Experimental economics

• 1. Foundations And History

  ›

  Edward Chamberlin conducted the first market experiment in 1948. He published his findings as "An Experimental Imperfect Market" in the Journal of Political Economy that same year. This work laid the groundwork for future research into how markets function under controlled conditions. Vernon Smith later built upon Chamberlin's initial efforts to establish laboratory methods as a legitimate tool in economics. Smith studied buyers and sellers who were told their values for fictitious commodities. These participants then bid or asked on these items following rules from real-world institutions like double auctions. His experiments showed prices converging toward theoretical competitive equilibrium values despite imperfect information. In 2002, Smith received the Bank of Sweden Prize alongside Daniel Kahneman for establishing this experimental approach. Charles Plott collaborated with Smith during the 1970s at California Institute of Technology. They pioneered experiments in political science while using data to inform economic design policies.

  135

• 2. Methodological Frameworks

  ›

  Experimental economists generally adhere to specific guidelines when designing studies. Researchers incentivize subjects with real monetary payoffs rather than hypothetical rewards. Full instructions must be published so other scholars can replicate the process exactly. Deception is avoided entirely within standard protocols. Specific concrete contexts are excluded to maintain focus on core mechanisms. These practices address two central critiques regarding internal and external validity concerns. Internal validity asks whether results hold true within the experiment itself. External validity questions if findings apply broadly across different settings. None of these critiques target experimental economics exclusively since they affect all empirical approaches equally. Software tools help researchers implement these frameworks efficiently. Urs Fischbacher developed z-Tree starting in 1998 as a network-based system. This program runs on multiple computers where one serves experimenters and others serve participants. The software allows variable setups defined through its imperative programming language. By February 2020, z-Tree had accumulated approximately 9460 citations according to Google Scholar records. Fischbacher received the Joachim Herz Research prize for Best research work in December 2016 due to this contribution.

  42629

• 3. Market Mechanisms And Games

  ›

  Smith studied how prices converge toward theoretical competitive equilibrium values in controlled environments. Participants acted as buyers or sellers who knew their own valuations for fictitious commodities. They competed by bidding or asking based on rules from institutions like double auctions. Results showed convergence even when conditions did not meet many assumptions of perfect competition. Auction designs such as English and Dutch formats were tested alongside centralized trading forms. Charles Plott collaborated with Smith during the 1970s at California Institute of Technology. Their joint work pioneered experiments in political science while informing economic design policies. Finance researchers now use simulation software to study international markets with varying settings. Experiments manipulate information asymmetry about bond holding values or share pricing. Stock market bubbles emerge when some agents lack sufficient data compared to others. Trading flows, information diffusion, aggregation processes, price setting mechanisms, and returns are all analyzed systematically. These studies help understand why certain market behaviors occur under different informational constraints. The goal remains observing agent behavior and resulting characteristics across diverse environments.

  81014

• 4. Social Preferences And Behavior

  ›

  Ultimatum game experiments revealed that people often sacrifice monetary rewards when offered low allocations. This behavior contradicts simple models assuming pure self-interest motives. Researchers measured deviations from self-interest across various cultures using canonical games including dictator and trust variants. Social preferences encompass altruism, spitefulness, tastes for equality, and reciprocity tendencies. Public goods scenarios test how individuals contribute voluntarily toward collective outcomes. Gift-exchange games examine reciprocal interactions between parties who cannot observe all variables directly. Fairness emerges as a constraint on profit-seeking activities within experimental settings. Daniel Kahneman, Jack Knetsch, and Richard Thaler published "Fairness as a Constraint on Profit Seeking" in 1986. Their findings highlighted entitlements existing within markets despite theoretical predictions suggesting otherwise. Cultural variations influence how much deviation occurs from standard economic assumptions. These insights challenge traditional views of rational decision-making by incorporating social dimensions into analysis frameworks.

  71117

• 5. Learning Dynamics And Models

  ›

  Until the 1990s, adaptive models like Cournot competition or fictitious play dominated learning research. Alvin Roth and Ido Erev demonstrated reinforcement learning could make useful predictions in experimental games during the mid-1990s. Their paper appeared in The American Economic Review in September 1998 titled "Predicting how people play games." Colin Camerer and Teck-Hua Ho introduced Experience Weighted Attraction (EWA) in 1999. This model incorporated both reinforcement and belief learning components mathematically. Fictitious play proved equivalent to generalized reinforcement when weights were placed appropriately on past history. Critics noted overfitting issues due to numerous parameters involved in EWA applications. Generality across different game types remained limited until self-tuning functions replaced fixed parameters. Modern researchers continue investigating learning without feedback scenarios raised by Roberto Weber. David Cooper and John Kagel explored similar strategy types while Greg Barron examined cognitive strategies used by participants. Dale Stahl characterized decision-making rules observed over time. Charles Holt studied logit learning patterns across various game configurations including multiple equilibria cases. Amaldoss applied EWA techniques successfully within marketing contexts. Rapoport, Parco, and Murphy investigated paradoxes known as centipede games using reinforcement-based adaptive models.

  91625

• 6. Computational Modeling Approaches

  ›

  Agent-based computational modeling represents a relatively recent method with experimental dimensions in economics. Scott Page described agent-based models in The New Palgrave Dictionary of Economics during 2008. Leigh Tesfatsion defined these systems as dynamic processes driven solely by interactions among computational objects called agents. Kenneth Judd contributed computationally intensive analyses featured in Handbook of Computational Economics volume two published in 2006. Agents represent social or physical entities interacting according to predefined rules set by modelers. Initial conditions determine how the system develops forward through time without external interference. Issues common to general experimental economics apply here alongside challenges unique to agent-based approaches. Validation frameworks remain under development for empirical testing purposes. Urs Fischbacher's z-Tree software supports networked experiments running on lab computers. Alternative tools include FactorWiz released in 2000, Wextor in 2002, EconPort in 2005, MIT Seaweed project launched in 2009, FRAMASI also from 2009, MWERT appearing in 2014, ConG introduced that same year, oTree arriving in 2014, CLOSE project starting in 2015, Breadboard debuting in 2016, and nodeGame released concurrently. These platforms enable researchers to simulate complex economic systems dynamically.

  192023