Launch · Methodology
A parallel measurement of American consumer prices that uses BLS's own household-spending weights but substitutes real retail data for CPI's quality-adjustment methodology. Households have faced 32% more cumulative inflation than the official statistic captures since 2000.
Today Grabien is launching a new economic research project at realityindex.co. It is called the Reality Index, and it is the first comprehensive measurement of American consumer prices we have published — built over five years, using the Bureau of Labor Statistics' own household-spending weights but substituting real retail dollar data for CPI's quality-adjusted methodology.
The headline finding is straightforward.
Reality Index
$246
What $100 of 2000 prices costs today by the real-retail measurement.
Official CPI
$187
What the same $100 costs today per BLS's Consumer Price Index.
Cumulative gap
32%
Twenty-five years of inflation the official statistic does not capture.
Extended back to 1980, the same gap compounds further: $516 versus $391, the same 32% gap stretched over 45 years.
That number is not a surprise to anyone who has been paying for their own groceries, rent, or family health insurance. It is the gap between the morning inflation headline and the receipt at the register. What is genuinely interesting, when you look at how the Consumer Price Index is constructed, is not that the gap exists. It is that the official statistic understates the gap as little as it does.
CPI is a sophisticated index built by serious economists. It is also, by any honest read of its methodology, an index that systematically understates the price growth of things American families cannot avoid buying. The reasons are technical, but the arithmetic effect of each one is not.
CPI does not measure home prices. It uses Owner's Equivalent Rent — what homeowners would theoretically pay if they were renting their own homes — instead. OER is roughly 24% of CPI by weight, the single largest line in the index, and it has lagged actual home prices by approximately three percentage points per year since 1995.
CPI does not measure family health insurance premiums. It uses a medical care subindex tracking provider list prices for hospital stays and doctor visits. According to Kaiser Family Foundation, the average family premium has grown 4.66 times since 1999. The CPI medical subindex has grown 2.32 times over the same period — exactly half as fast.
CPI does not aggregate prices using the arithmetic mean. It uses a geometric mean formula, adopted in 1999, which mathematically produces a lower number than the arithmetic mean. The Bureau's stated rationale is that families substitute toward cheaper goods when prices rise. The arithmetic effect is to reduce reported inflation.
CPI treats faster computers and better smartphones as deflation through hedonic adjustment, even when the dollar price on the receipt is higher than last year's. The methodological argument is that you are getting more product for your money. The arithmetic effect is to reduce reported inflation.
Each of these choices has a stated rationale. Each is defended in the academic literature. And each, since the 1980s, has had the arithmetic effect of producing a lower reported inflation rate than the methodology that preceded it. None has moved the number in the other direction. That asymmetric pattern is the thing worth examining honestly.
The idea for the Reality Index has been waiting on tooling for years. Building a defensible parallel inflation measurement requires pulling and reconciling roughly two dozen retail data series across 45 years of historical records, normalizing them against BLS subindexes, and resolving the countless small inconsistencies that come from chaining different methodologies together over time. In the pre-AI era, that was a multi-year, six-figure research project — closer to a doctoral thesis than a media product.
AI changed the math. What used to require a team of research assistants and a year of grant funding now requires careful editorial judgment and a methodology that holds up to scrutiny. The data work is no longer the bottleneck. The interpretive work is.
The result is the Reality Index v1. The methodology paper is roughly the length of a long magazine feature. Every component series, every weight, every data source is documented in full at realityindex.co/methodology. The point is to be cited and challenged, not to be sold.
The Reality Index publishes every component, every weight, and every methodology choice openly. Among the resources we are launching with:
We are also launching a sister project called the American Dream Index. The Reality Index asks how much prices have risen for an evolving basket of goods. The American Dream Index asks a different question: how much it costs, every year from 1980 to today, to live the same middle-class life that 1980 promised — a four-bedroom home, three children, two cars, employer-sponsored health insurance, modest discretionary spending — priced against median household income.
The headline finding is uncomfortable. The dollar cost of that lifestyle grew 5.16 times since 1980. Median household income grew 4.89 times. The aggregate ratio is roughly unchanged, but the composition shifted hard: health care grew 7.5 times, housing carrying costs 4.2 times, groceries 3.3 times. Families have less room to cut and more cost coming from things they cannot escape. The 1980 lifestyle cost 159% of median household income in 1980. It costs 145% today. The middle-class American life of 1980 has remained out of reach for the median family for 45 years, and that gap has not closed.
Reality Index v1 is the start, not the end. Three things are on the immediate roadmap:
First, regular updates. New BLS data releases monthly. New annual data — KFF family premiums, FHFA home prices, NCES tuition, AAA driving costs — releases on staggered schedules across the year. The Reality Index is built to ingest each of these as it releases, with the headline number updating accordingly. We are not freezing a snapshot.
Second, user customization. A future version of the site will allow visitors to build their own version of the index — different weights, different baskets, different anchor years. The methodological choices that go into CPI are not the only defensible ones. The point of the Reality Index is to make the construction transparent and the choices examinable. The next step is to make them adjustable.
Third, a series. The Reality Index is the first in a series of accountability-data projects Grabien plans to publish. The instinct is the same one that built our transcript indexing service: take a public-interest dataset, make it transparent, and put it where anyone can use it.
Read the methodology. If you read the methodology paper, you will know more about how CPI is constructed than 99% of journalists who report inflation numbers. You will also be in a position to evaluate whether the Reality Index is a serious measurement or a polemic. We have built it to be the former.
Challenge us. The standing commitment of the project is to revise the methodology when revision is warranted. If you find a more rigorous data source for a category, a methodology improvement, or a measurement error, send it to info@realityindex.co. The site will publish corrections with timestamped methodology notes.
The gap between the official inflation statistic and the personal reality of American families has been a credibility crisis for federal economic measurement for at least the last decade. It is time to have the conversation honestly. This project is our contribution to that.
Reality Index is a project of Grabien Inc. The full methodology is at realityindex.co/methodology_paper. The American Dream Index, our companion measure of total annual cost of middle-class life, is at realityindex.co/american-dream-index. Every individual chart, with full editorial framing and underlying data, is browsable from the homepage. To get every update, subscribe via RSS.
Get in touch. Methodology questions, basket-spec critiques, data corrections, and partnership inquiries: info@realityindex.co.