Skip to main content

Über dieses Buch

Simulation is used in economics to solve large econometric models, for large-scale micro simulations, and to obtain numerical solutions for policy design in top-down established models. But these applications fail to take advantage of the methods offered by artificial economics (AE) through artificial intelligence and distributed computing. AE is a bottom-up and generative approach of agent-based modelling developed to get a deeper insight into the complexity of economics. AE can be viewed as a very elegant and general class of modelling techniques that generalize numerical economics, mathematical programming and micro simulation approaches. The papers presented in this book address methodological questions and applications of AE to macroeconomics, industrial organization, information and learning, market dynamics, finance and financial markets.





Chapter 1. A Potential Disadvantage of a Low Interest Rate Policy: the Instability of Banks Liquidity

This paper joins the strands of literature supporting the idea that monetary policy should consider seriously the behavior of financial institutions. In a number of recent episodes, lowering the interest rate has been ineffective in bringing economies out of unfavorable conditions. We show how a low interest rate policy could have adverse effects that oppose the positive ones for which they are adopted. In particular, banks may choose to finance more risky borrowers who have an higher average return to cancel out the negative effects triggered by a low interest rate on their revenues. This behavior could seriously endanger macroeconomic performances.

Gianfranco Giulioni

Chapter 2. Keynes in the Computer Laboratory. An Agent-Based Model with MEC, MPC, LP

The present paper aims at taking the core of Keynes’s macroeconomics - as it is portrayed in the 1937’s


paper - into the computer laboratory, in the spirit of a

counterfactual history of economic thought

. We design an agent-based model in which the principal role in determining economic dynamics is played by the three pillars of Keynesian economics, namely the Marginal Efficiency of Capital, the Marginal Propensity to Consume and the Liquidity Preference. The latter magnitudes are modelled with particular attention to their behavioural foundations. Indeed, in Keynes’s thought, such behavioural foundations result greatly important in determining the development of the business cycle. Simulation results endorse this view, with our model being able to recreate economic fluctuations with interesting statistical properties.

Giulia Canzian, Edoardo Gaffeo, Roberto Tamborini

Chapter 3. Pride and Prejudice on a Centralized Academic Labor Market

The Academic Labor Market in France can be viewed as a constrained Stable Marriage problem, pairing universities and candidates according to their (elitist) preferences. A Multi-Agent based model, calibrated after the empirical evidence, is used to investigate how universities can recruit the best candidates with high confidence. Extensive simulations suggest that universities can be divided in four categories: top and medium universities have no difficulty in attracting the candidates they have selected, contrarily to good and bad universities. In this paper, a learning mechanism is presented: universities are allowed to tune their expectations depending on whether they did succeed to attract candidates in the previous recruitment rounds. The impact of over/under estimations is analyzed with respect to the hiring efficiency and quality.

Philippe Caillou, Michele Sebag

Industrial Organization


Chapter 4. U. S. Defense Market Concentration: An Analysis of the Period 1996–2006

The defense market in the United States has undergone a significant amount of merger activity over the past 20 years. Several sources claim an increasing level of market concentration to be occurring. This paper examines several measures of the structure of the U. S. defense market from 1996–2006. Firm size is established as being Zipf distributed with exponent stable during this period. Other measures also show that significant market concentration has not resulted from these mergers. Simple computational approaches used to generate similar distributions methods do not explain this observation, suggesting that market entry conditions, firm growth rates, and diffusion of sales associated with purchased firms may be a factor in maintaining market structure.

Wayne Zandbergen

Chapter 5. Operator’s Bidding Strategies in the Liberalized Italian Power Market

This paper studies the Italian wholesale electricity market by means of a realistic agent-based computational model of the day-ahead market session, of the thermal-power production pool and of the Italian high-voltage transmission network. The aim of the paper is twofold. Firstly, it studies how the strategic behavior of the thermal power plants can influence the level of price at a national level. Secondly, it performs an empirical validation of the computational model over a period of one month which enables to assess the validity of the proposed model. In particular, three scenarios are studied and compared, i.e., the historical performance, a marginal cost based case and a strategic case where generation companies learn according to a reinforcement learning algorithm their best strategy. Results show that the strategic model reproduces real price dynamic during low- and medium- demand periods, whereas during peak-hours the strategic model tends to underestimate historical performances.

