Undergraduates at Cornell were given a coffee mug bearing the university crest, worth about six dollars in the campus shop, and told it was theirs to keep. Then they were asked the price at which they would sell it. A second group, who had not been given mugs, were asked what they would pay to buy one. A third group were asked to choose between a mug and various sums of money.
Owners wanted a median of about seven dollars. Buyers offered about three. The gap did not close through bargaining, and the number of trades that occurred was a fraction of what standard economic theory predicts.
Daniel Kahneman, Jack Knetsch and Richard Thaler ran a series of these experiments through the 1980s, publishing the definitive account in 1990, partly to answer economists who suspected the earlier results were artifacts of misunderstanding or transaction costs. The effect survived every control they added: real money, real goods, repeated markets, experienced traders in some conditions.
The explanation runs directly through the previous two chapters. Once the mug is coded as mine, my reference point has moved. Giving it up is now a loss rather than a foregone gain, and losses are weighted more heavily. The buyer, who never owned it, is evaluating an acquisition on the gain branch of the value function. Both parties are looking at the same object across a discontinuity that ownership created about ninety seconds ago.
Later work sharpened the boundaries in ways that matter for the manipulation. Legal title turns out to be unnecessary; a sense of possession is sufficient. Simply touching an object raises willingness to pay. So does imagining owning it, and so does the experience of having chosen it. There is also evidence that the effect attenuates for goods held explicitly for exchange rather than for use — the trader's inventory does not feel like the trader's mug — which is why market experience reduces but does not eliminate it.
The commercial architecture built on this is enormous and mostly invisible because it is presented as generosity. The thirty-day free trial, which is not a sample but an ownership installation, and which requires you to move in — configure the settings, import the data, invite your team — because effort deepens possession. The test drive, particularly the overnight one. The home trial for a mattress or a pair of glasses. Try before you buy, no obligation, cancel any time. The obligation is not contractual. It is that cancelling now feels like giving something up rather than declining to acquire it, and those feel different by roughly a factor of two.
The same asymmetry appears whenever a negotiation involves something already in your hands. Trade-in valuations run into it constantly: the seller's car has a history, and the buyer's car has a mileage. Default enrolment in a service works partly for this reason. So does the pre-ticked box, and so does the retention offer that lists what you are about to lose.
The countermeasure is to deliberately restore the reference point that existed before possession, and the way to do it is to convert the decision into a purchase you have not yet made. You are not deciding whether to cancel the trial; you are deciding whether, today, with no history, you would pay this price for this thing. If the answer is no, the only thing keeping you in is thirty days of manufactured ownership.
Defense: The Money Is Already Gone: Escaping Sunk-Cost Entrapment