Reading the survey
A household, roughly. The survey's unit is the consumer unit: a family, a person living alone or sharing a home but financially independent, or two or more people living together who share major expenses. Groups by age are by the age of the reference person, the first person named when the household is asked who owns or rents the home.
Averages, not medians. Every number is a mean across the households in the group. A few households that bought a car or paid tuition this year pull the average up for everyone, so the typical household spends less than the average on lumpy lines like vehicles and education, and nothing at all on some of them.
Shares of spending, not of income. Each share is a line's part of the group's average annual expenditures. Income taxes are not an expenditure in the survey, so they are not in the bars. In earlier years the tables also reported personal taxes and income after taxes, but not for 2024: BLS says the provider of the tax model it uses did not update it for the 2024 tax year, so there are no federal or state tax estimates and no after-tax income this year. This page leaves taxes out rather than guess.
Giving is wider than charity. Cash contributions is cash given to people or organizations outside the household: gifts to relatives and friends, alimony and child support, support for students living away from home, and contributions to religious, educational, charitable and political organizations. The tithe is in there, but so is a lot else.
Saving for retirement counts as spending. Personal insurance and pensions holds life and other personal insurance, and also all Social Security contributions paid by employees and contributions to government, railroad and private pensions and to retirement plans for the self-employed. So the biggest line for high-income households is partly saving.
Some cells are thin. Where BLS marks an estimate with a relative standard error of 25% or more, the bar is hatched and the hover readout says so; read it as rough.
What stands out
Method. scripts/fetch.py downloads the three tables and
scripts/build_data.py reads each group's published annual mean for 22 spending lines that
together make up total expenditures. The script checks that the lines add up to BLS's total for every
group and that the shares it computes match BLS's published shares, then writes data.js.
Shares on the page are those means divided by the group's total; “your numbers” are arithmetic on what
you type.
This page illustrates published survey averages. It is not financial advice.
It illustrates Spending and contentment, a lesson in the personal finance guide.