The COLA against inflation, and the lag nobody mentions
The adjustment is computed from CPI-W, so over decades it tracks CPI-W almost exactly. Year by year it does not, and the reason is timing rather than measurement: the window closes in September and the money arrives the following January.
Years the adjustment ran below contemporaneous CPI-W
24
Out of 50 comparable years. It ran above in 26. The largest shortfall was 6.51% in 2021; the largest surplus 5.38% in 2023.
- Years compared
- 50
- Adjustment below inflation
- 24
- Adjustment above inflation
- 26
- Largest shortfall
- 6.51%2021
What is being compared, precisely
The adjustment for a payable year is set by the third quarter of the previous year. What this page puts beside it is the CPI-W movement from December to December over the year the adjustment was actually being paid — the price change a beneficiary lived through while receiving it.
Those two windows do not overlap. That is the entire finding: the adjustment is not mismeasured, it is displaced. When inflation is accelerating the adjustment is behind; when it is decelerating the adjustment is ahead. Neither is a failure of the computation.
| Payable year | Adjustment | CPI-W that year | Gap | On a $2,000 benefit |
|---|---|---|---|---|
| 2025 | 2.5% | 2.57% | -0.07% | -$1 |
| 2024 | 3.2% | 2.77% | +0.43% | +$9 |
| 2023 | 8.7% | 3.32% | +5.38% | +$108 |
| 2022 | 5.9% | 6.25% | -0.35% | -$7 |
| 2021 | 1.3% | 7.81% | -6.51% | -$130 |
| 2020 | 1.6% | 1.45% | +0.15% | +$3 |
| 2019 | 2.8% | 2.31% | +0.49% | +$10 |
| 2018 | 2.0% | 1.77% | +0.23% | +$5 |
| 2017 | 0.3% | 2.18% | -1.88% | -$38 |
| 2016 | 0.0% | 1.99% | -1.99% | -$40 |
| 2015 | 1.7% | 0.38% | +1.32% | +$26 |
| 2014 | 1.5% | 0.32% | +1.18% | +$24 |
| 2013 | 1.7% | 1.45% | +0.25% | +$5 |
| 2012 | 3.6% | 1.68% | +1.92% | +$38 |
| 2011 | 0.0% | 3.21% | -3.21% | -$64 |
| 2010 | 0.0% | 1.68% | -1.68% | -$34 |
The last column is the monthly dollar effect of the gap on a $2,000 benefit — what the displacement is worth in the year it happens. It reverses in the following year more often than not, which is why the long-run tracking is close even though the year-by-year figures diverge.
The Medicare premium is the larger effect, and it does not reverse
The timing gap above averages out. The Part B premium increase does not: it is deducted every year and it compounds against the benefit. On a $2,000 benefit in 2026 the 2.8% adjustment became an effective net raise well below the headline once the premium was taken out. The net-of-Medicare study computes it at every benefit amount.
The measurement argument, separately
Everything above holds CPI-W as the right yardstick and asks only about timing. Whether CPI-W is the right index for a retired population is a different question: it measures the spending of urban wage earners and clerical workers, who are working-age and whose basket weights medical care and housing differently from a retiree’s. The CPI-E proposals and where they actually stand.
Latest published CPI-W reading: June 2026. The series in full.
Common questions
Does the COLA keep up with inflation?
Over the long run it tracks CPI-W closely by construction — it is computed from CPI-W. Year by year it does not, because of timing: the adjustment measures a third quarter against a third quarter and is paid from the following January. Of the 50 years on this page, the adjustment was below the CPI-W movement in the year it was being paid 24 times and above it 26 times.
Why is there a lag at all?
Because the September index has to be published before the adjustment can be computed, and payments have to be reprogrammed before January. The measurement window closes in September and the money arrives four months later.
Is the lag a flaw?
It is a design consequence, not a bug — and it cuts both ways. In a year when inflation is falling, beneficiaries receive an adjustment set by the higher inflation of the previous window. In a year when inflation is accelerating, they are behind.
Does CPI-W measure what retirees actually buy?
CPI-W measures the spending of urban wage earners and clerical workers — a working-age population. Whether an index built on that basket is the right one for a retired population is a live legislative argument, not a settled fact.
Related
- The CPI-W seriesthe input
- CPI-W vs CPI-Ethe proposals
- Every adjustmentsince 1975
- Highest and lowestthe extremes
- The net raisethe study
- Next yearprojection
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 — 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