In 2016, Social Security paid a zero COLA. Measured on the seniors' own inflation index, the CPI-E, the increase would have been 0.6 percent. That gap is the whole argument for calculating the COLA on CPI-E instead of CPI-W: the index that sets retirees' raises is built from the spending of people who are still working.
The index in use now, the Consumer Price Index for Urban Wage Earners and Clerical Workers, has set the COLA since 1975. It covers about 30 percent of the US population, and retirees are mostly not in it. To qualify for the CPI-W population, a household has to get more than half its income from clerical or wage occupations, and at least one earner has to have worked 37 weeks in the previous year. A household where everyone is retired fails both tests by definition. Most Social Security beneficiaries live in households the index never looks at.
The alternative has been sitting on the shelf since 1982. That year the Bureau of Labor Statistics introduced the Consumer Price Index for the Elderly, which tracks spending by households headed by someone 62 or older. It uses the same categories as the CPI-W but weights them the way seniors actually spend. Health care, which takes a bigger bite of a retiree's budget than a worker's, gets a bigger weight. The result is consistent: the CPI-E runs about two tenths of a point higher per year, according to The Senior Citizens League.
CPI-E vs CPI-W: what each year would have paid
| Year | COLA on CPI-W | COLA on CPI-E |
|---|---|---|
| 2024 | 3.2% | 4.0% |
| 2023 | 8.7% | 8.0% |
| 2022 | 5.9% | 4.8% |
| 2021 | 1.3% | 1.4% |
| 2020 | 1.6% | 1.9% |
| 2019 | 2.8% | 2.6% |
| 2018 | 2.0% | 2.1% |
| 2017 | 0.3% | 1.5% |
| 2016 | 0.0% | 0.6% |
| 2015 | 1.7% | 2.0% |
The CPI-E came out ahead in 7 of those 10 years. It is worth saying plainly where it lost: 2023 and 2022, the two biggest COLA years of the decade. When energy and transportation prices spiked, the workers' index paid more, because working households weight those categories more heavily. CPI-E is not a ratchet. It is just aimed at a different basket, and the basket that wins depends on the year.
Over the full decade, The Senior Citizens League's math says the average senior would have collected $2,689.20 more under CPI-E, and a person who retired in 2024 would end up with over $12,000 more across a 25-year retirement. Their larger claim is about buying power: between 2016 and 2026, Social Security benefits lost about 13.7 percent of their purchasing power, so a dollar of benefits buys roughly 86 cents of what it bought in 2016. Restoring that ground would take a 15.8 percent raise, about $295.85 a month for the average recipient. Those are TSCL's figures, from their 2026 report, and even if you quibble with the methodology, the direction is hard to argue with.
What really fuels the anger, though, is Medicare. In 2026 the standard Part B premium rose from $185 to $202.90 a month, a 9.7 percent jump, against a 2.8 percent COLA. The premium is pulled straight out of the check, so the cost of Medicare rose about three and a half times as fast as the raise. A better index does not fix that on its own. But CPI-E at least measures the medical spending that drives it, instead of the commuting costs of people in their thirties.
Why hasn't Congress switched it?
Switching the COLA to CPI-E takes an act of Congress, and the bills exist. Rep. Ruben Gallego and Sen. Bob Casey have introduced legislation that would use the CPI-E whenever it runs higher than the CPI-W, so nobody's raise ever gets smaller under the new math. The Social Security Expansion Act, introduced in the House in February 2025, carries the same provision. None of them has gotten a floor vote. This is a decade-old debate, not a live bill.
My view: the strongest objection to the switch is not politics, it is the index itself. BLS publishes CPI-E as a research series drawn from a smaller sample than the main CPIs, which makes it noisier. That is a real weakness. But it is a fixable one. A research index becomes a principal index by getting sampled properly. And it is a strange reason to keep measuring retirees' inflation with a gauge that excludes them: roughly 70 percent of the US population, including most beneficiaries, is not in the CPI-W at all. The index that sets their raises is aimed, on average, at somebody else's grocery cart.
Run the actual COLAs since 1976 through the free calculator and you can see exactly how the flat years compounded into today's shortfall.
See every COLA since 1976 in the free dataset
Frequently asked questions
Is the CPI-E an official government index?
Yes. BLS introduced it in 1982 and publishes it monthly. But the bureau treats it as a research series on a smaller sample, not as one of its principal economic indicators, which is part of why switching the COLA to it requires legislation.
Would CPI-E always produce a bigger COLA?
No. It would have paid more in 7 of the last 10 years, per TSCL's figures, but the CPI-W paid more in the two big inflation years, 2022 and 2023. The seniors' index wins the quiet years; the workers' index wins the years when gas and transportation spike.
How much would the switch have been worth?
TSCL estimates $2,689.20 for the average senior over the last decade, and over $12,000 across a 25-year retirement for someone who retired in 2024.
Has Congress ever voted on this?
No. The Gallego and Casey bills and the Social Security Expansion Act (introduced February 2025) would all switch the COLA to CPI-E when it runs higher. None has reached a floor vote.
Does CPI-E fix the Medicare premium problem?
Not directly. It weights medical care more, which tracks seniors' real costs better. But Part B premiums are still deducted from the check, so a premium jump like 2026's 9.7 percent still eats a 2.8 percent COLA. We covered that mechanism in our piece on whether the Part B premium eats the COLA. For the full formula, see how the COLA is calculated, and for the long view, our 10-year and 20-year COLA averages.
Sources: Bureau of Labor Statistics, CPI-E program documentation; The Senior Citizens League, 2026 COLA report (COLA comparison table, buying-power estimates); SSA COLA announcement data; CRS R43363 "Alternative Inflation Measures for the Social Security COLA". Figures verified October 2026.