole profile. Finally, PFE measures a tail event rather than expected exposure as in CVA/DVA. This is why it is used for adjusting the market prices which can be interpreted as expected values of future cashflows. As a result, the front offices in banks uses CVA (Credit Valuation Adjustment) and its symmetric counterpart DVA (Debit Valuation Adjustment).
CVA can be expressed as the probability-weighted total loss incurred by the portfolio of the bank due to counterparty default. It is negative to the bank and is proportional to the bank’s Expected Positive Exposure (EPE) and probability of counterparty default. Symmetrically, DVA is the expected gain incurred by a portfolio of the bank due to its own default. It is positive to the bank and is proportional to the bank’s Expected Negative Exposure (ENE) and the probability of bank’s default. Together they constitute bilateral CVA which is what both parties of a bilateral trade usually agree on as an adjustment to trade value. Both CVA and DVA are usually included in Fair Value pricing prescribed by accounting standards such as IFRS 13.
Although from a pricing perspective CVA and DVA are symmetrical, there are some important differences. It is more difficult to hedge DVA because to do so, banks have to sell protection against its own default, which is more difficult than buying protection against counterparty default. Another, rather controversial, difference is that DVA increases during a period of stress when a bank’s credit worsens. Significant DVA gains were the main reason why in Q3 2011 many US banks reported profits of billions of dollars.
Both CVA and DVA are calculated using ‘Risk- Neutral’ Monte Carlo, where market factors are calibrated to today’s prices. The main challenges in calculating and managing CVA and DVA are accuracy and performance, wrong-way risk (WWR), and properly taking into account netting and all CSA (Credit Support Annex) features.
For cryptoasset non-deliverable forward contracts, exposures (and thus CVA/DVA) depend strongly on the volatility of the underlying cryptoasset, though the prices of these contracts are not. That means that to calculate them correctly one has to calibrate cryptoasset volatility either to market prices of Bitcoin options or to historical data. Due to poor liquidity of Bitcoin or Ethereum options, we may not be able to calibrate the exposure analytics using historical data. Diagram 3 shows the implied volatility surface for Bitcoin call option chains on May 30th, 2018. Only the near expiry options are traded with fair liquidity, as the implied volatility surface is flat for long-term options for all strikes.
Basel ll and lll
In Basel II, introduced in June 2006, BIS (Bank for International Settlements) introduced capital charge provisions for counterparty default in the form of Risk Weighted Assets (RWA). Details can be found in reference . These can be calculated either on present-value based formula (formerly CEM/SM, currently SA-CCR) or on one-year exposure calculated in Monte Carlo (IMA).
The GFC brought counterparty credit risk and CVA into the spotlight. The Basel III proposals first published in December 2009 introduced changes to the Basel II rules and the need for a new capital charge against the volatility of CVA. This ‘CVA VaR’ capital charge was always likely to be punitive since the Basel committee considered CVA volatility to be responsible for two-thirds of counterparty risk related losses during the crisis. There were other counterparty risk related changes in Basel III as well: calculating stressed exposure (taking into account systemic and specific WWR’s), applying extra multiplier due to systemic risk, including in CSA an extra margin period of risk and applying risk weighting of 2% for central counterparties (CCPs).
Currently, CVA capital can be calculated either using a formula-based approach or historical CVA VaR. In 2015, BIS  published a proposal to replace the latter methodology with a new one, which uses CVA sensitivities and FRTB (Fundamental Review of the Trading Book) type aggregation formulas, but there is no firm deadline on its implementation. Note that DVA is completely excluded from Basel III capital, which means that banks do not get any capital relief.
Cryptoasset transactions are not mentioned explicitly in the Basel II or III documents so, at this point, it is difficult to comment on what capital charges would be assigned to mitigate default of CVA risk. Though many cryptoassets have a direct analogy with foreign exchange trades, we expect their regulatory charges to be more punitive due to their high volatility.
