Splitting Rent Between Rooms That Are Not the Same

Equal thirds: the method that is only fair by accident

Start with the flat we will use throughout: $2,400 a month, three bedrooms. Room A is 180 square feet and plain. Room B is 120 square feet and plain. Room C is only 100 square feet, the smallest in the flat, but it has a private shower room — which is why this example is worth working, since area and desirability point in opposite directions.

The equal split charges $800 each. It has one genuine virtue, which is that it requires no information and no discussion, and it is the right answer in exactly one case: when the rooms really are interchangeable, as in some purpose-built student housing. Everywhere else it is a tax on whoever ends up in the worst room, collected quietly every month for the length of the lease. On our flat it prices 180 square feet and 100 square feet identically, which nobody would defend out loud, and which is why the arrangement usually survives by nobody mentioning it.

Floor area, and floor area weighted: better rulers, same question unanswered

Splitting by floor area is the obvious improvement and it is genuinely better than nothing. The three rooms total 400 square feet, so room A takes 180 ÷ 400 = 45 percent of the rent, room B takes 30 percent and room C takes 25 percent: $1,080.00, $720.00 and $600.00. The arithmetic is transparent, it is checkable with a tape measure, and it removes the crudest injustice of the equal split.

It also charges the smallest room the least, in a flat where the smallest room is the one with a private shower. The obvious patch is to weight the area for amenities: give room C a factor of 1.35 for the shower room, and the weighted units become 18, 12 and 13.5, totalling 43.5. Room A now takes 18 ÷ 43.5 = 41.379 percent, room B 27.586 percent and room C 31.034 percent, which is $993.10, $662.07 and $744.83, summing exactly to the rent.

That is a real improvement and it is still the wrong kind of answer, because the weight is invented. Why 1.35 and not 1.2, or 1.6? A private shower is worth a great deal to a housemate who works early and nothing at all to one who showers at the gym. The moment you write a weight you are asserting a preference on everyone's behalf, and the arithmetic that follows merely dresses your assertion in decimals. The next section shows what that costs in this particular flat: even the weighted split leaves one housemate strictly wanting somebody else's room.

Envy-freeness: the right definition of fair

Stop asking what each room is worth and ask a sharper question: is there any assignment of rooms and any set of prices, adding up to the rent, such that nobody would rather have somebody else's room at that room's price? A division with that property is called envy-free, and it is the right target because it is checkable. You do not have to agree with anyone else's valuations; you only have to confirm, for yourself, that you prefer your own deal to each of the others on the table.

The reassuring part is that such a division exists. Francis Su's 1999 paper in the American Mathematical Monthly proved rental harmony using Sperner's lemma, building on the earlier existence result of Alkan, Demange and Gale in Econometrica, and the practical version of the statement is easy to carry: if each housemate's own valuations total at least the rent, there is an assignment of rooms together with prices summing to the rent under which nobody envies anyone else, and nobody pays more for their room than they said it was worth. You are not hunting for something that might not be there.

Three housemates, three rooms, worked through

Ana, Ben and Clara each write down what every room is worth to them, with their own three numbers totalling the $2,400 rent. Ana works from home and would trade space for the private shower: A 750, B 600, C 1,050. Ben wants the biggest room and nothing else matters: A 1,050, B 750, C 600. Clara wants light and floor space and does not care about the shower: A 900, B 900, C 600. Nobody has to justify these; they are preferences, and the whole point is that the procedure does not need them to agree.

Now check all six ways of handing out the rooms and total the bids each time. Ana to A, Ben to B, Clara to C gives 750 + 750 + 600 = 2,100. Ana to A, Ben to C, Clara to B gives 750 + 600 + 900 = 2,250. Ana to B, Ben to A, Clara to C gives 600 + 1,050 + 600 = 2,250. Ana to B, Ben to C, Clara to A gives 600 + 600 + 900 = 2,100. Ana to C, Ben to B, Clara to A gives 1,050 + 750 + 900 = 2,700. And Ana to C, Ben to A, Clara to B gives 1,050 + 1,050 + 900 = 3,000, the clear winner. Everybody gets the room they most wanted, which is a pleasant coincidence in this example and is not what the rule is for — the rule maximises the total, not any individual.

