The aim of this thesis is to propose and illustrate an alternative approach to economic modeling and policy design that is grounded in the innovative field of agent-based computational economics (ACE). The recent crisis pointed out the fundamental role played by macroeconomic policy design in order to preserve social welfare, and the consequent necessity of understanding the effects of coordinated policy measures on the economic system. Classic approaches to macroeconomic modeling, mainly represented by dynamic stochastic general equilibrium models, have been recently criticized for they difficulties in explaining many economic phenomena. The absence of interaction among heterogeneous agents, along with their strong rationality, are two of the main of criticisms that emerged, among others. Actually, decentralized market economies consist of large numbers of economic agents involved in local interactions and the aggregated macroeconomic trends should be considered as the result of these local interactions. The approach of agent-based computational economics consists in designing economic models able to reproduce the complicated dynamics of recurrent chains connecting agent behaviors, interaction networks, and to explain the global outcomes emerging from the bottom-up. The work presented in this thesis tries to understand the feedback between the microstructure of the economic model and the macrostructure of policy design, investigating the effects of different policy measures on agents behaviors and interactions. In particular, the attention is focused on modeling the relation between the financial and the real sides of the economy, linking the financial markets and the credit sector to the markets of goods and labor. The model complexity is increasing with the different chapters. The agent-based models presented in the first part evolve to a more complex object in the second part, becoming a sort of complete ``artificial economy''. The problems tackled in the thesis are various and go from the investigation of the equity premium puzzle, to study of the effects of classic monetary policy rules (as the Taylor rule) or to the study of the macroeconomic implications of bank's capital requirement or quantitative easing.

The concept of ‘Industry 4.0’ signalises the rise of innovative manufacturing technologies, including industrial robots. Wider applicability of robotic automation and higher efficiency of production processes shift the profitability analysis of strategic relocation decisions. Despite the technological feasibility, diffusion of technology lowers the profitability threshold for robots. Consequently, competitive labour cost advantages, formerly motivating manufacturing firms to offshore production become less relevant. In fact, robots additionally gain importance in the case of shifted global economic realities, such as stricter environmental regulation on global trade and the convergence of the global wage gap. However, the heterogeneous levels of automation among manufacturing firms have not been taken into account when studying the macroeconomic phenomenon of reshoring. This study adds novelty by offering an agent-based perspective which has allowed insights on how the behaviour of firms, guided by simple economic rules on the micro-level, is dynamically influenced by their complex environment in regard to relocation, decision-making hypotheses. Testing various variables sensitive to initial conditions, increased environmental regulations targeting global trade and upward shifting wage levels in formerly offshore production locations have shown to be driving and inhibiting mechanisms of this socio-technical system. Therefore, the dynamic demonstrates a shift from predominantly cited economic reasoning for relocation strategies towards sustainability aspects, pressingly changing these realities on an environmental and social dimension. The popular debate is driven by increased environmental awareness and the proclaimed fear of robots killing jobs. In view of reshoring shaping the political agenda, interest in the phenomenon has recently been fuelled by the rise of populism and protectionism claiming to “bring jobs back home”.

Le marché aux poissons de Boulogne-sur-Mer se caractérise par l’organisation singulière de son système de vente. En effet, sur celui-ci, les acheteurs et les vendeurs peuvent choisir chaque jour de recourir à un mécanisme d’enchères ou à un marché de gré à gré (voire à ces deux possibilités en même temps), pour commercer entre eux. La coexistence de ces deux systèmes de vente est stable dans le temps, chacun d’entre eux représentant approximativement la moitié des quantités échangées. Cette singularité économique conduit à s’interroger sur les conditions nécessaires à l’émergence et à la stabilité de cette coexistence. Pourquoi les agents ne s’accordent-ils tous pas pour un unique mécanisme de transaction comme dans la majorité des marchés? pourquoi observe-t-on une si grande volatilité dans les choix individuels de marché? Afin de comprendre les conditions nécessaires à cette coexistence de mécanismes de marché, la présente thèse se déclinera comme suit. La première partie sera dédiée à l’étude empirique des transactions journalières ayant lieu sur chacun des deux sous-marchés. Nous commençons par une analyse statistique et économétrique afin d’extraire les faits stylisés représentatifs des propriétés du marché et de ses acteurs, avant de procéder à une analyse des réseaux sociaux existants sur ce marché,visant à déterminer l’influence des interactions dans la prise de décision. Fort de ces résultats, nous construisons des modèles informatiques multi-agents, capables de reproduire les comportements observés au niveau individuel, et, au travers ceux-ci,le comportement du marché lui-même au niveau agrégé. / Should I buy or should I bid ? The influence of market mechanism : the case of Boulogne-Sur-Mer fish market. The Boulogne-sur-Mer fish market is organized in a very specific way. Each day buyers and sellers can choose to use either an auction mechanism, a negotiated market, or evenboth, in order to sell and buy goods.A stunning fact observed is the stable coexistence of those two sub-markets throughout time, with no convergence of agents toward one of them, each one accounting for roughly half of the exchanged quantities.The present thesis aims at discovering the necessary conditions of the emergence andstability of such a coexistence.To do it, we will begin with an empirical study of daily transactions that have occurred on this market for a few years. We begin with a statistical and econometric study to extract the main stylized facts of this market, then we study the social networks influencing the outcomes. Once those facts determined, we build agent-based computational models able to reproduce the individual behaviours of agents, and through these, the emergence of the market’sbehaviour itself.

