Source: BLS, "Consumer Price Index – August 2026" (USDL-26-1496, Sept 11, 2026).
Canonical URL: https://www.bls.gov/cpi/news.htm (I could not fetch it directly — BLS returned an automated-access "Access Denied" page — so I read the identical release text via the FRASER mirror: https://fraser.stlouisfed.org/title/consumer-price-index-6838/august-2026-735977).
Locator: opening paragraphs of the release body, plus the single NOTE line below them.
Quoted: "The Consumer Price Index for All Urban Consumers (CPI-U) increased 0.4 percent on a seasonally adjusted basis in August... Over the last 12 months, the all items index increased 3.4 percent before seasonal adjustment." And: "The index for all items less food and energy rose 0.3 percent... The energy index increased 16.3 percent for the 12 months ending August. The food index increased 2.7 percent over the last year."
Two details that change how the headline number should be read:
The two headline percentages are built on different bases. The monthly figure is explicitly "on a seasonally adjusted basis"; the 12-month figure is explicitly "before seasonal adjustment." So "0.4 percent this month" and "3.4 percent over the year" are not the same statistic scaled to a different window — one removes routine seasonal swings, the other compares August directly to August without that adjustment. Quoting both as a single trend line quietly mixes conventions.
The one-line NOTE: "The Oct and Nov 2025 data values are not available due to the 2025 lapse in appropriations." This is the consequential omission. Because two months of observations are missing, the 12-month change "ending September" (2026 vs. 2025) and the seasonal factors that depend on those months are affected in a way the headline does not disclose. I am flagging this as a caution, not asserting a specific direction of bias — the release itself does not quantify the effect, and I have not recomputed anything.
My interpretation, kept separate from the author's claim: the release is a clean, conventional CPI report; its own structure (SA vs. NSA labels, the gap note) is what tells you the 3.4% is a year-over-year NSA figure driven substantially by energy (+16.3% y/y), with core at 2.4%. Reading 3.4% as "general price growth" without that decomposition overstates the persistence of inflation.
Limit: this is one release. The energy-driven pattern is consistent with the gasoline passage but should be checked against the detail tables (Table A onward), which I could not fully render.