Eric Guerci, Mohammad Ali Rastegar, Silvano Cincotti

Chapter 6. Selection Processes in a Monopolistic Competition Market

In this paper, we extend the traditional evolutionary model of homogeneous product market by incorporating a particular abstraction of imperfect monopolistic competition borrowed from Dixit and Stiglitz. Specifically, we analyze a formal model of an industry in which a set of heterogeneous firms produce differentiated products; consumers have a preference for variety, and therefore firms enjoy an imperfect monopolistic position in the market. We explore the system dynamics, focusing on how selection processes operate depending on the monopolistic intensity of the market and the heterogeneity of firms.

Jose I. Santos, Ricardo del Olmo, Javier Pajares

Market dynamics and auctions

Chapter 7. Symmetric Equilibria in Double Auctions with Markdown Buyers and Markup Sellers

Zhan and Friedman (2007) study double auctions where buyers and sellers are constrained to using simple markdown and markup rules. In spite of the alleged symmetry in roles and assumptions, buyers are shown to have the upper hand both in the call market and in the continuous double auction. We replicate the study and show that their formulation of the sellers’ markup strategies, while seemingly natural, exhibits a hidden asymmetry. We introduce a symmetric set of markup strategies for the sellers and show how it explains away the paradox of buyers’ advantage in three different double-sided market protocols.

Roberto Cervone, Stefano Galavotti, Marco LiCalzi

Chapter 8. Multi-Unit Auction Analysis by Means of Agent-Based Computational Economics

In this paper an agent-based computational economics (ACE) model has been developed in order to test the bidding behavior in a multi-unit auction, the Ausubel auction. The model has been studied in two scenarios. In the first one, bidders have weakly decreasing marginal values and the theory predicts that bidding sincerely is a weakly dominant strategy. The ACE model corroborates this finding. In the second scenario, agents present synergies among their valuations. This scenario has been tested for two environments. In the first one, bidders have the same synergy value, but it differs from one experiment to another. In the second one, bidders within the same experiment exhibit different synergy values. The ACE model finds that underbidding is the most frequent strategy to avoid the exposure problem and maximize bidders’ payoff in the presence of synergies.

Asuncion Mochon, Yago Saez, David Quintana, Pedro Isasi

Chapter 9. Social Learning and Pricing Obfuscation

We examine markets in which companies are allowed to obfuscate prices and customers are forced to rely on their direct experience and signals they receive from social networks to make purchasing decisions. We compare interventions by public regulators that impose constraints on the amount of price obfuscation with those that augment customers’ cognitive capacities in order to determine which class of policies enhances social welfare the most in such a setting. We implement the strategic behavior of companies by a recursive simulation of


-th order rationality and extend the Experience Weighted Attractions framework to incorporate information from social networks for adaptive customers. Therefore, we search for market designs that are robust with respect to bounded rationality of companies and customers.

Maciej Latek, Bogumil Kaminski



Chapter 10. Mutual Funds Flows and the “Sheriff of Nottingham” Effect

Investors in mutual funds appear to reward disproportionately the best performing funds with large inflows while, at the same time, avoid to withdraw similar amounts from the poorly managed funds. We show that this peculiar flat-convex shape of the flow-performance curve for mutual funds can be generally explained by a model where profit chasing customers punish the bad funds by switching a fraction of their wealth to the best ones (“Sheriff of Nottingham” effect). In the absence of external flows, the model provably produces a constant curve when the standard deviation of excess returns is much larger than the level of the returns. This for the most part explains the apparent insensitivity of flows to below-average returns. The introduction of exogenous injections of money invested in the top funds complete the model and provides a realistic increase in the flows of the funds yielding above-average returns. We finally show by simulation that our results are robust to variations in the values of the parameters of the model.

Lucia Milone, Paolo Pellizzari

Chapter 11. Foundations for a Framework for Multiagent-Based Simulation of Macrohistorical Episodes in Financial Markets

Questions about methodology and model design are subject to constant debate in the emerging field of agent-based modelling and simulation. This article intends to make progress on some of the more foundational aspects affecting method and design. Our primary focus is on a much neglected area of agent-based modelling, namely that of large-scale, spatio-temporal phenomena in financial markets. We argue that multiagent-based models are ideally suited to tackle this class of problems, but that as of yet simulation research has not delivered the methods and tools necessary for this task. The lack of a methodological framework tailored to this complex research object, due in part to a persistent coloration of research questions and interests in positivist shades, is an evident obstacle to progress. We hold that simulation research in finance should set its own methodological agenda, and propose that the mechanism-centric philosophy of critical realism is moved to the centre stage. Our framework encourages researchers to be mindful of existing knowledge and insight in finance without being kept hostage to it and to embrace the historic turn and look out for synergies in qualitative research.

