My frustration is boiling over. This is very serious. A lot depends on halfway accurate CPI inflation readings.
By Wolf Richter for WOLF STREET.
Today’s CPI report for December by the Bureau of Labor Statistics still didn’t fix the massive issues with the CPI for Owners Equivalent of Rent (OER), the biggest component of CPI, weighing 26.4% of the all-items CPI, 33% of core CPI, and 44% of core services CPI. And due to the unfixed issues in September through November, it continued to substantially push down year-over-year readings in December for the services CPI, core CPI, and all-items CPI.
The issue first cropped up for September, when the month-to-month increase of OER did a suspicious outlier-plunge. Then with no data for October due to the government shutdown and apparently no data for November either, the September outlier was carried forward through November, which substantially pushed down overall CPI.
The December index value for OER then jumped by 0.31% (+3.8% annualized) from November’s basement outlier level. BLS has apparently no intention of fixing the issues with the September through November OER (circled blue in the chart), and the year-over-year inflation readings will remain downwardly manipulated by this issue through August 2026.
A lot depends on halfway accurate CPI inflation readings, such as Treasury Inflation Protected Securities (TIPS), long-term commercial lease contracts with CPI riders, Social Security benefit levels for current and future beneficiaries…
Not to speak of the Fed’s monetary policy decisions, for which the Fed looks at the PCE Price index which uses the data from the CPI plus some data from the PPI, but with different weights and methods.
The CPI for rent of primary residence, which weighs 7.5% of the all-items CPI has a similar issue as OER, but not quite as bad since the outlier values didn’t start in September, but in October-November – for two months, instead of three months.
In December, the month-to-month increase was back in the normal range (+3.2% annualized). But the prior two months were below 1%, and unless fixed, they will continue to repress the year-over-year inflation readings for services CPI, core CPI, and all-items CPI through September.
CPI for OER and CPI for rent of primary residence account for one-third of overall CPI, and will falsify the year-over-year inflation readings by a substantial amount through August 2026.
You can see the effect of the manipulation of the data for September through November on the year-over-year change for OER and Rent of primary residence – they adjusted downward by a sudden and large amount: OER adjusted downward from the August year-over-year change of +4.0% to the November year-over-year change of +3.4%.
The CPI for OER is based on what a large group of homeowners estimates their home would rent for, with the assumption that a homeowner would try to recoup their cost increases by raising the rent. It should indirectly reflect the expenses of homeownership as a service: homeowners’ insurance, HOA fees, property taxes, and repair & maintenance – which are not included in CPI otherwise.
The CPI for Rent of primary residence is based on rents that tenants actually paid, not on asking rents of advertised vacant units for rent. The survey follows the same large group of single-family rental homes and apartments over time and tracks the rents that the current tenants, who come and go, pay in rent for these units.
The chickenshit health insurance CPI.
The health insurance CPI manipulation has been repressing CPI ever since it blew up under the Biden administration. I called it chickenshit back in 2023 for those reasons. And it has gotten worse.
So now, the health insurance CPI:
- Month to month: -1.1% (-13.8% annualized)
- Over the 3 months since September: -3.9% in (-14.8% annualized).
- Year-over-year: -0.5%
- Since September 2022: -32%
- 0% health insurance inflation since February 2019
The chart below shows the price level of the health insurance CPI. It had spiked starting in 2018 to a peak in September 2022, which had been up by 28% year-over-year. At that point, the BLS declared the index had gone awry and tweaked it.
With the current manipulations, the index value is now back where it had been in February 2019, pretending that health insurance costs hadn’t risen at all – 0% health insurance inflation – since February 2019.
It is outrageous to present to the American people this kind of chickenshit index of a big and ballooning expense that plays such a huge role in Americans’ daily lives:
The problem with the health insurance CPI is that it doesn’t actually track any aspect of health insurance that consumers pay. It is based on the “retained earnings method” that tracks health insurers’ financial metrics.
Core services CPI, core CPI, all-items CPI, year-over-year.
Core services CPI (blue in the chart below) accelerated a hair to +3.0% year-over-year in December after the massive deceleration through November, from August (when it was +3.6%). The manipulated-down housing indices account for nearly half of core services CPI. In turn, core services accounts for about 60% of the all-items CPI.
Core CPI accelerated a hair to +2.64% (red). It comprises core services plus core goods (not food and energy). It’s lower than core services CPI due to lower inflation in core goods prices.
The all-items CPI decelerated a hair to 2.68%, on dropping gasoline prices and accelerating food price inflation (yellow).
But the year-over-year CPIs are useless since they’re heavily skewed by the manipulated housing CPIs in September through November, and dogged by the health insurance CPI.
To fix this mess…
Abolish OER, increase the weight of the rent CPI to something like 15% of total CPI, and replace the remaining weight of OER (19%) with separate indices for HOA fees, homeowners’ insurance, property taxes, and repair & maintenance expenses. Canada’s statistics agency, Statcan, uses a system like that. It would still reflect the cost of housing as a service, and not as an asset price.
Scuttle the health insurance CPI method – the “retained earnings method” that tracks health insurers’ financial metrics. Instead, track actual changes of health insurance premiums, deductibles, co-pays, maximums out of pocket, drug formularies, benefits, etc. I know health insurance is complicated. But work with major health insurers, with a panel of companies that offer health insurance plans to their employees, and with a panel of consumers, and nail down the actual increases that Americans face – and they’re steep every year. Don’t tell Americans that their health insurance expenses have not risen at all since February 2019 – that’s just a ridiculous absurdity.