Cette thèse s'inscrit dans le cadre de systèmes complexes appliqués aux systèmes économiques. Dans cette thèse, un modèle multi-agent a été proposé, qui modélise le cycle de production. Il est consitué d'entreprises, ouvirers/foyers, et une banque, et repecte la conservation de la monnaie. Son hypothèse centrale est que les entreprises se basent sur une marge espérée pour déterminer leur production. Un scénario simple de ce modèle, ou les marges espérées sont homogènes, a été analysé dans le cadre de models de croissance stochastique. Les résultats sont la distribution de tailles d'entreprises rassemblant des lois de puissance, et leur distribution du taux de croissance de forme 'tente', ainsi qu'une dépendence de taille de la variance de la croissance. Ces résultats sont proches aux faits stylisés issus d'études empiriques. Dans un scénario plus complet, le modèle contient des caractéristiques supplémentaires: des marges espérées hétérogèges, ainsi que des paiements d'intérêts, la possibilité de faire faillite. Cela ramène le modèle aux modèles macro-économiques multi-agents. Les extensions sont décrites de façon théorique par des équations de replicateur. Les résultats nouveaux sont la distribution d'age d'entreprises actives, la distribution de leur taux de profit, la distribution de dette, des statistiques sur les faillites, et des cycles de vie caractéristiques. Tout ces résultats sont qualitativement en accord avec des résultats d'études empiriques de plusieurs pays.Le modèle proposé génère des résultats prometteurs, en respectant le principe que des résultats qui apparaissent simultanément peuvent soit etre générés par un même processus, soit par plusieurs aui qui sont compatibles. / The thesis is in the field of complex systems, applied to an economic system. In this thesis, an agent-based model has been proposed to model the production cycle. It comprises firms, workers, and a bank, and respects stock-flow consistency. Its central assumption is that firms plan their production based on an expected profit margin. A simple scenario of the model, where the expected profit margin is the same for all firms, has been analyzed in the context of simple stochastic growth models. Results are a firms' size distribution close to a power law, and tent-shaped growth rate distribution, and a growth rate variance scaling with firm size. These results are close to empirically found stylized facts. In a more comprehensive version, the model contains additional features: heterogeneous profits margins, as well as interest payments and the possibility of bankruptcy. This relates the model to agent-based macroeconomic models. The extensions are described theoretically theoretically with replicator dynamics. New results are the age distribution of active firms, their profit rate distribution, debt distribution, bankruptcy statistics, as well as typical life cycles of firms, which are all qualitatively in agreement with studies of firms databases of various countries.The proposed model yields promising results by respecting the principle that jointly found results may be generated by the same process, or by several ones which are compatible.

This study examines the claim that individuals misinterpret the mean of a dataset (displayed as a scatterplot) more when the convex hull of the dataset is less representative of the data. In addition, this study also tests whether outliers in the data can predict the magnitude of error that individuals make in interpreting the mean of the dataset. Lastly, the study investigates whether individuals’ interpretations are predicted better by the mean of the convex hull than by the full dataset’s mean. The method used to conduct these investigations is through a survey, followed by several linear regression analyses. Applications of this study include improving the communication of data in economic policy and business contexts, along with broader applications in extending models that heavily rely on agents’ interpretations of information: especially bounded rationality and social norm-based models. The results show that convex hull unrepresentativeness correlates positively with error in mean interpretation; however, that the convex hull mean is not predictive of the interpretations’ direction. Overall, the study contributes to the field of visual information interpretation by investigating the effect of data structure on its interpretation – an unexplored area of research. This is done while initiating the concretization of bounded rationality in economics, by exploring the idea that individuals perceive a general shape of the information presented to them rather than a detailed, full picture. This can lead to misinterpretations whenever the general shape (convex hull) is not representative of the dataset.

