The Endowment Effect: Valuing What You Already Own More

The endowment effect is the gap between what somebody will accept to give a holding up and what they would pay to acquire the same thing. Standard reasoning says those two numbers should match. The two numbers come apart. The gap opens as soon as the thing is held, so possession itself becomes a source of value. Thaler named it in the Journal of Economic Behavior and Organization in 1980.

The endowment effect needs no mechanism of its own. The whole of it rests on the reference pointThe level outcomes are measured from. Everything above it is read as a gain and everything below it as a loss.. A reference point is the level a person measures outcomes from. Taking a thing on quietly moves that level, and once it has moved, letting the thing go is measured downwards from the new level while acquiring it was measured upwards from the old one. Kahneman and Tversky set out the shape of that measuring in Econometrica in 1979, and the arm below the reference point is steeper than the arm above it. The endowment effectValuing something more once it is held than before it was held. is not a separate quirk bolted on to the rest of the subject; it is what happens when that steeper lower arm meets a thing somebody has in hand. Everything that follows is an unpacking of that one sentence, using two numbers that can be put side by side and checked.

What are the two numbers, and why should they be the same?

Any object will do. A second hand scooter, a wedding sari that has been worn once, a scheme unit, a share. Two separate questions can be put about it. The first: what is the least its owner would take to give it up? The second, asked of the same object on the same day: if the owner did not have it, what is the most they would hand over to get it? Both questions ask how much the object is worth. Neither mentions anybody's feelings, and neither depends on anything the object itself does not carry.

Standard reasoning gives a short and confident answer here. Because the thing being valued is identical in both, the two questions must produce the same number. A thing does not change its worth depending on which side of a counter it happens to be sitting on. If a scooter is worth Rs 40,000/- then it is worth Rs 40,000/- to the person who has it and Rs 40,000/- to the person who wants it, and the only reason a sale ever happens is that the two people disagree about that figure for reasons to do with what each of them needs. The prediction is not merely that the numbers will be close; it is that they are the same number. There is only one thing being valued, and it does not know who is being asked.

The same test runs on anybody. For any thing owned for a while that could be sold tomorrow, the first number is what its owner would take for it. Supposing that same thing had never been owned, and were merely seen in a shop this evening, the second number is what the same person would pay. Most people find that the two numbers are not close, and almost nobody finds the first number below the second. The asymmetry is the endowment effect itself. On the illustrative scale used throughout, that same scooter gives two answers of Rs 60,000/- to give it up against Rs 40,000/- to acquire it, a gap of Rs 20,000/- on a thing with no market anywhere near it.

What exactly is the gap, and how would anybody measure it?

The two numbers have names, and naming them shortens everything that follows. The first is willingness to acceptThe least somebody would take to give a thing up., the least somebody would take to part with a thing they have. The second is willingness to payThe most somebody would hand over to acquire the same thing., the most somebody would hand over to acquire the same thing. The endowment effect is the finding that the first number sits reliably above the second, on the same object, at the same moment, and often by a very large margin.

Measuring the gap needs care. Almost any careless design lets a second explanation in. Without a control, a difference in the answers has three candidate explanations rather than one: the two people, the two objects, or possession. Asking one person what they would take for their scooter and a different person what they would pay for a scooter leaves the difference open to being about the two people, or about the two scooters. The design that closes those doors is a control: one group of people is given the object and then asked what they would take for it. A second group, drawn from the same room at the same time, is given nothing and asked what they would pay for the identical object. Nothing differs between the two groups except who is holding something. The gap is not measured by asking one person two questions; it is measured by changing exactly one thing between two matched groups and watching the answers separate.

Kahneman, Knetsch and Thaler ran that design and reported it in the Journal of Political Economy in 1990, and it is the demonstration the effect is usually taught from. Their point was structural rather than arithmetical: the separation survived when the object was ordinary, when the people had no attachment to it, and when the exchange was made as easy as it could be made. The numbers used here are not theirs and measure nothing. The scale below is an invented illustrative one built on the Kesari Logistics case, chosen so that the arithmetic is easy to follow and the widening is easy to see.

How long must something be held before the gap opens?

Here is where most people guess wrong, and the wrong guess is a reasonable one. The natural story is attachment: a thing is kept for months, its owner grows used to it, memories collect around it, and eventually they would not part with it for what they paid. The attachment story predicts a gap that starts at nothing and builds slowly, so a thing held for a week should show no gap at all.

PossessionHolding a thing, as distinct from having paid anything for it. does not work that way. In the controlled design, the separation is there almost as soon as the object has changed hands, well before anybody could have grown fond of anything. Time is not what opens the gap; having the thing is what opens it, and time only widens what is already there. Attachment needs months, and the effect does not wait for them. The timing alone rules the attachment story out.

Be careful about what the illustrative scale below can and cannot show. The scale starts the two numbers together at the instant before the holding is taken on and separates them steadily from there. Drawn that way it shows the widening cleanly and says nothing at all about how fast the first separation appears. Read the shape as a picture of how far apart the two answers can travel, not as a measurement of the opening. On that scale the gap grows by a steady Rs 25,000/- for every month the holding is held.

