Index

Prices Are Signals: Don’t Mute Their Alarms
9 December 2025
Peter Coffee

The politics of climate policy have swung from science-based consensus to partisan posturing – but below that fog and froth, there’s a powerful current of pricing (and paying for) rapidly growing risks in the market for homeowner’s insurance.

To start things off, let’s run the highlights reel:

• In 1992, the United Nations Framework Convention on Climate Change was ratified as a formal treaty by the United States Senate, in a voice vote with no recorded dissents. Acknowledging reality, and cooperating with other countries to address it, was not considered a partisan issue.
• Two decades later, again by unanimous consent, both houses of the U.S. Congress approved flood insurance reforms that were largely inspired by rapid and costly change in climate-related risks. “Nationally, the Special Flood Hazard Area is projected to be 40%-45% larger by 2100, with approximately 70% of this increase due to climate change,” observed a report from the Congressional Research Service. Again, this was not considered a partisan issue: this was about the safety of people’s homes, jobs, and lives.
• Fast forward by one more decade, though, and in 2024 we find the Florida state legislature passing a bill that deleted all references to “climate change” from the statement of the state’s energy policy. One alteration revisedly instructed the Department of Agriculture and Consumer Services to “advocate for energy and climate change issues.” Apparently, even merely mentioning that climate change issues might exist has become something that’s Just Not Done.
• Most recently, in July of this year, a U.S Department of Energy report asserted that “models and experience suggest that CO₂-induced warming might be less damaging economically than commonly believed, and excessively aggressive mitigation policies could prove more detrimental than beneficial.” Wishful thinking codified as policy.

Now that we’re caught up—and perhaps, a little shocked to realize how much the flavor and texture of the conversation changed in just three decades—let’s look at some measures of current reality, and contrast them with some new and troubling behaviors of delusion.

Reality:

• “In 2023, private insurers lost money on homeowners policies, meaning these companies paid more in claims than they collected in premiums, in 18 U.S. states… That total is up from 2018, when insurers were in the red in 12 states, and 2013, when insurers lost money in only eight states.”
• Some estimates suggest that “Insured losses from natural disasters in the U.S. now routinely approach $100 billion a year, compared to $4.6 billion in 2000.” Between 2020 and 2023, insurance premiums in the top 20 percent of counties for climate risk increased by 22 percent and studies have found that insurance premiums have increased 40 percent faster than inflation.

The website began publishing climate risk ratings last year using data from the risk-modeling company First Street. The scores aimed to quantify each home’s risk from floods, wildfires, wind, extreme heat and poor air quality. But real estate agents complained they hurt sales. Some homeowners protested the scores and found there was no way to challenge the ratings. Earlier this month Zillow stopped displaying the scores…

We are unwinding far more than merely three decades of fact-based behavior. We’re behaving in a way that might have made 14^th-century Genoese merchants shake their heads – because it was at least as long ago as the year 1347 that risk pools and insurance policies began to emerge, not merely in connection with a single voyage (a la The Merchant of Venice) but in a more general marketplace of assessing and pricing risks.

It gets worse, though, because the burden of delusion is often being carried by those who can least afford it. Also from The New York Times,

The typical American household last year paid about $500 in home insurance premiums for every $100,000 of home value, or 0.5 percent… But in California, which suffered through more than 7,000 wildfires last year, the typical homeowner paid premiums as low as .05 percent of home value. By contrast, in parts of Alabama, Oklahoma, Louisiana and Texas, the average homeowner faced home insurance premiums greater than 2 percent of the value of local homes.

What would cause such an uneven pricing of risk? It’s not, as one might presume, the uneven distribution of risk itself. Rather, according to research from the Harvard Business School cited by the NYT,

In states where officials tightly control what insurance companies can charge, premiums tend to be priced below what they would be if they reflected the true likelihood of damage from storms, fires or other catastrophes… After big losses in those tightly regulated states, such as California, national insurers tend to raise rates in more loosely regulated states.

When we mess up the signaling behavior of prices, attempting in effect to slow down the car by holding back the needle of the speedometer, markets get correspondingly messed up. From that same NYT article,

In communities where insurance rates exceed the actual risk, homeownership can be unaffordable. And in places where insurance prices are too low, it encourages people to move into homes in areas likely to be hit by wildfires or other disasters that could deliver financial ruin. The market is “incentivizing all sorts of crazy behavior” [HBS researcher Ishita Sen]

Climate change already presents a structural challenge to the most basic model of how insurance works.

Insurance has worked when many people had a similar risk that was somewhat predictable for large numbers of people, but was not highly correlated among individuals. For example, any typical house has little chance of burning down in any given year, and a single house afire doesn’t usually burn down an entire neighborhood; it’s therefore been possible for people to pay a small (and predictable) insurance premium, rather than risking loss of the entire value of one’s home.

In short words, insurance works when

• many different parties are exposed to common and familiar risks,
• with magnitudes that can be estimated from past experience and present observation,
• but uncorrelated in that any one party’s damage from that risk is unrelated to any other party’s likelihood of the same thing happening at the same time.

• that are not familiar,
• that are not readily estimated by current observation,
• and that are likely to have impacts over an entire region in massive incidents such as wildfires and floods.

At the very least, we’ll do well not to introduce deliberate distortion, or indulge in pretend-it’s-not-real concealment, when our signaling systems are already hard pressed to reflect a complex and rapidly changing situation. We need all the help we can get to deal with what’s to come.