The Textbook Says Pocket It
Standard economic theory makes a clean prediction about a wallet full of cash: the bigger the temptation, the more likely the finder keeps it, because honesty has a price and a rational actor sells out once the price is right.
Before running their experiment, the researchers asked ordinary people and professional economists to predict the outcome, and both groups said the same thing: fuller wallets would come back less often.
That prediction is not a fringe view; it is the default setting of a discipline that models humans as self-interested agents who cheat when cheating pays, constrained only by the odds of getting caught.
The lost wallet is the purest test of that model: no contract, no interested witness, no enforcement. Just a stranger, a wallet, and a choice.
17,303 Wallets, One Script
Alain Cohn, Michel André Maréchal, David Tannenbaum, and Christian Lukas Zünd turned the thought experiment into fieldwork.
Thirteen research assistants walked into banks, hotels, museums, post offices, and police stations in 355 cities across 40 countries and delivered one script: I found this on the street around the corner, I am in a hurry, can you take care of it?
Every wallet held three business cards with a fictitious local man’s name and email address, a grocery list, and a key.
The only thing randomized was the cash: nothing, the local purchasing-power equivalent of $13.45, or $94.15.
The assistants left no contact details and asked for no receipt; the outcome was one binary event, did the owner get an email within 100 days?
In the United States, United Kingdom, and Poland, some wallets held money but no key, separating altruism from usefulness, and an audit of reported wallets found more than 98% of the cash came back.
The Results Ran the Wrong Way
Empty wallets were reported 40% of the time, wallets with $13.45 were reported 51% of the time, and wallets with $94.15 were reported 72% of the time, with every pairwise comparison clearing P < 0.0001.
The pattern held in 38 of 40 countries, and in no country did money significantly reduce reporting.
The team tested the cynical explanations first: reporting did not vary with onlookers, cameras, or lost-property law strictness, and surveys found no bigger expected finder’s fees for bigger wallets.
What remained were two motives that grow with the stakes.
Wallets with a key were reported 9.2 points more often than identical wallets without one: altruism, since the key is valuable to the owner and worthless to the finder. Respondents also said failing to return a wallet feels more like stealing as the money rises: self-image.
Nobody wants to meet the thief in the mirror.
The Arithmetic Nobody Did
Press coverage reported the percentages; the odds ratios tell the sharper story, because a wallet with $13.45 had about 1.56 times the odds of being reported as an empty one (0.51/0.49 divided by 0.40/0.60), while a wallet with $94.15 had about 3.86 times the odds (0.72/0.28 divided by 0.40/0.60), both converted directly from the paper’s own rates.
The honesty response also shows diminishing returns.
The first $13.45 of temptation bought 11 percentage points of extra reporting, about 0.82 points per dollar, while the next $80.70 bought 21 points, about 0.26 per dollar, so the marginal honesty of a dollar falls by roughly two-thirds once the wallet is already full.
That is the shape you would expect if the first dollar triggers the identity question, am I the kind of person who steals, while later dollars only raise the price of an answer already given.
The Strongest Counterargument
The strongest case against the headline starts with who was studied: front-desk employees at banks, hotels, museums, post offices, and courts, people at work, in professional roles, possibly on camera, with reputations to protect.
Cameras and onlookers did not change the results, but no regression can fix the sample: a receptionist is not a pedestrian, and honesty at a counter may not travel to the street.
The design also loads the dice toward conscience, with a transparent case, named business cards, a local-language grocery list, and a key the owner plainly needs.
Keeping this wallet means stealing from a specific person, not pocketing a windfall, so the authors’ mechanism is also the experiment’s boundary: self-image beats self-interest when the victim has a name and the theft has a mirror.
Then came the measurement fight.
An independent team replicated the experiment in China and confirmed the core result, but argued that emailing is a Western-flavored honesty measure and that collectivist cultures safeguard the wallet without writing.
The original authors replied with unusual force.
The challengers’ statistics collapsed under a multicollinearity error; 4,400 alternative models never found the cultural effect; their own survey data contradicted their safekeeping story.
The money effect survived; the national leaderboard did not.
Finally, the “economists were wrong” framing deserves a footnote: the surveys tested naive self-interest, while behavioral economics already puts self-image in the utility function.
The experiment buries the cartoon homo economicus, not the serious one, though the cartoon is what most textbooks still teach.
What We Didn’t Prove
This experiment does not prove people are honest in anonymous settings, because every wallet was handed to a staffed counter, never dropped on a sidewalk; it does not prove the pattern extends to taxes, insurance claims, or expense reports, where the victim is abstract; and it does not establish a reliable ranking of national honesty, since the email measure captures one culturally specific flavor of doing the right thing.
It does not test whether the effect survives when the finder, not a researcher, chooses the moment.
Readers should also know the paper carries a correction notice dated 11 October 2019: the findings were not withdrawn, the paper has not been retracted, and targeted searches found no PubPeer record, but the correction exists and its substance is not spelled out on the paper’s face.
The Bottom Line
More money made strangers more honest, not less, across 355 cities and 40 countries, and the honesty curve bends where the self-image story says it should. That is evidence about staffed counters, not human nature in the dark. The wallet on the street is still an open question; the wallet at the reception desk has an answer, better than the textbooks predicted.
What You Can Do
First, put your name in your wallet: the business cards were the entire return channel, and an owner who cannot be contacted cannot be helped.
Second, if you lose a wallet, report it fast and watch that email address: four in ten empty wallets made it back, and more than seven in ten full ones did, which are better odds than cynicism suggests.
Third, if you find one, know the base rate before your conscience negotiates: most recipients returned the wallet, and the fuller it was, the more likely they were to comply.
If you run a front desk, make returning frictionless, because cameras, onlookers, and lost-property law explained none of the variation, while a clear channel to the owner and an identity worth protecting explained plenty. Build for the mirror, not the camera.