CPI-U annual average inflation, 2026

Annual-average year-over-year for 2026. Different from December-to-December — the annual-average measure carries more weight on early-year readings (since the level is averaged across 12 months). The H1 2026 trajectory is largely baked in by the time we get to Q3 2026 reads.

CPI-U typically runs 0.2-0.4pp above core PCE due to shelter weights. A 2.4 core-PCE median translates to roughly 2.7 CPI-U headline before tariff effects. Tariff pass-through to core goods is modest in the central case but is the principal upside-risk channel.

Distribution is mildly right-skewed: more upside risk from tariff escalation and energy shocks than downside risk from a sudden disinflationary collapse. CI reflects this.

Packed estimate before refresh = base-rate 2.55 + tariff tail 0.08 - energy mean reversion 0.03 = 2.6.

The interval remains broad because the annual-average measure still has late-year goods and energy risk. It would miss high if tariff pass-through or energy shocks dominate, and miss low if shelter disinflation accelerates.

Packed estimate = base-rate 2.6 + shelter/services persistence 0.05 + tariff right-tail mean shift 0.08 - energy mean reversion 0.03 = 2.7.

The packs move the center slightly up and narrow the low side versus the no-pack control. The run would land outside the interval if goods pass-through is much stronger than observed or if shelter disinflation breaks sharply below the component path.

This run intentionally excludes Brier packs. It starts from the recent annual-average CPI-U reference class and uses only a simple mean-reversion adjustment.

Control blend = 0.55 x recent 3-year mean + 0.45 x pre-shock mean = 2.72; shrink for realized disinflation gives 2.5.

The interval is wider than the headline run because this ablation lacks component-level shelter, goods, tariff, and policy context. It would miss high if energy or tariffs reaccelerate, and miss low if shelter disinflation arrives faster than the aggregate history implies.