What do you actually have to divide when a roommate moves out?
Two things the rent split never covered: the stuff you bought together and the deposit you paid together. Rent is a recurring bill with a standing rule, and the lease itself is a separate problem with its own law; our guide to what happens to the rent when a roommate leaves mid-lease covers that. The couch and the deposit are different. Each is a single asset with two owners, and one owner is leaving. In economics that has a name: you are dissolving a partnership, and there is a forty-year literature on how to do it without a fight.
The advice in circulation stops one step short. A roommate-move-out guide published by Cohabby, a roommate-finder site, in March 2026 puts the shared belongings plainly: “The couch you bought together. The kitchen table. That stand mixer someone found on sale. Decide now who keeps what, or agree on buyout prices.” On the deposit: “If they paid half, how do they get their share back? Most landlords won’t split a deposit mid-lease. Your options: reimburse their portion now and reclaim the full amount at move-out, or agree to split whatever comes back when the lease ends.” Both instructions are correct about the timing and silent about the rule. “Agree on a buyout price” is the whole difficulty, and “split whatever comes back” leaves the hardest question open: who eats the deductions?
Source: Roommate Moving Out? Step-by-Step Action Plan, Cohabby, March 9, 2026
Why is “just agree on a price” the wrong instruction?
Because the two of you are holding different numbers, and the receipt is neither of them. The person staying knows whether the couch fits the room and whether they can face buying another one. The person leaving knows whether it fits the next place, what the movers charge per item and how much they ever liked it. Each of those is a private valuation. “Agree on a price” asks two people with private valuations to converge on one in a conversation where each has a reason to shade their number, and the guide’s own warning about that conversation, “mildly awkward today and genuinely hostile if you try to have it over text three months later,” is a description of what happens when there is no rule to converge on.
The receipt does not rescue this. The IRS, whose interest is valuing used household goods people give away, is blunt about it in Publication 561: “The FMV of used household items is usually much lower than the price paid when new. Household items include furniture, furnishings, electronics, appliances, linens, and similar items.” And: “Such used property may have little or no market value because it may be out of style.” A $900 receipt is the only number on paper, so it is where the argument starts. It says nothing about what a two-year-old couch is worth to either of the people arguing over it.
The framing that fixes it. Stop asking “what is the couch worth?” It has no single answer. Ask instead “who values it more, and how much should that person pay the other?” That is a question with a known procedure, and the procedure works precisely because you own the couch half each.
Sources: Publication 561, Determining the Value of Donated Property, Internal Revenue Service, December 2025 revision; Roommate Moving Out? Step-by-Step Action Plan, Cohabby, 2026
What does economics say about dissolving a 50/50 partnership?
That equal ownership is the easy case. In 1987, Peter Cramton, Robert Gibbons and Paul Klemperer published “Dissolving a Partnership Efficiently” in Econometrica, studying exactly this setup: “Several partners jointly own an asset that may be traded among them. Each partner has a valuation for the asset; the valuations are known privately.” Their result: “Despite incomplete information about the valuation of the asset, a partnership can be dissolved ex post efficiently provided no single partner owns too large a share.” Efficiently means the asset ends up with the person who values it most; the paper’s other requirement, individual rationality, means each partner, knowing only their own value, expects to do at least as well by taking part as by keeping their share. That is an expectation, not a promise about every pair of numbers.
The contrast they draw is the important part for a couch. When one person owns an object outright and is trying to sell it to another, a well-known result by Myerson and Satterthwaite says no mechanism can guarantee it ends up with the person who values it more; the seller’s incentive to ask high and the buyer’s to offer low leave good trades on the table. Cramton, Gibbons and Klemperer show that co-ownership breaks the impasse. In Benny Moldovanu’s 2002 survey of this literature, the lesson is stated in one line: the main obstacle to an efficient split “is not really the presence of information asymmetries but rather the presence of asymmetries in physical endowments.” In plain terms, the fact that you bought the couch together is what makes it solvable. Two roommates who each paid half are the textbook case.
