Category: Effects
Type: Labeling money by source, use, and closing time
Origin: Amos Tversky and Daniel Kahneman, Science (1981); Richard H. Thaler, Marketing Science (1985)
Also known as: Psychological account (1980); psychological accounting (1981)
Type: Labeling money by source, use, and closing time
Origin: Amos Tversky and Daniel Kahneman, Science (1981); Richard H. Thaler, Marketing Science (1985)
Also known as: Psychological account (1980); psychological accounting (1981)
Quick Answer — Mental Accounting is the set of operations that label money by where it came from or what it is for, judge a payment inside that label, and close the label on a schedule. In Tversky and Kahneman’s 1981 ticket questions, 88 percent of 183 people would still buy a $10 ticket after losing a $10 bill, and 46 percent of 200 would buy a second ticket after losing the first.
What is Mental Accounting?
Mental Accounting is the set of operations people use to label money by its source or its job, to judge a payment inside that label, and to close the label on a schedule, so that a dollar in one label is not treated as a substitute for a dollar in another. A paycheck and a birthday bill can be the same ten dollars at the bank. The label decides which purchase is allowed to touch them. Picture two jars on a kitchen shelf, one marked rent and one marked dinner. Moving a bill from one jar to the other is easy with cash. The label makes that move feel like a different act.Mental accounting is the set of cognitive operations used by individuals and households to organize, evaluate, and keep track of financial activities.That sentence opens Richard H. Thaler’s 1999 review in the Journal of Behavioral Decision Making. Economists call the rule “a dollar is a dollar” fungibility. Loss Aversion is the extra weight of a loss inside whatever label is open. Mental Accounting chooses the label, the price the deal is compared with, and the day the books close.
Mental Accounting in 3 Depths
- Beginner: A $10 bill lost on the way into the theater, and a $10 ticket lost at the door, are the same wealth change with two labels. Write down which label the $10 hits before you buy again.
- Practitioner: Before a trip to save $5, write the $5 beside the price it comes off, and name the category the purchase will deplete. In the 1981 calculator questions, 68 percent would drive 20 minutes to save $5 on a $15 calculator, and 29 percent would drive to save $5 on a $125 calculator.
- Advanced: The same inflexible label can keep rent money out of a night out and can also shut a category too early. Chip Heath and Jack B. Soll measured the second side in 1996, on stated spending. A later online sample still leaned the way Thaler’s value-curve rules lean, while its largest cell on two separate gains was “no difference.”
Origin
The name arrives in three steps, and the first counted results belong to the ticket and the calculator. Thaler’s 1980 paper, “Toward a Positive Theory of Consumer Choice,” Journal of Economic Behavior & Organization, volume 1, pages 39–60, is the paper he later points to for the earlier name. In the 1999 review he writes that Tversky and Kahneman used “psychological account” in 1981, following the terminology of that 1980 paper, and that Kahneman and Tversky in 1984 suggested “mental account.” This page follows that note. The words “mental accounting” are the 1984 and 1985 name. Amos Tversky and Daniel Kahneman, “The Framing of Decisions and the Psychology of Choice,” Science, volume 211, 30 January 1981, pages 453–458, print the ticket counts under the older name. Problem 8, N = 183: you have decided to see a play where admission is $10. As you enter, you find you have lost a $10 bill. Would you still pay $10 for a ticket? Yes, 88 percent. No, 12 percent. Problem 9, N = 200: you have paid $10, and at the door you find you have lost the ticket. The seat was not marked and the ticket cannot be recovered. Would you pay $10 for another ticket? Yes, 46 percent. No, 54 percent. They say the new ticket is entered in the account opened by the first ticket, so the show is entered as $20, while the lost bill is not linked to the ticket.Buying a second ticket increases the cost of seeing the play to a level that many respondents apparently find unacceptable. In contrast, the loss of the cash is not posted to the account of the play.Kahneman and Tversky repeat those two sentences in “Choices, Values, and Frames,” American Psychologist, volume 39, number 4, April 1984, pages 341–350. In that reprint the problem numbers are swapped. Problem 8 is the lost ticket, N = 200, 46 percent yes. Problem 9 is the lost bill, N = 183, 88 percent yes. The counts match the 1981 counts. The reprint is the same pair of questions, not a second sample. They write that their analysis owes a large debt to Thaler (1980, in press). The in-press item is cited there as a forthcoming Journal of Marketing