The winning total, 3,000, exceeds the rent by 600. That surplus is the gain from allocating rooms to the people who value them most, and it belongs to the household rather than to any one member, so split it three ways: 200 each. Each person's price is then their own bid minus their share of the surplus. Ana pays 1,050 − 200 = $850 for room C. Ben pays 1,050 − 200 = $850 for room A. Clara pays 900 − 200 = $700 for room B. The three prices sum to $2,400, exactly the rent, and that is not luck — it follows from the winning total having been 3,000 and 600 having been given back.

Now verify envy-freeness, one person at a time. Ana values her room C at 1,050 and pays 850, so her surplus is 200; room A would give her 750 − 850 = −100 and room B 600 − 700 = −100, both worse. Ben values room A at 1,050 and pays 850, surplus 200; room B would give him 750 − 700 = 50 and room C 600 − 850 = −250, both worse than 200. Clara values room B at 900 and pays 700, surplus 200; room A would give her 900 − 850 = 50 and room C 600 − 850 = −250. Nine comparisons, no envy, and the surpluses are equal by construction at exactly 200 each. One caution worth stating: the equal-surplus prices are the natural first candidate but they are not envy-free in every problem. When a check fails, you adjust the prices until all the inequalities hold at once — that is a small linear program, and it is what the public implementations solve.

Where the area split leaves someone genuinely envious

Run the same three people and the same assignment through the floor-area prices — room A at $1,080.00, room B at $720.00, room C at $600.00 — and the failure appears immediately. Ana in room C is delighted: she values it at 1,050 and pays 600, a surplus of 450. Clara in room B values it at 900 and pays 720, a surplus of 180. Ben in room A values it at 1,050 and pays 1,080, so his surplus is minus 30. He would rather have Clara's room B at 720, which is worth 750 to him and would leave him plus 30. He would even rather have Ana's room C at 600, which he values at exactly 600, leaving him at zero. Ben strictly envies both of the others, and the measurement that produced this outcome was perfectly accurate.

The amenity-weighted split does not save it either, which is the more interesting result. At $993.10, $662.07 and $744.83, Ana keeps a surplus of 305.17 and Clara 237.93, while Ben has 1,050 − 993.10 = 56.90 in his own room and would get 750 − 662.07 = 87.93 in Clara's. He still envies her by $87.93. Adding a coefficient made the numbers finer without making them right, because the weight expressed the author's view of what a shower room is worth, and the only views that matter are the three in the flat.

The equal split, curiously, does not produce strict envy in this particular flat — at $800 each, Ana keeps 250, Ben keeps 250 and Clara keeps 100, and Clara is exactly indifferent between her room at 800 and Ana's room A at 800. But look at those surpluses: 250, 250, 100. Two people are getting two and a half times the benefit of the third, and Clara sits one unit away from strict envy. Envy-freeness is a floor, not a ceiling; a division can clear it and still distribute the gain unevenly, which is why the bid-based method equalises the surplus as well as removing the envy.

Shared space, couples, and the rule that actually prevents the argument

Shared space handles itself, and this surprises people. Because each housemate bids on the whole package — a bedroom together with the same kitchen, the same living room and the same bathroom queue as everyone else — the common areas cancel out of the comparison, since they are identical for all bidders. What does not cancel is exclusive use: if the person in room A gets the only balcony, or the person in room C keeps a bicycle in the hall, that must be written into the room description before anyone bids, or the bids are not about the same objects.