The existence of sunspots or sunspot equilibria has been debated for several decades on its influence in the field of Economics. While models of sunspots or sunspot equilibria have fitted well for some subsets of empirical features, it comes at a cost of moving further away from economic believability and robustness. Studies on the theoretical plausibility of sunspot equilibria have been addressed extensively in several different economic models, but exist almost entirely within the framework of the homogeneous rational expectations equilibrium devised of representative agents. This framework shapes later arising learning approaches to sunspot equilibria. These models have proposed various ways of learning, but they deal mainly with the learning of representative agents. Models of adaptive learning with heterogeneous agents, however, enable us to explicitly tackle coordination issues, such as the coordination mechanism of expectations. This is certainly desirable since sunspots are often used as a coordination device of expectations. In this dissertation, we continue this line of research, investigating the plausibility of sunspot equilibria in stock markets within the framework of heterogeneous agents and the dynamic relationship between sunspot variables and stock returns. We adopt an Agent-based Computational Approach, now known as Agent-based Computational Economics or ACE, to study the plausibility of sunspot equilibria. More specifically, we deal with this issue in the context of an Agent-based Artificial Stock Market (AASM). We contemplate AASMs to be highly suitable to the issue we examine here. Currently, none of the theoretical, empirical, experimental, or simulation models of sunspot equilibria directly capture sunspots within a stock market composed of heterogeneous agents. We conducted three series of experiments to examine this issue. From the results of these three series of simulations, we observed that sunspot variables generally do not have influence on market dynamics. This indicates that sunspot variables remain largely exogenous to the system. Furthermore, we traced the evolution of agents' beliefs and examined their consistency with the observed aggregate market behavior. Additionally, this dissertation takes the advantage of and investigates the micro-macro relationship within the market. We argue that a full understanding of the dynamic linkage between sunspot variables and stock returns cannot be accomplished unless the feedback relationship between individual behaviors, at the micro view, and aggregate phenomena, at the macro view, is well understood / The existence of sunspots or sunspot equilibria has been debated for several decades on its influence in the field of Economics. While models of sunspots or sunspot equilibria have fitted well for some subsets of empirical features, it comes at a cost of moving further away from economic believability and robustness. Studies on the theoretical plausibility of sunspot equilibria have been addressed extensively in several different economic models, but exist almost entirely within the framework of the homogeneous rational expectations equilibrium devised of representative agents. This framework shapes later arising learning approaches to sunspot equilibria. These models have proposed various ways of learning, but they deal mainly with the learning of representative agents. Models of adaptive learning with heterogeneous agents, however, enable us to explicitly tackle coordination issues, such as the coordination mechanism of expectations. This is certainly desirable since sunspots are often used as a coordination device of expectations. In this dissertation, we continue this line of research, investigating the plausibility of sunspot equilibria in stock markets within the framework of heterogeneous agents and the dynamic relationship between sunspot variables and stock returns. We adopt an Agent-based Computational Approach, now known as Agent-based Computational Economics or ACE, to study the plausibility of sunspot equilibria. More specifically, we deal with this issue in the context of an Agent-based Artificial Stock Market (AASM). We contemplate AASMs to be highly suitable to the issue we examine here. Currently, none of the theoretical, empirical, experimental, or simulation models of sunspot equilibria directly capture sunspots within a stock market composed of heterogeneous agents. We conducted three series of experiments to examine this issue. From the results of these three series of simulations, we observed that sunspot variables generally do not have influence on market dynamics. This indicates that sunspot variables remain largely exogenous to the system. Furthermore, we traced the evolution of agents' beliefs and examined their consistency with the observed aggregate market behavior. Additionally, this dissertation takes the advantage of and investigates the micro-macro relationship within the market. We argue that a full understanding of the dynamic linkage between sunspot variables and stock returns cannot be accomplished unless the feedback relationship between individual behaviors, at the micro view, and aggregate phenomena, at the macro view, is well understood.