On October 15th, 2018, the Swiss Financial Market Supervisory Authority (FINMA) took a conservative stance and advised financial services providers to assign a flat risk weight of 800% to cryptoassets to cover market and credit risks, regardless of whether the positions are held in the banking or trading book, until the Basel Committee on Banking Supervision (BCBS) has issued global recommendations. Such risk weighting of 800% is at the high end of the range, putting cryptoasset trading at the same level as hedge fund activity. Furthermore, FINMA puts a cap on cryptoasset trading activities to the amount of 4% of total capital, including both long and short positions. Once this cap has been reached, the respective institution is obliged to report to FINMA. FINMA exclude cryptoassets from High Quality Liquid Assets (HQLA) when determining liquidity ratios.
Extending to Cryptoasset Markets
In January 2019, cryptoassets reached an estimated market capitalisation of $140bn. Whilst this is a rather modest sum in terms of global markets, it is a remarkable rise to prominence considering this asset class was only created exactly 10 years ago. In a recent study by the Financial Stability Board (FSB) , it was stated that cryptoassets “do not pose a material risk to global financial stability at this time”. The driving force behind this sentiment was the relatively small market share that cryptoassets currently possess. The report also recommended “vigilant monitoring is needed in light of the speed of market developments”.
Over the last year, cryptoassets have become more acceptable by large financial organisations. Currently, some notable banks, including Goldman Sachs and Morgan Stanley, have desks dedicated to trading cryptoasset products. An interesting area of research is the potential of cryptoassets as hedging tools against stocks, bonds and the American dollar (see reference  or  for example). This paper highlights the important role that cryptoassets, specifically Bitcoin, have in the areas of portfolio management and risk analysis. Research like this suggests that in the future a growing number of well-diversified portfolios will contain some form of cryptoasset. This raises new questions about how to manage these products from a risk management perspective.
Until the risk associated with cryptoassets are understood, it is hard to imagine them becoming a mainstream asset class. One current limitation for cryptoasset markets is the lack of clear regulation guidance. To make matters worse, due to the decentralised nature of cryptoassets, it is not clear whether it would even be possible to enforce any established regulatory requirements. This challenge and others are discussed in ‘Crypto asset markets – Potential channels for future financial stability implications’  by the FSB. The report concludes that there is a clear benefit to applying regulation to cryptoassets gained from improved investor confidence.
Challenges Calculating Regulatory Capital
In its current format, Basel II and III do not specifically handle cryptoassets as a separate asset class and capital requirements for cryptoassets are currently considered in line with regulatory requirement using the “Other Asset” specification. Whilst this treatment is currently sufficient to meet existing regulatory demands, it is a long way off the FINMA recommendation stated in the previous chapter. It could be argued that there is sufficient variety in the quality of cryptoassets to define cryptoassets as a separate asset class and grade each type of cryptoasset in terms of risk profile. The challenge then is how to develop an efficient rating system that is capable of estimating the riskiness and what weight should each rating receive in the context of RWA. These weightings need to be obtained using statistical analysis similar to that used to derive the weightings for other asset classes.
The other challenge is how to calculate CVA capital. This is currently calculated using either a formula-based approach, although it is unclear how to calibrate the forward curve for the underlying cryptoasset, or historical CVA VaR where in an evolving market historical data might not be totally useful. One potential solution to this problem would be to use an AI approach when calculating CVA VaR. The limitation here is that the underlying models mimicked by AI are not typically known, which makes it difficult to develop a rigorous mathematical understanding of the underlying system.
By September 2020 any counterparty with non-cleared derivatives products amounting to over €8bn will have to calculate and post initial margin for non-cleared derivatives. Since a number of derivatives products now exist that derive value from underlying cryptoassets, what now needs to be addressed is how to accurately calculate the initial margin for products with cryptoassets as risk factors. This leads to similar types of modelling problems discussed in the previous section when attempting to evaluate the expected exposures.
The main use case for cryptoassets is trading. A key component of trading is managing risk, with credit risk being one of the three main categories of risk that cryptoasset traders face. We expect that financial institutions and investors will continue to dedicate resources to building a robust infrastructure and developing risk solutions to solve counterparty risk in the cryptoasset industry.
Even though trading cryptoassets presents significant challenges, it is our view that through technology, regulatory development and comprehensive risk management these challenges can be minimised. With cryptoasset industry evolving at pace, firms must not only be smart about the risks they pose but also manage them accordingly.