Loading field map
Experimental economics is the branch of behavioral economics that studies how people make economic decisions by placing them in controlled situations and observing their behavior. It is distinguished from theoretical economics, which derives predictions from mathematical models, and from empirical economics based on naturally occurring data, such as tax records or supermarket scanner data. In an experiment, the researcher creates the decision environment, assigns participants to conditions, and controls the information and incentives they face. The goal is not merely to demonstrate that people behave in interesting ways, but to test, refine, and sometimes replace economic theories under conditions that allow causal inference.
The discipline’s central question is deceptively simple: under what conditions do economic theories describe actual behavior, and when do they fail? This question has several parts. One part concerns preferences: do people maximize their own material payoffs, as classical models assume, or do they care about fairness, reciprocity, and social status? Another concerns rationality: do people reason through decisions using probability and expected value, or do they rely on heuristics that produce systematic errors? A third concerns institutions: how do markets, auctions, and bargaining rules shape the outcomes that emerge from individual choices? Experimental economics thus sits at the intersection of psychology, economics, and institutional design, and its findings often carry direct implications for public policy and market design.
Experimental methods in economics emerged only slowly, in part because the discipline long treated controlled experimentation as impractical or irrelevant. The subject matter of economics—markets, firms, national economies—seemed too large and complex to be reproduced in a laboratory. Yet isolated experiments appeared as early as the mid-twentieth century. Edward Chamberlin, in the 1940s, ran classroom markets to show that competitive equilibrium did not necessarily emerge from bargaining. In the 1950s and 1960s, researchers such as Vernon Smith and Charles Plott began to run more systematic experiments on markets and resource allocation, testing whether the predictions of supply-and-demand theory held in small groups. These early efforts were initially met with skepticism; many economists doubted that results from a handful of student volunteers could say anything meaningful about real economies.
The field gained institutional footing over the following decades. By the 1980s, experimental economics had its own journals, regular conferences, and a growing set of standard methods. The turning point for wider acceptance came with a series of influential findings that challenged core assumptions. In the early 1980s, experiments on ultimatum bargaining showed that people frequently rejected offers they judged unfair, even at a cost to themselves—a result inconsistent with purely self-interested maximizing. Around the same time, experiments on public goods contributions showed that people often cooperate more than standard free-rider models predict. These results did not merely add empirical detail; they forced theorists to construct new models of social preferences and bounded rationality.
A second major development was the rise of behavioral economics proper, which imported psychological findings into economic theory. While experimental economics and behavioral economics overlap heavily, they are not identical. Experimental economics is a methodological orientation—the use of controlled experiments to answer economic questions. Behavioral economics is a substantive program—the attempt to make economic models more realistic by incorporating psychological evidence about human cognition and motivation. Many behavioral economists run experiments, but the two fields have different emphases. Experimental economics also includes experiments designed to test or refine traditional economic theories, not to overturn them. For instance, many experimental studies of auction and market mechanisms are designed to help design better institutions, not to challenge rationality assumptions.
By the 1990s and 2000s, the field expanded both methodologically and geographically. Experiments moved out of the laboratory into the field, yielding more naturally realistic settings. Researchers from continental Europe, particularly in Germany, the Netherlands, and Switzerland, contributed heavily to both experimental and behavioral economics, and experimental laboratories became standard features of economics departments worldwide. The 2002 Nobel Memorial Prize in Economic Sciences awarded to Vernon Smith (for experimental methods) and Daniel Kahneman (for behavioral economics) is often seen as marking the field’s full integration into the discipline.
The defining feature of experimental economics is the deliberate construction of a decision environment. This construction follows a set of norms that distinguish it from survey research or casual observation. The most important norm is incentive compatibility: participants’ choices should have real consequences for their payoffs, usually in cash. If a participant is asked what they would do in a hypothetical situation, their answer may be cheap talk; if they are paid according to the consequences of their choice, their behavior is assumed to be more revealing of their true preferences or decision processes. This practice is not without limitations, however. Payoffs in experiments are usually small relative to real-world stakes, and participants are often students who may not represent the broader population. Researchers respond with replications, with stakes varied across conditions, and with field experiments using non-student populations.
A second norm is free choice within a controlled environment. Participants are not told what to decide; they are given options and choose. The researcher controls the menu of options, the information available, the sequence of actions, and the rules of interaction, but does not direct the participant’s choice. This control allows causal inference: if two groups of participants face environments that differ in only one feature, any average difference in behavior between the groups can be attributed to that feature.
A third practice is pre-registration and replication. Because experimenters have considerable discretion in designing conditions and interpreting results, there is a risk of "cherry-picking" findings. Pre-registration—specifying hypotheses and analysis plans before data collection—has become increasingly common, as has the practice of direct replication by independent laboratories. These norms are not unique to experimental economics but have been embraced with particular strength due to the relatively low cost of running experiments.
The most common experimental formats include:
A crucial methodological tool is the dictator game and its relatives. In a dictator game, one participant is given a sum of money and decides how much, if any, to give to an anonymous other participant. Since the recipient cannot respond, any giving is usually interpreted as pure altruism or a preference for fairness. In the closely related ultimatum game, the recipient can reject the offer, in which case neither party receives anything. The comparison between these games isolates the role of strategic fear—the recipient’s ability to punish—from pure generosity. Together, these and other "games" form a toolkit for isolating components of social preferences.
The history of experimental economics can be organized around an ongoing tension between two ways of interpreting deviations from classical theory. The first, sometimes called the rational-choice or economic-theory testing program, treats experimental results as evidence to be explained by extending or modifying standard models. On this view, deviations from self-interest or perfect rationality are not signs of irrationality; they are evidence that the true utility function includes social concerns or that cognitive constraints are binding. This approach has produced models of social preferences, in which people care not only about their own payoffs but also about fairness relative to others, and models of bounded rationality, in which decision makers use simplified rules because the full optimization problem is too complex or costly to solve.