The mechanism they construct is an auction between the owners. “A bidding game is constructed that achieves such dissolution whenever it is possible,” and for the equal-shares case the paper shows something stronger: “an equal-shares partnership can always be dissolved efficiently” by a family of simple sealed-bid auctions in which the highest bidder takes the asset and the bid revenue “is divided equally among all the bidders.” The authors add the honest limit: “Such a simple auction only works, however, when partners’ shares are approximately equal.” Their worked case with uniformly distributed valuations puts a number on “approximately” for that example only: with two partners the rule still works “if no partner’s share exceeds 57.7%.” The general result is looser, that the set of dissolvable partnerships “is centered around the equal-shares partnership.” Read it as a shape, not a cutoff. A couch you split down the middle is the case the theorem was built for. A couch one roommate paid nearly all of is outside any reading of “approximately equal,” and for that one the problem is a reimbursement, not a dissolution, with the buyer-seller impasse back in play. (As between the two of you, payment share is the natural proxy for ownership share; if you agreed something else at the time, that agreement governs.)
Sources: Dissolving a Partnership Efficiently, Econometrica 55(3), 1987; How to Dissolve a Partnership, Journal of Institutional and Theoretical Economics 158(1), 2002
How do two roommates run the couch auction?
Each of you writes down, privately, a price for the whole couch. Reveal both numbers at once. The higher number keeps the couch and pays the other person half of the average of the two numbers. That is the entire rule. It needs no appraiser, and it is the two-person version of the auction Cramton, Gibbons and Klemperer proved dissolves an equal partnership efficiently. The number does not have to be your exact private value; the paper assumes both partners bid strategically and derives a symmetric bidding strategy under which the higher bid still comes from the person who values the couch more. That is a statement about the equilibrium, under the paper’s assumptions, not a guarantee about every pair of numbers two people could write.
Read the result from each side, taking the two numbers as true values for the sake of the arithmetic. Maya valued the whole couch at $500, so her half was worth $250 to her; she bought Dev’s half for $200 and is $50 ahead of walking away. Dev valued the whole couch at $300, so his half was worth $150 to him; he sold it for $200 and is $50 ahead. Both are better off than they were, the couch went to the person who wanted it more, and nobody had to negotiate. The receipt-based number, $450, would have made Maya pay more for Dev’s half than that half was worth to her. The “just take it” number, $0, would have handed Dev’s $150 share to Maya for nothing. Both of the intuitive answers are wrong in opposite directions; the two-number rule lands between them for a reason. The two $50 surpluses are what this illustration produces, not what the mechanism promises.
Why does it work? Not because it forces honesty; it does not, and the paper does not claim it does. Both partners will shade their numbers, and Cramton, Gibbons and Klemperer derive the symmetric strategy each of them ends up using. In that symmetric equilibrium both partners shade by the same strategy, so the ranking of the numbers matches the ranking of the values and the couch lands with the person who wants it more. What the paper does not establish is that any two bids whatsoever will do that; it establishes that the rule gives neither of you a reason to bid in a way that breaks it. “Just agree on a price” has no such property. The mechanism was designed for partners who cannot see each other’s valuations, which is the situation two roommates are in, however long they have lived together.
One number belongs on the table before either of you writes yours: what the couch would fetch if you sold it. Sometimes the honest answer is that neither of you wants it, and a buyer on a resale listing values it more than both of you. McAfee builds that case into the rule. “In the event that an outside option, the possibility of sale to a third party, exists, both the WBA and the LBA dictate sale to the third party if neither bidder’s bid exceeds the third party’s value.” His model treats that value as known; yours is an actual offer, net of what selling costs you, not a listing price. So the gate is: get a real offer, and if neither private number beats it, sell the couch and split the proceeds in half. If one number does beat it, run the auction, and use the winner’s number rather than the average, because that is the version McAfee finds still allocates correctly with an outside buyer in the picture: “The WBA is ex post efficient,” while “the LBA is ex post efficient only if no outside option exists.” That is a result inside his model; in a living room it is the sensible default, not a proof. Either way, the offer, not the $900 receipt, is the number the argument should start from. Publication 561 already told you why.
