Atlas › The world in numbers

Case study 3 of 198

Lost wallets: does money make people more honest?

In the 40-country lost-wallet experiment, does Jev know how often wallets were returned in each country, and does it know the surprise: wallets with money came back more often than empty ones?

result84 questions

Asked what share of lost wallets were returned in each of 40 countries, Jev puts the countries in a reasonable order (rank correlation 0.68 with money in the wallet, 0.69 without) but guesses roughly half everywhere. It also misses the study's surprise: money raised returns in 95% of countries, by 11 points on average, yet Jev's estimates don't move with money, even though, asked directly, it says wallets with money came back more often (82% of its answers).

0255075100
Denmark
82%
Sweden
82%
New Zealand
80%
Switzerland
79%
Norway
79%
Czech Republic
78%
Netherlands
74%
Australia
69%
Poland
68%
Croatia
67%
Germany
65%
Canada
64%
Romania
61%
UK
61%
Russia
59%
Spain
58%
France
58%
USA
57%
Serbia
56%
Greece
54%
Israel
50%
Argentina
49%
Brazil
49%
Italy
48%
Portugal
45%
India
44%
Turkey
42%
Thailand
40%
South Africa
39%
Chile
37%
UAE
35%
Indonesia
33%
Malaysia
22%
Morocco
22%
Ghana
22%
China
22%
Kazakhstan
21%
Kenya
19%
Mexico
18%
Peru
13%

Jevpeopleother settingsshare returned, wallet with money (magenta Jev, ink observed; ticks: no money)

How to read this: One row per country, for the wallet with money: the ink diamond is the real return rate, the magenta square is Jev's estimate, and the thin ticks are the same two numbers for the wallet without money. Jev's squares bunch near the middle; the diamonds spread from under 20% to over 75%.

40 countries x 2 conditions; 90% interval on the mean error [3.86, 10.86] points.

In short

  • Jev guesses that about half of lost wallets come back everywhere, 45% to 58% by country, while the real rates ran from 7% to 82%.
  • It still orders the 40 countries moderately well (0.68 with money), placing Switzerland and Sweden above Peru and Morocco.
  • Money raised real returns in 95% of countries, yet Jev's estimate rises with money in only 18%, though asked directly it says money helps (82%).

What the data shows

the world in numbers
Jev: about half of lost wallets come back, everywhere (45-58%). Real: 7-82%.
Doge meme: Jev: about half of lost wallets come back, everywhere (45-58%). Real: 7-82%.
How funny is this meme? Jev: 3/5, funny11%234%361%44%50%
  • The order is partly right: 0.68 with money, 0.69 without. It knows Switzerland and Sweden returned more wallets than Peru or Morocco.
  • The levels aren't: every estimate sits between 45% and 58%. It overestimates honesty by 7 points on average and misses by 17. Peru's real rate with money was 13%; Jev guessed about 51%. Sweden's was 81.5%; Jev guessed 57%.
  • Money doesn't move its estimates: in the experiment money raised returns in 95% of countries, by 11 points on average. Jev's with-money estimates are, if anything, slightly lower, and higher than its no-money guess in only 18% of countries.
  • Yet it knows the finding: asked directly which wallets came back more often, it says the ones with money (82%).

What it means, and what it doesn't

Asked directly, Jev gives this study's punchline, but its country-by-country picture of people doesn't use it. Its estimates look like "about half, adjusted a little for the country's reputation", which is a guess, not knowledge of how people behave.

It doesn't mean Jev thinks people are dishonest; if anything it's slightly too optimistic. It means its sense of honesty around the world is flat, while the real world varies a lot.

Caveats

  • A middle-of-the-scale guess. Estimates that hug the middle of a 0-100% scale can earn a moderate rank correlation and a large average miss at the same time. The 0.68 says Jev has the order partly right, not that it knows the rates.
  • It may know the study. The study was covered widely in 2019. Jev knowing the headline (money helps) when asked directly, but not applying it country by country, looks like remembering a fact rather than reasoning from it.
  • Pooled across places. Each country's rate pools several cities and kinds of institutions (banks, hotels, post offices, museums). Jev was told the setting in general terms, not which city or desk.
  • Human predictions not used. The study also surveyed economists and ordinary people, who predicted that money would reduce returns. Those predictions are cited from the paper; their data weren't used here.

Jev on this experiment

Would a person find it interesting to read?
Yes84%
Does it describe you?
No58%
Would you have predicted it?
No52%
How fair is the comparison?
The comparison is reasonable
How much should a reader rely on it?
A little
Which caveat matters most?
Every estimate near 50%97%

Why ask this

In 2019 a team of economists (Cohn, Maréchal, Tannenbaum and Zünd) handed 17,303 "lost" wallets to strangers at reception desks in 355 cities across 40 countries and counted how many got returned. Return rates ranged from under 20% to over 75% by country. And in 38 of the 40 countries, a wallet with money in it was returned more often than an empty one, the opposite of what economists and ordinary people predicted.

A model that reasons from self-interest ("more money, more temptation") would make the same wrong prediction. A model that knows people would get it right.

How this was done

The people and the data

The study's own data (public, CC0), covering 17,303 wallets in 355 cities of 40 countries. A wallet counts as returned when the staff member emailed its owner. For each country the data give the share returned without money and with about US$13 in local currency, plus a larger amount (about US$94) in the US, the UK and Poland. The "people" here are the staff at banks, hotels, post offices, museums and public offices who received the wallets, and each country's rate pools several cities and kinds of institution.

What Jev was asked

The experiment described in full, per country and condition:

In a field experiment in large cities in Peru, researchers handed lost wallets to staff at reception desks of banks, hotels, post offices, museums and public offices, saying they had found it on the street and asking the staff to take care of it. Each wallet was a clear card case with business cards showing the owner's name and email address, a grocery list, a key, and about 13 US dollars in local currency. What share of the staff emailed the owner to return it?

0% · 5% · 10% · ... · 100%

(In Peru, about 13%.) Plus one direct question: which wallets were returned more often, with money or without? 84 questions in all, each asked with the options in shuffled orders and averaged.

How it was measured

Jev's estimate is the middle of its answer. It is compared with the real return rate: how well Jev orders the 40 countries (rank correlation: 1 same order, 0 no relation), the average miss in percentage points, and, for each country, whether Jev's estimate with money is higher than without, as the real rates almost always are.

Where these questions live

84 questions across 1 topic of the map. Each opens on the map with every question in it.

Countries
Lost Wallets Around the World84

Every question

All 84 questions behind this result, the telling ones first: the examples the analysis points to, then the ones where Jev misses, biggest gap first.

Jev’s own answer
    Showing 0 of 84