Government Data and Inflation Measures

Government Data and Inflation Measures

February 20, 2024

What is the CPI?

One of the most important and most used government measures of inflation is the Consumer Price Index (CPI). The Consumer Price Index (CPI) serves as a crucial government measure of inflation, offering insight into the average change over time in the prices paid by urban consumers for a basket of goods and services. Comprising a diverse range of items, including food, housing, clothing, and transportation, the CPI reflects the cost-of-living fluctuations that impact households. Policymakers and economists utilize CPI data to gauge inflationary trends, make informed decisions on economic policies, and adjust various financial parameters, such as interest rates and social security benefits, to ensure stability and address the evolving economic landscape.

The CPI breaks up the overall inflation measure into different weights broken out of 100. The current weights and report look like the following:

This chart to the left shows that out of every 100 dollars one might spend, 14 of those dollars will be spent on food, 6.9 will be spent on energy and so on. These weightings are the expected “average” consumer will spend on each of the following sectors.

While there can be arguments to overweighting / underweighting certain sectors it should be noted that medical services & commodities only make up 8.8%. We will touch back on this later!

Diving into the Food portion of the CPI sector I wanted to note the numbers that they post.

Based on the 2023 October CPI report, it was reported that the Food portion of the CPI rose 3.3% over the last year. This meant that on average food prices rose by 3.3% from October 2022 – October 2023 (this accounts for both food at home and away). While this seems highly unlikely given the price changes in grocery stores and restaurants in the last few years, let’s give them the benefit of the doubt and assume this is close to what the actual increase is.

Components of the CPI: Shelter

The largest component of the CPI measure is Shelter. Shelter is broken up into 3 categories but most of the component, 24% of the total CPI weighting, is given to “Owners Equivalent Rent of Residences”. As you might be confused by this title, let me assure you the way they measure this is even more confusing.

Instead of measuring Shelter or Rent by changes in house prices or average rent through Apartment listings or rental listings they came up with this “alternative” way to measure the inflation in housing. In the CPI, they view owned housing units as capital or investments therefore they do not measure the change in prices of those investments. Therefore, any increases in housing prices have little to no effect on the change in CPI. What they do instead, in general terms, is they ask said owners of the investments (houses) how much they would rent out said homes. Instead of getting actual numbers from sales or actual rental agreements, they choose to ask select participants how much they would rent their own house/residence and base the shelter component of hypotheticals.

The following chart shows some of the issues with the shelter component as measured in the CPI. The orange line measures the asking rent by getting the average changes in rent of apartments whereas the green line shows the CPI reported number. As shown in this chart the orange line shows actual increases in inflation in real time from actual listed rents and contracts. The CPI listed number lags the overall data by roughly 12 months, meaning that any change in actual inflation is delayed roughly a year which hurts the individuals who have COLA based adjustments.

While one may argue that the orange line only shows the price of apartment listings, we can also look at the average increase in the price of homes (which the CPI does not account for). The chart below (in purple) shows the Case-Shiller home price index from the beginning of 2020 through the end of October 2023.

In this time, it shows the average house has appreciated roughly 45% from the levels in 2020. While some areas of the country show more and others less, this is a much more accurate number as this index uses direct data to calculate the index rate.

So regardless of which type of chart or index one would look at, the CPI’s use of “Owners Equivalent Rent” measure for the Shelter component of inflation greatly understates and lags the true market.

Components of the CPI: Health Insurance

If you were asked how much health insurance was weighed on the CPI calculation what percent of overall spending would you guess? If you correctly guessed 0.525% then you are correct. Understanding that none of you guessed this, it seems incredibly low since insurance costs have skyrocketed in the last few decades. While this lower number might make sense as many Americans do not pay for health insurance and it is covered through their work plan. It still only makes up half a percent of the total CPI calculated inflation rate. When looking at the October CPI report it showed a very questionable change in data in this category.

The photo shown is a snapshot of the 2023 CPI report released by the BLS. It shows that the Health Insurance portion of the CPI dropped 34% year-over-year (YOY) in October 2023… Take a minute and process this report. Look at all the other numbers on the page as they all signify changes in price for said goods over the previous years. All the numbers fall in the single digits, yet health insurance shows a drop of 34%. While some may argue that this is seasonal change, or that they adjusted how they measure health insurance adjustments (which they did in October 2023 by changing it from annually to every 6 months). It still leaves a lot of questions about how they could get a 34% decrease.

This next portion is going to get bland and really go into the weeds on how the BLS calculates the health insurance portion of the CPI but is intended to show the complexity and limitations of the overall index.

Digging into this, one discovers that the way the BLS calculates the Health Insurance portion of the CPI is not how one would think or expect they would. The BLS derives this number from the insurers and not the actual rate they charge the public. The way they calculate the Health Insurance number starts by breaking down the health insurance premiums into two categories based on how they are used by the insurance company: earnings retained by the insurance company and the benefits paid out on behalf of the customers. The graphic on the right shows how they look at the health insurance premiums based on the BLS data. The BLS only calculates the retained earnings of the insurance companies and must reallocate the health insurance weight representing benefits paid out.

Going through the worksheet on how they calculate the actual number (you can find it here I summarize that the basis for what they are trying to calculate through this health insurance portion of the CPI is the actual change in health insurance earnings made by the company each year. They are not calculating how much additional money or inflation one occurs by increasing medical costs but rather finding out how much money the insurance company charges and makes based on the insurance premiums they charge and the payouts they must pay out. Therefore, this portion of the CPI does not actually account for increasing insurance premiums but calculates the inflation in “insurance company earnings”. This data is also lagged by around 10 months.

