Foundations

Proceedings of the 1987 Santa Fe Institute workshop that first brought physicists (Anderson) and economists (Arrow) together to treat the economy as an adaptive nonlinear system. Widely regarded as the founding document of complexity economics, it framed the agenda of out-of-equilibrium dynamics, increasing returns, and emergent aggregate behavior that the field has pursued since.

Arthur's compact manifesto arguing that the economy is an ongoing computation by agents who form expectations about outcomes their own expectations create. It crystallized the contrast between equilibrium economics and an economics of process, and remains the standard short statement of the complexity-economics viewpoint.

The most influential book-length synthesis of complexity economics for a broad audience, recasting wealth creation as an evolutionary process of differentiation, selection, and amplification among business plans and technologies. It gave the field its name recognition outside academia and organized its critique of Traditional Economics into a coherent alternative research program.

Written in the aftermath of the 2008 crisis, this commentary argued that DSGE macroeconomics failed because it assumes equilibrium and representative agents, and called for large-scale agent-based simulation of the economy. It became the rallying citation for the post-crisis ABM research program in macroeconomics and policy institutions.

The field's most authoritative recent self-definition: Arthur surveys three decades of work and argues that complexity economics is economics done out of equilibrium, with algorithmic rather than purely mathematical foundations. Serves as the canonical entry point and consolidates the field's intellectual history, methods, and open problems.

Used the survival of the QWERTY keyboard to argue that small historical events can lock economies onto persistent, possibly inefficient paths. The founding empirical parable of path dependence, it made 'history matters' a precise economic claim and provoked a literature-defining debate over lock-in and market efficiency.

Formalized how technologies competing under increasing returns to adoption generate multiple equilibria, path dependence, and potential lock-in to inferior outcomes, using nonlinear Polya urn processes. The theoretical cornerstone of positive-feedback economics and the analytical companion to David's QWERTY narrative.

Introduced the El Farol bar problem to show that when deductive rationality is self-referentially impossible, agents must reason inductively with an evolving ecology of predictive hypotheses. The paper became the template for modeling expectation formation in complexity economics and spawned the minority-game literature in physics.

Collects Arthur's papers on self-reinforcing mechanisms, adoption externalities, and industry location, establishing the mathematics of positive feedback in economics in one volume. The standard reference for how increasing returns produce unpredictability, inflexibility, and non-ergodicity in economic outcomes.

Modeled binary choice with stochastic recruitment, inspired by ant foraging experiments, to show how herding among interacting agents produces persistent aggregate switching that no representative agent could exhibit. A foundational demonstration that aggregate outcomes reflect interaction structure rather than individual optimization, widely applied to opinion dynamics and financial markets.

Demonstrated that fat-tailed returns and clustered volatility emerge endogenously from interactions between noise traders and fundamentalists, without fat-tailed exogenous shocks. The landmark proof-of-concept that agent-based models can reproduce the statistical regularities of real markets from behavioral heterogeneity alone.

The definitional statement of agent-based computational economics (ACE): economies modeled as open-ended dynamic systems of interacting agents, grown from initial conditions rather than solved for equilibrium. Codified the methodology, vocabulary, and research objectives that organized the ABM community for the following two decades.

The authoritative survey of heterogeneous agent models, showing how interacting fundamentalists and chartists with adaptively switching beliefs generate bubbles, crashes, and the stylized facts of financial data. Bridged analytically tractable behavioral models and full agent-based simulation, defining the HAM research program.

First demonstration that a data-driven agent-based model of a national economy (Austria) can match or beat DSGE and VAR benchmarks in out-of-sample macroeconomic forecasting. Answered the long-standing objection that ABMs cannot be taken to data, opening their use in central-bank forecasting practice.

The field's comprehensive stocktaking in the mainstream's flagship survey journal, covering forty years of ABM across markets, macroeconomics, and finance, and assessing computational advances from large-scale microdata to machine learning. Its publication in the JEL marks the method's arrival in canonical economics and sets the agenda for its next phase.

Developed an analytical model of contagion on interbank networks showing a robust-yet-fragile property: greater connectivity reduces the probability of contagion but increases its severity when it occurs. Became the benchmark framework for central-bank analysis of interbank stress and the reference point for the financial-networks literature after 2008.

Proved that idiosyncratic shocks to individual firms or sectors need not average out: in asymmetric input-output networks they propagate into aggregate fluctuations, with volatility decaying far more slowly than the law of large numbers suggests. Founded the granular-networks literature in macroeconomics and made production-network structure a first-order concern of business-cycle theory.

Modeled cascades of failures among organizations linked by cross-holdings, showing non-monotonic effects of integration and diversification on contagion. The mainstream economic-theory treatment of financial network fragility, complementing the physics-style cascade models with a general-equilibrium foundation.

