In October 2026, California began requiring the insurers it supervises to write down what they expected 2030, 2040 and 2050 to do to them, in confidential plans prepared for state examiners.1 The first plans were due by January 1, 2028. Most were long. Nearly all were written by people who would be retired, dead, or running someone else's company by the time the furthest horizon showed up.
Calder Mutual's ran 412 pages. Its chief risk officer, Walter Brennecke, signed it in the fall of 2027 and retired in 2034.
That horizon is here now, and the person living in it is Prudence Ashworth, 47, Calder's chief executive since 2045. We met on the first Tuesday of November, on the fourth floor of the company's Sacramento headquarters, in a records room kept at sixty-four degrees for the paper's sake. It was late enough in the year, she noted, that the windows no longer needed their afternoon film.
A digital copy of the plan exists. It has survived three archive migrations and lost its footnotes in two, so she prefers the binder. It is green and coffee-ringed, with a sticky note on the spine that reads, in capitals, DO NOT RECYCLE. EXAMINERS. Brennecke's marginalia, in blue felt-tip, surfaces every few pages like a man clearing his throat.
I read the plan aloud. She listened with her eyes closed, sometimes mouthing a sentence half a second ahead of me.
Page 61: "The scenarios that follow are not forecasts of the most likely future outcomes."2
Prudence: Walt's crucifix. He hung it over every door in the building.
Did it keep anything out?
Prudence: Nothing. (laughs) Nobody reads a disclaimer the way it's meant. My board read Scenario B like a horoscope they happened to like. Late action, moderate physical risk. That one became "the plan." Scenario C, the bad one, became "Walt's pessimism," a phrase I personally heard said in this room in, um, 2037? About a man who was right about the... well. Partly right.
Then let's grade him. Page 88: "By 2040 the Company will have reduced its exposure in the highest-hazard foothill territories through non-renewal and pricing, consistent with solvency targets."
Prudence: Non-renewal. The polite word for a letter saying we won't be back next year. We finished in 2039. He was off by a year. In our favor, if you're the kind of person who thinks that's in anybody's favor.
Are you that kind of person?
Prudence: On Mondays. (pause) Look. Back in the twenties the French ran a stress exercise that let insurers manage their books all the way out to 2050, and the participants barely touched the levers. Hardly anyone pulled out of a region. Hardly anyone cancelled anybody.3 Walt cited it. He left a note in the margin of page 90, underlined twice:
Reluctant in exercise. Will not be reluctant in practice.
He understood that a stress test is a room where nobody's career is on the line. Our exit took one board meeting. Forty minutes. There were muffins.
The plan saw the retreat. Did it see what came after?
Prudence: It saw the households. The Australians had done the arithmetic. If your premium eats four weeks of income or more, you're counted as uninsured, and by that measure they got roughly one freestanding home in four by 2050.4 Walt ran our version, and the foothills came out worse. Page 141. Lovely table, two decimal places.
What it doesn't have is a column for what happens to the one in four. They leave our model the second they stop paying us. Actuarially, that's correct. They don't leave the county.
Appendix D, describing the catastrophe model: "while keeping all other modules and assumptions constant, including vulnerability, exposure and insurance conditions."5
Prudence: (She takes the binder and reads the line to herself, twice.)
Yeah. That's an honest sentence. A good one. It's how you isolate a variable: hold everything else still so you can see what the weather alone does to you. Swiss Re did it that way. Everybody did. And I want to defend it, because people read these old plans now as if they were written by idiots. Walt was a careful man who told you exactly what he was doing.
But everything he held still is what moved. Vulnerability? Half our remaining book has ember vents and a gravel ring now, because we priced for it. Exposure? People left. People rebuilt bigger. People rebuilt smaller somewhere else. Insurance conditions? We became one of the insurance conditions. He held us constant, and then we moved.
And the part I can't... okay. I keep a printout in my desk. An AP story from January 2026, about a man in Altadena after the fires. He liquidated most of his retirement savings while his payout was tied up, so he could hire contractors and rebuild fast. Partly because his golden retriever was thirteen and he wasn't sure she'd live long enough to get home. She made it. The reporter found her lying in her old spot in the yard while he drank his coffee.6
There's no module for Daisy Mae. You can't hold constant how long a dog has left.
And the plan wasn't wrong to leave her out. The story itself said most people didn't have his options. But every claim we've paid since 2028 had a Daisy Mae in it somewhere. A school year. A mother who wouldn't move. A guy who needed to stand on his slab for a month before he could decide anything at all.
