CPI-W, CPI-E, CPI-U and chained CPI — which one actually sets the COLA
One of these four is the law. One is enacted law for something else entirely and gets confused with it constantly. One is a research series that no statute uses. And one is the number in the news.
The index the statute names
CPI-W
Series CWUR0000SA0, all items, U.S. city average, not seasonally adjusted, 1982-84=100. Nothing else has ever been used.
The four indexes
| Index | Population it measures | Status | Used for the COLA? |
|---|---|---|---|
| CPI-W | Urban wage earners and clerical workers | Published, official | Yes — every adjustment since 1975 |
| CPI-U | All urban consumers | Published, official | No. This is the headline inflation figure in the news |
| CPI-E | Households headed by someone 62 or over | Experimental research series | No. No enacted statute uses it for anything |
| Chained CPI | All urban consumers, substitution-adjusted | Published, official, enacted in law | No — it indexes federal income tax brackets |
The chained-CPI confusion, cleared up
Chained CPI is enacted law, which is why the confusion is so persistent. The 2017 tax act switched federal income tax bracket indexation to it. It has never applied to the Social Security cost-of-living adjustment. A proposal to use it for benefits was debated in the early 2010s and was not enacted. Anyone telling you your benefit is indexed to chained CPI is describing your tax brackets.
Where the CPI-E proposals actually stand
Two bills before the current Congress would compute the adjustment as the higher of CPI-W and CPI-E. Both have been referred to committee. Neither has had a vote. Both contain a fallback provision that uses a research version of CPI-E “until an official CPI-E index is available” — which tells you the central problem with the proposal: the index it names does not exist in publishable form.
Nothing about the adjustment’s index is scheduled to change. Any page telling you the formula is changing next year is describing a bill, not a law.
A more useful correction than the index argument
The adjustment is not “last year’s inflation rate”. It is a quarterly average against a quarterly average, which is a different quantity from the twelve-month change everyone quotes. The two are usually close and sometimes are not — and the gap has nothing to do with which index is used.
| Payable year | Adjustment | CPI-W twelve-month change that September | Difference |
|---|---|---|---|
| 2026 | 2.8% | 2.94% | -0.14% |
| 2025 | 2.5% | 2.25% | +0.25% |
| 2024 | 3.2% | 3.56% | -0.36% |
| 2023 | 8.7% | 8.46% | +0.24% |
| 2022 | 5.9% | 5.94% | -0.04% |
| 2021 | 1.3% | 1.50% | -0.20% |
| 2020 | 1.6% | 1.49% | +0.11% |
| 2019 | 2.8% | 2.34% | +0.46% |
| 2018 | 2.0% | 2.31% | -0.31% |
| 2017 | 0.3% | 1.22% | -0.92% |
The quarterly averaging smooths a single month’s noise out of the answer, which is almost certainly why the statute specifies it. The computation, step by step.
Why this site stays on national all-items CPI-W
Beyond the statute, there is a data-quality reason. BLS affirmatively measured the effect of its 2025 collection reductions on the national all-items series and found it negligible; it explicitly states it did not measure the effect on subnational or item-level indexes. Staying on national all-items keeps every figure here inside the one zone BLS has vouched for. The collection changes in detail.
Latest published CPI-W reading: June 2026 at 327.075. The 2026 adjustment came out of the same series.
Common questions
Is the COLA switching to CPI-E?
No. CPI-E is an experimental BLS research series, not an official index, and no enacted statute uses it for anything. Two bills currently before Congress would take the higher of CPI-W and CPI-E; both have been referred to committee, neither has been voted on, and both contain a fallback because an official CPI-E does not exist to point at.
Is the COLA already using chained CPI?
No, and this is the most common conflation in the whole subject. Chained CPI is enacted law — it indexes federal income tax brackets under the 2017 tax act. It has never been used for the Social Security cost-of-living adjustment.
What is the difference between CPI-W and CPI-U?
CPI-U covers all urban consumers, about 93% of the population, and is the headline inflation number in the news. CPI-W is a subset — urban wage earners and clerical workers, roughly 29% — and it is the one the statute names for the adjustment. The two move closely but not identically, because the populations spend differently.
Would CPI-E have produced bigger adjustments?
This site does not publish an answer, because it does not hold a CPI-E series. Producing one from secondary summaries would be exactly the kind of derived-from-nothing figure this site exists to avoid. What can be said is that BLS classifies CPI-E as experimental and does not maintain it to the standard of the published indexes.
Related
- The CPI-W seriesthe input
- Data qualitywhat changed in 2025
- Against inflationthe timing lag
- The computationfour steps
- Methodologythree regimes
- Every adjustmentsince 1975
Sources and freshness
- U.S. Bureau of Labor Statistics — CPI-W (CWUR0000SA0), all items, U.S. city average, not seasonally adjusted — verified 10 August 2026
- SSA Office of the Chief Actuary — CPI-W table — verified 10 August 2026
- SSA Office of the Chief Actuary — Cost-of-Living Adjustments — verified 10 August 2026
Page figures last verified against the sources above on 10 August 2026. BLS data retrieved 10 August 2026. BLS.gov cannot vouch for data or analyses derived from these data after retrieval. Corrections: the correction log · support@inventum.com.au