The index that sets a 71-million-person raise runs on a sample from 1981

Every month, a few thousand government interviewers phone or visit tens of thousands of households, and a smaller crew of price collectors call or walk into stores, to ask two questions: is anyone working, and what does this cost.▸ Those answers become the unemployment rate and the Consumer Price Index (CPI) — two numbers that move interest rates, wages, and, every January, the size of the check that goes to nearly 71 million Social Security beneficiaries. In October 2025, the Social Security Administration announced a 2.8% cost-of-living adjustment for 2026, based on the CPI for Urban Wage Earners and Clerical Workers — raising the average retirement benefit by about $56 a month. The survey infrastructure that number rides on is smaller, and in places thinner, than it was a year ago.

The Bureau of Labor Statistics is a small agency with an outsized signal.▸ Congress funds it in five activity lines; two of them — labor force statistics and prices — are the ones that produce the headline numbers.

Of the enacted $708.5 million, per COSSA's analysis of the final appropriations agreement, $320.0 million funds Labor Force Statistics — the Current Population Survey (CPS) that produces the unemployment rate, and the Current Employment Statistics survey behind the monthly jobs report.▸ $248.0 million funds Prices and Cost of Living — the CPI and Producer Price Index. The administration's budget request had proposed cutting BLS by 8.6%, to $648.0 million; Congress rejected that cut and enacted $708.5 million instead, per COSSA's reading of the final conference agreement. The U.S. Senate Committee on Appropriations confirms the full-year Labor-HHS-Education bill carrying that funding passed the House 217–214 and the Senate 71–29, and was sent to the President on February 3, 2026.

Money restored in February 2026 does not undo what happened the year before.▸ Starting in April 2025, BLS suspended CPI data collection entirely in Lincoln, Nebraska and Provo, Utah; in June 2025 it suspended collection entirely in Buffalo, New York. In the other 72 areas still priced, roughly 15% of the sample was also suspended, on average, affecting both the Commodities and Services survey and the Housing survey that measures rent. BLS's own explanation is unadorned: "BLS makes reductions when current resources can no longer support the collection effort." The agency ran a simulation back to 2018 and found the effect on the published national CPI was small — differences of less than 1/100th of a percentage point on average — but it also found the suspensions increase the volatility of the local and item-level indexes many state and local programs rely on. The number of imputed (estimated rather than collected) prices rose the month the suspensions began. The staffing math behind that decision is not hidden either: BLS's own FY2026 budget justification shows total agency staffing already down from 2,062 full-time positions in FY2024 to 2,019 in FY2025, before the administration's now-rejected request to cut further, to 1,851.

The Current Population Survey has a separate, older strain that no single year's appropriations bill touches.▸ Every month it interviews about 60,000 households — and per BLS's own April 2026 report to the House and Senate Appropriations Committees, that sample size "is the same as it was in 1981, even though the population used for the CPS has increased by 61%." The same report states that response rates — historically in the low-90% range — have declined "to the upper 60 percent range over the last 15 years." The combined effect is that each respondent's answer now has to represent far more people than it used to.

BLS states the consequence in its own numbers, not a critic's: at the current sample size, "a net change of fewer than 50 survey responses could move the headline unemployment rate by 0.1 percentage point."▸ Twenty years ago, the CPS could detect a month-over-month move in the unemployment rate as small as 0.18 points as statistically meaningful in a single month; today it takes two months of data to reach the same confidence. BLS and the Census Bureau are jointly developing an internet self-response option to control collection costs without shrinking the sample further, but the report describes that work as still in limited field testing, with a larger 45,000-household parallel test survey proposed, not yet running in production.

Budget figures reflect the enacted FY2026 Labor-HHS-Education appropriations act as reported by COSSA; the underlying bill text was not independently retrieved. CPI collection-suspension figures and CPS sample and response-rate figures are BLS's own published statements, current as of the cited notices and report. The Social Security COLA figures are SSA's own October 2025 announcement and apply to the average retired-worker benefit, not every beneficiary category.