What is Your Inflation Rate?
On August 12, the Bureau of Labor Statistics (BLS) released July 2025 inflation statistics: “In July, the Consumer Price Index for All Urban Consumers rose 0.2 percent, seasonally adjusted, and rose 2.7 percent over the last 12 months, not seasonally adjusted.”
These are national statistics, however, and do not necessarily represent the price increases your household is experiencing. Let’s look at how the BLS calculates inflation, then consider how to compute a personal inflation rate.
How the Consumer Price Index for All Urban Consumers (CPI-U) is calculated¹
CPI-U statistics are intended to measure “the average change over time in the prices paid by consumers for a representative basket of consumer goods and services.” Every month the BLS collects price data for hundreds of goods and services (the representative “market basket”) in each of 72 urban areas throughout the country. Some of the prices are collected in personal visits to retail stores or from the vendor’s website or app. Other prices come from external sources: for example, airline fares come from a Department of Transportation database and information about vehicle prices is purchased from J.D. Power.
The retail outlets and items in the market basket are selected using information from the Consumer Expenditure Survey (CES), which asks households how they spent their money and where they purchased the goods and services. The CES has two components: a Diary Survey that asks about frequently purchased items such as groceries, meals eaten away from home, and personal care services (small purchases may be difficult to recall so the respondent is asked to record every purchase for two weeks), and an Interview Survey that collects information on larger purchases such as rent, utilities, automobiles, and appliances. It would be impossible to obtain prices on every item that could be purchased, so CES information is used to determine the market basket, which is intended to be representative of the types of expenses a household might have.
Expenditure items are classified into eight major categories, shown in Table 1. Price data for these items are then combined to produce the CPI. The BLS cannot just average the price increases for the different items since people spend more on some categories than others. Instead, the BLS assigns each item or category a “relative importance” based on its share of expenditures estimated from the CES.
| Category | Examples |
|---|---|
| Housing | Rent of primary residence, owners’ equivalent rent, utilities, floor coverings, appliances, housekeeping supplies, tools |
| Food and beverages | Groceries, meals at restaurants, take-out food |
| Apparel | Clothing, footwear, jewelry, watches |
| Transportation | Vehicles, gasoline, auto insurance, motor vehicle fees, tires, airline fares, public transportation |
| Medical care | Physicians’ and dentists’ services, medical equipment, eyeglasses, home health care, prescription drugs, health insurance |
| Education and communication | College tuition, educational supplies, telephone services, computer software, telephone hardware |
| Recreation | Televisions, cable and streaming services, toys, pets and pet products, sports equipment, musical instruments, newspapers, museum admissions |
| Other goods and services | Haircuts, cosmetics, dry cleaning, financial services, tobacco products, funeral expenses |
Figure 1 shows the relative importance scores of the eight categories for 2025. Each score represents that category’s percentage of spending, and the scores sum to 100 percent. The BLS also publishes separate CPI statistics for selected metropolitan areas; Figure 1 shows that the relative category importances for the Phoenix metropolitan area are similar to those used for the national estimates. An additional CPI is published for persons aged 62 and over, with higher relative importance for medical expenses.
Figure 1. Relative importance of each expenditure category for calculating three versions of CPI: national, for the Phoenix metropolitan area, and for persons aged 62 and over. For each version of CPI, housing has the highest relative weight, accounting for more than 40 percent of the index.
Price Increases of My Purchases
It seemed to me that the published inflation rate of 2.7% was lower than the price increases I had seen in stores. But this judgment is unreliable, because one tends to notice prices that increase (chocolate! eggs!) and not pay attention to those that stay the same or decrease.
I need to look at data to see if my impression is correct. Comparing total expenditures in July 2024 and July 2025 will not give a good estimate since I purchased different items in the two time periods. Instead, I created my own “market basket” of the items my household purchased in July 2024 and compared the prices of those items in July 2024 and July 2025.
Step 1. Find the prices paid for all items purchased in July 2024 (my market basket). I was able to find or reconstruct the exact price paid for about 98 percent of the items we purchased in July 2024. The grocery store and online merchants I patronize archive my itemized receipts in my online accounts; I was able to infer other prices from banking and credit card statements (since, for example, we always order the fish fry when we go to Tom’s BBQ). I allocated 1/12 of large annual expenses such as auto insurance and tree trimming to July.²
Step 2. Find the July 2025 cost for each item in the basket. This is a little tricky, since items that were bought on sale in July 2024 would not necessarily be on sale a year later. I would overestimate price increases if I compared the sale prices of my purchases in 2024 to the retail prices in 2025. To reduce this potential bias, I calculated 2025 item prices as the lesser of (a) the 2025 sale price on the store website or (b) the 2025 retail price with the same percentage price reduction as for my 2024 purchase.³
Figure 2 shows my relative weights for each category, compared with those used for the CPI-U. Figure 3, which displays the percentage increase in costs for each category for my market basket, shows that my costs increased more than the CPI-U estimates for every category. The total cost of my July 2024 market basket was 5.3 percent higher in 2025 — nearly twice the increase measured by the CPI-U.
Figure 2. Relative importance of each category for national CPI-U and my July 2024 expenditures.
Why the Difference?
In part, my price increases differ from the CPI-U because I had a different market basket. New and used vehicles account for 7.5 percent of CPI-U but nothing in my index because we bought no vehicles in 2024. I used our actual expenses for housing; the CPI-U does not include direct expenses such as property taxes but instead estimates the amount that the homeowner would pay to rent the home, called “owner’s equivalent rent of primary residence.”
