You may have seen some recent stories about inflation “moderating.” That’s BS. The Fed’s preferred inflation measure has increased back to 3% over the last year, up from about 2.5% when Trump’s liberation day tariffs were imposed last April.
Paul Krugman thinks the tariffs imposed so far have added about 1% to price levels over time. That’s either price increases we’ve seen so far, or increases that are in the pipeline, regardless of the Supreme Court decision striking down the most egregious “emergency” tariffs. So absent the tariffs, inflation would probably be getting down toward the Fed’s 2% target!
So what about those recent news stories about inflation moderating? They are based on the Consumer Price Index (CPI), which is less comprehensive than the Fed’s preferred price index, and which has an important glitch in one of it’s largest components, the CPI for Shelter.
So why was there a glitch in the CPI-Shelter? The CPI for Shelter includes so-called owners’ equivalent rent (a proxy for the cost of home ownership) and rental housing.
Why is the CPI-Shelter Messed Up (caution wonky)?
Last October there was a brief government shutdown. The data for the September CPI was incomplete, but the Bureau of Labor Statistics did report a September CPI result — they had to report something, since the September CPI is required for the Social Security COLA!
However, the September CPI, based on incomplete data, had a very low outlier result for CPI-Shelter, including owners’ equivalent rent and rental housing. That reduced the 2025 Social Security COLA fractionally, by the way!
Then, there was a longer shutdown later in the Fall, and there was no new data for many components of the CPI for a couple months. So what did BLS do? Instead of interpolating the missing data based on average results, they extended the low September outlier! And there will be no catch up — that arbitrary extension of the low September outlier number is baked in to the official CPI forever. (See that flat section circled in the following graph? That’s the fudged data.)
Here’s a longer story from Wolf Street that explains the details:
PCE Inflation (2.9%, 3.0%) and GDP Inflation (3.7%) Got Hotter, but CPI Inflation (2.4%) Cooled. Why? The Fed Needs to Pay Serious Attention (by Wolf Richter, Feb 21 2026)
How Does the CPI Mess-up Affect Your COLA and Mine?
So there are two problems with the CPI mess up. First, as Wolf Street highlights, some people might get the wrong impression that inflation is moderating when it’s actually worsening!
Second, the CPI-Shelter mess-up permanently reduced COLAs and other index measures that depend on the CPI.
For example, I calculate that my Social Security COLA will be reduced annually by $51.32 by the CPI fudge. Assuming I live another 30 years (wishful thinking maybe!) that’s $1,500 in lower benefits. That’s a new bike!
Here’s the math: Take your monthly Social Security benefit (mine is about $2700). Then multiply by 12 for your annual benefit. Then multiply by .158% (that’s the impact magnitude of the CPI-Shelter fudge on the Social Security COLA by my calculation, using data from this excellent Wolf Street article and details in the comments). Presto, if you’re a Social Security beneficiary, that’s how much the CPI-Shelter fudge will cost you for each year of the rest of your life! Thanks Trump!