Issue 14: DeFi insurance
Around the Block sheds light on key issues in the crypto space. In this edition Justin Mart and Ryan Yi share a deep dive on DeFi insurance.
Where are all the DeFi insurance markets?
Insurance may not be the most exciting part of crypto, but it is a key piece that’s missing in DeFi today. The lack of liquid insurance markets prevents the maturation of DeFi and holds back additional capital from participating. Let’s take a look at why, and explore the different paths to providing insurance protection.
Insurance: The “so what?”
Insurance empowers individuals to take risks by socializing the cost of catastrophic events. If everyone was nakedly exposed to all of life’s risks, we would be much more careful. Readily available insurance coverage gives us confidence to deploy capital in emerging financial markets.
Let’s look at the relationship between risk and yield. If you squint, risk and yield are inextricably linked -- higher yields imply more implicit risk. At least, this is true for efficient and mature markets. While DeFi isn’t a mature market today, the significant yields are still an indication of higher latent risk.
Principally, this risk comes from the complexity inherent to DeFi and programmatic money. HIdden bugs-in-the-code are nightmare fuel for investors. Even worse, quantifying this risk is a mix of rare technical skill combined with what seems like black-magic guesses. The industry is simply too nascent to have complete confidence in just how risky DeFi really is. This makes insurance even more critical.
Clearly, strong insurance markets are a critical missing primitive and would unlock significant new capital if solved. So why haven’t we seen DeFi insurance markets at scale?
What are the challenges to DeFi insurance markets?
There are a few challenges in sourcing liquidity:
Who acts as underwriter, and how is risk priced? No matter the model, someone has to underwrite policies or price insurance premiums. Truth is, nobody can confidently assess the risks inherent in DeFi, as this is a new field and protocols can break in unexpected ways. The best indication of safety may well be the Lindy Effect -- the longer protocols survive with millions in TVL (total value locked), the safer they are proven to be.
Underwriter yield must compete with DeFi yield. When DeFi yields are subsidized by yield farming, even “risk-adjusted” positions often favor participating directly in DeFi protocols instead of acting as an underwriter or participating in insurance markets.
Yield generation for underwriters is generally limited to payments on insurance premiums. Traditional insurance markets earn a majority of revenue from re-investing collateral into safe yield-generating products. In DeFi what is considered a “safe” investment for pooled funds? Placing them back in DeFi protocols re-introduces some of the same risks they are meant to cover.
And there are a few natural constraints on how to design insurance products:
Insurance markets need to be capital efficient. Insurance works best when $1 in a pool of collateral can underwrite more than $1 in multiple policies covering multiple protocols. Markets that do not offer leverage on pooled collateral risk capital inefficiency, and are more likely to carry expensive premiums.
Proof of loss is an important guardrail. If payouts are not limited to actual losses, then unbounded losses as a result of any qualifying event can bankrupt an entire marketplace.
These are just some of the complications, and there is clearly a lot of nuance here. But given the above we can start to understand why DeFi insurance is such a challenging nut to crack.
So what are the possible insurance models, and how do they compare?
We can define different models by looking at key parameters:
Discrete policies or open markets: Policies that provide cover for a discrete amount of time and with well-defined terms, or open markets that trade the future value of a token or event? These coincide with liquid vs locked-in coverage.
On-chain or off-chain: Is the insurance mechanic DeFi native (and perhaps subject to some of the same underlying risks!) or more traditional with structured policies from brick-and-mortar underwriters?
Resolving claims: How are claims handled, and who determines validity? Are payouts manual, or automatic? If coverage is tied to specific events, be careful to note the difference between economic and technical failure, where faulty economic designs may result in loss even if the code operated as designed.
Capital efficiency: Does the insurance model scale beyond committed collateral? If not, there may be natural constraints on the amount and price of available coverage.
Let’s look at a few of the leading players to see how they stack up:
Specific DeFi insurance models
Hybrid insurance markets: Nexus Mutual
Straddling the DeFi and traditional markets, Nexus Mutual is a real Insurance Mutual (even requiring KYC to become a member), and offers traditional insurance contracts with explicitly defined coverage terms for leading DeFi protocols. Claim validity is determined by mutual members, and they use a pooled-capital model for up to 10x capital efficiency.
This model clearly works, and they carry the most coverage in DeFi today with $500M in TVL underwriting $900M in coverage, but still pales in comparison to the $50B+ locked in DeFi today.
Bundling these models together, there are several projects building either prediction markets or futures contracts, both of which can be used as a form of insurance contracts.
In the case of futures contracts, short selling offers a way to hedge the price of tokens through an open market. Naturally, futures contracts protect against pure price risk, paying out if the spot price declines beyond the option price at expiry. This includes the whole universe of reasons why a token price could decline, which includes exploits and attacks.
Prediction markets are a kind of subset of options markets, allowing market participants to bet on the likelihood of a future outcome. In this case, we can create markets that track the probability of specific kinds of risks, including the probability that a protocol would be exploited, or the token price would decline.
Both options and prediction markets are not targeting insurance as a use case, making these options more inefficient than pure insurance plays, generally struggling with capital efficiency (with limited leverage or pooled models today) and inefficient payouts (prediction markets have an oracle challenge).
Automated insurance markets: Risk Harbor
Exploits in DeFi protocols are discrete attacks, bending the code to an attacker’s favor. They also leave an imprint, stranding the state of the protocol in a clearly attacked position. What if we can develop a program that checks for such an attack? These programs could form the foundation for payouts on insurance markets.
This is the fundamental idea behind Risk Harbor. These models are advantageous, given that payouts are automatic, and incentives are aligned and well understood. These models can also make use of pooled funds, enabling greater capital efficiency, and carry limited to no governance overhead.
However, it may be challenging to design such a system. As a thought experiment, if we could programmatically check if a transaction results in an exploit, why not just incorporate this check into all transactions up front, and deny transactions that would result in an exploit?
Tranche-based insurance: Saffron Finance
DeFi yields can be significant, and most users would happily trade a portion of their yield in return for some measure of protection. Saffron pioneers this by letting users select their preferred risk profile when they invest in DeFi protocols. Riskier investors would select the “risky tranche” which carries more yield but loses out on liquidation preferences to the “safe tranche” in the case of an exploit. In effect, riskier participants subsidize the cost of insurance to risk averse participants.
For everything else, traditional insurance companies are underwriting specific crypto companies and wallets, and may someday begin underwriting DeFi contracts. However this is usually rather expensive, as these underwriters are principled and currently have limited data to properly assess the risk profiles inherent to crypto products.
The fundamental challenges around pricing insurance coverage, competing with DeFi yields, and assessing claims, in combination with limited capital efficiency, has kept insurance from gaining meaningful traction to date.
These challenges collectively result in the largest bottleneck: capturing enough underwriting capital to meet demand. With $50B deployed in DeFi, we clearly need both a lot of capital and capital efficient markets. How do we solve this?
One path could be through protocol treasuries. Most DeFi projects carry significant balance sheets denominated in their own tokens. These treasuries have acted as pseudo-insurance pools in the past, paying out in the event of exploits. We can see a future where this relationship is formalized, and protocols choose to deploy a portion of their treasury as underwriting capital. This could give the market confidence to participate, and they would earn yield in the process.
Another path could be through smart contract auditors. As the experts in assessing risk, part of their business model could be to charge an additional fee for their services, and then back up their assessments by committing proceeds as underwriting capital.
Whatever the path, insurance is both critical and inevitable. Current models may be lacking in some areas, but will evolve and improve from here.