# Money Left After Essentials — US households by income fifth **What is left of a household's take-home pay after five categories of spending: food at home, housing, transportation, healthcare, and personal insurance & pensions.** Nationally, by income fifth. Derived from the US Bureau of Labor Statistics Consumer Expenditure Surveys. - **Dataset page:** https://efficientdollar.com/data/leftover-money-after-essentials/ - **The article built on it:** https://efficientdollar.com/blog/leftover-money-after-essentials/ - **Licence:** CC BY 4.0 — reuse it, including commercially; just credit us - **Version:** 1.0.0 - **Coverage:** dollars 2013–2023, spending shares 1984–2024, United States, annual Everything here is **derived**. No BLS file is re-hosted. Every row is our construction — a basket, a residual, a deflation — over figures BLS publishes. --- ## Why this exists Plenty of people chart what Americans earn, and plenty chart what things cost. Far fewer subtract one from the other on **the same households**, which is the only way to get a residual that means anything. Income and spending here come from one survey measuring one set of consumer units, so the subtraction is legitimate. That is the whole contribution. --- ## Files | File | What it is | Rows | |---|---|---| | `data.csv` / `data.parquet` | The dollar table: after-tax income, the five spending categories, and what is left — nominal and in constant 2024 dollars. Six income groups x 11 years. | 66 | | `essentials-share-of-spending-1984-2024.csv` / `.parquet` | The long table: those five categories as a share of everything a household spends, back to 1984. Six income groups x 41 years. | 246 | | `data-dictionary.csv` | Every column in both tables — name, type, units, definition, source. | one row per column | | `datapackage.json` | Frictionless descriptor, with a SHA-256 for every file above. | — | | `methodology.md` | Our full working methodology, published unedited. | — | | `LICENSE.txt` | CC BY 4.0, full legal code. | — | CSV and Parquet carry identical values; they are written from one table in one pass and checked against each other before release. --- ## Quick start ```python import pandas as pd d = pd.read_csv("data.csv") # or pd.read_parquet("data.parquet") # What a decade did to each fifth's monthly leftover money, in 2024 dollars w = d[d.year.isin([2013, 2023])].pivot( index="income_group", columns="year", values="leftover_broad_real_2024_usd" ) ((w[2023] - w[2013]) / 12).round(0) ``` --- ## Read this before you quote a number These are not gotchas; each one changes what a figure means. **1. "Essential" labels a spending category, not a necessity.** Every dollar inside those five categories counts however discretionary it was. In 2023 the highest fifth spent 2.5x the lowest on food at home, 3.1x on housing and 5.1x on transportation. A high essentials share is **not** by itself evidence of financial pressure, and no figure here may be published as though it were. **2. Which income group got squeezed REVERSES on the basket, and only the dollar findings survive the swap.** Personal insurance & pensions is mostly Social Security contributions and retirement deposits — arguably saving, not consumption. Count it and the five fifths converge (spread 3.7pp in 1984 to 3.1pp in 2024); exclude it and they diverge sharply (11.6pp to 18.0pp), with the top fifth's share flat across 41 years. Both baskets are in this dataset and neither is privileged. **Do not publish one alone.** The dollar findings hold under either. **3. The peak is 2020, not the most recent year — and "highest since" needs checking per group.** Essentials as a share of spending peaked in 2020 in every single income group, on both baskets: a pandemic artifact, as discretionary spending collapsed. So never write "an all-time high". For **all US households** — the series our article headlines — 2024 is the highest in 41 years apart from 2020, on both baskets. **That does not generalise to the individual fifths.** Excluding 2020, the latest year is the maximum in only 5 of the 12 group-and-basket series. On the narrow basket the lowest fifth peaked in 2022, the second, third and fourth fifths in 2021, and the highest fifth in **2013**. If you quote one income group, take that group's own maximum from the data rather than assuming the last row is it. **4. Two large things are in neither the income nor the spending side.** Employer-paid health insurance appears in neither — the employer share of family coverage was $20,143 in 2025 (KFF), about 3.25x the entire CE healthcare line. This exact unit mismatch accounted for over half of the Cost-of-Thriving Index's headline finding and discredited it. Mortgage principal is excluded too, because CE treats it as asset accumulation — so "total expenditures" is not money out the door. **5. The reference household changed over the span.