Bàrbara Llacay, Gilbert Peffer

Chapter 12. Explaining Equity Excess Return by Means of an Agent-Based Financial Market

The observed values of equity premium, i.e., the excess return required by investors to hold equities instead of risk-free securities, are usually far larger than values foreseen by consumption capital asset pricing models with realistic aversion to risk. In order to tackle the problem form a different point of view, we present a model of an artificial economy, where different heterogeneous agents are interacting in the financial market. Households, firms, and a commercial bank make endogenous financial decisions which involve portfolio investments for households, capital structure and dividends policy for firms, and lending and borrowing rates for the commercial bank. In particular, households are characterized by behavioral rules derived from prospect theory. Labor income for households and earnings for firms are exogenous determined, according to independent stochastic processes. From simulation experiments it emerges that the model offers new interesting insights on the issue, confirming some hypothesis about the influence of households psychological features on the equity premium dynamics. In particular, the model shows that the length of time over which agents aggregate and evaluate returns, called evaluation period, has a significant role in explaining equity excess returns.

Andrea Teglio, Marco Raberto, Silvano Cincotti

Financial Markets


Chapter 13. Bubble and Crash in the Artificial Financial Market

In this paper, we investigate bubble and crash in the artificial financial market. Based on Ball and Holt (1998)’s experiment in the laboratory with real human beings, we create a simple artificial financial market using an agent-based simulation. In this simulation, we model each agent with different characteristics with respect to expectation formation and time discounting. We found that the case of prospect theory plus exponential time discounting is most resemble with the price dynamics found in Ball and Holt’s experiment and real world price bubble and crash. We also examine whether Ball and Holt’s and our experiment are really judged as a bubble and crush with some indexes so far proposed. Then, Ball and Holt’ experimental result is judged not as a price bubble but as a divergent oscillation, while our result is judged as a bubble.

Yuji Karino, Toshiji Kawagoe

Chapter 14. Computation of the Ex-Post Optimal Strategy for the Trading of a Single Financial Asset

In this paper we explain how to compute the maximum amount of money one investor can earn in trading a single financial asset under a set of trading constraints. The obtained algorithm allows to identify the ex-post optimal strategy


* over a set of (known) prices, which is unconventional in Finance. We deliberately adopt such a simplification to show that


if one posits a complete knowledge of the “future”, the determination of


* is far from triviality, especially in a framework with transaction costs. We review some solutions that are exponential and propose a new polynomial algorithm. Among others, our results shed light on a not so documented aspect of financial markets complexity, propose an absolute boundary for the profits one can realize in a specific time window and against which any investment strategy can be gauged.

Olivier Brandouy, Philippe Mathieu, Iryna Veryzhenko

Chapter 15. A Generative Approach on the Relationship between Trading Volume, Prices, Returns and Volatility of Financial Assets

The relationship among trading volume, prices, returns, etc., of financial assets is complex, but its proper understanding may be of great influence on the development of financial theories. Throughout the last half century, many researchers have faced the issue, but a general consensus has not been reached. In this paper, we propose to use agent based simulation, a methodology that allows us to recreate different scenarios to reproduce the observed behavior in financial markets.

José Antonio Pascual, Javier Pajares

Information and learning

Chapter 16. Comparing Laboratory Experiments and Agent-Based Simulations: The Value of Information and Market Efficiency in a Market with Asymmetric Information

In this paper agent-based simulations are employed to deepen our understanding of results from experimental asset markets with asymmetric fundamental information. Beside the experimental treatment, we implement two simulation settings: a base-case simulation with all agents using their fundamental information and an equilibrium solution in which agents can choose from a set of three different strategies. We find that the behavior of the human subjects closely matches a strategy based on using the fundamental information provided, even when other strategies would have resulted in higher earnings. As a consequence, efficiency in the human markets is lower than in most of the simulated markets.

Florian Hauser, Jürgen Huber, Michael Kirchler

Chapter 17. Asset Return Dynamics under Alternative Learning Schemes

In this paper we design an artificial financial market where endogenous volatility is created assigning to the agents diverse prior beliefs about the joint distribution of returns, and, over time, making agents rationally update their beliefs using common public information. We analyze the asset price dynamics generated under two learning environments: one where agents assume that the joint distribution of returns is


, and another where agents believe in the existence of regimes in the joint distribution of asset returns. We show that the regime switching learning structure can generate all the most common stylized facts of financial markets: fat tails and long-range dependence in volatility coexisting with relatively efficient markets.