There is a simpler rule you will hear recommended: one person names a price, and the other chooses whether to buy the couch at that price or sell their half at it. Moldovanu calls it “commonly known as the ‘Texas shoot-out’” and credits its appeal honestly: it “is easy to use, since its rules do not depend on the parties’ values.” McAfee’s 1992 paper in the Journal of Economic Theory analysed it, under the name cake-cutting, and found the catch. “With symmetric information,” when both people can see what the couch is worth to each other, it “produces an efficient allocation, and agents prefer to be the proposer.” When information is private, “the mechanism is ex post inefficient.” The person who names the price has an edge and the couch can land with the wrong person. If you use it anyway, flip a coin for who proposes. If you would rather not have an edge to flip for, write two numbers.
Sources: Dissolving a Partnership Efficiently, Econometrica, 1987 (the k+1-price auction, whose bid revenue is “divided equally among all the bidders”, and “split-the-difference bidding (the average of first- and second-price)”); Amicable divorce: Dissolving a partnership with simple mechanisms, Journal of Economic Theory 56(2), 1992 (the Winner’s Bid and Loser’s Bid auctions, and the cake-cutting result); How to Dissolve a Partnership, JITE, 2002 (the Texas shoot-out)
What about the twenty things nobody wants to pay for?
Take turns picking. The couch and the TV are worth an auction; the colander, the bookshelf and the seven mismatched mugs are not, and running a sealed bid on each would be a parody of fairness. For a pile of small items where no money will change hands, McAfee’s paper studies exactly this procedure under the name “alternating selection”: one person picks the item they want most, then the other, and so on until the pile is gone. He gives conditions under which “a myopic strategy of picking the most preferred available item forms an equilibrium,” meaning nobody gains by picking strategically rather than honestly, and he shows those conditions “are satisfied under independence,” that is, when what one of you wants tells you nothing about what the other wants. For a box of mismatched kitchenware that is a reasonable assumption; it is not one you can verify, so treat honest picking as the sensible default rather than a guarantee.
Two boundaries keep this honest. The alternating-picks result is for items with no money attached; the moment someone says “I’ll pay you for the bookshelf,” it is an auction item and goes in the first column. And the ownership question comes before either rule. The sorting step, which items were bought together and which were not, is where the argument usually lives, and it is a question of fact, not of fairness. Bank statements do not prove what you agreed, but they show who actually paid, which is the fastest place to start.
Source: Amicable divorce: Dissolving a partnership with simple mechanisms, Journal of Economic Theory, 1992 (the alternating-selection mechanism, “when there are many items to be allocated, agents are risk neutral, and transfers are prohibited”)
What happens to the security deposit when one roommate leaves mid-lease?
In every state we checked, it stays where it is. Deposit law is state statute, so this guide reads three of them rather than all fifty, and the pattern across the three is the same: the deposit is held against the tenancy, not against a person, and the rules for returning it are written around the tenant vacating and the tenancy ending; none of the three provides for a partial refund mid-lease when one of several tenants leaves. California’s Civil Code section 1950.5 requires the landlord to return the deposit with an itemized statement “no later than 21 calendar days after the tenant has vacated the premises,” and when several adults are on the lease, the default is one payment to all of them: the landlord “shall return the remainder of the security by a check made payable to all adult tenants on the rental or lease agreement at the time the tenancy terminates.” New York’s General Obligations Law, in the section covering non-rent-stabilized units, caps the deposit at “the amount of one month’s rent” and gives the landlord “fourteen days after the tenant has vacated the premises” to return it with an itemized statement. Massachusetts allows 30 days, and its guidance is the only one of the three that addresses the mid-lease leaver directly: “if a tenant has a written lease and moves out before the lease ends, the landlord has up to 30 days after the lease ends to return the security deposit.”