After reading this, I have tried to do some mental gymnastics to understand the reason they would put this into the CPI calculations but have yet to find a reason as for why they would base the inflation off the insurance companies retained earnings over increased policy premiums… But regardless of their reason it still doesn’t explain a 34% decrease in one month.

Government Data

The CPI report for October was estimated to come in at 3.3%. The report came in at 3.2%, this was a 10-basis point (bp) beat which caused stocks and bonds to rip higher as market expectations of a December Fed rate hike fell from 14% to 0%. This “news” event caused a massive surge in prices across the board all because of a beat of lower inflation from a projection that the market had already made. This 10 bp beat showed that inflation is falling more than market consensus and the Fed might be done hiking.

Before you get excited about the news, I wanted to breakdown the CPI beat. Looking at the 34% drop in healthcare expenses we can do a little calculation to see how much that affected the CPI in October. With a weight of 0.525% a 34% drop in healthcare expenses created a drop of 17.85 bp in the total CPI. This means that this one data point caused not only a beat in inflation but kept inflation from surprising to the upside. The massive surge in stocks and bonds were likely caused by this “questionable” data point. By the time you are reading this, this point might be explained as seasonal adjustments or something else however, the damage was already done, and the effect was already made on the market.

The point of government data is to give the market information on how the economy is doing. This data is supposed to be accurate and depict the health of the economy. However, this is not always the case. This chart below shows the Non-Farm Payroll numbers in 2023 as reported by the BLS. The chart shows the blue number which is the initial report then followed by the brown number which is the revised number reported a month later. It shows that the initial report has “Over-Reported” the total employment numbers 8 of the 9 months this year.

The odds of this occurring are very slim, if looking at this from a statistical point of view where the odds of overstating and understating are the same, then the odds of understating 8/9 times has a probability of 2% in a perfect world.

While the statistics don’t matter in this case, it is interesting to note the odds of this occurring in the world are slim. Therefore, there are serious questions as to how they are calculating this data as there seems to be a trend or issue in their calculations which cause a consistent “overstated” employment number. It should be noted that the larger news event is the initial data on employment and the revisions are not as widely publicized or talked about.

Changes in the CPI

There seems to be a lot of questionable data collection occurring in the BLS which seems appropriate since they put the BS in BLS however, it seems as though they purposefully misrepresent data to benefit the current narrative and downplay inflation. One point of discussion which has spurred some debate over the last few years is the changes that have occurred in the CPI since the 1970’s. A site called “shadowstats” reports the data of current CPI along with prior decade-based CPI reports. The site pulls data from how the CPI was calculated based on prior years rules regarding how the CPI is calculated.

The two charts shown by shadow stats show a much different perspective on how inflation is measured from prior years until now. It might be argued that current measures of CPI understate true inflation due to how much lower the CPI depicts inflation and how much inflation has been felt by actual measures. To show a real-world example of this we are going to look at the price of used cars depicted by the CPI and the Manheim Used Car Index.

Used Car Index

This is a graph of the “Used Car Index” used by the CPI for calculating the inflation rate in the US. In the chart you can clearly see a stagnation of prices from 1995 until 2021 where prices moved sideways to down. This caused a depreciating effect on the overall CPI index as prices of used cars fell over a span of 25 years until they started rising in 2021.

Zooming in on the years 1997 – 2023 we can see that at a base rate of 154.8 measured at the beginning of 1997 used car prices increased to 187.253 over the last 27 years. This shows a 20.96% increase in total prices or an average annual increase of 0.77% a year.

I chose to run the period from 1997 as a different methodology for calculating used cars was introduced that year, it is called the Manheim Used Vehicle Value Index. This index is calculated by using the average sale price of used cars based on model year/make/body and is adjusted for sale price and mileage by market class. You can find the methodology here (

This chart shows a completely different story than the one used by the CPI to calculate used car prices. Compared to the other chart, which was largely flat to downward sloping, this chart shows a steady incline. The difference is substantial. The change in price from 1997 – 2023 is 109.4% or an annual change of 4.05% a year. When comparing the two methodologies you see a difference of 88.44% in total or 3.28% a year.

When looking at the period from 1997 – 2020 the CPI calculation actually shows a depreciation in used car prices (154.8 - 138.428 = 16.372 or a drop of 10.5%). As the value of nearly everything has risen from 1997 to 2020 just in terms of price it seems to the untrained eye that the calculations used by the CPI grossly understates actual inflation in used cars and trucks as it reports that prices fell 10.5% from 1997 to 2020.

Why This Matters

As stated in the beginning, the CPI, which should be renamed the CP-LIE, is used by the government to affect policy and direct increases in benefits like social security. If the CPI is under reporting inflation, then people who rely on social security benefits will lose out on buying power as actual inflation will far outpace the true increases felt by the average American. This understating of inflation greatly benefits the government as its liabilities (social security benefits) that must be paid out are reduced over time as true inflation eats away at the overall debt burden felt by the government and its entitlements. While there are certain arguments that can be made on every side of this discussion, this report was to point out that the inflationary measure used by our government does not seem to accurately depict true inflation. Therefore, there is even greater responsibility placed on the individual to ensure they have enough financial support for one’s retirement and have less reliance on government income as a source of retirement stability.