Why this exists
About
Households at the bottom of the income distribution cope hardest when firms unbundle, tier, and redesign menus — they drop to Basic, skip the add-on, take the ads. Any measure built on average spending erases their coping the most, because their coping is the biggest downward pull. This project prices the experience they had, not the spending they settled for.
The motivation, without conspiracy
Unbundling is real, directional, and persistent — not because anyone is cooking books, but because shrouded add-on pricing is profitable and stable (the economics goes back to Gabaix–Laibson and Ellison; the field evidence includes a platform-scale experiment where back-loaded fees raised spending by a fifth). Airline ancillary revenue grew from $67B to roughly $150B globally in eight years. The Council of Economic Advisers put junk fees near $90B a year for US consumers. Meanwhile the last carrier bundling bags surrendered in 2025, and the leading fee-disclosure rules were rolled back in 2026. None of that requires bad faith at any statistics agency — BLS's own Producer Price Index corrected for airline fee unbundling in 2009 because matched-fare pricing missed it. This measure generalizes that correction into a standing, consumer-side, auditable design.
What this is not
- Not a "true inflation" number. That genre adds an opinion to CPI and calls it measurement. Every figure here is built from collected menu prices, published line item by line item, reconciled to the official series it differs from.
- Not a rival CPI. The shipped product is a spread — the Restoration Gap over the matching CPI series — because the two answer different questions, and the difference is the finding.
- Not symmetric-blind. The measure prints negative when fees fall or features re-bundle, and the launch set includes sectors chosen because they should print zero or negative.
- Not comprehensive — on purpose. What has no market price gets counted in the Exclusion Register, never invented.
How it stays honest
- Pre-registered methods, frozen before live collection; a retrospective validation gate that can kill the project before it publishes a single live number.
- External review before the first print, again before the composite.
- Open code, archived hash-committed menu snapshots, documented data-access rules, and a public correction log.
- Falsifiable predictions stated in advance — including one where the measure's own thesis can fail in public.
Roadmap
| Phase | When | What |
|---|---|---|
| Sprint 0 | Jul 2026 | Preservation captures (fragile menus archived before they change), codebook v0 |
| Phase 0 | Aug–Oct 2026 | Retrospective validation against the 2008–09 bag-fee episode; public pre-registration |
| Phase 1 | → Q1 2027 | First print: air travel + streaming + controls |
| Phase 2 | 2027 | Rental housing flagship, wireless, lodging — the holistic turn |
| Phase 3 | 2028 | Composite Gap, annual cadence |
FAQ
Why “restoration”?
Because the measure prices the cost of restoring — getting back — the experience you had last year, once a firm has unbundled, tiered, or shrunk it. When your carry-on stops being included, restoring it has a price; when the ad-free feed becomes a premium tier, restoring it has a price. The Restoration Gap is how much those restorations add up to, over and above what headline CPI shows. In one line: what it costs to keep last year’s life.
Is this saying BLS is wrong?
No. CPI is a matched-specification measure with substitution logic, built to answer a compensated cost question, and it does that well — including capturing packaged-goods shrinkflation and most checked-bag fees, which we say plainly. This measure holds the experience fixed instead. Different question, different number; the gap between them is the information.
Won't this always read higher than CPI?
Within a covered sector during an unbundling wave — usually, and by construction: it's an upper bound, labeled as such. But the trigger rules make zero and negative prints not just possible but expected in specific sectors, and the launch set includes them deliberately.
Why the bottom 40%?
Because coping is concentrated there, and because BLS's own research series shows the lowest income quintile experiencing measurably faster inflation than the highest — 0.28 points a year, compounding. Weights follow BLS's published income-quintile methodology, and every release separates the effect of the pricing concept from the effect of the basket.
Who is behind this?
Independent research project. Contact and contributor details will appear here with the pre-registration note.