Fungibility
A dollar is a dollar — in theory, but not in the mind.
What it means
Fungibility is the economic principle that money is perfectly interchangeable: any unit of currency is a flawless substitute for any other, so a dollar should carry the same value and prompt the same decisions regardless of where it came from or what it is mentally earmarked for. This property is foundational to standard economic reasoning, which treats wealth as a single pool to be allocated optimally. Behavioral evidence shows that people routinely violate fungibility through mental accounting: they sort money into separate, non-substitutable categories — by source (salary versus a windfall or gift), by intended use (a 'vacation' budget versus a 'groceries' budget), or by form — and then make choices that depend on the label rather than treating all money alike. The mechanism is psychological bookkeeping that imposes meaning and constraints on money that economics says it should not have, which can lead to inconsistencies such as simultaneously holding low-yield savings and high-interest debt, or spending a 'gift card' freely while guarding identical cash. It matters because recognizing where money is and is not treated as fungible illuminates real saving, spending, and budgeting behavior, and informs the design of financial products, mental-accounting nudges, and policy such as how rebates or transfers are labeled.
How it works
In its strict economic sense, fungibility is a property of assets: two units are fungible when each substitutes perfectly for the other, so that only the quantity matters and never the identity or origin of a particular unit. Money is the canonical fungible good, and standard consumer theory leans on this heavily. A household's resources are modeled as a single budget constraint, and the source of a dollar or any label attached to it is treated as irrelevant to how it ought to be spent. An unexpected tax refund, a pay rise, a gift, and money moved out of savings are all interchangeable at the margin; each should change consumption only through its effect on total wealth, not through the story of where it came from. This assumption is not merely a modeling convenience, because it makes a sharp, testable prediction: the marginal propensity to consume out of an extra dollar should not depend on which mental or physical pocket the dollar arrived in. If two sources of income of equal size and permanence produce different spending, fungibility has failed. Much of the behavioral literature is organized around constructing situations where that comparison can be made cleanly, so that a violation cannot be explained away as an ordinary income or wealth effect.
Why it happens
The dominant explanation is mental accounting, the framework Richard Thaler developed to describe how people track and evaluate financial activity. Rather than pooling all resources, people keep an implicit set of accounts, coding income and spending into categories with their own informal budgets and rules. A sum tagged for rent, groceries, or retirement is not felt to be freely available for a spontaneous purchase, even though nothing physically prevents its use. The behavior is less a bookkeeping error than a heuristic for self-control and planning. Earmarking shields money from impulsive spending, and quarantining a windfall from hard-earned salary can license a small indulgence without guilt. The cost of these conveniences is that money stops behaving as if it were fungible. The same dollar carries different marginal value depending on the account it lands in, so choices come to depend on the label rather than on total wealth. Two mechanisms tend to be distinguished. Labeling attaches meaning to money by its source or intended use, as when a benefit named for children feels like children's money. Narrow bracketing evaluates each account in isolation instead of integrating it into the whole, which is what lets a person guard low-yield savings while carrying costly debt. Both push in the same direction: they let the framing of money, rather than its amount, drive the decision.
What the evidence shows
The cleanest field test comes from Hastings and Shapiro (2013), who studied gasoline purchases using detailed household panel data. When overall gasoline prices rose, households substituted toward lower octane grades far more than the modest reduction in their real income could justify, and far more than they adjusted quality in response to comparable changes in their broader budget. Because grade choice within the gasoline account responded strongly to gasoline prices but weakly to unrelated income changes, the pattern is difficult to reconcile with fungibility and fits a model in which drivers budget for fuel in a separate account. The same authors later examined nutrition assistance (Hastings and Shapiro, 2018). Standard reasoning holds that for most recipients such benefits should be fungible, because a household spending more on food than the benefit covers could simply reallocate. Instead they estimated a marginal propensity to spend on food out of benefits of roughly 0.5 to 0.6, against something closer to 0.1 out of equivalent cash. That gap is a large violation of fungibility and revives the older labeling literature, including Kooreman (2000), who found that spending on children's clothing responded far more strongly to the portion of income arriving as a child benefit than to other income of the same size. The record is not uniform. Milkman and Beshears (2009) found that a ten-dollar grocery credit raised spending by only about 1.59 dollars, concentrated on items shoppers did not usually buy, a directionally clear but small windfall effect. Abeler and Marklein (2017) tested labeling directly with a restaurant-voucher field experiment and laboratory tasks; they rejected full fungibility, but the estimated departures were modest and stronger in some conditions than others. Taken together, the evidence supports the existence of non-fungibility while cautioning that its size is context-dependent rather than universal.