Why does the reference point explain the gap?

Go back to the shape underneath. A person does not score outcomes against their total wealth; they score them against a level, and everything is read as a distance above or below that level. The level is the reference point, and the single most useful thing to know about it is that it moves. The level moves when circumstances change, it moves when a question is worded differently, and it moves when somebody takes a thing on.

Now put the two questions on that picture. Ask somebody who does not have the object what they would pay for it, and the object sits above their reference point: acquiring it is a gain, and they are deciding how much of a certain loss of money to accept in exchange for it. Ask somebody who does have it what they would take, and the object has already been absorbed into the level they measure from: parting with it is a loss, and it is being measured down the steeper arm. Nothing about the object changed between the two questions; what changed is which arm of the value function it is being measured on, and the lower arm is steeper than the upper one.

The endowment effect is therefore best described as reference dependence meeting possession rather than as a bias in its own right. Tversky and Kahneman made exactly this argument for riskless choices in the Quarterly Journal of Economics in 1991, and it is the reason the gap has a predictable direction. Losses are felt more heavily than equivalent gains, and the invented Palash cohort measurement puts that at 2.2 for a fifty-fifty gamble against a Rs 10,000/- loss. The 2.2 figure is a measurement of something adjacent, taken on the same afternoon from the same 60 people, and it is not the size of this gap. The figure says only that the steeper arm is not a small effect and that the direction is always the same way round.

What do the two numbers look like on one holding?

Meera Sundaram opened a holding on 4 January with four positions of Rs 3,00,000/- each, and one of them is Kesari Logistics Limited. At the eight quarter valuation struck on 30 September, that position stood at Rs 1,95,000/-, down 35.0 per cent from its cost. She has had it for the whole period, and she still has it.

The pair is then put to her. Today, what is the least she would take to give up the Kesari Logistics Limited position? And separately: if she did not hold it at all, what is the most she would hand over this afternoon to acquire exactly that position? Both questions are about the same position, in the same market, on the same date, with the same information behind them. On the invented scale used here, at six months held the accept to pay ratio is 1.5, so the first answer is Rs 4,50,000/- and the second is Rs 3,00,000/-, a gap of Rs 1,50,000/-. Pushing the holding period to twelve months takes the ratio to 2.0, the first answer becomes Rs 6,00,000/- and the gap becomes Rs 3,00,000/-. The gap has then reached the entire original cost of the position.

Two things about those figures need saying plainly. Either one misread makes the illustration useless. The pay figure of Rs 3,00,000/- is fixed at the original cost as the anchor of this illustration; it is not the 30 September valuation of Rs 1,95,000/-, and it is not a claim about how much the position is worth to anybody. The Palash decision log records neither number. A decision log records decisions rather than valuations. Nothing about Kesari Logistics Limited differs between the two questions, so the gap is not a fact about Kesari Logistics Limited at all.

Endowment Effect vs Sunk Cost Fallacy: which one is actually operating?

The endowment effect and the sunk cost fallacy get merged constantly, in classrooms and in review meetings, and the merger destroys both. So define each one properly first, on its own terms, before putting them anywhere near each other.

The endowment effect, as set out here, is about present possession changing a valuation. Somebody has a thing. Asked what they would take for it, they name a number above what they would have paid for the identical thing. The mechanism runs through the reference point: holding it has moved the level they measure from, so releasing it is measured as a loss. Nothing in that account mentions money, and nothing in it mentions the past.

The sunk costMoney already spent that cannot be recovered whatever is decided next. fallacy is about a past outlay changing a forward looking decision. Somebody has already spent money that cannot be got back, and that unrecoverable spending pushes them into continuing rather than stopping. Stopping would make the spending feel wasted. Arkes and Blumer set this out in Organizational Behavior and Human Decision Processes in 1985. The mechanism runs through the spending: no outlay, no fallacy. Nothing in that account mentions holding a thing, and nothing in it requires that anything be valued at all.

Now the contrast, and it comes down to one test each. Give somebody a thing for nothing at all, an unasked for gift or a sample handed over at a counter, and ask the pair of questions. The gap opens anyway. Not a single rupee has been spent, so there is no unrecoverable outlay to point at and no wasted spending to feel bad about, and the sunk cost account has nothing whatever to work with. A holding acquired for nothing still opens the gap, and no sharper proof exists that the endowment effect is not the sunk cost fallacy in different clothes.

The test run the other way separates them again from the opposite side. Somebody who spent a great deal on a course, then abandoned it, then finds themselves arguing for signing up to the follow on because of everything already put in, is reasoning from sunk cost with nothing in hand at all. There is no possession, nothing is being valued, and there is no accept number and no pay number anywhere in the situation. One of these effects is about a past outlay and the other is about present possession, and an account that merges them can explain neither.

What do the two have in common, and where exactly do they part?

The two do share something real, and the shared part is why the confusion is so persistent. Both are departures from the same rule: that a decision taken today should depend only on what happens from today onwards. Sunk cost breaks that rule by letting a past payment into a forward looking choice. The endowment effect breaks it by letting a past event, namely the moment of taking the thing on, into a valuation that ought to depend only on the thing. Both are the past reaching forwards, and both were once explained away as carelessness before it turned out that each one repeats in the same direction every time it is looked for.

