Insurers plan for extreme events that could crater their solvency. Shouldn’t all levels of government do the same?

What if a storm like Melissa hit the northeast US?

Remember Hurricane Melissa? It hit Jamaica and Haiti just over a week ago, killing more than 60 people and causing billions in property damages, only a fraction of which will be covered by insurance. It was one of the two strongest Atlantic storms recorded in modern times. The US was mostly spared this time — the northeast coast experienced some dangerous swells, but that was it.

What if a major hurricane, or series of major storms, hit the nine states of the Northeast, a region that accounts for about a fifth of the nation’s economic output? We keep focusing on Miami and the southeast coast, but the Northeast US has experienced the largest increase in extreme precipitation in the nation over the last 30 years — and tropical cyclones and atmospheric rivers have played a significant role in that trend. Flood risk and infrastructure weaknesses are a major problem in Northeast states.

I learned recently at NCAR’s Research Applications Laboratory that we should be thinking not only about the risk of major storms, but also extreme volatility across all scales. Climate whiplash (key Daniel Swain paper here) in the form of wild year-to-year swings between clusters of multi-billion dollar events and quiet seasons, as well as between overwhelming rain and sweeping wildfires, is a particularly big issue in the Northeast. Scientists aren’t just looking at the individual mechanics of a specific event — they’re noticing that the collective behavior of storms and fires is changing from year to year and from decade to decade in our changing world. Sharp shifts between punishing flooding, drought, and wildfire risks are already a defining feature of this region.

Insurers respond to all of this with enormous caution. They’re wary about buying up risk in the Northeast, because they know volatility there is accelerating at the same time that climate modelers know much less about the hurricane risk there than they do about, say, storms that might hit Florida. There are fewer historical storms to study, and storms farther north are weird — they have different structures than classic hurricanes do, even as we know that hurricanes these days are tracking further north than they used to.

Large insurers, in particular, need to show that they have enough capital to survive in the event of extreme but plausible scenarios. I was told that insurers are interested in knowing their 1 in 200-year (0.5 percent annual probability) risks. The National Association of Insurance Commissioners has an Own Risk and Solvency Assessment, known as ORSA, that requires reporting on risks, and large insurers often model and report on 200-year (or worse) events.

Because whiplash in the Northeast is strong and growing stronger, because the storms there are strange, and because scientists have lower confidence in their predictions, insurers get jittery. So they back away, sending out non-renewal notices and withdrawing from future business.

A 0.5 percent annual risk of anything might sound pretty low to you. But it’s a key metric for running a large insurance business in the Northeast. Why isn’t at least that level of risk the planning lever for local governments in that region, driving building codes, land use decisions, and investment priorities? The American Society of Civil Engineers has proposed risk-based standards for new construction within 500-year (0.2 percent annual risk of happening) floodplains, but getting these standards into actual building codes that are enforced by local jurisdictions is an extraordinarily difficult and prolonged process at this point. Local governments don’t understand they need to do this.

We should be anticipating and planning for the worst, not the average. There’s a saying in education circles that “When we design for the average, we design for no one,” and in infrastructure design the whole point is to plan for peak loads, not placid mornings. We are planning for human wellbeing, or we should be, which is the essential infrastructure of a functioning country and economy. That means we need to take into account the substantial consequences of less likely events—the thick statistical tails of weather as the climate warms.

Hurricane Melissa’s effects were catastrophic in Jamaica. Dutch experts, who live in a land that plans for 1 in 10,000 year risks, not one in 200, are warning that the Netherlands should be planning for critical regions to experience six feet of sea level rise by 2100. Major coastal storms are more destructive when they ride in on higher seas. And you can think of a warming climate like a giant ever-growing sponge hovering above us, absorbing ever more moisture from the soil, sparking crazy wildfires, and then quickly dumping torrential rains on our heads and houses. This is happening to our friends in Australia all the time.

It is fine to hope for the best. Our entire stock market is a hopeful bet on the success of a few large American tech companies in an AI-driven future. We also need to expect the worst — and we’re not very good at that.