Where it shows up
Recognizing where money is and is not treated as fungible clarifies a range of otherwise puzzling behavior. Households simultaneously hold savings earning almost nothing and revolving debt at high interest because the two sit in separate mental accounts with separate purposes. Windfalls such as bonuses, rebates, and gift cards are spent more freely, and often on more frivolous categories, than identical amounts of ordinary income. Budgeting apps and envelope systems deliberately exploit the same tendency, partitioning income into labeled buckets to make overspending in one category feel costly. The design of transfers and incentives turns on the same lever. A payment framed as belonging to a category, or delivered as a restricted voucher rather than cash, tends to be spent more on that category than an unrestricted transfer of equal value, which is why the wording and packaging of rebates, subsidies, and benefits can change how they are used. The effect cuts both ways for policy. Labeling can steer spending toward an intended purpose, but it can also lock resources into suboptimal uses, and marketers can use the looseness of windfall accounts to prompt spending that a fungible view of money would not predict.
Limits and caveats
Fungibility is violated in degree, not wholesale, and the degree varies. Some settings show behavior close to the fungible benchmark, and even the strongest field estimates leave much of spending explained by conventional income effects. Magnitudes are sensitive to the size and salience of the labeled sum, to how binding the earmark feels, and to whether the account is one a person can easily reallocate. Small windfalls tend to produce small departures. Because several key results rest on particular products, populations, and time periods, and because a few labeling effects are modest or condition-specific, claims that money is broadly non-fungible should be stated as a robust qualitative pattern rather than a fixed quantitative law. It is also worth noting that treating money as non-fungible is not simply a mistake. The same earmarking that produces the anomaly serves self-control and planning, helping people commit to saving or to a budget they might otherwise abandon. From that angle, the interesting question is less whether people violate fungibility, which they demonstrably do, than when the resulting structure helps them and when it quietly costs them, as with the saver who guards an emergency fund while paying interest on a card.
Examples
A 'gift card' burns a hole in the pocket while identical cash sits untouched in savings.
Facing a jump in fuel prices, a household downgrades from premium to regular gasoline far more than the small dent to its total income would justify, as though the money set aside for fuel were a sealed budget of its own.
A person keeps several thousand in a savings account earning almost nothing while carrying a credit-card balance at a high interest rate, because the savings is mentally reserved as an untouchable emergency fund.
A shopper who receives a ten-dollar store credit spends it more loosely, and on items she would not normally buy, than she would spend the same ten dollars in cash from her wallet.
A benefit paid into an account named for children is spent more heavily on children's goods than an unrestricted cash transfer of exactly the same amount would be.
First described in Standard economics; violations from Thaler's mental accounting.
Key references
- Thaler, R. (1985). Mental accounting and consumer choice. Marketing Science, 4(3), 199-214. doi.org/10.1287/mksc.4.3.199
- Thaler, R. H. (1999). Mental accounting matters. Journal of Behavioral Decision Making, 12(3), 183-206. doi.org/10.1002/(SICI)1099-0771(199909)12:3<183::AID-BDM318>3.0.CO;2-F
- Hastings, J. S., & Shapiro, J. M. (2013). Fungibility and consumer choice: Evidence from commodity price shocks. The Quarterly Journal of Economics, 128(4), 1449-1498. doi.org/10.1093/qje/qjt018
- Hastings, J., & Shapiro, J. M. (2018). How are SNAP benefits spent? Evidence from a retail panel. American Economic Review, 108(12), 3493-3540. doi.org/10.1257/aer.20170866
- Kooreman, P. (2000). The labeling effect of a child benefit system. American Economic Review, 90(3), 571-583. doi.org/10.1257/aer.90.3.571
- Abeler, J., & Marklein, F. (2017). Fungibility, labels, and consumption. Journal of the European Economic Association, 15(1), 99-127. doi.org/10.1093/jeea/jvw007
- Milkman, K. L., & Beshears, J. (2009). Mental accounting and small windfalls: Evidence from an online grocer. Journal of Economic Behavior & Organization, 71(2), 384-394. doi.org/10.1016/j.jebo.2009.04.007