This year, California wildfires have torched 1.4 million acres, an area bigger than Connecticut. Nine thousand homes and other structures have burned. While brand-name insurers like Allstate and State Farm are suffering from a surge in homeowner insurance claims, another industry is gasping for air behind the scenes: reinsurance. Reinsurers cover companies like Allstate when they see big losses, typically paying out 60% of claims for catastrophic events. To make up for lost profits and rising risks, reinsurers in California have raised their prices by about 600% over the past five years.
San Francisco startup Kettle thinks its sophisticated statistical models can predict wildfires better than any other reinsurer, and it plans to offer lower prices as a result. It officially launched today and recently closed $4.7 million in seed funding from True Ventures, Acrew Capital, Homebrew Ventures, Anthemis and Inspired Capital.
Nathaniel Manning, 35, cofounded the company with Andrew Engler, 33. Before Kettle, Manning worked in climate research, and as a White House Presidential Innovation Fellow in 2012, he spent three nights sleeping on the floor at FEMA after Hurricane Sandy. “I began to have this amazing respect for insurance companies,” he says. “It’s pooling our money together as a community and saying, ‘If you’re the one who has gotten hurt with this, I got your back.’” For several years afterward he ran Ushahidi, an open source software nonprofit that organizes data for crisis response.
In 2018, Manning knew nothing about insurance but wanted to start an insurtech company. He was introduced to Engler, an insurance veteran who had worked at Allstate and was then vice president of digital products at Bermuda-based reinsurer Argo Group. The two met and founded Kettle in early 2020, with Engler as chief executive and Manning as chief operating officer.
Kettle uses artificial intelligence and machine learning to predict the likelihood of a wildfire in any given square mile. It ingests data from 47 different sources, including NASA satellites, weather satellites and laser-based Lidar mapping sensors. Circumstances like brown-colored brush, high winds, little rainfall and hot temperatures make fires more likely and are picked up by Kettle’s models. For every analysis, their models run 42 million simulations using a statistical approach called “swarm neural networks.” Manning says most reinsurers use stochastic modeling that runs roughly 100,000 simulations.
“Of the 14 largest fires in California this year, which together have burned 3.5 million acres, our model has predicted 11 of these to be in the top-10%-chance areas of burning in California,” he says.
Eight of the 10 biggest fires in California’s history have happened in the past 10 years, erasing decades of profits for reinsurers and driving them to set exorbitant prices. This year Pacific Gas and Electricity, a California utility hit hard by the wildfires, paid a stunning $749 million to insure potential losses of just $758 million. Last year, it paid $212 million for $430 million of coverage.
These rising costs eventually reach consumers. A January 2020 report from the Sacramento Bee summarized, “Homeowners who used to pay about $2,000 a year for coverage can find themselves paying $6,000 or more.”
As climate change escalates and brings more fires with it, Manning and Engler think their models can help keep insurance affordable. They say the going rate for California reinsurance 30-35%, which means a company would pay $30 million in annual premium fees to insure $100 million worth of property. That’s up from a 3-5% rate in 2015. Mathematically, a 30% rate means there’s nearly a one-in-three chance that any acre in California will burn this year. One-third of California isn’t going to burn, and Kettle’s cofounders think they can better predict risk and offer insurance at an average 20% rate across the state.
Before taking on clients, the startup needs to secure financing from outside parties who would help pay out claims when disasters hit. It’s trying to get funding from other reinsurance companies and hedge funds, aiming to secure between $50 and $100 million of backing this year. Engler says they’re in negotiations with 30 capital providers.
To give a sense for how large of a business they expect to build in the near term, if Kettle issued $100 million in reinsurance while charging customers a 20% rate, it would bring in $20 million in premium revenues, or roughly $5 million in net revenue.
Among the biggest uncertainties about Kettle’s prospects are whether it can truly predict wildfires. “They’re better suited than any team I’ve seen to predict catastrophic risk,” says Lauren Kolodny, a partner at venture firm Acrew and Kettle investor. “At the same time, catastrophic risk has been so catastrophic because it surprises society.”
Rick McCathron, president of fast-growing homeowner’s insurance startup Hippo, says he’s impressed by Kettle’s sophistication, although he’s not ready to buy their product just yet. He says there are many factors that make wildfires unpredictable. “The shift of population matters. Where people are building matters. Weather patterns are changing. And fire seasons have pushed later and later in the years.”