** It aged (46.7 to 52.2) and shrank (2.6 to 2.4 people). Older, smaller households need less, so part of any rising residual is demography rather than affordability. ### Every blind spot we know of, in full 1. **Employer-paid health insurance.** Appears in NEITHER income nor spending. The employer share of family coverage was $20,143 in 2025 (KFF) — about 3.25x the entire CE healthcare line. This exact unit mismatch accounted for over half of the Cost-of-Thriving Index's headline finding and discredited it. The employer's Social Security match has the same blind spot. 2. **Mortgage principal.** CE treats principal as asset accumulation and excludes it, so 'total expenditures' is not money out the door. 3. **Household composition drift.** The reference household aged (46.7 -> 52.2) and shrank (2.6 -> 2.4) across the published span. Older, smaller households need less; part of any rising residual is demography, not affordability. 4. **Spending category, not a necessity test.** The five categories are a SPENDING taxonomy, not a test of what a household could not avoid. Every dollar inside them counts however discretionary it was: in 2023 the highest fifth spent 2.5x the lowest on food at home, 3.1x on housing and 5.1x on transportation. A high essentials share is therefore not by itself evidence of financial pressure, and no figure derived here may be published as though it were. --- ## Breaks you must not compare across | Year | Affects | Consequence | |---|---|---| | 2004 | income | No income-based figure may be compared across 2003/2004. | | 2013 | income | The real-dollar window starts at 2013 for this reason. | | 2014 | healthcare | Healthcare levels are not comparable across 2013/2014. | | 2024 | income | The after-tax construction is a permanently closed window and can never become a recurring series. | The `income_comparable_across_breaks` column in the share table flags the 1984–2003 rows this applies to. --- ## Why the dollar series starts in 2013 and stops in 2023 **It starts in 2013** because that is when CE switched to modelling taxes rather than asking households to report them. Across 2012–2013 real personal taxes step from $3,041 to $10,008 — a 3.3x artifact, not a tax increase. One methodology regime only. **It stops in 2023, permanently.** After-tax income is discontinued after 2023 because the tax model behind it is no longer updated. This dataset can never gain another year. Spending shares continue to 2024, which is why that table runs longer. **There is deliberately no 40-year dollar series.** CE's coverage of national consumption fell from roughly 0.795 of PCE in 1984 to 0.536 in 2024. Under-captured spending mechanically manufactures a rising residual, so a long-run dollar chart would show a trend that is mostly survey drift. The share series is a ratio and cancels part of that; it does not eliminate it. --- ## How we know it is right The 2023 figures reconcile against a Federal Reserve Bank of St. Louis primer that used the same construction and published its own numbers ("Federal Reserve Bank of St. Louis, 'That Extra Money: A Primer on Discretionary Income', August 2025"): | Figure | Ours | Published | |---|---:|---:| | After-tax income, all households | $87,869 | ~$88,000 | | Leftover, all households | $27,491 | ~$27,000 | | Leftover, lowest fifth | −$10,648 | ~−$11,000 | The five-category basket is theirs, not ours — which is the point. A Reserve Bank owns the normative call about what counts as essential; we own only the arithmetic. --- ## Source and reproduction - **Source:** US Bureau of Labor Statistics, Consumer Expenditure Surveys, via LABSTAT bulk time series (download.bls.gov/pub/time.series/). Public domain; attribution requested, no licence required. - **Deflator:** CPI-U annual averages (CUUR0000SA0, period M13), base 2024. **This choice is ours, not BLS's** — BLS publishes CE in nominal dollars and has no house method for deflating it. Any real-dollar figure taken from here must say so. - The exact BLS files used are pinned by SHA-256 in the methodology, because BLS silently revises published CE figures. Pull one vintage, in one go. --- ## Citation Efficient Dollar, "Money Left After Essentials, US Households by Income Fifth" (1.0.0), https://efficientdollar.com/data/leftover-money-after-essentials/ — CC BY 4.0. Derived from US Bureau of Labor Statistics, Consumer Expenditure Surveys. --- ## Corrections If you find an error, say so: jared@efficientdollar.com. Corrections are published on the dataset page with the date and what changed. A number that reaches you and turns out to be wrong is the failure this whole bundle exists to make less likely.