CPI and HICP in Depth: Weights, Shelter, Trimmed Means and Germany versus the Euro Area

What you will learn

• How to calculate a component's contribution to headline inflation from its weight.
• How US shelter and owners' equivalent rent are measured and why they lag market rents.
• What trimmed-mean, median and supercore measures tell you that core inflation does not.
• Where German CPI and euro-area HICP differ and why the gap can be large in some months.

Explanation

Headline versus core inflation was covered in the course Macro Basics. Here we open the index itself.

Weights and contributions. A consumer price index is a weighted average of price changes for a basket of goods and services; each item's weight is its share of household spending (the relative importance in US terminology). The contribution of a component to headline inflation is approximately weight × the component's own inflation rate. Adding the contributions reproduces the headline, which tells you which part of the basket is driving a print. The BLS updates CPI weights every year and HICP weights are updated every January from national accounts data.

Shelter and owners' equivalent rent (OER). In the US CPI, shelter is by far the largest component: about 35.6 percent as of December 2025, of which OER is about 26.2 percent and rent of primary residence about 7.8 percent (weights change every year; check the current BLS table). OER is the rent a homeowner would pay to rent their own home. The BLS does not price houses; it applies the measured change in rents of comparable rental units. Each sampled unit is surveyed only every six months, so the index moves slowly and lags new-lease rents by roughly a year. Readers who know this can anticipate the direction of the largest component.

Underlying inflation measures. Core inflation excludes food and energy by fixed rule; statistical alternatives let the data decide. The trimmed mean ranks component price changes by size, discards the extremes (the Cleveland Fed trims 8 percent from each tail) and averages the rest; the median CPI takes the middle of the weighted distribution. Supercore means core services excluding housing, watched because it tracks wage costs most closely. When these measures disagree, the disagreement tells you whether a print was driven by outliers or by broad pressure.

German CPI versus HICP. Destatis builds two indices from the same price collection: the national Verbraucherpreisindex (CPI) and the HICP that feeds the euro-area aggregate. The main differences, per Destatis and the Bundesbank: the German CPI includes owner-occupied housing (imputed rents), the HICP does not; the HICP also leaves out games of chance; and because of these exclusions services carry a smaller weight in the German HICP than in the CPI. In the 2026 German HICP weights, services are about 47 percent, non-energy industrial goods 28 percent, food including alcohol and tobacco 16 percent and energy 9 percent. Because the same price changes carry different weights, the two can diverge by several tenths in a month. Markets quote the German HICP, because the European Central Bank (ECB) targets euro-area HICP; domestic news quotes the CPI. The euro-area flash appears at the start of the following month, the final figure about two weeks later.

Worked example

Hypothetical example: a simplified four-component basket with invented y/y price changes.

Headline inflation is the sum: 0.34 − 0.20 + 1.42 + 0.67 = 2.23 percent. Shelter alone contributes 1.42 points, almost two thirds. Core inflation is (1.42 + 0.67) / (0.356 + 0.444) = 2.09 / 0.80 = 2.61 percent. If shelter inflation falls to 3.0 percent with everything else unchanged, the headline drops by 0.356 × 1.0 = 0.36 points to about 1.9 percent. A reader who knows new-lease rents slowed a year ago can expect this.

Now a trimmed mean. Ten equally weighted components show changes of −6.0, −1.0, 0.5, 1.5, 2.0, 2.5, 3.0, 3.5, 4.5 and 9.0 percent. The plain mean is 1.95 percent. Trimming one component from each tail gives 16.5 / 8 = 2.06 percent; the median is (2.0 + 2.5) / 2 = 2.25 percent. The outliers almost cancel in the mean, but the trimmed measures show the centre of the distribution is higher.

Common mistake

Practice activity

On paper: food (weight 15 percent) rises 3.0 percent, energy (10 percent) falls 8.0 percent, shelter (30 percent) rises 5.0 percent, other core (45 percent) rises 2.0 percent. Calculate the headline rate, the core rate and the headline rate if energy had been flat.

Key takeaways

• Contribution = weight × component inflation; the sum of contributions is the headline.
• US shelter is about a third of the CPI and lags market rents by roughly a year.
• Trimmed-mean, median and supercore measures show whether pressure is broad or driven by outliers.
• German CPI includes owner-occupied housing, the HICP does not; the ECB and markets watch the HICP.

Next step

The next lesson explains why the US Federal Reserve targets a different index, the PCE price index, and how it is linked to the GDP deflator.