A collective statement by leading complexity scientists and economists arguing that systemic financial risk is an emergent network phenomenon requiring tools from epidemiology and ecology: stress tests on networks, early-warning indicators, and macroprudential design. Canonical as the policy-facing synthesis of the complexity approach to financial stability.

The definitive review of statistical-physics approaches to financial networks: network reconstruction from partial data, DebtRank-style stress propagation, and systemic-risk metrics. Consolidates two decades of econophysics work on interconnected balance sheets into the standard reference for the subfield.

The founding textbook of econophysics, applying scaling analysis, Levy-stable processes, and correlation-matrix methods from statistical physics to financial time series. Established the empirical program of treating markets as complex systems with measurable universal properties and trained the first generation of physicists working on finance.

Using census microdata on all U.S. firms, established that firm sizes follow a Zipf (power-law) distribution across the entire size range, robust to measurement choices. A benchmark empirical regularity that any theory of firm growth must reproduce, and a touchstone result for scaling approaches to economic organization.

Codified the stylized facts of financial returns — heavy tails, volatility clustering, absence of linear autocorrelation, aggregational Gaussianity — into a canonical checklist. These regularities became the empirical targets against which agent-based and econophysics models of markets are validated.

The canonical survey of power laws in economics — city sizes, firm sizes, income and wealth, trading volume, and returns — and of the random-growth and optimization mechanisms that generate them. Gave scaling phenomena a unified theoretical treatment inside mainstream economics and remains the standard starting point for the topic.

Mapped the product space — the network of relatedness between traded products — and showed that countries diversify by moving to nearby products, so a country's position in the network conditions its development path. Made structural transformation a network phenomenon and provided the empirical substrate for economic complexity theory.

Introduced the Economic Complexity Index, inferring the unobserved capabilities of economies from the network structure of what they export, and showed it predicts future growth. Founded the economic-complexity research program and its policy apparatus (the Atlas of Economic Complexity) now used in development economics worldwide.

Placed Wright's and Moore's laws on rigorous statistical footing by building distributional forecasts of technology costs, validated out-of-sample on decades of data across dozens of technologies. Established that technological progress is forecastable with quantifiable uncertainty, the foundation for subsequent energy-transition cost projections.

Applied probabilistic experience-curve forecasting to solar, wind, batteries, and electrolyzers to show that a rapid green energy transition is likely cheaper than continued fossil dependence — inverting the standard cost framing of climate policy. The most consequential policy application of the technology-forecasting program in complexity economics.

The Santa Fe artificial stock market: agents co-evolve expectational rules by induction, and the market self-organizes into either a rational-expectations regime or a rich complex regime with technical trading, bubbles, and realistic return statistics. The archetype of computational finance experiments and the direct ancestor of modern market ABMs.

Introduced the adaptive belief system in which traders switch between costly fundamentalist and cheap chartist predictors according to realized profits, generating bifurcation routes to chaotic asset-price dynamics. The analytically tractable core model of heterogeneous-belief finance, underpinning much of the subsequent HAM and market-ecology literature.

Proposed treating trading strategies as species in a market ecology, with capital as population, deriving market impact, price dynamics, and the analogue of trophic interactions among strategies. The founding paper of market ecology, which reframes market efficiency as an evolving ecological balance rather than a fixed-point condition.

The definitive treatment of market microstructure from the econophysics tradition: order-book dynamics, the square-root law of market impact, and the statistical mechanics of liquidity. Synthesizes two decades of empirical high-frequency research into the standard reference on how prices actually form.

The K+S (Keynes-meets-Schumpeter) model: an evolutionary agent-based macroeconomy in which Schumpeterian innovation drives growth while Keynesian demand dynamics generate cycles and crises, reproducing a long list of micro and macro stylized facts. The workhorse of the European evolutionary-ABM school and the platform for a large family of fiscal, monetary, and climate policy studies.

A central banker and a theoretical ecologist applied lessons from ecosystem stability — modularity, diversity, and the fragility of highly connected systems — to interbank networks and derivatives markets. Canonical for legitimizing complexity-science reasoning inside financial regulation, directly influencing post-crisis macroprudential design.

Built a data-calibrated agent-based model of the UK housing market — first-time buyers, owner-occupiers, buy-to-let investors, and banks — to evaluate loan-to-income caps and other macroprudential tools. A landmark of ABM adoption inside a major central bank, demonstrating the method's use for policy experiments infeasible in equilibrium models.

Farmer's synthesis of fifty years of complexity science applied to economics, arguing that simulation-based complexity economics can forecast crises, guide the energy transition, and underpin better policy. The field's current programmatic statement for both scientific and general audiences, consolidating its claims to practical relevance.