The American Household Economy · 2026
Methods
How to read this atlas: observation versus interpretation versus boundary; survey year versus publication year; person versus household versus PEU; stock versus flow; nominal versus real; mean versus median; revisions; and the correction policy.
Data as of
Observation, interpretation, boundary
Every figure in this atlas is presented in three layers, and they are never blurred together.
The measured number, its unit, and the period it describes. This is what the data say.
What the observation reasonably supports. This is analysis built on the number, clearly marked as such.
What the number does not establish — the denominator it excludes, the concept it does not measure, the claim it cannot carry.
Distinctions this atlas keeps
- Survey year versus publication year
- A report published in 2026 about outcomes observed in 2025 or 2024 is labeled by its observation year. The Federal Reserve’s 2025 household survey was published in May 2026; it is 2025 data.
- Person versus household versus PEU
- Survey measures use different units. SHED measures adults; income and poverty use households; the Survey of Consumer Finances uses the primary economic unit (PEU), and its age is the age of the reference person, not every adult in the unit.
- Stock versus flow
- A stock is a balance at a point in time (total debt, the share of dollars delinquent). A flow is a rate of change (the annualized transition into serious delinquency). A high flow and a low delinquent stock can both be true at once.
- Nominal versus real
- Nominal values are in the dollars of their day; real values are inflation-adjusted so different years compare in constant purchasing power. Pay can rise nominally while staying flat in real terms.
- Mean versus median
- The median is the middle; the mean is pulled upward by the top of a skewed distribution. The mean net worth is not the typical household’s net worth, and this atlas never uses it as shorthand for “typical.”
- Net worth versus liquidity
- Net worth is assets minus liabilities over a lifetime. Liquidity is cash reachable this week. Positive net worth — often locked in home equity or retirement accounts — does not guarantee cash for a $400 emergency.
- The 2022 wealth vintage
- No nationally representative 2026 wealth survey exists yet. The net-worth-by-age table is the 2022 Survey of Consumer Finances translated into June 2026 dollars by CPI. It is a structural distribution, never a mark-to-market revaluation and never a newly observed 2026 balance sheet.
Evidence timeline
When each reading in this atlas was published by its issuing agency.
Release dates across the ledger, newest first. Longer bars carry a publication lag; shorter windows are near-real-time.
- 3 readings Freddie Mac PMMS · Federal Reserve H.15
- 1 reading U.S. Treasury Fiscal Data
- 2 readings Census Bureau and HUD
- 5 readings Bureau of Labor Statistics · Bureau of Labor Statistics / FRED
- 7 readings Bureau of Labor Statistics · Bureau of Labor Statistics / FRED
- 2 readings Bureau of Labor Statistics JOLTS / FRED · U.S. Census Bureau / FRED
- 1 reading Federal Housing Finance Agency
- 4 readings Bureau of Economic Analysis
- 3 readings Bureau of Economic Analysis · Bureau of Economic Analysis / FRED
- 3 readings Federal Reserve Distributional Financial Accounts / FRED · Federal Reserve Distributional Financial Accounts
- 1 reading Federal Reserve FOMC
- 5 readings Federal Reserve Financial Accounts (Z.1) · Federal Reserve Financial Accounts / FRED
- 2 readings Federal Reserve / FRED · U.S. Treasury / BEA / FRED
- 23 readings Federal Reserve SHED
- 10 readings Federal Reserve Bank of New York
- 1 reading University of Michigan / FRED
- 1 reading U.S. Census Bureau / FRED
- 2 readings USDA Economic Research Service
- 1 reading U.S. Treasury / FRED
- 3 readings U.S. Census Bureau / FRED · U.S. Census Bureau
- 2 readings U.S. Census Bureau CPS ASEC
- 1 reading Federal Reserve Survey of Consumer Finances
Adversarial caveat register
The strongest objections to this atlas are included here rather than buried in footnotes.
- SHED is October 2025, not July 2026
- The 2026 publication date does not make its household interviews real-time. Spring and summer 2026 price and labor changes are only partly reflected.
- The newest full wealth survey is still 2022
- Quarterly Distributional Financial Accounts bridge the gap, but the 2025 SCF microdata and detailed demographic estimates were not yet published as of this atlas.
- Aggregate net worth is not spendable cash
- Housing and equity revaluations dominate the $183 trillion headline. Selling costs, taxes, account restrictions and household-level ownership are not visible in the aggregate.
- The NY Fed sees the credit-report universe
- Its Consumer Credit Panel can miss informal obligations, while policy-driven student-loan reporting changes create breaks in recent delinquency comparisons.
- Inflation rates are not price levels
- A slower annual rate does not reverse cumulative price increases. Households can still feel worse even as month-to-month inflation cools.
- Mean wealth is not the middle
- Top-end holdings pull averages sharply upward. Age-conditioned medians and percentiles are more honest self-benchmarks.
- Food-security continuity is uncertain after 2024
- The USDA annual household measure is the latest clean national series used here; the SHED one-month adult measure has a different denominator and cannot replace it.
- The 2024 SPM rate is provisional
- Census flagged threshold-coding corrections affecting 2019–2024. This atlas preserves the published 12.9% with that warning rather than treating it as final.
- Real wage signs depend on the deflator
- CPI-U, PCE and subgroup-specific baskets can produce different real-growth readings. The atlas names the deflator and treats near-zero changes as essentially flat.
- Low mortgage delinquency does not mean renters are fine
- Locked-rate owners and renters are different populations. Housing wealth for one group cannot be averaged into affordability for the other.
- Sentiment is not the same as household well-being
- Michigan sentiment and SHED self-assessment use different questions, populations and frames. Their divergence is informative, not a data error.
- “Americans” is not one economic actor
- Age, tenure, income, race, education and region produce opposite lived economies at the same moment. National summaries must preserve that distribution.
Revisions and corrections
Revisions. Monthly and quarterly series are estimates that their issuing agencies revise as more data arrive. Payroll changes, GDP, starts and permits are especially prone to revision; a figure marked preliminary is an early estimate. The trend across several prints is more reliable than any single one.
Correction policy. When a figure here is superseded or found to be in error, the canonical data file is corrected, the downloadable feed is regenerated by scripts/export_data.py, and the SHA-256 manifest hash changes. Corrections are made in the data, not silently in prose, so the data feed and the site always agree.