Skip to content

Latest commit

 

History

History
21 lines (12 loc) · 5.34 KB

README.md

File metadata and controls

21 lines (12 loc) · 5.34 KB

Weather Insurance

If you have the MetaMask Wallet as browser plugin or mobile app, you can interact with the contract! To just play around on Ethereum's Ropsten Testnet, get some free Testnet ETH. It's also live on Ethereum's Layer 2 networks Optimism and Arbitrum, which you can add to your Metamask wallet here.

The contract

This contract allows anyone to sell a trustless draught insurance policy to anyone else on earth, based on NASA's ongoing precipitation data feed measured by the GPM satellite constellation via their microwave radiometers. Individuals can request an insurance by specifying the conditions under which they want to receive some payout, say "if it rains less than 10L during April 2022 at coordinates XY, pay out 2 ETH", while depositing some insurace fee they would be willing to pay, say 0.5 ETH. Any insurer can accept these conditions by depositing enough funds into the contract to cover the payout, and immediately claim the 0.5 ETH fee. If the insured individual can eventually generate an enclave proof that the draught happened, i.e. it rained less than 10L at XY in April 2022, they can claim their 2 ETH payout. If instead the insurer can prove that it didn't happen, they can reclaim their 2 ETH deposit back.

Here are some ideas for how you could improve the contract logic to create a more compelling insurance product:

  • Add support insuring other types of risks, e.g. fire via this real-time NASA data feed
  • Invent a mechanism that allows the contract to estimate payout probabilities for all policies (so it can automatically price and sell insurance policies), e.g. by incentivizing prediction markets via dedicated token rewards, or create a incentivized machine learning compentition like Numerai, with an on-chain leaderboard that rewards the publisher of the best weather-forcasting model, as validated via oracle enclaves
  • Once the contract itself can estimate payout probabilities, it could fund all payouts from a single pool of capital which only needs to be roughly as large as the expected, rather than the maximum possible total payout. This would dramatically increase capital efficiency and reduce insurance fees.

The big picture

Every agriculture business needs to insure against weather risk. Most people in low income countries work in the agriculture sector. But in most low income countries, affordable weather insurance is almost nonexistent. One reason is that the size of an individual insurance policy, and thus the fees you could charge per person, are rather small. Therefore, to become affordable, an insurance solution must efficiently operate at a huge scale. Another reason is that people need to be able to trust that they will actually receive the payout if a damage occurs, even if they have no way to take the insurance provider to court. So it's not surprising that the idea of using smart contracts to provide parametric weather insurance to the world has been around for many years, and was already explored before.

But as far as we know, that vision hasn't really taken off yet. We don't have a simple answer for why that is, other than that it's just damn hard to profitably provide a compelling weather insurance product in low income countries. And while the problem doesn't immediately get easier if you bring smart contracts into the mix, it's also clear that if there's any way to solve it, it must be online, highly automated, globally trustworthy, and scale efficiently across countries regardless of their local payment infrastructure or institutional context - so there's a good chance it will end up being a smart contract.

Fortunately, Open Contracts' Python oracle capability could solve a lot of problems developers will encounter along the way: it could be used not just to verify whether a given damage occured, but also to directly compute a fair price for a requested policy by estimating the payout likelihood via a machine learning model. It also could be used to evaluate and reward proposals for more accurate ML models, and so on. And other Open Contracts could be helpful along the way: a version of Fiat-Swap could make it easy for farmers quickly purchase the tokens needed to buy an insurance policy via their local currency, and convert the payouts back into cash to pay for their damages, for example.