paper. The model that appeared is Thaler, “Mental Accounting and Consumer Choice,” Marketing Science, volume 4, number 3, Summer 1985, pages 199–214. The 1981 calculator question is their Problem 10. One group, N = 93, received the prices in parentheses: a jacket for $125 and a calculator for $15, on sale for $10 at a branch 20 minutes away. The other group, N = 88, received the prices in brackets: a jacket for $15 and a calculator for $125, on sale for $120 at that branch. The paper says 68 percent would make the trip to save $5 on a $15 calculator, and 29 percent would make it to save $5 on a $125 calculator. It does not print the clause “68 percent of the 93.” The 1984 reprint says 68 percent of 88 respondents would drive to save $5 on a $15 calculator, and 29 percent of 93 would drive to save $5 on a $125 calculator. Mengfei Li and Gilad Feldman, in a 2025 table of the original samples, list 93 people with the $15 calculator and 88 with the $125 calculator. The two published assignments of those counts disagree. This page keeps the shares 68 and 29, and the sizes 88 and 93, and it keeps the disagreement in view. Buying both items at the first store costs $140 in either wording. The trip saves $5. They say people judge that $5 inside an account that contains the calculator and leaves the jacket out. Thaler’s 1985 model has three moves. The prospect-theory value function, from Kahneman and Tversky, Econometrica, 1979, replaces a utility of total wealth, and it is applied to gains and losses that can be split or joined. Price enters through a reference price, which yields transaction utility. Fungibility is relaxed, so a category can have its own budget. The 1999 review states the same three components in household language: how outcomes are perceived and then evaluated, which activities are assigned to which accounts, and how often those accounts are balanced. For the third, he uses the term choice bracketing from Read, Loewenstein, and Rabin (1998). Accounts can be balanced daily, weekly, or yearly, and they can be drawn narrowly or broadly.
Key Points
Four operations do the work. A cost is posted to a topic. A price is compared with the price that seller is expected to charge. A typical purchase is tracked against its category. Two outcomes are felt as one event or as two.1
Post the cost to a topic
The lost bill and the lost ticket each remove $10 before the play. In 1981, 88 percent of 183 people would still buy the ticket after the bill, and 46 percent of 200 would buy a replacement after the ticket. Framing Effect is two wordings of one option. Here the wealth change matches, and the topic of the account changes. The second ticket is posted to the play. The lost bill is not.
2
Compare the price with the seller's reference price
Thaler separates acquisition utility, the good against the price, from transaction utility, the price against the reference price for that seller. In the 1985 beach question, regular beer drinkers in an executive development program stated the most a friend should pay. The median was $2.65 when the beer would come from a fancy resort hotel, and $1.50 when it would come from a small, run-down grocery store. The beer, the beach, and the drinking are the same, there is no bargaining, and the buyer does not consume the shop’s atmosphere. The paper does not print how many of those drinkers answered. The 1999 review calls the two medians 1984 dollars. A later line in that review, a drinker who would pay $4 at a resort and $2 at a grocery and who faces a grocery price of $2.50, is an illustration. It is not a third pair of medians. Anchoring Effect is a later number pulled toward a first number. The reference price here is the price the buyer expects, or calls fair, for that seller.
3
Let a typical purchase deplete its own category
Heath and Soll, Journal of Consumer Research, volume 23, pages 40–52 (June 1996), asked 29 M.B.A. students, paid $7 for a session of about 30 minutes, how much they spend per week on entertainment and food and per month on clothing. Each person then restated the remainder of that week or month after an event. Underconsumption is the cut that remains after two benchmarks are removed: a gift of the same item, and the average of four unexpected $20 costs outside the category (replacing a dropped dish, a flu shot, a parking ticket, and buying a $20 gift for a friend’s birthday). A separate group of 26 students rated typicality from 1 to 7. A mark of “not a member” was entered as zero in the mean. The sports ticket’s entertainment mean was 5.88, and 25 of the 26 rated it at least somewhat typical of entertainment. After buying that ticket, 69 percent of the 29 showed underconsumption in entertainment, 17 percent in food, and 17 percent in clothes. The 69 is a share of people, not a share of a budget. The 29 and the 26 are different groups. A summary that says 209 students is the three studies together.
