Social Security Purchasing Power: Have Benefits Kept Up?
The official yardstick: COLA vs. CPI-U by decade
| Decade | Average annual CPI-U change | Average COLA (determination years) | Difference |
|---|---|---|---|
| 1980s | 5.55% | 5.72% | +0.17 pts |
| 1990s | 3.00% | 2.89% | -0.11 pts |
| 2000s | 2.57% | 2.78% | +0.21 pts |
| 2010s | 1.77% | 1.52% | -0.25 pts |
On the government's own numbers, the COLA formula has done roughly what it promises over long horizons: adjustments land within a few tenths of a point of all-urban inflation in most decades, sometimes ahead, sometimes behind. The year-by-year comparison — including the important caveat that the COLA is actually set from third-quarter CPI-W, not the CPI-U annual average — is worked through on the COLA vs. inflation page. And the compounding check is inescapable: a benefit that began at $1,000 in 2000 is $1,988 in January 2027 on COLAs alone (cumulative calculator) — whether that feels like "keeping up" depends entirely on what the household buys.
Why the official answer can still feel wrong
The basket problem. The CPI-W measures the spending of urban wage earners — a working population. Retiree households spend a materially larger share on medical care and shelter, categories whose prices have frequently outrun the overall index, and a smaller share on transportation, education and apparel. A beneficiary whose personal inflation runs hotter than the CPI-W loses buying power every year even though the formula is working exactly as written. The Bureau of Labor Statistics' experimental retiree index, the CPI-E, exists precisely to measure that difference (about 0.2 points a year faster, long-run).
The fixed-cost problem. Deductions and thresholds around the benefit do not all move with prices. The Medicare Part B premium has risen much faster than benefits over long spans (from $3.00 in 1966 to $202.90 in 2026 — premium history), the benefit-tax thresholds have been frozen since 1983/1993, and the SSI resource limits since 1989. Each is a small, steady leak in real take-home value that no COLA table shows.
The advocacy estimate: TSCL's Loss of Buying Power study
Everything in this section is The Senior Citizens League's own estimate, not an SSA or BLS figure. TSCL, a nonpartisan seniors advocacy organization, publishes an annual study that reprices a basket of goods and services weighted toward what retirees actually buy — food, housing and healthcare costs carry more weight than in the official indexes — and compares the result with the COLAs paid over the same period. Its 2026 edition concluded that Social Security benefits lost 13.7% of their buying power between 2016 and 2026, and that the average benefit would need to rise by roughly $296 a month to restore it.
The dispute. The result is highly sensitive to choices a reader should see: the basket (retiree-heavy categories have inflated faster), and the start year (2016 follows the 0.0% determinations of 2009, 2010 and 2015 and precedes the large 2022–2023 adjustments' full effect on the comparison window). Analysts who defend the official indexes note that the CPI-W comparison over the same span shows benefits roughly keeping pace, and that no official statistical agency endorses the 13.7% figure. Both statements can be true at once: they are answers about different baskets. This site reports TSCL's number only in this attributed form and never blends it into official tables.
What gets proposed about it
The recurring proposals in Congress map directly onto the two yardsticks: switch the COLA index to the CPI-E (raises adjustments by roughly the 0.2-point annual gap, at a real long-range cost to the trust funds), add a supplemental increase for long-time beneficiaries (an "age bump" at 82 or 85, aimed at the years when savings run out), or leave the formula and address the leaks (indexing the tax thresholds, or restraining premium growth). None has been enacted; the current-law position is set out on the CPI-W vs. CPI-E page.
Frequently asked questions
Have Social Security benefits lost buying power?
It depends on the yardstick. Measured against the official price indexes, COLAs have roughly kept pace on average across decades — the decade table on this page shows COLAs close to CPI-U inflation in most eras. Measured against a retiree-weighted basket of goods, The Senior Citizens League estimates benefits lost 13.7% of their buying power between 2016 and 2026 alone. The two answers differ because the baskets differ.
Why do retirees feel inflation more than the COLA covers?
The COLA is indexed to the CPI-W — the spending pattern of working households. Retirees spend proportionally more on healthcare and housing, whose prices have often risen faster than the overall index, and less on categories like education and apparel. A household whose personal basket inflates faster than the CPI-W loses ground even when the official math is exact.
What is the CPI-E and would it fix this?
The CPI-E is the Bureau of Labor Statistics' experimental index for households aged 62 and over. It weights medical care and shelter more heavily and has risen about 0.2 points a year faster than the CPI-W over the long run. Bills to base the COLA on the CPI-E are introduced regularly; none has been enacted, and the change would add to the program's long-range cost.
What does TSCL's $296 figure mean?
The Senior Citizens League's 2026 study concluded that the average benefit would need to rise by about $296 a month to restore the buying power it estimates was lost since 2016. It is an advocacy organization's estimate built on its own basket of retiree expenses — presented here with its methodology, not as an official statistic.
Sources: BLS CPI-U annual data and SSA COLA series (data/cpi_comparison.csv, data/cola_series.csv) for the official sections; The Senior Citizens League, "Social Security Loss of Buying Power" study (2026 edition) for the attributed estimate in the marked section only.