Concept · The Lens

Real inflation vs the official 3%.

~8 min read · Numeracy

When the government says inflation was 3% last year, what does that actually mean? Not what most people think. CPI is a specific basket of goods, weighted in a specific way, computed with specific adjustments. Change any of those choices and the number changes — sometimes by a lot. This article walks through what the official number measures, what alternative methodologies show, and why your investing math depends on knowing which one you're using.

This matters because every return number you ever see is either nominal (before inflation) or real (after inflation). The cost calculator and compounding visualizer on this site use 7% as the default annual return — that's a real number, roughly the S&P 500's long-run average after subtracting BLS CPI. If real-life inflation is actually higher than BLS CPI for your basket, the same nominal returns leave you with less purchasing power than the math suggests.

What CPI actually measures

The headline US inflation number is CPI-U — Consumer Price Index for All Urban Consumers — published monthly by the Bureau of Labor Statistics (BLS). It's a weighted basket of ~80,000 prices across eight major categories: housing, transportation, food, medical care, education and communication, recreation, apparel, and "other." Each category gets a weight based on what the average urban household spends, sampled from the Consumer Expenditure Survey.

Three things about this construction matter for understanding why people argue about it:

None of these adjustments are conspiracies. Each has a defensible methodological rationale. But each also has the effect of producing a smoother, lower headline than a "what does life actually cost more this year?" survey would. That gap is the entire argument.

The alternative methodologies

Several private indexes attempt to measure inflation differently. None is "the truth"; each is a different lens on the same question. The best practice is to know all three and read them together.

Chapwood Index
The "what's actually in your cart" index
Created by wealth manager Ed Butowsky, who started in 2008 by collecting over 4,000 items people around the country actually buy, then narrowed to the 500 most frequently purchased — and has tracked their raw prices, quarter by quarter, across the 50 largest US metros ever since: groceries, gas, insurance, utilities, restaurant meals, household goods. Crucially: no substitution and no hedonic adjustments. If beef goes up, the index goes up — it doesn't assume you switch to chicken. Historically reports inflation ~10% per year (7-13% by city) over stretches where BLS CPI printed 1.5-3% — implying a CPI-indexed pension or salary quietly loses purchasing power every year the gap persists. Methodology and the 500-item list are public on chapwoodindex.com. Honest caveat the other direction: Chapwood publishes no auditable raw-price archive, its basket skews toward urban discretionary spending (which inflates faster than a full household budget), and refusing ALL quality adjustment overstates constant-standard-of-living inflation just as aggressive hedonics understate it. Read it as the honest ceiling, not the true rate.
ShadowStats
CPI computed with 1980 methodology
John Williams' index. Takes the current BLS data and re-applies the pre-1980 CPI methodology (before substitution + hedonic adjustments were introduced). Useful as a historical-continuity gauge: what would the CPI number look like if we hadn't changed how we measure it? Typically reports inflation roughly 4-6 percentage points higher than the current BLS number. Subscription-based access to the long series; methodology described publicly.
Truflation
Real-time price data from 30+ sources
Aggregates ~13 million prices daily from retail, real-estate, and commodity datasets. Updates every 10 seconds (vs. BLS's monthly cadence). Heavier weight on housing and food than CPI; lighter on the "other" residual. Public dashboard at truflation.com. Tends to land between BLS CPI and Chapwood — closer to CPI in calm regimes, closer to Chapwood when food + housing are spiking.
Billion Prices Project (Cavallo / Rigobon, MIT)
Academic web-scraping benchmark
A 2008-2017 research project that scraped daily prices from hundreds of online retailers across 60+ countries. Found that online-retailer CPI tracked official CPI very closely in most developed economies — meaning the headline number is roughly right when applied to the goods online retailers actually sell. Gaps showed up for the categories online retailers don't sell well (housing, services, healthcare). Project archived in 2017 but the academic papers are the gold-standard reference for cross-checking official statistics.

Honest read: the truth is somewhere in the convergence. BLS CPI is rigorous and consistent — useful as a benchmark — but it under-measures the inflation experienced by households heavy on housing, food, and services. Alternative indexes typically show 2-6 percentage points higher than BLS. The exact gap depends on your personal basket.

Why this matters for your math

The S&P 500's nominal long-run total return is roughly 10% per year. After subtracting BLS CPI (~3% long-run), the real return is about 7%. That's where the 7% default on this site's calculators comes from. But:

Worked example · housing-heavy household

A subscriber lives in a major metro. ~40% of their take-home pay goes to rent or mortgage. Housing prices in that metro rose 8% over the year; BLS measures the same metro at ~4.5% via OER (smoothed); their personal experienced inflation on the housing line is therefore much higher than what hits the headline. Same person sees grocery and insurance lines rising similarly above CPI. Their personal inflation rate is closer to 7-8% even though the headline reads 3%.

Now run that household's investment math: at 10% nominal returns, BLS-CPI real is 7%, but their personal-basket real is 2-3%. The implication isn't that they shouldn't invest — investing at 2% real still compounds — it's that "I beat inflation" requires a much higher bar than the headline suggests, and the calculators on this site (defaulting to 7%) are overestimating their purchasing-power growth.

What to do with this

Three concrete steps:

  1. Know which CPI number any return is measured against. "Real return" without specifying the deflator is half-information. The default in finance is BLS CPI; that's a reasonable benchmark but not the only one.
  2. Estimate your personal inflation rate. Track your actual spending for 3-6 months. Compare against the same categories a year later. The number you get is the one that matters for your purchasing-power math, not the BLS headline.
  3. Adjust the calculator defaults accordingly. If your personal inflation runs 5% (vs. BLS 3%), use 5% rather than 7% as the return rate on the compounding visualizer — that gives you a more honest projection of your purchasing power 20-30 years out.

Bottom line: CPI is one specific way of measuring inflation. It's not wrong, but it's not the only number, and it's not necessarily your number. The official 3% is a useful benchmark for cross-period comparison and Fed policy; it's a less useful benchmark for the question "is my portfolio actually growing my purchasing power?" That question requires you to estimate your personal rate — and the calculators on this site will give you honest answers as long as you feed them an honest input.

Sources: Chapwood Index methodology, 500-item basket, ~10% five-year average vs CPI — chapwoodindex.com, About/methodology, Ed Butowsky interview (Peak Prosperity); BLS CPI methodology (substitution, hedonic adjustment, owners' equivalent rent) — BLS Handbook of Methods: CPI; ShadowStats methodology — shadowstats.com; Truflation real-time index — truflation.com; Billion Prices Project — Cavallo & Rigobon, Journal of Economic Perspectives (2016).

Educational content, not personalized financial advice. Private inflation indexes are unaudited and have their own methodological weaknesses (noted above) — treat all of them, including the official CPI, as lenses rather than ground truth.