Social Security's cost-of-living adjustment is not broken. It is doing exactly what it was designed to do

SOCIAL SECURITY benefits have lost 13.7% of their buying power since 2016, according to a study published in May by the Senior Citizens League, a lobby group for retirees. The average check today, about $2,083 a month, can purchase roughly 86 cents worth of what it could a decade ago. Advocates want Congress to do something about it. Politicians want to be seen doing something. The 2027 cost-of-living adjustment, forecast at between 3.6% and 3.9%, is supposed to help.

The trouble is that the problem is not a glitch in the adjustment mechanism. It is a feature of the system's design. Social Security's COLA is not broken. It is doing exactly what it was designed to do: track the inflation experienced by people who are still employed.

The price index in question

The COLA is calculated using the Consumer Price Index for Urban Wage Earners and Clerical Workers, or CPI-W, first adopted in 1972 when automatic annual adjustments were introduced. CPI-W covers roughly 30% of the population — urban workers and their households. The index was chosen not because it describes retirees but because it was the only one available. By 1978 the Bureau of Labor Statistics had published the broader CPI-U, which covers 93% of urban residents and tracks closely with CPI-W, but Congress never bothered to update the formula.

The result is a mismatch that grows worse with every ageing cohort. Retirees do not buy the same basket of goods as urban wage earners. They spend more on housing and medical care, both of which tend to inflate faster than the economy-wide average. They spend less on transportation, food and clothing. The CPI-W does not weight these differences.

A better index, a small improvement

In 1987 Congress asked the Bureau of Labor Statistics to build one. The result is the Research CPI for Americans 62 and Older, commonly called the CPI-E. It uses the same outlets and prices as the CPI-U but re-weights spending categories to reflect an older population. Housing is 48% of the CPI-E basket versus 42% in CPI-W. Medical care is 11% versus 7%. Transportation drops from 19% to 14%.

The CPI-E has generally run ahead of the CPI-W. From 1985 to 2024 it rose by 211% compared to the CPI-W's 188%. In the most recent comparison, the hypothetical CPI-E COLA for 2025 would have been 3.0%, half a percentage point above the actual 2.5%. That would have put an extra $9 a month into the average retiree's pocket.

To be sure, the gap sounds trivial. Over a 25-year retirement the compound difference can exceed $12,000 for the average person, according to the Senior Citizens League's own projections. Switching to the CPI-E would also increase Social Security's 75-year deficit by roughly 12%, says the Congressional Research Service, an arm of the Library of Congress. The index is also experimental — the Bureau of Labor Statistics itself cautions that it may not accurately reflect elderly spending patterns — and no current legislation to adopt it has gained serious traction.

The real measurement problem

Even the CPI-E does not capture the full story. The 13.7% buying-power decline cited by the Senior Citizens League comes from a proprietary index tracking 70 products and services, sourced from the most affordable prices available for each item. Its 10-year inflation estimate of 43.6% exceeds the CPI-W's official figure of 37.6% by nearly six points. The gap is not merely statistical: 39% of seniors rely on Social Security for all their income, and fixed incomes mean less room to substitute away from expensive items when prices rise.

The deeper erosion comes from Medicare, which sits outside the COLA mechanism entirely. The standard Part B premium — covering doctor visits and outpatient care — rose to $202.90 a month in 2026, a jump of nearly 10%. A further increase to approximately $209.50 is projected for 2027. Premiums are deducted directly from Social Security checks. For a typical retiree receiving about $2,083, a 3.8% COLA yields roughly $79 more per month. The Part B premium increase eats up $6.60 of that. The rest of the healthcare cost problem — dental, vision, hearing care that Medicare does not cover — is not indexed to anything at all. A survey by the Senior Citizens League in 2025 found that 58% of seniors had skipped such care to save money.

This is where the system begins to creak. The COLA adjusts benefits for the goods and services a typical wage-earner buys. It does not adjust for the healthcare costs that define a retiree's budget. The two mechanisms are legally and institutionally separate. Merging them would mean Social Security's growth rate became partly dependent on the actuarial projections of the Centers for Medicare & Medicaid Services, a different bureaucracy with its own budget politics. Nobody has tried.

The solvency wall

Even if a politically coherent fix could be designed, there is another constraint: money. The 2026 Social Security Trustees Report, published in June, projects that the Old-Age and Survivors Insurance trust fund will be exhausted in the fourth quarter of 2032, one quarter earlier than last year's forecast. At that point continuing payroll taxes will cover only 78% of scheduled benefits. The combined disability fund extends the depletion date to 2034.

A one-time restoration of lost purchasing power would cost roughly $296 a month per average beneficiary, or $157bn annually in new spending. Permanent adoption of the CPI-E would add less per year but widen the 75-year deficit. A CPI-BEST proposal favoured by the Senior Citizens League — which would guarantee a minimum COLA of 3% and use whichever index, CPI-W or CPI-E, is higher — would raise the average annual adjustment from 2.8% to about 4.0% over the past decade. None of these options is fiscally costless. All of them arrive in a year when the programme is sliding, not gaining, ground.

What a 2027 COLA actually does

The 2027 COLA will be announced in October, based on CPI-W data from the third quarter of this year. Forecasts range from 3.6% to 3.9%, up from 2.8% in 2026. If it comes in at the mid-point of 3.7%, the average check rises by about $77 a month. After Medicare's premium adjustment, the net gain is closer to $70.

That is a genuine increase. It is not a correction.

The 13.7% hole in buying power was not dug in a single year. It accumulated through a decade in which the CPI-W consistently lagged behind the inflation seniors experienced, punctuated by a pandemic-era spike in 2022 and 2023 when even the CPI-W-based COLA surged but healthcare costs did something similar. No single adjustment can reverse it. The 2027 COLA will be larger than last year's, but it will not be the turning point that advocates hope for.

The structural choice

The real question is not whether the 2027 COLA will be big enough. It is whether the government should continue pretending that a wage-earner's price basket adequately describes a retiree's life. The CPI-W is an anachronism, not a conspiracy. Nobody switched to it to hurt seniors. It simply stopped being updated when a better tool was available.

Switching to the CPI-E is the least painful reform. It would cost the system roughly $5,000 to $12,000 in additional lifetime benefits per person and widen the 75-year deficit by 12%. The trade-off is modest, but it is a trade-off. In a programme whose trust fund is projected to run dry within eight years, even a modest increase in spending requires a decision about what to cut elsewhere: higher taxes on earnings above the current $184,500 cap, slower benefit growth for high earners, or a higher retirement age. None of these is politically popular.

To be sure, there is an argument that the CPI-W's flaws are offset by the index's failure to account for consumer substitution — the tendency of shoppers to switch to cheaper alternatives when prices rise. Economists at the Center for Retirement Research at Boston College have noted that switching to a chained CPI, which does account for substitution, would lower COLAs by about 0.3 percentage points annually and reduce the 75-year deficit by 17%. The net effect of the current CPI-W, viewed as sitting between the CPI-E and the chained CPI, is "just about right," they argue. That view has the virtue of intellectual honesty but the defect of ignoring that substitution is harder for a retired person with a fixed income and chronic medical conditions than for a mid-career professional.

The better answer is to index benefits to the reality of the people receiving them, then deal with the fiscal gap honestly. The CPI-E is not perfect. It is better. And eight years of political avoidance will not make the arithmetic go away.

Consumers pay first. In this case, it is a generation of retirees who are paying for a formula nobody has the courage to update.