News · Methodology
The Reality Index prices a fixed 1980 household basket with real retail data and compares it to the official CPI. After a fair public critique, we rebuilt the headline on more conservative ground. Here is exactly how the number is made.
Most Americans do not believe the official inflation numbers. That instinct is not crazy, but it is usually expressed badly — with a vague sense that “the government is lying” and a number pulled from somewhere on the internet that cannot be checked. The Reality Index was built to do something more disciplined: to put a defensible, fully reproducible figure next to the official Consumer Price Index and show, category by category, where the two agree and where they part ways.
This post explains how that number is made. It also explains why, after a fair critique from an economist who took the time to check our work, we just moved the headline figure down — and why we think that makes the project stronger, not weaker.
The Reality Index takes a basket of goods and services that defined middle-class American life in 1980 — groceries, housing, transportation, health care, utilities, education, dining out, communications, recreation, and the rest — and asks a single question: what does that same basket cost now, priced with the actual prices families pay at retail?
As of May 2026, the answer is that the Reality Index rose 4.95% year over year, against the official CPI’s 4.25% — running about seven-tenths of a point hotter. Stretch the lens back to 1980 and a basket that cost $100 then costs about $476 priced the Reality Index way, versus $407 by the official CPI: a cumulative gap of roughly 17% over forty-five years.
That headline is the most recent monthly reading. We also publish the full 1980–2025 annual series as long-run context: on that series the gap is 18.5% — the Reality Index up 4.63 times since 1980 against CPI’s 3.91 times. We label the annual figure clearly so it does not get confused with the live monthly number.
Two choices separate the Reality Index from the official CPI.
First, the prices. Wherever an independent retail or consumer dataset exists, we use it instead of a constructed government subindex. Family health premiums come from the Kaiser Family Foundation’s employer survey. Home prices come from the Federal Housing Finance Agency. Gasoline comes from the Energy Information Administration, vehicles from AAA, and retail food and electricity from the BLS Average Price series — the actual dollar-and-cents tables, not the modeled indexes. Only where no independent series exists — dining out, communications, recreation — do we fall back to a CPI subindex.
Second, the weights. The Reality Index fixes each category at its share of household spending in 1980 and holds that basket constant across every year — what economists call a Laspeyres index. This is the honest way to ask what the 1980 cost of living has done: you fix the basket and let the prices move.
Housing is where the most careful work lives, because housing is the largest single piece of the basket. We do not simply track home prices, which rose 6.84 times since 1980. Most households are not buying a house in any given year; most owners carry older mortgages or own outright. So we build a tenure-weighted carrying cost — what owners and renters actually pay each year, weighted by the national homeownership rate. That composite rose 5.10 times since 1980, close to CPI’s own shelter measure. The gap between the Reality Index and CPI comes from how much weight housing carries in the basket, not from a dispute over what a house costs.
When the Reality Index launched, its headline applied modern expenditure weights — the 2024 shares — to every year going back to 1980. The economist Jeremy Horpedahl pointed out, publicly and correctly, that this is a mistake. Applying today’s basket to the past over-weights categories like housing and health care that only became a large share of spending later, which mechanically inflates the historical gap. It makes the past look more expensive than it was for the families actually living in it.
He was right. We adopted the fix. The headline now uses the fixed 1980 basket described above, and the cumulative gap fell from the roughly 32% we had been reporting to about 17% on the latest reading.
An index that revises when a critic is right is more trustworthy than one that never moves.
It is worth being precise about what that change was and what it was not. It was almost entirely a weighting decision, not a change in the prices. If you take today’s corrected price data and re-apply the retired 2024 weights, the gap climbs right back to about 39.5% — essentially the index’s original published figure. The prices did not move; the basket did. We show all of this in the open, side by side, in the weighting explorer, so anyone can reproduce each version and watch the number move with the assumption.
We would rather publish the smaller, more defensible figure than defend a larger one we no longer stand behind.
The natural objection is that we simply picked the weighting that flatters our story. The opposite is true: the fixed 1980 basket produces the smallest gap of every defensible scheme we tested. Run the same corrected prices through the Federal Reserve’s preferred PCE weights and the gap rises to about 65%. Run them through 2024 consumer-expenditure shares and it lands near 37–40%. Across five legitimate weighting schemes, the gap ranges from 18.5% to about 65%. The size changes; the direction never does. Under every reasonable weighting, the goods and services families actually buy have outrun the official CPI.
The weighting explorer is the first step toward something larger. The cumulative gap is not really one number; it is a family of numbers, one for every defensible way of weighting the basket. Rather than pick one and ask readers to take it on faith, we are building an interactive version that lets you — or a client running an analysis of your own — choose the weighting basis and watch the headline recompute in real time. Three constructions will sit side by side:
The fixed 1980 basket, carried forward. Today’s headline — the 1980 spending shares held constant across every year, a Laspeyres index. It answers the cleanest question: what has the 1980 cost of living actually done?
The 2024 basket, carried backward. Modern spending shares applied to the whole history — the method we retired as the headline. It over-weights what only grew large later, but it is a legitimate view, and we will keep it on the table rather than bury it.
Year-by-year actual weights. The basket updated every year to match how households really spent that year — a chained index, and the construction closest to how the official CPI is itself built. It is the most apples-to-apples comparison to CPI, and the one we are working hardest toward.
That last version is the reason we want to be candid now: building it honestly means assembling consumer-expenditure weights for every year from 1980 to the present, and it turns the Reality Index from a fixed-basket measure into a chained one. We do not yet know where that lands the cumulative number — and we would rather say so than imply today’s figure is permanent. The roughly 17% cumulative gap is a reference point on one defensible basis, not a monument. It is a CPI-adjacent number by construction, and as these options come online — the chained version especially — expect the headline to move. When it does, we will show the work.
The project’s commitment is to be principled, not partisan — and that means publishing what does not flatter the thesis. Several things we track ran slower than CPI: chicken, milk, bread, electricity, gasoline. Food as a whole rose more slowly than the official index. We publish those with the same prominence as the categories that ran hot.
We are also honest about the index’s blind spots. The clearest is quality. When an airline seat shrinks from 35 inches of legroom to 31, or a package quietly loses two ounces, you are paying the same for less — and that erosion is largely invisible to any price index, ours included. We call this unmeasured erosion. It is not evidence that the CPI is dishonest; it is a real limit on what any index can capture, and naming it is part of being a serious reference rather than a grievance.
The whole point of the Reality Index is that you do not have to take our word for it. Every series, every weight, every data source is published. When someone shows us a methodology choice that is wrong, we change it and we date the change. That is what happened here, and it is what will keep happening.
The full methodology, the cumulative chart, and the weighting explorer are all available on the site.