Sometime in mid-October the Social Security Administration will announce the cost-of-living adjustment for 2027, and the current projection from the advocacy group that tracks it most closely sits around 3.6 percent, revised down from 3.8 percent earlier in the summer. The number will be reported as a raise, and in a narrow sense it is one. What almost never gets explained alongside it is that the figure is already fixed by the time it is announced, that it was calculated from three specific months, and that it measures the spending of households that look very little like the people receiving it.
The calculation itself is mechanical. The adjustment is the change in the average Consumer Price Index for Urban Wage Earners and Clerical Workers, known as the CPI-W, across July, August, and September, compared with the same three months a year earlier. That is the whole formula. Whatever happened to prices in the other nine months does not enter it, and whatever happens between the September reading and the January payment does not either. A summer of soft prices produces a small raise for the following year even if costs accelerate in the fall, and the reverse holds too.
Why the Index Is Named After Wage Earners
The name is not decorative. The CPI-W is built from the spending patterns of households where more than half of income comes from clerical or hourly wage employment, meaning working households, and households consisting only of retired people are outside its scope entirely. That distinction changes the weights. Working households spend proportionally more on transportation to and from a job and on the goods that go with raising a family. Older households spend proportionally more on medical care and on housing, which are also the two categories that have been rising fastest for most of the past decade. The result is a benefit indexed to a basket that systematically underweights the two largest and fastest-growing line items in the budget it is meant to protect.
The Mechanics of a Measure Nobody Chose to Get Wrong
The government is aware of this. The Bureau of Labor Statistics has published an experimental index for Americans aged 62 and older for decades, reweighted toward how older households actually spend, and in most years it runs slightly higher than the CPI-W. It has never been adopted for the cost-of-living adjustment, and the reason is not a conspiracy but ordinary institutional inertia: the CPI-W was written into the law in the 1970s, the experimental index is smaller and less statistically robust, changing the formula raises long-run program costs, and no year is ever a convenient one to make that change. So each year the adjustment is calculated correctly under a rule that was set before most of today's beneficiaries retired, and the small annual gap between the two indexes compounds quietly across a twenty-five year retirement into something that is no longer small.
The Deduction That Happens Before the Check
Then there is the second subtraction, which is more visible but still routinely misread. The standard Medicare Part B premium is deducted directly from the Social Security payment, so the number that matters is not the announced percentage but what lands in the account afterward. Last year illustrates the arithmetic plainly. The adjustment came in at 2.8 percent while the standard Part B premium rose from 185 dollars to roughly 203, an increase near 10 percent, which absorbed more than a quarter of the raise before it ever arrived and pushed the premium to about 9.4 percent of the average retired worker's benefit, the highest share on record. The projected premium for 2027 is higher again. There is a protection here, the hold harmless provision, but it is narrower than most people assume: it prevents the dollar amount of a check from falling because of a Part B increase, and it does not apply to everyone, including those paying income-related surcharges and some new enrollees.
What the Announcement Will Not Say
None of this is an argument that the adjustment is worthless. An indexed benefit that rises every year is a genuinely rare and valuable thing, and most private retirement income has no such feature at all. But the announced percentage describes the change in a price index for working households, not the change in what retirement actually costs, and the difference between those two things is deducted quietly and never appears as a line anywhere. The practical move when the number comes out in October is to skip the headline and work out the net: the announced percentage applied to the current benefit, minus whatever the new Part B premium turns out to be, which is the only figure that describes the actual deposit.
— John Stone