Oracle (parametric insurance)
An external data source whose reading is binding for triggering a parametric payout. Its reliability becomes the contract's weak point.
Haiti earthquake, January 12, 2010: the CCRIF facility paid the Haitian government within days, with no loss adjustment, because the index computed from seismic data crossed the contract's threshold. In parametric insurance, the oracle is the source of the reference data that triggers the payout or not. The term comes from computing and was popularized by the blockchain world. It names an entity, whether an institution, a device or a network, whose published index value the contract recognizes as irrevocable and sufficient to trigger it. Institutional oracles include national meteorological agencies, geological services, hurricane observation centers and official statistics offices. Decentralized oracles such as Chainlink, API3 or UMA aggregate several data sources and compute a reference value by consensus, which removes the single point of failure of an exclusive source. The oracle problem, a term from smart contract theory, is how to bring real-world data into an automated system without having to trust a third party again. In parametric insurance it has three dimensions: the technical reliability of the source, its publication latency, and its geographic coverage relative to the loss site. A weather station several tens of kilometers from a farm can record rainfall very different from what fell on the plot, and basis risk gets worse.
A parametric drought contract for Indonesian rice farmers uses the national meteorological agency's data as its primary oracle. In 2025, a cyclone knocks the reference station out for six weeks, and its data is then revised downward in the final publication. The contract was set to trigger on the provisional value. On the revised value, it does not trigger. A decentralized oracle network fed by three additional stations and a satellite could have prevented this.
Follow the research
One letter a week. First, the week's underwriting note, when one is published: a dated, argued decision, signed by Paul Bernardski. Then what the observatory published on cyber risk, political risk and insurance. No account, no tracking.
No spam, one-click unsubscribe, no data ever sold.