Couples sharing one room are the case people get wrong most often, because two questions are being run together. For the rent, a couple in one bedroom is one bidder with one budget: they occupy one room and they bid one set of numbers. For everything that scales with the number of people, they are two — water, heating, electricity, cleaning products, the queue for the bathroom. The workable convention is to split rent by room using the method above and to split consumption per head, and to say so in writing at the start, because the household that discovers this disagreement in month seven has a harder conversation than the one that settled it in week one.

Two more provisions save trouble later. Decide in advance what happens when someone leaves: either the incoming housemate simply takes the departing person's room at the departing price, which is simple and slightly unfair to a newcomer with different tastes, or the whole flat is re-divided with fresh bids, which is fairer and means everybody's rent can move. Either is fine; not deciding is not. And check how your tenancy is actually written, because who owes what to the landlord is a separate legal question from who owes what to whom — a single joint lease with a solidarity clause can make each tenant liable for the whole rent regardless of any private arrangement, so read the lease and, where it matters, take local advice.

Frequently asked questions

Why must my bids add up to the total rent?

Because that constraint makes different people's numbers comparable. Without it, one housemate could inflate everything and another deflate everything, and the totals would measure enthusiasm rather than preference. Requiring each person's three numbers to add to the rent turns the exercise into a distribution of the same amount of money, so what carries information is the shape of your bids — how much more one room is worth to you than another — not their level. It also underwrites the guarantee: when every person's values total at least the rent, an envy-free set of prices exists in which nobody pays more for their room than they said it was worth.

What if two people bid highest on the same room?

Nothing breaks; that is exactly the case the procedure is built for. The assignment is chosen to maximise the total of the bids across all rooms, not room by room, so one of the two gets it and the prices adjust so that the other does not mind. In practice the person who gets the contested room pays visibly more for it and the other is compensated by a lower price on theirs, and the envy check confirms that the compensation is enough. If the check fails at the equal-surplus prices, you shift the prices until every inequality holds — the existence result guarantees a solution is there to be found.

Can someone game the process by bidding strategically?

In principle yes, and in practice the incentives are unpleasant enough to discourage it. Overstating a room raises the price you pay if you win it, and understating one risks being handed a room you did not want at a price computed from your own understatement. Two safeguards matter more than any theory. First, bids must be sealed and revealed simultaneously, since a sequential reveal lets the last bidder play against people rather than rooms. Second, everyone should be told before bidding how the prices will be computed, so that no one is guessing at the mechanism while trying to guess at the rooms.

Does the method work for four, five or more housemates?

Yes, and the mathematics is unchanged; only the search gets bigger. Three people give six possible assignments, four give twenty-four, five give a hundred and twenty, and at some point checking them by hand stops being sensible. The step you are performing is the classic assignment problem, which standard algorithms solve efficiently at any household size, and the pricing step remains a small system of inequalities. Public implementations exist — Spliddit's rent-division tool and the interactive built around Su's method are the best-known — so with more than four rooms, use a solver rather than a spreadsheet and a Saturday afternoon.

How should a couple sharing one bedroom be counted?

Split the two questions. For the rent, they are one bidder with one budget: they occupy one room, they submit one set of values totalling the rent, and they pay one price. For anything that scales with the number of people in the flat — water, heating, electricity, cleaning supplies, and the practical pressure on the bathroom and the kitchen — they are two, and a per-head split of those costs is the honest treatment. Agree this before anyone moves in and write it down, because it is one of the two or three disagreements that reliably surfaces later and is very hard to settle once someone feels they have been overpaying for months.

Does agreeing a private split change what we owe the landlord?

No, and this is the point at which fair division stops and tenancy law starts. Who owes what to whom inside the flat is a private arrangement between housemates; who owes what to the landlord is determined by the lease. A single joint tenancy, particularly one carrying a solidarity or joint-and-several clause, can make each tenant liable to the landlord for the whole rent irrespective of any internal split, so a housemate who stops paying leaves the others exposed for the full amount. Separate leases per room work differently again. Read the actual document, and where the stakes justify it, take advice in the country where you are renting rather than relying on a general description.