Splitting trip costs with kids: how many shares is a child?

Where does the “kids count as half” rule come from?

Nowhere in particular, which is the problem. It circulates as received wisdom. A 2026 group-travel guide aimed at US groups of four to fifteen people states it plainly — assign shares like “adults = 1, kids under 10 = 0.5, or a family of 4 = 3 shares” — crediting the figures to another travel blog rather than to any derivation. Nothing explains where the 0.5 came from, or why the cutoff is ten years old rather than eight or fourteen.

Source: How to Split Group Trip Costs Fairly: Workflows and Spreadsheet Templates, Spark, 2026.

That’s not a criticism of the advice so much as a description of a whole genre. The rule survives because it is directionally obvious — a six-year-old does eat less and use less house than an adult — and because 0.5 is the easiest fraction to compute in your head at a kitchen counter. It is a folk equivalence scale: a real economic object, invented independently by every group that has ever needed one, and never once checked against the literature that exists on precisely this.

Why can’t you just itemize what the kids used?

Because the costs that start the argument are the ones that cannot be cut into portions. This is the structural difference between a restaurant dinner and a shared house, and it’s why the weighting question only bites in one of them.

At a restaurant, the kids-versus-adults problem has a clean solution: itemize. When two couples share a table and one brought children who ordered off the kids’ menu, splitting evenly makes the childless couple subsidize meals they didn’t eat — and the fix is simply to put each dish on whoever ordered it. The receipt already contains the answer. Nobody needs a theory of childhood consumption; they need to read the lines.

A house has no lines. A $3,200 week in a rental is one indivisible good, consumed jointly, with no per-person meter. So are the rental car, the Wi-Fi, and the cleaning fee. Some of these have a better basis than headcount if you look for one — a house can be split by bed and night, groceries by category — and where such a basis exists, use it; household weighting is the fallback for what’s left.

Shared grocery runs are not a hypothetical stand-in for real costs, either — they show up constantly in what people actually scan. When the cost can’t be itemized, the only lever left is how much each person counts — and that is exactly the question an equivalence scale answers.

Grocery share: splitty’s own scanned receipts (US-leaning sample, 2026 snapshot), share of receipts by merchant category, rounded. This is splitty’s sample, not a national spending survey.

What is an equivalence scale?

An equivalence scale converts a household of any size and composition into a number of “equivalent adults,” so that households of different shapes can be compared on the same footing. As the OECD describes it, each household type is assigned a value in proportion to its needs — which is how statistical agencies adjust household income before measuring poverty and income inequality. It is, structurally, the same question as: how much of this house does each family owe?

The logic rests on economies of scale in consumption. As the US Bureau of Labor Statistics puts it, a family of two is assumed to need more income than a single person, but not double — because they share. Two people share one kitchen, one router, one heating bill. In a rental house, they also share one living room and one view.

Sources: The OECD Approach to Measuring Income Distribution and Poverty, Förster & Mira d’Ercole, OECD; and Differences across Place and Time in Household Expenditure Patterns: Implications for the Estimation of Equivalence Scales, Daley, Garner, Phipps & Sierminska, U.S. Bureau of Labor Statistics Working Paper 520, 2019.

That single idea — sharing makes each additional person cheaper than the last — is what the folk rule misses. “Adults = 1” charges the second adult in a household exactly as much as the first, as though they arrived in separate houses. Every published scale disagrees.

The four scales, and what each says a child is worth

Four scales cover most of the practical ground — the OECD lists the first three among the most commonly used, and the fourth is the one the US government actually runs on. They were built by different institutions, for different countries, using different methods — and they give different answers.

The OECD-modified scale, adopted by Eurostat in the late 1990s, assigns 1.0 to the first adult, 0.5 to each additional person aged 14 or over, and 0.3 to each child under 14. The older “Oxford” scale — which the OECD had floated in 1982 for countries without one of their own — is more generous to extra bodies: 1.0, then 0.7 per additional adult and 0.5 per child. The square root scale, used in recent OECD cross-country comparisons, throws out age entirely and simply divides by the square root of household size, so a household of four is assumed to need twice what one person needs.

Sources: Glossary: Equivalised disposable income, Eurostat; and The OECD Approach to Measuring Income Distribution and Poverty, Förster & Mira d’Ercole, OECD.

The American entry is the most interesting for our purposes. The US Census Bureau’s Supplemental Poverty Measure uses a three-parameter scale, and its formula for a two-parent family is:

Source: Supplemental Poverty Measure: Technical Documentation, U.S. Census Bureau — which is also where the official measure’s own approach is described: it adjusts thresholds by family size, number of children and adults, and whether the householder is 65 or older, a different mechanism. The SPM is a supplement, not a replacement: it “does not replace the official poverty measure” (Fox & Burns, The Supplemental Poverty Measure: 2020, U.S. Census Bureau).