We paid. We stayed solvent through all of it, which is the one thing the plan was asked to promise. But even back in '26 the state was ordering lenders to account for survivors' insurance money still sitting unreleased after the checks went out.7 Walt's plan stops at "claim paid," the way a wedding video stops at the kiss. Everything after that, nobody films.
Did Brennecke think about any of that?
Prudence: Page 203, the limitations section. "Material limitations and known weaknesses." That was the professional language, straight out of the actuarial standards.8 Migration is on his list. Lender behavior is on his list. So he knew those things lived outside the model. Knowing something is outside your model and never having thought about it are different crimes. I've stopped charging him with the second one.
What did it never think to ask?
Prudence: (long silence) Whether anyone could live in it.
I mean us. The plan is full of sentences like "Management in 2050 will have discretion to..." and I am management in 2050. I first read that as a junior analyst in 2031 and thought, who is that? That person sounds like they have it handled.
The standards say the documentation should let another qualified actuary judge whether the work was reasonable. It does. I can check every number in here. It just can't tell me whether I'm allowed to feel bad about page 88.
Are you?
Prudence: (laughs) Is that a regulatory question? I sign an attestation every year saying the process was applied. I haven't found the line for that.
Have you talked to him?
Prudence: Once. He's in his eighties, living in Placerville. Up in the foothills. Which, yes, I know. We don't write there anymore; he's on the state's last-resort plan. (pause) I called to ask about... no. I called to ask why he wrote "will not be reluctant in practice" and then never put the reluctance anywhere in the plan. He said:
"Because I was the reluctance, Prudence. I figured you'd notice."
Then he asked whether the rains had started.
Your 2080 plan is due soon. What goes in your limitations section?
Prudence: Everything I can name. (She closes the binder, then opens it again to page 61, the way you'd check a stove.) Plus one line I'm still fighting legal over. It says the person reading this will know things we don't, and we've tried to leave them room to disagree with us. Legal says that isn't a material limitation. I asked them what it is, then. They're getting back to me.
When we finished, she slid the binder back onto its shelf, between a carton labeled ORSA 2049 and an empty space. Her phone showed the county smoke index flipping from amber to clear, the first clear reading since August. She didn't mention it.
Calder Mutual's examiners are advised that Ms. Ashworth, Mr. Brennecke and the year 2050 have not yet arrived. As page 61 would remind them, the scenario above is not a forecast of the most likely future outcome.
Calder Mutual's plan is invented. Where it is quoted, its language borrows from real documents that anyone writing such a plan in 2028 would have had on their desk.
Footnotes
-
California Department of Insurance, "California sets global standard for insurance resilience…," October 27, 2026. The regulation took effect October 26, 2026, with planning horizons of 2030, 2040 and 2050. https://www.insurance.ca.gov/0400-news/0100-press-releases/2026/release040-2026.cfm ↩
-
Bank of England, Results of the 2021 Climate Biennial Exploratory Scenario, May 24, 2022 ("The CBES scenarios are not forecasts of the most likely future outcomes"). https://www.bankofengland.co.uk/stress-testing/2022/results-of-the-2021-climate-biennial-exploratory-scenario ↩
-
ACPR, "Main results of the climate exercise for the insurance sector." Participants made relatively limited use of available management actions. https://acpr.banque-france.fr/en/publications-and-statistics/publications/main-results-climate-exercise-insurance-sector ↩
-
APRA, Mind the Gap: An Insurance Climate Vulnerability Assessment, March 24, 2026. https://www.apra.gov.au/news-and-publications/mind-gap-insurance-climate-vulnerability-assessment ↩
-
Swiss Re, Annual Report 2025: Sustainability Report extract, pp. 57–58. https://www.swissre.com/dam/jcr:5b44d56a-8502-401a-bf8e-66178660d563/2025-ar-sustainability-report-extract.pdf ↩
-
Alex Veiga and Gabriela Aoun Angueira, Associated Press, "Fewer than a dozen homes have been rebuilt a year after being burned down in LA-area wildfires," January 7, 2026. https://www.clickorlando.com/business/2026/01/07/a-year-after-la-area-wildfires-destroyed-thousands-of-homes-fewer-than-a-dozen-have-been-rebuilt/ ↩
-
California BCSA and DFPI, "California Orders Financial Institutions to Report Data on Insurance Payouts to LA Fires Survivors…," October 15, 2026. https://www.bcsa.ca.gov/newsroom/2026/10/20261015.html ↩
-
Actuarial Standards Board, ASOP No. 56, Modeling, §§3.1, 3.7, 4.1. https://www.actuarialstandardsboard.org/asops/modeling-3/ ↩