4
Split or join two outcomes along the value curve
Because the value of gains flattens as the gain grows, two gains feel larger when kept apart. Because the pain of losses flattens as the loss grows, two losses feel worse apart than joined. A loss cancelled against a larger gain is felt as the net gain. A small gain kept off a large loss is what Thaler calls the silver lining. He asked 87 students in a Cornell undergraduate statistics class which frame felt better. Two lottery wins, $50 and $25, against one win of $75: 56 said the two wins were happier, 16 said the single win, 15 said no difference. Two tax bills, $100 and $50, against one bill of $150: 66 said the two bills were more upsetting, 14 said the one bill, 7 said no difference. A $100 lottery win plus an $80 rug repair, against a $20 win: 22 said the mixed pair was happier, 61 said the $20, 4 said no difference. A $200 car repair plus a $25 pool win, against a $175 repair: 19 said the repair-plus-win was more upsetting, 63 said the $175 repair, 5 said no difference. Each row sums to 87.
Applications
Use the label as a line you can read. Keep each figure inside the question that produced it.Write the topic beside a second payment
Before you replace a lost ticket, write whether the new $10 posts to the play or to cash, and read the play’s total. In the 1981 questions the play-posting group, 200 people, said yes 46 percent of the time. The cash-posting group, 183 people, said yes 88 percent of the time.
Write the seller beside the thirst
If the most you will pay moves from $1.50 to $2.65 because the seller changed and the beer did not, the gap is the deal. Ask what number you would state if the seller line were blank. Those two medians are stated maxima from regular beer drinkers. The head count is not in the paper.
Check the other category after a typical buy
A sports ticket is a typical entertainment expense in Heath and Soll’s rating group. Among the 29 students, buying it was followed by underconsumption in entertainment for 69 percent, and in food and in clothes for 17 percent. Before you cancel a dinner because the ticket landed, write that the ticket did not buy the dinner.
Add a rebate back onto the bill
Thaler’s silver lining keeps a small gain apart from a large loss. On an invoice, add the rebate to the price and read one number, then decide. In the Cornell class, 63 of 87 said a person with a $175 repair was more upset than a person with a $200 repair and a separate $25 win. That count is a judgment of who is more upset. It is not a store receipt.
Case Study
In the 30 January 1981 issue of Science, Tversky and Kahneman gave two groups a night at the theater. One group of 183 had decided to see a play with a $10 admission and, at the door, discovered a lost $10 bill. One group of 200 had already paid $10 and, at the door, discovered that the ticket was lost, the seat was not marked, and the ticket could not be recovered. Each group was asked whether they would pay $10 now. The bill group said yes 88 percent of the time and no 12 percent. The ticket group said yes 46 percent of the time and no 54 percent. The authors’ reading is topical. The replacement ticket is posted to the account the first ticket opened, so the evening is entered as $20. The lost bill is not posted to that account. It changes the decision only by leaving the person slightly less affluent. The 1984 reprint in American Psychologist uses the same counts and swaps the problem numbers. It is the same pair of questions. The lesson is the label. Ten dollars left the wallet either way. The play’s account absorbed the second ticket and did not absorb the bill. The boundary is in the same 1984 paper. When the same people saw both versions, willingness to replace the lost ticket rose if the lost-bill question came first. Willingness to buy after losing the bill did not shift when the ticket question came first. They print no percentage for that rise. They also write that paying twice may be genuinely more annoying than losing $10 in cash. If that annoyance counts as part of the outcome, the two questions are not only a relabeling of one wealth change.Boundaries and Failure Modes
