Managing Correlation Risk

Background

The mutual needs to manage its exposure to single events that could cause large losses to make sure it is sustainable in the future. The current products manage this by limiting exposure to any one risk to a global maximum of 20% of MCReth. This works quite well from a risk perspective but is challenging from a UX perspective as the user needs to buy many covers to get full coverage.

New products like Yield Token Cover (soon) will address the UX issues at the expense of making it a lot harder to manage correlation risk.

Generally, this makes sense as complicated aspects should be managed by a few experts rather than forced on to end consumers, so the challenge becomes how do we manage correlation risk well.

Approach

Overall, correlation risk will be managed via several levers rather than hard rules, and it will be up to members to adequately adjust system parameters to arrive at the appropriate risk/reward settings for the mutual.

Lever Description How it will be Used
Global Capacity Limit 20% of MCReth, limits the total amount of cover on any one risk. As the mutual grows and writes more diverse covers this limit should gradually decrease. A 10% limit is more long term sustainable.
Capacity Multiple Multiple of stake to open up capacity. Currently ranges from 1x - 4x. All covers should converge to 1x over time. This is likely implemented by not increasing any above 1x, especially for Yield Token Cover, and then when Staking 3.0 is introduced this factor will be removed and be effectively 1x for each risk.
Pricing Ability to set prices independently from capacity. Have the ability for Super Users to set very high prices (effectively turning off future sales) to allow overall limits to be managed if too much correlation risk accumulates. This is particularly important as correlation risk occurs across risks, so individual risk levers are required.
Reinsurance Build a reinsurance layer, as described separately that tops up the capital pool after large claims. Probably the best overall way to manage correlation risk, with the amount of capital available being linked to the probability of it being called on, so there is an alignment of interest between reinsurance capital and the mutual managing correlation risk well.

Ongoing Monitoring

Even with these levers it is likely that the mutual accumulates risk on particular events, for example:

  • USDT significantly failing
  • MakerDAO failure causing a DAI de-peg
  • Compound/Uniswap or other low level protocol failure

As a result we need to develop the skills/resources for the community to run risk analysis across the mutual on an ongoing basis. The reports/output can be used to influence decision making like pricing/capacity across different risks. This is likely to take the form of an analysis that tags each risk with particular extreme events and then adding up the PML (probable maximum loss) assuming each event occurs.

Understanding the PML’s across various scenarios will become a necessary resource for members to manage the overall risk position of the mutual.

Implementation

The mutual should work towards building these aspects out after the launch of Yield Token Cover, in particular pricing and a reinsurance layer.

More generally the mutual should favour continuous mechanisms (like pricing) rather than relying on on-off governance votes to adjust parameters to adjust risk settings.

5 Likes