Credit Loss Distributions: How Crypto and TradFi Are Converging

When Voyager, a digital asset lender, defaulted in late 2022 because they’d loaned money to 3AC, they landed the regrettable honour of being the first major example of default correlation in crypto. As we’ve discussed before, prior to 2022, counterparty risk was exclusively restricted to hacks or frauds, like Mt. Gox.  But 2022 witnessed the first counterparty defaults from trading and operating losses. Given this is the typical cause of TradFi defaults, it could be seen as something of a coming of age.

Credit risk correlation, when the default of one entity increases the probability that another will default, is an important concept in risk management.  These correlations can be direct, as was the case with Voyager, or indirect, so called third cause correlation.  An example of the latter also occurred in2022 when Vauld and Celsius were both taken down by Luna’s collapse and the subsequent market volatility--the third cause. Whilst this inter-relationship doesn’t affect expected losses, it does impact the distribution of losses and thereby risk exposure and capital requirements.

The graphic illustrates this point.  We’ve simulated a portfolio with exposure to five large crypto-counterparties including Coinbase and Huobi.  We’ve also included a multiplier effect, defined as the impact that one default has on the probability of another default.  When the multiplier is 1.0 there is no correlation since the default probabilities of the other counterparties stay the same. Where the multiplier is 3.0 the default of one counterparty triples the chances of the other counterparties failing.

In the base case, when the multiplier is 1.0, there’s a 79% chance of having no defaults in a given year, a 19% chance of one and a 2% chance of two or more.  Adding correlation, then changes the distribution.  The expected loss stays the same, by definition.  But the variance increases as the tails of the loss distribution fatten.  Specifically, the chances of no loss rises to 83%.  Conversely, the chances of two or more defaults more than doubles to 5%. Whilst most crypto-firms can survive one of their counterparties getting into trouble, as was the case for the majority in 2022, they would struggle to weather two or more. In this sense, allowing for correlation doubles a firm’s risk.

Later this month we’ll be releasing a risk simulator that combines Agio Ratings’ individual counterparty default probabilities with customisable correlation assumptions and user-defined exposure levels to calculate a firm’s loss distribution.  This can be used to measure credit value at risk (i.e. the worst loss that can occur with a given probability) or the probability of credit losses exceeding a given amount, such as the capital base of the firm.  These can then be used to calculate counterparty loss contributions, set exposure limits, measure risk-adjusted yields and so on.