The Fundamental Problem With Risk
The past is less useful than ever in our efforts to imagine the future
Saddleridge Fire [source: New York Times]
This piece in the NY Times seems to be about insurance issues related to the California wildfires. But it’s really about something much larger. How can we assess risk in the postnormal era, where history does not shed light on the trajectories of the future? [emphasis mine]
California Bans Insurers From Dropping Policies Made Riskier by Climate Change | Christopher Flavelle and Brad Plumer
The state’s homeowners insurers lost a total $20 billion in the 2017 and 2018 wildfires, according to an analysis published in October by Milliman, an actuary and consulting firm. That’s twice the industry’s cumulative profits since major wildfires in 1991. A line of business that was until recently profitable is now unprofitable, the authors wrote, “exposed to a severe peril that is neither easily measured nor fully understood.”
Eric Xu, an actuary at Milliman’s San Francisco office and one of the report’s authors, said that the shock of the California wildfires echoes Florida after Hurricane Andrew in 1992, which caused $28 billion in damage and caused the failure of a dozen insurers.
But the threat facing insurers in California is in one sense trickier: After Andrew, many national insurers stopped writing coverage in Florida. But Mr. Xu said California represents too great a share of total revenue for most national insurance companies to just walk away from the state altogether.
Unable to leave, insurers have sought other solutions to protect themselves from rising wildfire costs. But those changes highlight the obstacles facing insurers as climate change worsens.
One fix is for insurers to buy what’s called reinsurance — a sort of insurance for insurers — providing payments if claims rise beyond a certain level. But as the risks from climate change have grown, reinsurance companies have raised the cost of the protection they offer.
For insurance companies, the most obvious response is to pass the costs on to customers in the form of higher prices. California insurers filed 80 requests for rate increases in 2018, more than double the number of requests in 2015, according to data provided by the state.
But California, like many states, gives regulators the power to reject those requests. And the state forbids insurance companies from setting rates based on what they expect in future damages. Insurers are allowed to set rates only based on prior losses.
“That works, until it doesn’t,” said Mr. Frazier, of the insurers’ trade group. He said the state should change the rules so that insurers can base premiums on more than just past experience.
Ricardo Lara, California’s insurance commissioner, said he’s wary of letting insurers use models that may not be accurate.
“I want to be very cautious about opening the rate-approval process to anything that compromises the transparency and objectivity that exists today,” Mr. Lara said in a statement. “Protecting consumers is our top priority, and that is the lens we will use to evaluate any catastrophe risk models in the future.”
Mr. Frazier acknowledged that the ability to predict wildfire damage remains imperfect.
“There are a lot of people scrambling to really understand the nature of this catastrophic risk,” Mr. Frazier said. He called the wave of fires starting in 2017 “a complete wake-up moment for an industry that thought it knew what was on its books, but actually didn’t.”
If you think about the quote from Ricardo Lara, who says the state wants to use models that are accurate. But in a time of accelerating climate change, basing the models only on the past means that the models will be wildly inaccurate. He is confronted by a paradox, a Hobson’s choice: the insurers are effectively blocked by the state from even trying to make better risk assessments because they can’t use models of the future, but models based on the past are useless.
But even if they could use projections of the future, would the models be accurate? Not really. But they would certainly lead to a huge spike in insurance fees as insurers would likely adopt worst-case projections.
