Bounded self-interest
Real people care about fairness and others, not just their own payoff.
What it means
The behavioral-economics observation that humans, unlike the purely selfish agent of standard theory, have other-regarding preferences: they cooperate, reciprocate, give to strangers, and sacrifice their own resources to punish unfairness. It is the third bound — alongside bounded rationality and bounded willpower — that distinguishes real Humans from idealized Econs. Evidence comes from the ultimatum, dictator, trust, and public-goods games, where behavior departs systematically from the self-interest prediction. The concept matters because so much economic life — contracts, teamwork, tipping, charitable giving — depends on motives that pure self-interest cannot explain.
What 'bounded' claims, and does not
The word bounded is doing careful work. It does not say people are altruists or that self-interest is weak; it says self-interest is one motive among several, decisive in many settings and overridden in others. Real people still respond to prices and incentives. They simply also weigh fairness, reciprocity, and the welfare of others, sometimes at a real cost to themselves. The claim is descriptive, not moral: it predicts systematic, patterned departures from the payoff-maximizing choice, not random noise and not universal generosity. Missing this scope produces the two opposite errors, treating every choice as selfish or treating people as reliably kind. Both fit the data worse than a bounded account, in which the context decides which motive dominates a given decision.
What the evidence shows
The workhorse is the laboratory game. In the dictator game, where a proposer simply divides a sum and the other party cannot respond, a meta-analysis of over a hundred studies finds dictators give away about 28 percent of the pie on average, with roughly a third giving nothing and a clear spike at the even split. In the ultimatum game, responders routinely reject low offers, forgoing real money to deny a proposer they consider unfair. Fehr and Gachter showed people will pay to punish free-riders even when they gain nothing and never meet the person again. But magnitudes are not universal: Henrich and colleagues ran the same games across fifteen small-scale societies and found offers and rejections varied widely, tracking how market-integrated and cooperative each economy was.
Where it breaks down
The lab result is real but fragile to framing. When List added a simple option to take money rather than only to give it, generosity collapsed and many dictators took, suggesting part of measured giving is an artifact of a choice set that offered no way to stay neutral, plus a pull to do what the experimenter seems to want. Designs that hide the proposer's choice from the experimenter also cut giving sharply. None of this shows self-interest is unbounded; people still rarely grab everything. It shows the size of other-regarding behavior is highly sensitive to context, anonymity, and how the options are described, so an effect measured in one paradigm travels poorly to another, and to the field.
Why it matters for markets and policy
The pull of fairness reshapes how institutions must be modeled. Fehr and Schmidt's inequity-aversion model, and related reciprocity models, formalize the idea that people dislike unequal outcomes and will spend to correct them; these models predict wage rigidity, incomplete contracts, and cooperation that the pure self-interest baseline misses. In practice the third bound underwrites tipping, gift exchange between employers and workers, tax compliance beyond what audits compel, and consumer boycotts. It also cuts both ways: the same machinery that sustains cooperation drives costly punishment, spite, and resistance to reforms seen as unfair. A designer who assumes only goodwill is as wrong as one who assumes only greed, because the operative motive shifts with the framing of the choice.
Examples
Diners leave tips at restaurants they will never visit again, where self-interest predicts leaving nothing.
Strangers hand in lost wallets with the cash still inside, and are more likely to do so the more money is in them — the reverse of what pure self-interest predicts.
Customers who will never meet the offender still switch away from a firm caught underpaying its staff, paying more elsewhere to punish a wrong that cost them nothing.
Programmers spend unpaid evenings fixing bugs in open-source tools their competitors also rely on, sustaining shared code they could simply free-ride on, a form of cooperation pure self-interest cannot rationalize.
Most taxpayers declare income that third parties never report to the authorities, and comply more readily where they see the system as fair, behavior that audit-and-penalty models built on pure self-interest badly underpredict.
First described in Mullainathan & Thaler (2000); Fehr & Schmidt (1999).
Key references
- Engel, C. (2011). Dictator games: a meta study. Experimental Economics, 14(4), 583-610. doi.org/10.1007/s10683-011-9283-7
- List, J. A. (2007). On the interpretation of giving in dictator games. Journal of Political Economy, 115(3), 482-493. doi.org/10.1086/519249
- Fehr, E., & Gachter, S. (2002). Altruistic punishment in humans. Nature, 415(6868), 137-140. doi.org/10.1038/415137a
- Henrich, J., Boyd, R., Bowles, S., Camerer, C., Fehr, E., Gintis, H., & McElreath, R. (2001). In search of Homo economicus: Behavioral experiments in 15 small-scale societies. American Economic Review, 91(2), 73-78. doi.org/10.1257/aer.91.2.73
- Fehr, E., & Schmidt, K. M. (1999). A theory of fairness, competition, and cooperation. Quarterly Journal of Economics, 114(3), 817-868. doi.org/10.1162/003355399556151