Where they part is the two conditions, and the clean way to see it is to stop treating them as a single sliding scale and lay them out as two separate questions. Was money spent that cannot be recovered? Is the thing in hand right now? The two questions cross, and crossing them gives four cases rather than one blurred one. Two of those four cases contain exactly one of the effects, and that is what makes them separately diagnosable rather than two words for one muddle.

What happens when the gap is wider than the two sides are apart?

So far the gap has been a curiosity about one person's numbers. An exchange needs the least one side will take to be below the most the other side will hand over. The gap becomes a real cost the moment two people are involved. If the holder will take nothing under Rs 4,50,000/- and the other side will go no higher than Rs 3,00,000/-, then every price between them is refused by one of them and there is no price outside them that either will consider. The exchange does not happen. Turn the two numbers the other way round. With the most a buyer would hand over sitting above the least the holder would take, every price between them works for both of them and the exchange goes through.

A gap that wide is a trade barrierA gap between the two numbers wide enough that no price exists which both sides will accept, so nothing is exchanged., and it has a nasty property: it leaves no trace. A completed sale produces a record with a price on it. A sale that never happens produces nothing at all, so nobody can count the exchanges the gap prevented, and neither side learns anything from it. The cost of a wide gap is an exchange that never takes place. With no transaction to record, no record anywhere shows it.

The trade barrier happens in plain view. A household tries to sell a scooter that has sat unused for two years, turns down three offers as insulting, and still has it. A shopkeeper will not clear old stock at the only price anybody offers. A brother and sister cannot agree on a share of a house that neither of them has lived in since school. In every case both sides walk away certain that the other one was being unreasonable, and in every case the thing sits where it was.

When is valuing what is held more simply correct?

The temptation is to treat every higher number as an error, and that is wrong often enough to matter. Somebody who has had a thing for two years knows things about it that no buyer can know. The owner knows it starts on the first pull in December. The owner knows the third gear is stiff. The owner knows which of its uses turned out to matter and which never came up. The owner's knowledge is real information, it has been paid for in time and attention, and it belongs inside the number.

The same holds in the other direction, and it is why buyers discount. A buyer who cannot inspect a thing properly should offer less. The range of things it might turn out to be includes some bad ones. A holder who knows exactly what it is has no reason to accept that discount. Some of the gap between the two numbers is nothing more mysterious than the two sides knowing different amounts, and correcting for it would be a mistake rather than a discipline.

The endowment effect is what is left over after every scrap of genuine private knowledge has been accounted for, and that is exactly why the controlled design matters so much. Hand out an ordinary object that nobody has had time to learn anything about, to people picked from the same room, and there is no private knowledge left to explain anything. The gap opens anyway. The residue, and not the whole of the difference between a holder's number and a buyer's number, is what the effect names.

How does somebody deciding for others actually use this?

Devika Rao, the adviser at the invented Palash Advisory Services Private Limited, cannot measure this effect on a client and does not try. She can convert a feeling into a pair of numbers, and that is a much smaller and much more useful job. When a client says a holding is worth keeping, she asks the two questions in the order that makes them hard to blend: first what is the least the client would take for it today, and then, separately, if the client held none of it this morning, what is the most they would put into it this afternoon.

The value of asking both is that the second question is nearly impossible to answer defensively. A client who says they would take nothing under Rs 4,50,000/- and then, asked cold, would not put more than Rs 3,00,000/- into the identical position has produced a Rs 1,50,000/- number out of their own mouth, and it is now something to look at rather than something to argue about. The check does not tell anybody what to do; it converts a conviction into a figure that can be examined, and that is the entire contribution.

For somebody deciding alone, with no adviser and no committee, the same two questions work. Nothing in ordinary life prompts the second one, so it has to be asked on purpose. Whatever any of this means for a duty of suitability or disclosure is a matter for the Securities and Exchange Board of India at sebi.gov.in.

What does the endowment effect not explain?

A named effect earns its keep by being narrow, and this one is narrower than its reputation. The endowment effect explains why one person gives two different answers about one thing depending on whether they have it. The effect explains that and nothing else, and three neighbouring questions get handed to it constantly that it cannot answer.

The effect does not say what somebody will sell and when. Selling through time is a question about behaviour, and it needs a record of decisions rather than a pair of valuations. The effect does not measure how much heavier a loss feels than a gain. The gap between an accept number and a pay number mixes that weighting together with everything else the two questions differ on. And it says nothing about somebody who simply does nothing. Doing nothing is not a valuation at all and needs its own account; Samuelson and Zeckhauser gave that one its name in the Journal of Risk and Uncertainty in 1988. The effect explains one thing well, and the discipline of refusing the three questions next to it is what keeps the explanation worth having.

Meera Sundaram, Devika Rao, Palash Advisory Services Private Limited, the Palash decision log, the Palash cohort measurement and Kesari Logistics Limited are invented.
Educational material. Not advice on any investment, tax, budget or market position.