DeFi Risk Transfer: Towards A Fully Decentralized Insurance Protocol
TLDR
This paper proposes a decentralized insurance mechanism for insuring/covering smart contracts without using oracles or external dependencies. It does so by pooling assets, and splitting pool redemption rights into risk tranches. It also provides full Solidity implementation for this mechanism.
Key learnings
One of the most interesting mechanisms within this proposal is the pricing of risk premia. This rate is set by the ratio of two coverage tokens which are used to tranche risk. This pricing mechanism is novel and interesting as it requires no modeling or forecasting of any kind. Using self-contained pricing mechanisms is difficult to achieve in crypto, but valuable.
The design choice to have two modes of redemption to address potential liquidity risk is practical and simple. The fallback redemption mode splits liquidity risk amongst tranches in a way that naturally assigns the liquidity risk to correlated tranches. This mechanism is worth considering for Concrete.
Applicability to Concrete
This paper is highly applicable to Concrete, and it generalized such that it can be directly implemented for Concrete's use case. The pricing model is of particular interest as it is one of the most difficult aspects of any derivatives application and this paper's approach is simple, self-contained, and robust.
The distribution of risk to multiple other protocols according to a tranching structure is not novel, but the specific implementation proposed is novel in a smart contract context. This mechanism is directly applicable to future LP designs. We should and must explore implementation of something similar to this mechanism in the future.
Methods and outputs
This paper does not perform a study per se, but rather submits a protocol design and implements it in Solidity. The code implementation provided is applied to AAVE and Compound.
The design relies on transferring risk from one protocol to at least 2 other external protocols. Risk is tranched and distributed to these protocols accordingly. The paper describes this well.
"The general idea of our insurance protocol is to pool assets from two third-party protocols, and allow users to split the pool redemption rights into two tranches: A and B. If any of the third-party protocols suffer losses during the insurance period, those losses will be primarily borne by the B-tranche holders. A-tranche holders will only be negatively affected if 50% or more of the pooled funds are irrecoverable, or if both protocols become temporarily illiquid and face (partial) losses. We effectively split the redemption rights into a riskier and less risky version and allow the market for A- and B- tranches to determine the fair risk premium in line with the users’ expectations."
There are 3 key steps in this process
Risk splitting
Anyone can deposit any amount of the collateral token C.
Users receive equal denominations two tranches - A and B - via tokens A and B.
Invest/divest
First step is to split the total collateral value to two lending protocols (AAVE and Compound in their code example). The insurance protocol gets wrapped liquidity shares from each protocol - X and Y.
The insurance protocol tries to liquidate the wrapped shares.
Redemption
There are two potential modes of the redemption phase - liquid and fallback modes. Mode selection depends on the success of the liquidation at the end of the invest/divest phase
If the liquidation of X and Y is successful, the insurance protocol distributes earned interest equally among all A and B tranche holders. and the potential losses are be computed.
If liquidation is unsuccessful, redemption is conducted through the fallback mode. The primary reason for this is because of liquidity shortages.
In fallback mode, users redeem their tranche tokens directly for their preferred mix of X and Y tokens.
It is vital to note that A tranche holders are only negatively affected if more than 50% of the pooled funds are lost. All losses below this value are absorbed by B tranche holders.
The seniority of the A tranche is enforced in fallback mode via a timelock-based redemption sequence. First, A-tranche holders decide on the mix they prefer of X and Y tokens. After the time lock has expired, B tranche holders can claim the remainder.
The tranches' risk premia is priced according to the relative price between the two tranches. The price emerge naturally and does not depend on preset parameters or statically implemented risk spreads.
Key benefits of this mechanism
Allows for collateral (B tranche) to be held in an interest-bearing asset instead of a stablecoin.
Extreme capital efficiency is achieved because this mechanism doesn't over insure. The risk and capital are linked through our tranches and cannot be separated without the use of another protocol.
No oracles or other external dependencies of any kind. This is especially impressive and important as the risk premia are priced in a completely self-contained manner.
Questions/challenges/concerns
This model insures entire protocols and as such, is not directly applicable to Concrete's insurance use case but can be modified/adapted to fit this.
Otherwise, this mechanism is well-designed and merits close reading and potential direct application to Concrete.
Links and further reading
Last updated