When a shoe the Bureau of Labor Statistics (BLS) has been pricing disappears from a store, the agency replaces it with the most similar item that outlet sells and then decides how much of any price difference is real inflation and how much is a different product. For men’s, women’s, boys’ and girls’ footwear, BLS lists hedonic quality adjustment, a statistical method that puts a dollar value on features, as one of the tools it uses. This is the reason a shoe price index can run for decades even though no single pair stays on the shelf that long.
The problem a shoe index has to solve
The Consumer Price Index (CPI) works by pricing the same items in the same outlets month after month. Data collectors follow one specific pair, in one store or on one website, and record its price each time it is due. If the price of that pair moves from $60 to $63, the index records a 5% increase for that observation.
Shoes make this hard. Styles turn over each season, materials change, and a pair can vanish when a line ends. The BLS Handbook of Methods says that when an item is no longer available, the standard procedure is to replace it with the most similar item sold in the same outlet. It also notes that the rules prevent replacing a seasonal item when it is out of season, and that most apparel items are treated as seasonal. Footwear sits inside the apparel group in the CPI.
Once a replacement is chosen, BLS has three broad ways to handle it.
Comparable replacement
If the new pair is essentially the same as the old one, the commodity analyst, the BLS specialist who reviews every replacement for that item, treats them as directly comparable and the price difference goes into the index as a price change.
Direct quality adjustment
If the new pair differs in ways that can be valued, BLS adjusts the price of the old item for those differences before comparing. The Handbook calls this the most explicit method for dealing with a replacement of different quality. The value of the difference can come from manufacturer cost data or from hedonic regression models.
Imputation
If the two items can be neither compared directly nor adjusted, BLS calls the replacement noncomparable. The price change for that observation is then estimated from other items. The Handbook says class-mean imputation, which borrows the price change of other replacements in the same item and area that were either directly comparable or quality adjusted, is used mainly for vehicles, other durables and apparel.
What hedonic adjustment means for footwear
A hedonic model is a regression that relates the prices of many shoes to their characteristics. Fit on a large sample of CPI prices, it estimates how much each feature, such as a leather upper or a boot shaft, is associated with price when the other features are held constant. When a pair in the sample is replaced by one that differs in those features, the estimates are used to remove the part of the price gap that the model attributes to the change.
The BLS quality adjustment page lists four footwear items under hedonic adjustment: men’s footwear (AE011), boys’ footwear (AE021), girls’ footwear (AE022) and women’s footwear (AE031). BLS publishes model details for three of them.
| Model | First used in CPI | Data behind current model | Observations |
|---|---|---|---|
| Men’s footwear | 1991 | August–September 2022 | 645 |
| Women’s boots, shoes, sandals, slippers | 1991 | March–April 2022 | 1,523 |
| Athletic footwear | 1995 | May–June 2026 | 1,151 |
According to BLS, the men’s and women’s models have each been updated seven times since 1991, and the athletic model five times since men’s and women’s athletic shoes were combined into one model in 2000. Both of the 2022 models were estimated from web-based information because of restrictions on data collection during the COVID-19 pandemic.
Which features count
The features differ by model, which tells you what BLS found to matter for prices.
- Men’s footwear: footwear type (shoe, sandal, slipper and three boot heights), upper material (leather, man-made, fabric, combination), outsole material, brand category, and terms for rain boots and electrical hazard protection.
- Women’s footwear: type, including boots above the knee, upper material, outsole, brand category, heel height in four bands, and closure (buckle, zipper, hook-and-loop, laces or none). BLS says the closure variable returned for the first time since 2011 and heel height was split into more bands, with higher heels linked to higher prices.
- Athletic footwear: upper material (leather, canvas, man-made), brand category with separate terms for six large national brands, high-top versus low-top, hiking and water shoe types, closure, and visible support features. Outsole material is left out because most athletic shoes use synthetic soles.
All three models also include controls for type of store and region, which BLS says are not used in adjustments. They help the model fit but do not change a quality adjustment.
In the men’s model, a man-made upper carries a negative coefficient relative to leather, and an exclusive brand category carries a large positive one relative to national brands. In the women’s model, a heel over three inches carries a positive coefficient relative to a heel under one inch. These signs line up with what a shopper would expect.
A worked example from BLS
BLS does not publish a shoe example, but its hedonic questions and answers page shows the mechanics with a men’s shirt. A $20 shirt is replaced by a $30 shirt with different sleeve length and fabric. Using the model, BLS estimates that the old shirt, if it had the new features, would have cost about $29.84. The index then records a change of only $0.16, not $10. The same logic applies when a leather shoe in the sample is replaced by a fabric one, or a low boot by a taller boot. The page also says BLS re-estimates hedonic models about every two years.
Comparing a 1985 pair with a 2026 pair
Our calculator scales an old price by the footwear index. For all footwear, the 1985 annual average was 102.358 and the September 2026 index was 154.506. So $50 spent on shoes in 1985 equals $75.47 in September 2026 terms on the footwear index. On the all-items CPI the same $50 becomes $156.62. From 1985 to 2025, the footwear index rose 42.4% while all items rose 199.3%.
Here is what that result means and what it does not.
- It is a quality-constant answer. Because replacements are adjusted for features where BLS can value them, the index tries to price the same kind of shoe across time. The $75.47 is what a pair with 1985-like characteristics would cost, not what the typical pair people buy today costs.
- New features are partly excluded. If shoes today have cushioning or materials that the models value, the index treats that difference as more shoe, not as a higher price. A 2026 pair that costs much more than $75.47 may be a different product, not proof that the index missed inflation.
- Mix shifts are not prices. A move by shoppers toward pricier styles is not a price change. The index follows comparable items, so that kind of shift does not by itself raise it.
- Categories differ. On the same $50 from 1985, the men’s index gives $77.83 and the women’s index $70.40. You can try other years on our 1985 price page or the home page calculator.
The Handbook is direct about the limits: substitutions and quality adjustments are listed among the possible sources of error in CPI estimates, and BLS supplements its own data with secondary sources because complete information is never on hand. Hedonic adjustment narrows the problem; it does not remove it.
Why this matters for reading our charts
The long, gentle climb of the footwear lines on our women’s shoe page and elsewhere is partly the product of these methods. A raw average of shoe price tags would mix in every shift toward pricier or cheaper styles. The CPI aims to strip that out. Producer price indexes for footwear, such as the one on our athletic and rubber factory prices page, follow their own sampling and quality rules, so a gap between factory and retail lines can reflect method as well as margins. Our method page lists every series the site uses.