Look closely at that first term. Inside the parentheses, the US Census Bureau’s Supplemental Poverty Measure counts a child as exactly half an adult. The group-chat rule is not a folk corruption of the expert one — on that parameter, it is the expert one. What the folk rule drops is everything to the right: the exponent that then discounts the whole household for sharing. (Because the exponent is non-linear, 0.5 is an input, not the child’s finished weight — and a single parent’s first child enters at 0.8 instead.)

Running one beach house through all four

Take an illustrative case: a $3,200 week, split between two households. The Reyes family is two adults and two children under 10. The Chens are two adults, no kids. Every scale below is applied the same way — compute each household’s share value, then divide the house in proportion.

The share values swing wildly — the family is worth 3.00 units under the folk rule and 2.00 under the square root scale, a 50% difference. And yet:

Every method puts the family between 58.3% and 61.4% of the house. The folk rule’s 60.0% is not an outlier; it’s the median. On a $3,200 rental, choosing the most family-friendly published scale over the least moves about $97 — roughly 3% of the total, or one dinner out.

Illustrative scenario, not survey data. Share values computed from the scale definitions cited above; dollar figures rounded to the nearest dollar. The SPM row uses (2 + 0.5 × 2)^0.7 = 2.16 for the family and the two-adult form (2)^0.5 = 1.41 for the couple.

Why do such different scales give such similar answers?

Because equivalence scales are relative. Splitting a fixed cost doesn’t care about the absolute size of anyone’s share value — only about the ratio between them. A scale that is stingy toward extra people is stingy toward both households, and most of the effect cancels in the division.

The residual differences come from where each scale places children relative to adults. The Oxford scale is the most expensive for families (children at 0.5 against extra adults at 0.7) and the OECD-modified is the cheapest (0.3 against 0.5). That gap is real, but between two similarly-shaped households it is worth less than the cost of the argument. The cancellation is a property of this composition, though, not a law — it depends on both households having the same number of adults. Break that symmetry and the scales stop agreeing, as the next section shows.

Where the folk rule actually goes wrong

Not on the children. On the second adult.

Each of the four scales here discounts additional adults: 0.5 under OECD-modified, 0.7 under Oxford, and implicitly under both the square root and the SPM exponent. (A strict per-capita split, which the OECD lists as the no-sharing extreme, is the one rule that doesn’t.) The folk rule discounts none of them either. Its adults are 1.0 each, all the way up.

This cancels out in the beach house above, where both households happen to have two adults. It stops cancelling the moment household sizes diverge — and then it moves real money. Take the same $3,200 house shared by three couples and one person travelling alone. The solo traveler is worth 1.0 under every scale, because the first adult always is. Only the couples move.

Note which way this runs. Discounting the second adult shrinks every couple’s weight while the solo traveler’s stays at 1.0, so the published scales push the solo traveler’s share up, not down — from 14.3% under the folk rule to 19.1% under the square root scale. The folk rule is not harsh on the person who came alone. It is unusually generous to them, and correspondingly hard on everyone who arrived in a pair.

Same illustrative $3,200 house; share values computed from the scale definitions cited above, dollars rounded. The square root and Census SPM rows coincide here because both value a two-adult household at √2 ≈ 1.41.

There is some evidence the true discount is steeper still. Estimating equivalence scales for eight countries from household expenditure data, Daley, Garner, Phipps and Sierminska report considerable variation in economies of scale across countries and find them generally larger than the square root scale implies — though that is a finding about measured expenditure patterns, not about how much a specific group of friends actually shares in a rental house.

What this borrows — and what it doesn’t

Two honest caveats, because this article is applying a tool outside the job it was built for.

The underlying measurement problem is genuinely hard. In their 1986 study of how to measure what children cost, Angus Deaton and John Muellbauer examined the two classical methods — Engel’s food-share approach and Rothbarth’s adult-goods approach — and concluded that true costs are generally overstated by Engel’s method and understated by Rothbarth’s. Under the most extreme assumptions, they found, the Engel method makes children about four times as expensive as the Rothbarth method does — a spread they illustrate with Sri Lankan and Indonesian survey data, and one that measures disagreement between two estimation methods rather than uncertainty about any particular American holiday.

Source: On Measuring Child Costs: With Applications to Poor Countries, Deaton & Muellbauer, Journal of Political Economy, 1986.

That is a fourfold disagreement between two respectable methods, in the peer-reviewed literature, on the central quantity — and it is a disagreement about measuring what children cost their parents, not about what anyone owes for a beach house. The transferable lesson is narrow but real: if the specialists cannot pin this number down, a group chat is not going to either, and precision is the wrong thing to chase. Pick a rule, say it, move on.

How to actually split a trip with mixed households

The neighboring problems have their own answers worth borrowing: a group house splits by bed and night before it splits by person, which handles the couple who took the master suite; shared groceries have their own category logic; and whoever fronts the deposit is carrying a real cost until everyone settles up. Household weighting sits on top of those, not instead of them.