Sources: California Civil Code § 1950.5, 2025 text as reproduced by Public.Law; N.Y. General Obligations Law § 7-108, as reproduced by Public.Law; Learn about returning or getting back a security deposit, Mass.gov. Deposit law is state statute and varies; these three are illustrations, not a survey, and none of this is legal advice.
So the leaver’s half of the deposit is not money they can collect. It is a claim on a refund of uncertain size, payable months from now, in a check that, in California at least, carries every adult name on the lease. The size is uncertain because the landlord may deduct, in California’s words, for unpaid rent, for “the repair of damages to the premises, exclusive of ordinary wear and tear, caused by the tenant or by a guest or licensee of the tenant,” and for “the cleaning of the premises upon termination of the tenancy necessary to return the unit to the same level of cleanliness it was in at the inception of the tenancy.” Every one of those deductions comes out of the one pooled deposit at the end, whether the damage or the missed rent dates from month two or month eleven. The statute names the cause, damage “caused by the tenant or by a guest or licensee of the tenant,” but it does not divide the deduction between co-tenants, and it does not care which month the damage happened. Sorting that between you is your problem, not the landlord’s.
That timeline is why “split whatever comes back” is the worse of Cohabby’s two options. It makes the leaver’s refund depend on how carefully the people who stayed treat a bathroom the leaver will never see again, and then routes the money through a check the leaver may need to be present to endorse. The other option is the better one, in our view, and it is a judgement, not a statute: the incoming roommate, or the ones staying, buy the leaver’s share of the deposit, whatever they paid in, on the day they leave. Net out anything already known to be the leaver’s, such as unpaid rent or damage they caused, and pay the rest at face value. The leaver walks away paid. The stayers are paying par for a claim that may come back smaller, and that is the point, not an oversight: from that day on they are the people whose conduct has the most say in what the refund will be. The risk of everything that happens after the departure ends up with the people who control most of it.
Two lines of paper. First, a dated note between the roommates that the departing roommate has been paid their share of the deposit and, as between the roommates, gives up any claim on it. It binds the roommates, not the landlord: the refund check may still carry the leaver’s name if it is still on the lease when the tenancy ends (California’s rule names the adult tenants on the agreement at that moment), so the note should also say the leaver will endorse it if it does. Second, photographs of the unit on the day they leave. The note settles the money; the photographs record the unit’s condition on that date, which is the evidence you would need if a deduction ever has to be argued into the tenancy’s later months. In California, a third option exists on paper: the landlord and every adult tenant can sign a written agreement specifying “how any remaining portion of the security will be returned, including whether it will be returned to a specific adult tenant or divided among multiple tenants, with the allocation percentages.” It needs the landlord’s signature, which is why the buyout is the practical route.
Sources: California Civil Code § 1950.5, subdivisions (b) and (h); Roommate Moving Out? Step-by-Step Action Plan, Cohabby, 2026
What should the move-out settlement look like on paper?
One page, six lines, finished before the box truck arrives. The order matters: sort ownership first, because every rule below assumes you already know which things were bought together.
List what was bought together, with the receipt price beside each item
Bank statements, not memory. The receipt price is the anchor for the argument you are about to avoid, not the value; it goes on the list so nobody has to look it up mid-conversation.
Auction the items both of you might want
Get a resale quote first. Then each writes a private number for the whole item and you reveal together. If neither number beats the quote, sell it and split the proceeds. If one does, the higher number keeps it and pays the other half the winner's number. Decide the version before writing, never after; the choice must not depend on who won.
Alternate picks through the rest
Coin flip for first pick. Each turn, take the item you want most. Nothing changes hands for these; if someone offers money for one, it moves up to step 2.
Buy out the leaver's deposit share
The incoming roommate, or the stayers, pay the departing roommate whatever they put into the deposit, less any unpaid rent or damage already known to be theirs, and take over the whole claim on the refund. The risk of everything after departure day moves to the people who control it.
Photograph the unit and write the two-line note
Date-stamped photos of every room, plus a signed line between the roommates that the deposit share is paid, that the leaver gives up any claim on it, and that they will endorse the refund check if it carries their name. The statutes itemize deductions against the tenancy, so the tenancy's later months need their own record.
