Social Security COLA Reduction Proposals: What a Flat-Rate Raise Would Really Cost Retirees

Retirees got a 2.8% cost-of-living adjustment for 2026. For the average beneficiary that works out to roughly $57.90 more per month. It is not a large number, and plenty of people receiving it have said out loud that it does not match what they are actually paying at the grocery store.

Which makes the current policy conversation an odd one. Because the proposals getting the most attention in Washington right now are not about making that raise bigger. They are about making it smaller.

Two ideas are on the table, and they work in completely different ways. One changes the formula quietly. The other rewrites it outright. Here is what each would actually do to a monthly check.

First, how the raise is set today

Social Security’s annual increase is tied to a price index called CPI-W, which tracks what urban wage earners and clerical workers spend money on. Every year the Social Security Administration compares third-quarter prices against the same quarter a year earlier, and whatever the percentage difference is becomes the COLA.

Notice the mismatch built into that. CPI-W measures the spending habits of working people. The benefit it adjusts goes overwhelmingly to people who have stopped working. Retirees spend a much larger share of their income on medical care and housing, and a much smaller share on things like transport to a job. That gap is the reason this argument never really goes away.

Proposal one: give everyone the same dollar amount

The flat-rate COLA is the more aggressive of the two. Instead of applying a percentage to each person’s benefit, it would calculate one dollar figure and hand that same figure to every beneficiary. The amount would be pegged to the raise a fairly low earner would have received, around the 20th percentile of benefits.

The Committee for a Responsible Federal Budget pushed the idea into the mainstream, and the Washington Post editorial board endorsed it in July 2026. The pitch is that it is progressive: low earners are held roughly steady while higher benefits grow more slowly, and the programme saves money.

The arithmetic tells a harsher story. In 2026 terms, the average monthly increase would have been about $34.20 instead of $57.90. That is $285 less over a single year.

Why a small annual gap becomes a large one

The problem with any COLA change is that it compounds. A percentage raise is applied to a benefit that already includes every previous raise. A flat dollar raise is not. Skip that compounding for twenty or thirty years and the two lines drift far apart.

Take a worker who retired at 65 in 1998. Under the current formula, by age 93 that person would be collecting about $22,600 a year. Under a flat-rate COLA, the same person would be collecting about $18,000. That is $4,600 gone in a single year, a cut of roughly 20%.

Across the whole retirement the shortfall adds up to around $77,900 in inflation-adjusted terms. Purchasing power falls about 28% between 65 and 93, meaning a check would buy less than three-quarters of what it did at the start.

Where it pushes people over a line

The uncomfortable part is what happens near the poverty threshold. One example in the analysis follows a retired woman whose benefit sits at $15,960, about 13% above the poverty line. Under the flat-rate formula it lands at $15,800, which puts her below it.

A disabled man in the same modelling goes from $17,000, comfortably above the line, to $15,200, roughly 5% below. For a proposal marketed as protecting low earners, those two outcomes are hard to explain away.

Proposal two: chained CPI, the quiet version

Chained CPI has been circulating for well over a decade and it is far less dramatic on paper. It uses a price index that accounts for substitution, the assumption being that when beef gets expensive people buy chicken instead, so the real cost of living rises a little slower than a standard index suggests.

In practice it grows roughly 0.25 percentage points more slowly per year. That sounds like nothing. Over a full retirement it works out to an average benefit reduction of about 2%, according to analysis from the Center on Budget and Policy Priorities.

The cut is not spread evenly. It deepens the longer you live, reaching about 3.2% by age 75. Some versions of the proposal include a bump-up at 85 to soften the effect for the oldest beneficiaries, which is a tacit admission that the formula bites hardest exactly where people can least absorb it.

Critics have called chained CPI a benefit cut dressed up as a technical correction. Supporters say it is simply a more accurate measure. Both descriptions can be true at once, and that is precisely why the fight over it is so durable.

The proposal pointing the other way

Not every idea on the table shrinks the raise. CPI-E, an experimental index the Bureau of Labor Statistics maintains for Americans aged 62 and over, weights medical care and housing more heavily. Because those costs tend to climb faster, CPI-E would usually produce a larger COLA than CPI-W.

Senator Bernie Sanders and a number of retiree advocacy groups have pushed for it for years. It has never made it through Congress, for the obvious reason: it costs money rather than saving it, and the programme’s long-term funding gap is the pressure driving this entire debate.

Why all of this is surfacing now

Social Security faces a long-run shortfall between what it takes in and what it has promised to pay out. Closing that gap means raising revenue, reducing benefits, or some combination. Raising revenue is politically expensive. Cutting benefits outright is worse.

Changing the inflation formula is the third path, and it is attractive to policymakers for one specific reason. Nobody’s check gets smaller on the day it passes. The reduction shows up years later as a raise that was never quite big enough, and by then it is nearly impossible to point at a single decision and blame it.

What this means if you collect a benefit

Nothing has passed. Both proposals are ideas being argued over, not law, and no COLA change applies retroactively to money already received.

• If you are already retired, a flat-rate COLA would matter more the longer you live. The first year barely registers. Year twenty is where it lands.
• If you are still working, treat any formula change as a reason to assume Social Security replaces slightly less of your income than the current projection suggests.
• If your benefit sits near the poverty threshold, you are in the group these proposals affect most, despite being the group they are marketed as protecting.
• Watch what happens to CPI-E. If it gains traction, the debate has shifted direction entirely.

Frequently asked questions

Is the Social Security COLA being cut in 2026?

No. The 2026 COLA is 2.8%, worth about $57.90 a month for the average beneficiary. The reduction proposals being discussed would apply to future adjustments and none of them has been enacted.

What is a flat-rate COLA?

It would replace the percentage increase with a single dollar amount paid to every beneficiary, set near the raise a 20th-percentile earner would have received. In 2026 that would have been roughly $34.20 a month instead of $57.90.

How much would chained CPI reduce benefits?

Roughly 0.25 percentage points a year, which compounds into an average reduction of about 2% across a retirement and around 3.2% by age 75.

What is CPI-E and would it raise benefits?

CPI-E is an experimental index tracking spending by Americans 62 and older. It weights medical care and housing more heavily, so it would generally produce a larger COLA than the current CPI-W formula.

Why is CPI-W used when most recipients are retired?

It is the index written into law when the automatic adjustment was created, and changing it requires an act of Congress. The mismatch between what CPI-W measures and how retirees actually spend is the core of the ongoing argument.

None of these formulas is neutral. Each one is a decision about who absorbs inflation, written in a language technical enough that most people will never see the bill arrive.