TradingFuse
Market research, published in the open
Reference 10 June 2026 · 9 min

A plain-English guide to CPI components and the supercore.

The four categories that make up the CPI basket, the OER trick that makes shelter dominate the index, why the "supercore" (services ex-housing) is the cleanest demand read, and the math behind the sub-component weights.

The Consumer Price Index headline is one number that comes out of a basket with several hundred sub-components. A desk that reads only the headline misses where the inflation is actually happening, which is the part that matters for Fed policy. Today's CPI print is the cleanest case in point: headline at 4.2 percent year-on-year was the highest since 2023, but the composition matters more than the headline, and the composition is the reason the Fed reaction function did not immediately price more hikes. This is the explainer on what is in the basket, what the supercore is and why it matters, and how to read a CPI print end-to-end.

The basket, in four lines

The CPI-U basket (the headline CPI for urban consumers) is built by the Bureau of Labor Statistics from the Consumer Expenditure Survey, which asks households what they actually buy. The basket is reviewed every two years and re-weighted with each review. The five top-level groupings are:

  • Shelter: roughly 35 percent of the basket.
  • Food: roughly 13 percent.
  • Energy: roughly 7 percent.
  • Services ex-housing: roughly 25 percent.
  • Goods ex-food and energy: roughly 20 percent.
35.0% Shelter 13.4% Food 6.7% Energy 25.2% Services ex-housing (supercore) 19.7% Goods ex-food and energy
CPI-U basket weights, top-level grouping. Shelter dominates; the "supercore" sub-component lives inside services ex-housing and is roughly 18-19 percent of the headline basket on its own. Source: BLS CPI relative-importance table. Chart by TradingFuse.

The OER trick that makes shelter dominate

Shelter at 35 percent of the basket sounds intuitive until you look at how it is measured. About three-quarters of the shelter weight comes from owners' equivalent rent (OER), which is what homeowners would pay to rent their home from themselves. The BLS estimates this by collecting actual rent paid by renters, then applying a regression to translate those rents into an implied owner-equivalent value.

Two consequences of the OER methodology that desks need to understand:

  1. Lag. OER changes much more slowly than market rents, because BLS samples rent rolls on a rotating six-month cycle. Today's OER reflects market rents from one to two years ago. In a fast-moving rent cycle, this gap can be large.
  2. Dominance. Because shelter is 35 percent of the basket and OER is most of that, a single methodological lag accounts for a sustained component of headline CPI in either direction. The 2022-2023 OER catch-up to actual market rents added roughly 100 basis points to headline CPI for over a year.

What "supercore" actually is

The supercore is BLS shorthand for core services excluding housing. The exclusion list takes core CPI (everything except food and energy), then strips out shelter, transportation services, and a few smaller items, leaving the portion of services prices that are most directly tied to wage growth and labour costs. The Atlanta Fed's "wage growth tracker" maps roughly onto the supercore inflation series; the correlation is high enough that supercore is widely treated as the cleanest market-readable proxy for underlying inflation pressure from the labour market.

Three reasons supercore matters more than core:

  • Strip out the OER lag. Core CPI is heavily affected by the rent-measurement lag described above. Supercore removes shelter entirely.
  • Strip out the noisy goods cycle. Goods prices move with global supply chains, FX-driven import prices, and one-off product cycles. Supercore removes goods entirely; it is services only.
  • Capture the labour-market signal. Services ex-housing is the part of the basket where wage costs pass through to consumer prices most directly. Wages rise, service prices rise; wages slow, service prices slow.

Today's supercore reading was the closest thing to a dovish detail in the CPI release. Headline ran hot on energy; core was lifted by shelter (the OER pass-through); supercore was the softest component. The Committee's framing under Warsh is likely to weight that detail heavily, because the supercore signal rhymes with his stated preference for trimmed-mean PCE, which excludes extreme tails by construction.

The sub-component weights inside supercore

  1. Medical services: roughly 7 percent of headline CPI and the largest single piece of supercore. The methodology is unusual; the index uses producer prices for hospital services and retail prices for outpatient and prescription drugs.
  2. Transportation services (ex-fuel): roughly 5 percent. Includes airfares, vehicle maintenance, car insurance. Car insurance has been the noisiest component since 2023.
  3. Recreation services: roughly 3 percent.
  4. Education and communication services: roughly 4 percent. Includes phone and internet services and college tuition.
  5. Other services: roughly 1 percent.

How to read a CPI print end-to-end

A working sequence for parsing a CPI release:

  1. Headline and core, year-on-year and month-on-month. The headline; market reaction usually responds to this first.
  2. Energy and food contributions. Strip them out mentally; check whether the headline is being moved by one-off shocks (oil, weather, tariffs).
  3. Shelter, OER specifically. Is the shelter contribution running above or below the trend? The pass-through from market rents to OER is the most predictable medium-term component.
  4. Supercore. The cleanest demand read. A print above the recent trend is the actionable signal for Fed framing.
  5. Sub-components. Check car insurance, airfares, and medical services; these are the noisiest pieces and can meaningfully move the headline.
  6. Composition. A 4 percent headline driven by energy and shelter is a different story for the Committee than a 4 percent headline driven by supercore.

What CPI does not tell you

  1. The Fed target. The Committee targets PCE, not CPI. PCE runs about 30 basis points below CPI on average. A 4.2 percent CPI print maps to roughly 3.9 percent PCE.
  2. The distribution. The headline is an aggregate; price changes are not uniformly distributed across households. The BLS publishes an experimental "research" CPI for elderly households that runs higher than the headline, reflecting heavier shelter and medical weighting.
  3. Inflation expectations. CPI tells you what prices have done. Breakevens tell you what the market thinks prices will do. Both are necessary.
  4. The cause. A hot CPI can be a demand story (services-ex-housing leading) or a supply story (energy leading). The composition tells you which; only the composition does.

Related reading