Settle the last shared receipts the same week
The final grocery run, the cleaning supplies, the pizza on box day. These are itemized receipts with the usual problem: not everyone had everything. Scan them and settle them while everyone is still in the same group chat.
Step six is where splitty fits, and only step six. The buyout in step 2 and the deposit transfer in step 4 are single payments between two people that you calculate on paper; splitty does not value a couch, hold a deposit or run a household ledger over months. What it does is split the receipts a move-out generates. Scan the receipt, and every line starts shared among everyone in the apartment. Tap to remove whoever did not have that item, tax lands in proportion to each person’s share, and each roommate gets a pre-filled payment request in their preferred payment app. Only one person needs the app, which matters when one of the people you are settling with has already changed their address.
Honest boundary. A household that has been running a shared ledger for a year should close that ledger in the app that holds it; Splitwise tracks, splitty settles. Use the two-number rule for the couch, the face-value buyout for the deposit, and splitty for the paper the last week produces.
FAQ
Roommate move-out FAQ
Common questions about dividing shared belongings and a shared deposit when one roommate leaves.
01 How do you decide who keeps shared furniture when a roommate moves out?
Run a two-number auction for anything both of you might want. Each roommate privately writes down what the whole item is worth to them, you reveal the numbers together, and the higher number keeps the item and pays the other person half of the average of the two numbers. The item goes to whoever values it more, and the payment lands between what each person's half was worth to them. For the small items nobody would pay for, flip a coin and alternate picks. Economists studying partnership dissolution (Cramton, Gibbons and Klemperer, 1987; McAfee, 1992) analysed both procedures for owners who hold roughly equal shares and cannot see each other's valuations; for the auction they show the item still goes to the person who values it more even though both bid strategically, and for alternating picks they show only that picking your favourite each turn is an equilibrium under stated conditions.
02 Should the roommate who is leaving get paid the receipt price for their half of the couch?
No. The receipt price is what the couch cost new, and the IRS's own valuation guidance for used household goods says their fair market value is usually much lower than the price paid when new. Paying half the receipt usually overcharges the person staying; paying nothing takes the leaver's half for free. The two-number rule replaces the receipt with what the couch is actually worth to the two people arguing over it.
03 Does the landlord have to return part of the security deposit when one roommate moves out?
Generally not mid-lease. Deposit statutes are written around the end of the tenancy: California requires return within 21 days after the tenant vacates, by a check payable to all adult tenants on the lease unless everyone and the landlord sign a written agreement setting percentages; New York requires return within 14 days of vacating; Massachusetts guidance says a tenant who leaves before a written lease ends waits until 30 days after the lease ends. Deposit law is state-specific and this is not legal advice, but across the three states this guide reads the pattern is consistent: the return rules are keyed to vacating and to the end of the tenancy, none provides a partial mid-lease refund when one co-tenant leaves, and Massachusetts says outright that an early leaver waits until the lease ends.
04 Who should reimburse the departing roommate's share of the deposit?
The incoming roommate, or the roommates staying, on the day the leaver goes: whatever the leaver paid in, less any unpaid rent or damage already known to be theirs. That puts the full claim on the refund in the hands of the people whose conduct will decide the later deductions, and lets the leaver exit without waiting on a check that may carry every name on the lease. Write a dated note between the roommates that the share was paid and that the leaver gives up any claim on it and will endorse the refund check, and photograph the unit the same day.
05 What if one roommate paid most of the cost of a shared item?
Then it is closer to that person's item than to a partnership, and the auction rule weakens. Cramton, Gibbons and Klemperer's simple auction works when shares are approximately equal; their general result says only that the dissolvable range is centered on equal shares, and their one worked example, two partners with uniformly distributed valuations, puts the edge at 57.7%. There is no universal cutoff. For a lopsided purchase, treat the item as belonging to the majority payer as between the two of you, unless you agreed otherwise at the time, and settle the minority payer's contribution as a reimbursement rather than a bid.
