Budget Percentages by Income: 50/30/20 Does Not Scale
One set of percentages cannot fit every income, because the cost of keeping a roof and getting to work does not scale down. Realistic bands at five income levels, from spending data rather than a slogan.
Here are the numbers. They come from what US households actually spent in 2024 rather than from a rule of thumb, and the reason they are grouped by income is that a single set of percentages cannot fit every income. The cost of keeping a roof over your head and getting to work does not scale down.
The bands, by income
Each category as a share of total household spending, by income fifth. Boundaries in 2024: the second fifth started at $29,932, the third at $57,452, the fourth at $94,511 and the top at $155,925.
| Category | Lowest | Second | Third | Fourth | Highest | All |
|---|---|---|---|---|---|---|
| Housing | 41.6% | 38.3% | 36.0% | 32.6% | 29.3% | 33.4% |
| Transport | 14.6% | 16.8% | 17.4% | 17.7% | 16.9% | 17.0% |
| Food | 15.7% | 14.8% | 13.6% | 13.2% | 11.3% | 12.9% |
| Healthcare | 9.8% | 9.6% | 8.5% | 8.1% | 6.5% | 7.9% |
| Those four together | 81.6% | 79.6% | 75.5% | 71.6% | 64.0% | 71.2% |
| Everything else | 18.4% | 20.4% | 24.5% | 28.4% | 36.0% | 28.8% |
And the two figures people actually come looking for, this time against pre-tax income rather than against spending:
| Lowest | Second | Third | Fourth | Highest | All | |
|---|---|---|---|---|---|---|
| Average pre-tax income | $16,658 | $42,925 | $74,474 | $121,548 | $264,510 | $104,207 |
| Housing as % of income | 87.4% | 44.7% | 32.4% | 24.2% | 16.6% | 25.2% |
| Shelter alone as % of income | 55.1% | 27.7% | 20.5% | 14.7% | 10.3% | 15.7% |
| Retirement, pensions and insurance as % of income | 4.3% | 6.3% | 8.4% | 9.8% | 10.3% | 9.4% |
How to read your own row
Find your household's pre-tax income in the second table and read down that column. Then compute your own figures the same way: twelve months of each category divided by twelve months of total spending.
Two rules for interpreting the result, and they are the opposite of how percentage tables are usually used.
A gap is information, not a verdict. Housing at 44% when your column says 36% does not mean you have overspent. It usually means you live somewhere expensive, or you bought recently at a high rate, or your household is smaller than average for your income. Each of those has a different response and none of them is "try harder".
Look at what the gap forces. Because the shares sum to 100%, being eight points high on housing means being eight points low somewhere else, and where it comes from is the real finding. If it comes out of the saving line, that is a structural problem needing a structural answer. If it comes out of entertainment, the household has already absorbed it.
What changes as income rises
Read the first table left to right and three things move together.
Housing falls, by twelve points. From 41.6% to 29.3% of spending. The absolute figure rises, from $14,563 to $44,033, but far more slowly than income, which is the whole reason the percentages cannot be universal.
Healthcare falls, from 9.8% to 6.5%. The dollar amount nearly triples across the range. Its share falls because everything else grows faster.
Everything else doubles, from 18.4% to 36.0%. This is where the entire difference in flexibility lives. The highest fifth is not more disciplined than the lowest. It has twice the proportion of its spending available to redirect.
The saving row is the one worth sitting with. It rises from 4.3% to 10.3% of pre-tax income and never comes close to the 20% that rules of thumb prescribe. If you are saving 8% of a $74,000 household income, you are doing slightly better than the typical household at your income level, and every article you have read has told you that you are failing. Both of those statements are true, and only one of them is useful. That is the argument in the 50/30/20 rule, tested against real incomes.
What does not change
One row barely moves at all. Transport sits between 14.6% and 17.7% at every income level, and it is the second-largest category for four of the five groups.
That is worth more attention than it gets. Households scrutinise the grocery bill, which is 11.3% to 15.7%, and treat the car as a fixed fact. But transport includes the payment, the insurance, the fuel and the maintenance, and it is the one large category where a single decision, the car you buy, sets the figure for years. Cutting $40 a month from groceries takes constant effort; buying a car costing $90 a month less takes one afternoon and never needs doing again.
Note also what the lowest fifth's 14.6% conceals. That group runs 1.0 vehicles per household against 2.7 in the top fifth, so the low share reflects not having a car rather than spending less on one, and the cost of that shows up as constrained work options instead.
Why one set of percentages cannot work
The mechanism is arithmetic rather than behavioural.
Essential costs have a floor. A small flat, a food shop for two, and getting to work cost roughly the same whether the household earns $35,000 or $135,000. Call that floor $28,000, which is close to what the lowest fifth actually spends on housing, food, transport and healthcare combined. As income rises above the floor, the floor becomes a smaller share, automatically, with nobody making a decision.
So any fixed percentage rule is really a statement about one income level, generalised. 50/30/20's needs line works at the top fifth, approximately, and nowhere else.
The useful replacement is not a better set of percentages. It is to work in dollars, which is what zero-based budgeting does, and to use percentages once a year to check the shape.
When this does not apply
If you live in a high-cost or very low-cost area. These are national averages. Housing at 45% of spending in an expensive metro can be perfectly rational, and 22% in a low-cost area can still be too much house. The national table cannot see your rent.
If you are in the lowest income fifth. That group's figures show spending of $35,046 against pre-tax income of $16,658, which is not a description of monthly budgeting. It includes retirees drawing down assets, students, and households having one bad year. Read the share-of-spending row rather than the share-of-income row, and treat the second table's 87.4% as a statistical artefact rather than a target you are missing.
If your household is an unusual shape for its income. The bands are averages of households containing 1.6 people at the bottom and 3.2 at the top. A single person earning $121,548 will look nothing like the fourth quintile row, and should compare against a higher one.
If you are self-employed. Business expenses and quarterly tax do not appear in these categories at all, and mixing them into a household percentage produces a number that means nothing. Separate the two sets of books first.
If you are carrying high-rate debt. Debt payments are not in this table, because BLS treats principal repayment as a transfer rather than an expenditure. A household putting 12% of spending at a card balance will look strange against every row here and may be doing exactly the right thing: avalanche or snowball.
Next steps
- Total twelve months of housing, transport, food and healthcare from your own statements.
- Divide each by your total spending over the same period, and write the four percentages down.
- Find your income column and compare. Note only the categories more than about five points out.
- For each of those, work out where the difference is coming from. That is the actual finding.
- Then go back to dollars. The percentages have done their job and cannot help you decide anything else.
Sources and assumptions. All figures are from the BLS Consumer Expenditure Survey, 2024 (USDL-25-1586, released 19 December 2025) and its Table 1101 by income quintile, and the shares are computed from the published means. BLS "housing" includes shelter, utilities, household operations, supplies and furnishings; "shelter" is the rent or mortgage component alone. The saving row uses the personal insurance and pensions line, which includes Social Security contributions and life insurance premiums, so it overstates voluntary saving. Quintile figures describe groups rather than individuals, and income is before taxes. The same table underlies the 50/30/20 test, so the two agree by construction. Checked 20 September 2026.