Even with the different market baskets, however, my price increases were higher than the CPI-U statistics in every category. CPI-U calculations have three features that would make their price increases lower than mine. First, the CPI-U formulas use geometric means to estimate price ratios for most area-item categories because consumers tend to shift their spending toward items that have fallen in relative price (for example, Phoenix shoppers might buy more of an apple variety that has fallen in price, and less of a variety that has risen; see Greenlees and McClelland, 2008). Using the geometric mean reduces the effect of outliers, so that the goods with the most rapid cost increases count less toward the CPI-U.⁴ The simple price difference I used does not downweight the outliers.
Second, the CPI-U employs a “hedonic adjustment” to adjust for quality differences (literally, differences in pleasure) when new items are substituted in the market basket. For example, the running shoes I purchased in July 2024 were no longer available in July 2025; the closest substitute cost 14 percent more. A hedonic adjustment would say that part of that cost increase can be attributed to improved quality of the new model and therefore the true cost increased by less than 14 percent.⁵
My calculations assumed that I would buy the same items in 2025 that I purchased in 2024 — that my market basket and relative importance scores would be the same in July 2025 as in July 2024. In reality, people shift spending patterns when prices go up. They may forego a vehicle purchase, eat out less, discontinue subscriptions, or substitute flaxseed for eggs.⁶ The relative importance scores of the CPI-U change every year (based on previous CES data) to reflect spending pattern changes. For example, the current relative importance of meat, poultry, fish, and eggs is 1.621; last year, this category had a relative importance of 1.722.
My price index and the CPI-U are thus measuring different quantities. My index measures the price increases for the items I bought in July 2024. The CPI-U accounts for item substitutions and for changes in consumption patterns over time. My actual expenses in July 2025 increased by less than the percentages shown in Figure 3 because I shifted my purchasing away from discretionary items with the largest price increases. I experienced lower expenses than my index projected, but also had no chocolate.
Copyright (c) 2025 Sharon L. Lohr
Footnotes and References
¹ This is a very rough description of the methods used to calculate CPI-U. The full technical description, with details about the household surveys, market basket selection, and calculation formulas, is given by the Bureau of Labor Statistics (2025). Nagaraja (2020) and Salwati and Wessel (2021) give clear and entertaining summaries of the methods along with history of the CPI-U and descriptions of other measures of inflation. Inflation statistics are used to determine Social Security cost of living adjustments, to set interest rates for Treasury Inflation-Protected Securities, to adjust income tax brackets, and to guide the Federal Reserve’s monetary policy.
Note that the Bureau of Labor Statistics (2025) handbook says 75 urban areas are sampled. The sample size decreased from 75 to 72 in 2025 because of federal workforce reductions; the number of prices sampled also decreased, leading some economists to have concerns about future data quality. The response rate for each component of the CES is about 40 percent.
² For apparel, restaurant meals, utilities, and recreation, I looked at the average expenses for June and July 2024 to smooth the month-to-month variation in those expenses. Our largest monthly expense was health insurance, which results in medical care having a much higher relative importance in my index than in the CPI. My market basket does not include expensive, infrequently purchased items such as vehicles and large appliances. The CPI-U does include such items; the BLS is trying to estimate the price increases that would be experienced by an “average” household.
³ I calculated electricity prices for 2025 using the cost per kWh with the electricity amounts consumed in 2024, then added the fixed fees and taxes, with similar calculations for other utilities. For items with package size changes, I calculated the 2025 price for the same package size purchased in 2024. I substituted a similar item for each good that was no longer offered for sale (the CPI-U also substitutes items). I imputed the one-year CPI-U price increase for items I could not identify individually (for example, I used the CPI-U price increase for “tools, hardware and supplies” for the credit card expenses from Ace Hardware).
⁴ Geometric means are used because they account for compounding effects. A geometric mean of n numbers is the nth root of their product. For example, the arithmetic mean of the numbers 1, 2, 3, and 50 is 14, showing the huge influence of the outlier 50; the geometric mean is 4.16. The CPI-U uses a weighted geometric mean of price ratios, with weights equal to the relative importances of the items.
⁵ The BLS uses regression models to assign a utility value to each characteristic of an item. For athletic footwear, characteristics include the upper material, the brand name, type of closure, and waterproofing. Hedonic adjustments are used for apparel, electronics, appliances, housing, and vehicles. The hedonic adjustments have only a small effect on the overall CPI-U, but have had substantial effects on estimated price increases for items such as televisions.
⁶ To replace one egg in baking recipes or pancakes, mix 1 Tbsp of ground flaxseed with 3 Tbsp water and let sit for 5 minutes. Not recommended as an egg substitute for omelets.
Bureau of Labor Statistics (2025). Handbook of Methods: Consumer Price Index. https://www.bls.gov/opub/hom/cpi/home.htm.
Greenlees, J.S. and McClelland, R.B. (2008). Addressing misconceptions about the Consumer Price Index. Monthly Labor Review, August, 3-19.
Nagaraja, C. (2020). Measuring Society. Boca Raton, FL: CRC Press.
Salwati, N. and Wessel, D. (2021). How does the government measure inflation? https://www.brookings.edu/articles/how-does-the-government-measure-inflation/