After the health insurance CPI is fixed and reflects the actual price increases consumers struggle with, its weight in the overall CPI needs to be increased to levels that represent the portion of consumer spending that goes into health insurance. Currently the weight of the health insurance CPI in overall CPI is less than 1%, which is another ridiculous absurdity.
Enjoy reading WOLF STREET and want to support it? You can donate. I appreciate it immensely. Click on the mug to find out how:
“And due to the unfixed issues in September through November, it continued to substantially push down year-over-year readings in December for the services CPI, core CPI, and all-items CPI.”
Hmm…intentionally unfixed perhaps….I mean how else are you going to continue the pressure the FED to lower rates again and again..
or just like KellyAnn Conway used to say….alternative facts baby….
I was waiting on pins and needles for this one! I saw they had “fixed” OER, but was sure the mathematical order-of-ops for getting to annualized data (just point-to-point Dec 25 vs Dec 26 for all contributions, then weighted into annualized. Or what it appears they do, some cumulative math around the prior 12 months M-o-M inflation.)
Thanks for clearing up.
The health care is insulting to the American people. Both premiums, deductibles, and out-of-pocket maxes are clearly significantly up since 2019.
The health insurance metric is absurd. Isn’t the purpose of the CPI to look at things from a consumer perspective?
Can you estimate the true yoy cpi if the modifications you suggest are made?
Anyone can estimate it. Results may vary. People come up with all kinds of crazy stuff all the time. But coming up with a system to replace OER and with a health insurance CPI that actually tracks health insurance expenses for most Americans halfway accurately is a very tough undertaking, requiring lots of really smart people, and the cooperation of many companies (corporate data). But that’s what a government agency is for. This is not the job for a guy sitting at a desk. Health insurance is immensely complicated because each insurance company offers different plans in different states to different employers and people, and they’re so many aspects to health insurance.
Obviously better numbers would paint a better picture and could lead to better policy decisions but in the end the majority of people will continue to believe what they feel or perceive to feel with finances. I’m sure the low approval and trust ratings of politicians translates down to almost anything the government publishes. I’m sure most Americans believe the government will do what it sees is in its best interests of it and its donors at the expense of tax payers. Not like there isn’t a dozen solid examples just in last few decades.
I work in state government where when software projects don’t just completely fail, they simply redefine the definition of success and pat themselves on the back. I don’t see any aspect of government any differently.
Wolf, I’m attempting to be more informed economically and have enjoyed reading your perspective. I was hoping you could comment on the CPI’s increased different cell imputation rate, which I just learned about recently this past semester. I couldn’t find mention of this on your website. Apologies if I missed it. The data is here: “https://www.bls.gov/cpi/tables/imputation.htm”
What’s your perspective on the different cell imputation rate increasing over the past year or so while the home cell imputation rate has decreased in turn? Is this part of a concerning trend, e.g., cherry picking from a different cell? Or much ado about nothing, e.g., BLS doesn’t have the manpower it used to and is just doing its best? Thanks!
I talked about it at length in this article: It’s what they did with OER’s missing data for October and November: They extended the outlier month-to-month increase in September to October and November. Same with the rent CPI.
Excellent reporting. Classic case of cooking the books by those at the top.
Insurance is the biggest component of my budget. My auto, home, and health insurance have steadily increased and is noticeable. The cost to home and auto repair keeps going up and up. Healthcare costs aren’t dropping. Consumers know and will vote with their pocketbook even if the data is manipulated to look better. Couple this with threat to Fed independence, Republicans need to worry about the midterms.
“Health insurance” is less than 1% of the CPI. It includes the cost of “insurance” but not the cost of actual medical services. BLS says that the total weight for “medical services and commodities” (including insurance) is 8.273%
“BLS has apparently no intention of fixing the issues with the September through November OER (circled blue in the chart), and the year-over-year inflation readings will remain downwardly manipulated by this issue through August 2026.“
How convenient.
Aside from the fact it’s an outlier, why do they duplicate a single data point to fill in missing data? Is that standard practice in someone’s line of work? I’d interpolate.
Based on what Wolf says in this article, would someone like to venture what the correct YoY CPI-U for December would be (should be)?
Sorry, at the time I wrote this the previous question and reply by Wolf hadn’t been posted yet.
Is it any wonder that our social fabric is fraying (and not at the edges but in the center), when the Federal government gaslights the public at every opportunity? I believe the proper term, as coined by Emile Durkheim, is anomie.
“In the physical world, there are constants that serve as dependable benchmarks against which to observe natural phenomena. Examples are the velocity of a falling object, the freezing point of water and the time taken for one rotation of the earth on its axis. In the economic and financial world, this degree of precision is lacking. Instead, we content ourselves with approximations, indices and averages. We pride ourselves in knowing the difference between an inflation rate of 2% per annum and 2.5% per annum. Small deviations of outcomes from expectations can trigger dramatic trading in financial instruments and result in the transfers of billions of dollars between investments. Yet, in the financial realm, can we really be sure of the value of anything?” — From Peter Warburton: The debasement of world currency: It’s inflation but not as we know it
Retired since 2018. My 2019 Medicare Part B premium was 135.50/mo, my deductible was 185/yr. My Part G supplement health insurance was 110.45/mo. My 2026 Medicare Part B premium is 202.90/mo, my deductible is 283/yr, my Part G supplement health insurance is 165.27/mo.
That makes Part B premium up 49.7% since 2019, deductible up 53%, my Part G supplemental is up 49.6%. Bald face lies by BLS on these health costs, let alone any eye or dentist or drug cost increases. This is getting old fast.