Mental Accounting is the wrong name when a second feeling is part of the outcome, when a classroom majority is carried into a later room as if the cells had stayed put, and when a stated budget is treated as a year of receipts. The first boundary is the ticket itself. Kahneman and Tversky write that it may be more pleasurable to save $5 on a $15 purchase than on a larger one, and more annoying to pay twice for one ticket than to lose $10. They treat those feelings as possibly legitimate. The within-subject order changes the lost-ticket answer and leaves the lost-bill answer in place, and the size of that change is not printed. [Sunk Cost Fallacy](/effects/sunk-cost-fallacy) is the choice to continue because the money is already spent, as when a prepaid seat pulls someone through a storm. The ticket question on this page is about where a new $10 is posted. The second boundary is the 1985 classroom against the 2025 sample. Li and Feldman, Department of Psychology, University of Hong Kong, Royal Society Open Science, volume 12, number 9, article 250979 (17 September 2025), repeated problems reviewed in Thaler’s 1999 paper. The sample is a US online group recruited on Mechanical Turk through CloudResearch, 1,007 people in all, about 500 on each problem. Ages for the 1,007: median 40, mean 43.28, standard deviation 12.61, range 20 to 80. On the two-gain item, $50 and $25 against one win of $75, n = 505: 178 people, 35 percent, said the two wins were happier; 62, 12 percent, said the single win; 265, 52 percent, said no difference. On the two-loss item, $100 and $50 against one bill of $150, n = 505: 213, 42 percent, said the two bills were more upsetting; 69, 14 percent, said the one bill; 223, 44 percent, said no difference. In the Cornell class those modal cells were the split frame, 56 of 87 and 66 of 87. In the later sample the largest cell on both items is “no difference.” The authors still report support for the four rules, with Cramer’s V of 0.26 (interval 0.20 to 0.33) for segregating gains, 0.23 (0.17 to 0.29) for integrating losses, 0.43 (0.37 to 0.49) for cancelling a smaller loss against a larger gain, and 0.31 (0.25 to 0.37) for keeping a small gain off a large loss. They count this problem among 11 of 17 with support, and they report mixed support for 3 and no support for 3. The direction among people who pick a side is still the 1985 direction. The majority sentence stays with the Cornell class. The third boundary is what a budget study measures. Study 1 of Heath and Soll is what 29 M.B.A. students said they would spend next, after the gift benchmark and the four outside costs were removed. It is not a week of card receipts, and it is not a household’s year. Thaler treats a category rule as a second-best way to keep spending from running on. He does not print a rate at which that rule saves a household. The same inflexibility is the protection and the early closing. Denomination Effect is the physical unit in the hand, a large bill against several small ones. The label on this page is the source or the job, whatever the unit looks like. A gain already in hand that makes the next risk feel cheap, and a reluctance to close a falling investment, are different questions from which jar a new dollar enters. This page does not use the ticket, the beer, or the Cornell rows to stand for either question.Common Misconceptions
Three stories travel with the ticket and the beer. The name is dated to the wrong paper. The silver lining is folded the wrong way. A stated median or a share of a classroom is carried out as a price or a budget.Thaler named it in 1980 and ran the ticket study
Thaler named it in 1980 and ran the ticket study
Thaler’s 1999 note puts “psychological account” in his 1980 paper and in Tversky and Kahneman’s 1981 paper, and “mental account” in Kahneman and Tversky’s 1984 paper. The ticket counts, 88 percent of 183 and 46 percent of 200, are those 1981 questions. The 1984 reprint swaps the problem numbers and keeps the counts. The 1985 Marketing Science paper is the model: the value curve, transaction utility, and budgets.
A silver lining folds a small loss into a larger gain
A silver lining folds a small loss into a larger gain
Folding a smaller loss into a larger gain is Thaler’s cancellation, the mixed gain. In the Cornell class, 61 of 87 said a person who simply won $20 was happier than a person who won $100 and paid $80 for a rug. The silver lining is the other mix: a small gain kept apart from a large loss. On that item, 63 of 87 said the person with a $175 repair was more upset than the person with a $200 repair and a separate $25 win.
Related Concepts
Mental Accounting sits next to the slope inside a label, two wordings of one option, a cost already spent, the unit in the hand, and a number stuck to a first number. Each neighbor keeps the figures from its own page.Loss Aversion
A loss weighs more than a same-sized gain. Mental Accounting chooses which outcomes share the label inside which that slope is applied.
Framing Effect
Two descriptions of one option. The ticket pair keeps the wealth change and changes the topic the $10 is posted to.
Sunk Cost Fallacy
Continuing because the money is already spent. The lost-ticket question is about posting a new $10 to the play.
Denomination Effect
A large bill against several small bills of the same total. The label on this page is the source or the job.
Anchoring Effect
A later number pulled toward a first number. The beach medians move with the seller’s reference price, $2.65 against $1.50.