Keeping Score

Core idea

Money is fungible: $100 from a birthday gift and $100 earned at work buy exactly the same things. But psychologically, people treat money from different sources and in different contexts as if it were not fungible. This is mental accounting: the tendency to assign money to separate psychological accounts — entertainment budget, savings, gambling winnings — and apply different rules to money in different accounts.

Mental accounting has several consequential manifestations. Sunk costs (money already spent, psychologically charged to the "investment account") influence future decisions even when they are rationally irrelevant. "House money" (gambling winnings) is spent more freely than saved money because it is categorized differently. Regret about outcomes differs dramatically depending on whether the outcome was the result of action or inaction, commission or omission — because mental accounting tracks these as separate categories.

Why it matters

Sunk cost fallacy

The sunk cost effect is the most consequential application of mental accounting. Rational decision theory says past investments are irrelevant to current decisions — only future costs and benefits matter. But loss aversion creates a powerful psychological force to "close the account" by continuing to invest in failing projects to avoid locking in the loss.

• A manager who approved a project continues to fund it past the point where objective analysis would call it a failure — because stopping means acknowledging the sunk investment as a loss.
• A moviegoer who paid $20 for a ticket sits through a terrible film rather than walking out — because leaving means acknowledging the $20 as a wasted loss.
• A country continues a failing military campaign because the lives already lost create a sunk cost that stopping would seem to dishonor.

The sunk cost fallacy is decision-making in service of mental account closure rather than in service of good outcomes. The correct response to a sunk cost: ask "Would I start this project today, given where it is, if I had not invested what I have?" If no, stop.

House money effect

Money won in gambling or otherwise obtained unexpectedly is categorized in a separate mental account — "house money" — and treated as less real than earned income. People take larger risks with house money than with "hard-earned" savings, even though the economic utility of each dollar is identical.

The house money effect creates systematic patterns in gambling behavior (escalating bets after wins), investment behavior (reinvesting dividends into riskier assets), and consumer spending (bonus spending on luxuries that would not be bought from regular income).

Regret aversion and action vs. inaction

Regret is not symmetric: actions that produce bad outcomes generate more regret than equivalent inactions that produce the same bad outcome. "I took the action that caused the loss" is psychologically worse than "I failed to take the action that would have prevented the loss," even if the monetary outcome is identical.

This regret asymmetry produces omission bias — preferring inaction when outcomes are uncertain, to protect against the more intense regret associated with action. It also explains conservative professional behavior: doctors who prescribe standard treatments over experimental ones are less liable to regret if patients worsen.

Key takeaways

Mental model

Copy sourceZoom

Practical application

Designing against mental accounting errors:

• Budget structures: fixed-category budgets exploit mental accounting (people underspend in one category rather than reallocating from another even when the reallocation is obviously better). Flexible budgeting reduces this inefficiency.
• Investment policy statements: writing down investment criteria before evaluating specific decisions prevents the sunk cost and house money effects from distorting individual choices.
• Project review cadence: regular reviews that force a "would we start this now?" question counteract sunk cost persistence. The question should be answered before evaluating project progress — not after reviewing all the past investment.

Example

A pharmaceutical company has spent $300M developing a drug that has just failed Phase III trials. Two options: abandon the program (sunk cost = $300M, no further investment) or run an additional $50M study with 30% chance of finding a new application. Expected value of the additional study: 0.3 × $200M potential = $60M > $50M cost. The study should proceed — the $300M is sunk and irrelevant.

But in the meeting, discussion focuses heavily on "we've already spent $300M on this." The framing makes the abandonment option feel like locking in a $300M loss and makes the additional study feel like a chance to "make the $300M worth it." The sunk cost drives the discussion even after it is identified. The corrective: evaluate the $50M study as if the prior $300M did not exist — pure positive expected value.

Related lessons