The second approach, often identified with behavioral economics proper, is more willing to abandon the assumption that people optimize at all. Drawing on cognitive psychology, it proposes that humans use heuristics—mental shortcuts—that work well in most daily situations but produce systematic biases in unfamiliar or abstract settings. Where a rational-choice modeler might write a utility function that includes regret or social comparison, a behavioral modeler might describe a process of anchoring, availability, or loss aversion that does not resemble optimization at all. These two programs are not mutually exclusive; many models combine features of both. But they emphasize different questions. The rational-choice program asks: how far can we stretch the optimization framework to accommodate the data? The behavioral program asks: what are the actual psychological processes, and do they produce predictions that standard models cannot?
This distinction matters for how experimental results are used. Under the rational-choice approach, a finding such as the rejection of unfair offers in the ultimatum game is taken as evidence that people have a concern for fairness, which can be modeled as a preference. Under the behavioral approach, the same finding might be attributed to emotional responses or norm-driven behavior that is not well captured by any preference ordering. The choice between these interpretations is not merely semantic; it affects predictions in other settings, such as bargaining, market entry, and labor contracts, and it affects whether policy interventions should focus on changing incentives or changing decision frames.
A third approach within experimental economics is market and auction design, sometimes called mechanism design experimentation. This tradition is less concerned with testing behavioral theories and more concerned with practical questions: given a set of goals, what rules should an auction, matching market, or regulatory system use? Experimenters in this tradition test whether different auction formats produce efficient outcomes, whether matching algorithms produce stable assignments, or whether emissions trading schemes behave as designed. This tradition has been influential in real-world policy, including the design of spectrum auctions, the allocation of school places, and the structure of electricity markets. It often treats participants as boundedly rational but does not seek a general theory of human decision making; it seeks institutional fixes that work under realistic conditions.
Several questions have sustained the field for decades, and they continue to organize its research agenda. One is the scope of social preferences. Decades of experiments show that people are not purely selfish, but the shape of social preferences is still debated. Some results suggest people care about equality; others suggest they care about efficiency; still others suggest they care about reciprocity—rewarding kind behavior and punishing unkind behavior. The current consensus, to the extent that one exists, is that behavior is context-dependent: people are more egalitarian in some settings and more self-interested in others, and the social framing of the decision matters. No single utility function captures all findings.
A second enduring question is the role of stakes and real-world relevance. Laboratory findings are often obtained with small monetary sums, and critics have argued that behavior at low stakes may not scale to high-stakes settings. Experiments that vary the stakes—sometimes by an order of magnitude—generally find that behavior is qualitatively similar, but the issue is not fully settled. Field experiments have helped by studying real decisions with substantial consequences, and they often confirm laboratory findings, but they cannot fully reproduce the laboratory’s control.
A third question concerns heterogeneity. Early experiments treated participants as a homogeneous group and reported average effects. More recent work investigates how behavior varies with gender, age, culture, socioeconomic background, personality, and even mood. Cross-cultural experiments have shown substantial variation in fairness norms and cooperation rates, but the sources of that variation—institutions, religion, economic development—remain unclear. The field has moved from asking "what do people do?" to asking "who does what, when, and why?"
The current landscape of experimental economics is characterized by several overlapping trends. One is the growth of large-scale replication efforts, following broader concerns about reproducibility in the social sciences. A second is the integration of experimental methods with neuroeconomics, which uses brain imaging and physiological measures to observe the neural correlates of economic decisions. Neuroeconomics remains a minority pursuit within the field, but it has contributed to debates about whether decisions are driven by deliberative or automatic processes. A third trend is the expansion of experiments into development economics, where randomized controlled trials (RCTs) have transformed the study of poverty, education, health, and microfinance. This application of experimental methods has been hugely influential, but it differs from laboratory experiments in relying on naturally occurring behaviors rather than induced incentives.
Despite its acceptance, experimental economics continues to face important criticisms. External validity—the concern that laboratory results do not generalize to real markets—remains a recurring challenge. The use of student participants, small stakes, and artificial decision settings raises legitimate questions about applicability. Proponents respond that the goal of an experiment is not to replicate the world but to test a theory under controlled conditions; if the theory fails where it should hold, that is informative regardless of external realism. This defense is persuasive in many cases, but it does not eliminate the need for field-based confirmation.
A more subtle criticism concerns the interpretation of experimental results as evidence about the "true" nature of human preferences. If experiments show that people reject unfair offers, does that mean real-world markets are unfair? Not necessarily. Market institutions often filter out non-selfish behavior; people may behave selfishly in a competitive market even if they care about fairness, because punishment is impossible or costly. Experimental economists are aware of this gap between individual behavior and market outcomes, and a large body of work examines exactly how institutions affect the expression of social preferences. This "aggregation" question—how individual behaviors combine to produce market outcomes—is one of the field’s most intellectually demanding and practically important topics.
Experimental economics is now a mature, permanent fixture of the economics discipline. It does not claim to replace theoretical or empirical economics; it supplies a distinctive kind of evidence—cleanly controlled, causally interpretable, and often surprising—that has forced economists to take human behavior seriously. Its findings have reshaped microeconomics, labor economics, public finance, and development economics, and its methods have been exported to political science, psychology, and sociology. The field’s greatest contribution may not be any single discovery but its demonstration that economic behavior can be studied with the same rigor as chemistry or physics—by manipulating variables, recording outcomes, and letting the data speak.