Elena Catanese, Andrea Consiglio, Valerio Lacagnina, Annalisa Russino

Chapter 18. An Attempt to Integrate Path-Dependency in a Learning Model

The absence of information on the state of the resource is considered as one of the main reasons of resource collapses. In the current study, we propose a solution to this problem stemming from the resource users. They can perceive the resource dynamics by the impact it has on their profits. At a given time step, the state of the resource depends on its previous states and hence on the agents’ past decisions. In this perspective, different perceptions are characterized by different weights that the resource users assign to the current and past actions in the profit formation. In order to capture these individual differences, we consider Schaefer-Gordon dynamic model. On its basis, we develop a learning model, adapted from Roth-Erev model. The simulation results show that the resource can be exploited in a sustainable manner if the past action is taken into account.

Narine Udumyan, Juliette Rouchier, Dominique Ami

Methodological Issues


Chapter 19. A Model-to-Model Analysis of the Repeated Prisoners’ Dilemma: Genetic Algorithms vs. Evolutionary Dynamics

We study the properties of the well known Replicator Dynamics when applied to a finitely repeated version of the Prisoners’ Dilemma game. We characterize the behavior of such dynamics under strongly simplifying assumptions (i.e. only 3 strategies are available) and show that the basin of attraction of defection shrinks as the number of repetitions increases. After discussing the difficulties involved in trying to relax the “strongly simplifying assumptions” above, we approach the same model by means of simulations based on genetic algorithms. The resulting simulations describe a behavior of the system very close to the one predicted by the replicator dynamics without imposing any of the assumptions of the analytical model. Our main conclusion is that analytical and computational models are good complements for research in social sciences. Indeed, while on the one hand computational models are extremely useful to extend the scope of the analysis to complex scenarios hard to analyze mathematically, on the other hand formal models can be extremely useful to verify and to explain the outcomes of computational models.

Xavier Vilà

Chapter 20. Impact of Tag Recognition in Economic Decisions

In this paper we replicate the model by Axtell

et al

. (2000), a game where two agents ask for proportions of the same pie. After simulating the same scenarios, we get the same results, both in the cases of one-agent and two-agent types (tag model). Once we know the model has been properly replicated, we go one step further, by analyzing the influence in the observed behavior of the ‘rational’ decision rule and of the matrix payoff. First, we change the agent’s decision rule, so that agents could decide playing a heuristic which is not so ‘rational’ as the original rule. We also evaluate the dependence of the results on the selected payoff matrix. We conclude that both the decisions rules and the payoff matrix could affect how and when the equilibrium and the segregation emerge in the system. This is particularly interesting for the tag model, as it is related to the role of group recognition in economic decisions.

David Poza, Félix Villafáñez, Javier Pajares

Chapter 21. Simulation of Effects of Culture on Trade Partner Selection

The criteria that traders use to select their trade partners differ across cultures. The rational criterion of expected profit of the next contract to be negotiated dominates the decision in individualistic, egalitarian, uncertainty tolerant cultures. In other cultures, criteria like personal relations, group membership, status difference and trust may strongly influence trade partner selection. There also exist differences in the level of information about potential partners that traders require before entering into business contacts. This paper models the role of culture at the level of individual agents, based on Hofstede’s five dimensions of culture. The model is applied in multi-agent simulations, that are designed as a research tool for supply chain research. The model is implemented as a random selection process, where potential partners have unequal probabilities of being selected. The factors influencing the probabilities are: expected profit and trust (learnt from previous contacts with potential partners or reputation), common group membership, societal status, and personal relations. Results are presented, that indicate that Hofstede’s model can be used to simulate the effect of culture on the formation and maintenance of business relationships.

Gert Jan Hofstede, Catholijn M. Jonker, Tim Verwaart
Weitere Informationen

Premium Partner

Stellmach & BröckersBBL | Bernsau BrockdorffMaturus Finance GmbHPlutahww hermann wienberg wilhelm

BranchenIndex Online

Die B2B-Firmensuche für Industrie und Wirtschaft: Kostenfrei in Firmenprofilen nach Lieferanten, Herstellern, Dienstleistern und Händlern recherchieren.



Künstliche Intelligenz und der Faktor Arbeit - Implikationen für Unternehmen und Wirtschaftspolitik

Künstliche Intelligenz und ihre Auswirkung auf den Faktor Arbeit ist zum Modethema in der wirtschaftswissenschaftlichen Politikberatung avanciert. Studien, die alarmistisch die baldige Verdrängung eines Großteils konventioneller Jobprofile beschwören, leiden jedoch unter fragwürdiger Datenqualität und Methodik. Die Unternehmensperspektive zeigt, dass der Wandel der Arbeitswelt durch künstliche Intelligenz weitaus langsamer und weniger disruptiv ablaufen wird. Jetzt gratis downloaden!