The Federal Reserve Bank of San Francisco published Economic Letter 2026-31 on November 2, by Molly Ardren, Tobin Graf and Brigitte Roth Tran. It estimates that the current El Niño may already have raised the Personal Consumption Expenditures (PCE) price level by about 0.3 percentage points between April and July 2026, and expects that bump to reverse by year-end. The authors then flag a “meaningful risk” of higher inflation in 2027: a net increase of up to 1 percentage point, but only if a La Niña follows. They call it a risk, not a forecast.
This piece does not argue the forecast is right. It asks what a self-insured should do with a conditional inflation risk when the same shock moves two reserve assumptions in opposite directions.
What the letter does and does not say
Three details matter for how much weight to give it. The price effects historically came mainly from core services and food, not from the medical or auto repair categories that drive claim cost. The sample holds only 22 El Niño events, four of them very strong, and the current event is projected to exceed all of them, so the authors scale past effects linearly and acknowledge that may not hold. And the letter finds payroll employment unaffected in the first year, then rising in the second.
The measure is PCE, not the Consumer Price Index (CPI) that the Bureau of Labor Statistics publishes. Neither is a claims index. Medical care services and motor vehicle repair run on their own trends, as our coverage of the headline versus medical CPI split and motor vehicle repair severity showed.
Who it affects
Long-tail programs with the most payments after the valuation date: self-insured workers’ compensation employers with lifetime medical and cost-of-living-adjusted indemnity, hospital and public-entity captives, and excess casualty books paid out over a decade or more. Short-tail lines barely register.
The reserve mechanism: trend and discount rate move together
Severity trend raises the nominal cost of every future payment. The discount rate, if you discount, lowers the present value of those payments, and the Federal Open Market Committee (FOMC) is the policy lever that follows inflation. The offset is real but partial. The numbers below are my own illustrative arithmetic, not figures from the letter. They assume $100 of unpaid claims paid in equal annual amounts over 15 years, with payments at year-end. The 4.0% starting point and the $74.12 base present value follow the September reserve-discount piece.
- Base: 4.0% discount rate, no added trend. Present value $74.12.
- Trend shock only: payments inflate one extra point a year, rate unchanged. Present value about $79.8, up 7.6%. An undiscounted reserve rises about 8.4%.
- Trend shock plus a 50 basis point rate rise to 4.5%: present value about $77.0, up 3.9%. Half the shock remains.
- Trend shock plus a 100 basis point rise to 5.0%: about $74.3, essentially flat.
The rule of thumb: a discounted reserve responds to the gap between the rate and the trend. A one-point trend shock is worth about one point of discount rate. The FOMC held at 3.75% to 4% on October 28 with two dissents for a hike, so a full 100 basis point offset is not on the table, and any increase would follow the inflation rather than precede it.
Three reasons the offset fails in practice. Undiscounted reserves, including many statutory and tax-basis programs, take the whole trend hit and none of the rate benefit. Lifetime medical and indemnity tied to benefit indexing inflate from the claim’s own schedule, whatever the Fed does. And many discount rates are set at a valuation date, so a December hike reaches your books only at the next reset (our discounting explainer covers the bases).
One offsetting item: if the letter’s payroll finding holds, workers’ compensation exposure bases would rise a year later, which lifts premium and the expected loss for the new accident year, not the unpaid estimate on old years.
Where this shows up in your reserves
Open the actuarial report’s trend selection exhibit (the severity trend used to project the expected claim ratio for the next accident year) and the discount-rate exhibit. Find the unpaid estimate for 2027 payments. Check whether the report states a sensitivity, and whether trend and rate are shown together.
What this means for your next review
Ask your actuary for three columns: base, base plus one point of trend, and one point of trend with the rate you consider plausible by year-end. Treat the letter as a risk flag. The five leading indicators of adverse development are a better trigger: change the trend when paid medical per claim moves, not when a forecast says it might.
Our read
Do not book the letter. Do book the sensitivity. If a La Niña does not follow, the letter’s own estimate puts the 2027 effect near zero and the trend assumption should not move. If it does, claim-cost inflation would reach medical and repair components with a lag, so the November and December CPI medical care and motor vehicle repair readings are the first real evidence. Programs that discount have the cushion of a rate response; programs that do not should assume the full trend effect on 2027 payments.