Which costs does the income ratio actually belong on?
The income ratio belongs on costs you consume together — rent, utilities, insurance, the couch. It does not belong on costs that are assignable to one of you: their commute, their subscription, their half of a restaurant check they ordered twice as much from. Apply one ratio to one combined total and you stop splitting by income and start subsidizing by income.
The income-split tools mostly get the first half of this right and stop there. Halfway draws a clear line between “strictly shared expenses (like rent, utilities, and groceries)” and “personal expenses (like hobbies or solo travel),” and it lets you set custom rules per category. That is the joint-versus-personal cut, and it is the easy one.
The harder cut sits inside the shared bucket. Halfway’s own guidance says many couples “start with rent and then extend the same ratio to utilities and groceries,” and Cino’s list of what to split proportionally runs “rent, groceries, dinners, holidays, utilities.” Rent is jointly consumed in a way groceries and dinners are not. Both partners attend the dinner; only one of them ordered the steak. A category can be shared and still itemise by person — and where it does, the receipt already answers the question the ratio is being asked to guess at.
Sources: Halfway Fair Split Calculator; Cino, “Split bills based on income,” 2025
Joint costs and assignable costs are different objects
Economists have a precise version of this distinction, and it has been around for decades. It comes out of the collective model of the household — the approach that begins with Pierre-Andre Chiappori’s 1988 paper in Econometrica, and which a 2018 survey of the field describes as having produced one of the largest bodies of work in microeconomics over the three decades since. In that literature a household’s spending is not one pot. Some goods are public: both partners consume the same unit. Some are private: the unit goes to one person. Browning, Chiappori and Lewbel put the two cases side by side. For a purely private good, what the household buys equals what he consumes plus what she consumes. For a purely public good, one purchase does for both.
The illustrations they reach for are usefully mundane, and they are illustrations rather than measurements — worked cases for setting up the model. A car is joint to the extent that the couple ride in it together: the paper parameterises exactly that, as the fraction of distance travelled as a pair. Food is joint in a smaller way, in their example, because two people living apart each waste some and living together halves the waste. Housing is where the everyday intuition is easiest, though that extension is ours rather than theirs: your partner’s presence in the apartment does not use up your share of the roof.
The mirror-image term is assignable. Chiappori and Meghir define it exactly: “A good is assignable when it is consumed by both members, and the consumption of each member is independently observed.” An itemized restaurant check is the everyday artifact that gets you closest to that condition. It lists what was ordered, which is not quite the same as who consumed it — somebody still has to say that the second espresso martini was yours — but it is far more than a single total gives you.
One honest caveat, because it matters. Economists draw this line to identify who consumes what inside a household — it is a measurement distinction, not a fairness rule, and none of these papers says who ought to pay. The fairness argument is ours: if you have already agreed the split should track consumption where consumption is visible, then a line item that names a person is exactly where a ratio built on income has nothing to add.
Sources: Chiappori, “Rational Household Labor Supply,” Econometrica 56(1), 1988; Donni & Molina, “Household Collective Models: Three Decades of Theoretical Contributions and Empirical Evidence,” IZA DP 11915, 2018; Browning, Chiappori & Lewbel, The Review of Economic Studies 80(4), 2013; Chiappori & Meghir, “Intrahousehold Inequality,” NBER Working Paper 20191, 2014
How much of a household budget is actually joint?
Enough of it that proportional splitting earns its place as a default — and not so much that the exception is rounding error. In 2024 the average US consumer unit spent $78,535, of which housing took $26,266, or 33.4%. Food away from home came to $3,945. Two caveats on those figures: a consumer unit is a household, not necessarily a couple, and the BLS classifies spending by category, not by who consumes it — the joint-versus-assignable reading is ours, applied to their totals. Housing is the cleanest case for jointness on anyone’s reading; a restaurant tab is the cleanest case against it.
33.4%of average US household spending is housing — on our reading, the clearest joint cost; BLS classifies by category, not by who consumes
$3,945average annual spend on food away from home — the category most likely to arrive itemised
$78,535average annual expenditures per US consumer unit, 2024
Source: U.S. Bureau of Labor Statistics, Consumer Expenditures — 2024
The restaurant check earns its own rule because it is where assignability concentrates. In splitty’s own US-leaning scanned receipts, 15 of the 18 most frequently scanned restaurant line items are single-serving drinks — a diet coke, an espresso martini, a coffee, an Aperol spritz. That counts how often item names recur across all scanned restaurant checks — not couples specifically, and not what share of any one bill they represent. It cannot tell you who drank which one. What it does show is that the line items a receipt records most often are single-serving by nature, which is the property that makes them assignable at all.
Source: splitty first-party receipt data, 2026-11-02 snapshot (US-leaning receipts scanned in splitty; not a national sample)
What the one-ratio shortcut costs
Run the page’s own example forward. Partner A earns $95,000 and Partner B earns $42,000, so A’s income share is 69%. Say the joint costs — rent, utilities, insurance — come to $3,000 in a month, and the couple also spend $400 eating out, of which A ordered $250 and B ordered $150.
Illustrative, using this article’s own income figures:
One ratio on everything: $3,400 x 69% = A pays $2,346, B pays $1,054
Two steps: joint $3,000 x 69% = $2,070, plus A’s own $250 = A pays $2,320, B pays $1,080
Difference: $26 a month, moving from B to A under the shortcut.
Twenty-six dollars is not a scandal. The mechanism is the point, and it generalizes: once the income ratio is applied to assignable spending, each partner is charged by what they earn rather than by what they ordered. The higher earner overpays whenever their share of the itemized spending is smaller than their share of income. Whether that happens on your bill is an empirical question about your own orders, not something any study settles — but it is the direction to check, and the worked example above shows how to check it. Widen the gap between the incomes, or between the orders, and the same $26 becomes a number people argue about.
Apply the ratio to: rent or mortgage, utilities, insurance, property tax, furniture, the joint streaming plan — anything where one person’s use does not reduce the other’s.
Itemize first, then split what’s left: restaurant checks, individual subscriptions, one person’s commute, solo travel, a birthday gift from one of you — anything a receipt already attributes to a person.
The two-step rule is not more admin, it is different admin, and the second step is the one a receipt does for you. Itemize what the check already assigns, then put the income ratio on the residue. It fits alongside the other refinements this system needs: which income figure goes in the denominator, and how the same gap plays out at a restaurant table where the itemization is sitting right there in front of you. If you have ever felt the pull of the even split’s hidden tax, this is its proportional twin: a rule that is fair on average and wrong on the line item.