I have developed a multiagent system platform that provides a valuable complement to the alternative research methods. The platform facilitates the development of heterogeneous agents in complex environments. The first application of the multiagent system is to the study of sealed-bid auctions with two-dimensional value signals from pure private to pure common value. I find that several auction outcomes are significantly nonlinear across the two-dimensional value signals. As the common value percent increases, profit, revenue, and efficiency all decrease monotonically, but they decrease in different ways. Finally, I find that forcing revelation by the auction winner of the true common value may have beneficial revenue effects when the common-value percent is high and there is a high degree of uncertainty about the common value. The second application of the multiagent system is to the study of English auctions with two-dimensional value signals using agents that learn a signal-averaging factor. I find that signal averaging increases nonlinearly as the common value percent increases, decreases with the number of bidders, and decreases at high common value percents when the common value signal is more uncertain. Using signal averaging, agents increase their profit when the value is more uncertain. The most obvious effect of signal averaging is on reducing the percentage of auctions won by bidders with the highest common value signal. The third application of the multiagent system is to the study of the optimal payment rule in Treasury auctions using Canadian rules. The model encompasses the when-issued, auction, and secondary markets, as well as constraints for primary dealers. I find that the Spanish payment rule is revenue inferior to the Discriminatory payment rule across all market price spreads, but the Average rule is revenue superior. For most market-price spreads, Uniform payment results in less revenue than Discriminatory, but there are many cases in which Vickrey payment produces more revenue.

I have developed a multiagent system platform that provides a valuable complement to the alternative research methods. The platform facilitates the development of heterogeneous agents in complex environments. The first application of the multiagent system is to the study of sealed-bid auctions with two-dimensional value signals from pure private to pure common value. I find that several auction outcomes are significantly nonlinear across the two-dimensional value signals. As the common value percent increases, profit, revenue, and efficiency all decrease monotonically, but they decrease in different ways. Finally, I find that forcing revelation by the auction winner of the true common value may have beneficial revenue effects when the common-value percent is high and there is a high degree of uncertainty about the common value. The second application of the multiagent system is to the study of English auctions with two-dimensional value signals using agents that learn a signal-averaging factor. I find that signal averaging increases nonlinearly as the common value percent increases, decreases with the number of bidders, and decreases at high common value percents when the common value signal is more uncertain. Using signal averaging, agents increase their profit when the value is more uncertain. The most obvious effect of signal averaging is on reducing the percentage of auctions won by bidders with the highest common value signal. The third application of the multiagent system is to the study of the optimal payment rule in Treasury auctions using Canadian rules. The model encompasses the when-issued, auction, and secondary markets, as well as constraints for primary dealers. I find that the Spanish payment rule is revenue inferior to the Discriminatory payment rule across all market price spreads, but the Average rule is revenue superior. For most market-price spreads, Uniform payment results in less revenue than Discriminatory, but there are many cases in which Vickrey payment produces more revenue.

I have developed a multiagent system platform that provides a valuable complement to the alternative research methods. The platform facilitates the development of heterogeneous agents in complex environments. The first application of the multiagent system is to the study of sealed-bid auctions with two-dimensional value signals from pure private to pure common value. I find that several auction outcomes are significantly nonlinear across the two-dimensional value signals. As the common value percent increases, profit, revenue, and efficiency all decrease monotonically, but they decrease in different ways. Finally, I find that forcing revelation by the auction winner of the true common value may have beneficial revenue effects when the common-value percent is high and there is a high degree of uncertainty about the common value. The second application of the multiagent system is to the study of English auctions with two-dimensional value signals using agents that learn a signal-averaging factor. I find that signal averaging increases nonlinearly as the common value percent increases, decreases with the number of bidders, and decreases at high common value percents when the common value signal is more uncertain. Using signal averaging, agents increase their profit when the value is more uncertain. The most obvious effect of signal averaging is on reducing the percentage of auctions won by bidders with the highest common value signal. The third application of the multiagent system is to the study of the optimal payment rule in Treasury auctions using Canadian rules. The model encompasses the when-issued, auction, and secondary markets, as well as constraints for primary dealers. I find that the Spanish payment rule is revenue inferior to the Discriminatory payment rule across all market price spreads, but the Average rule is revenue superior. For most market-price spreads, Uniform payment results in less revenue than Discriminatory, but there are many cases in which Vickrey payment produces more revenue.

Orientador: Fernando Antônio Campos Gomide / Tese (doutorado) - Universidade Estadual de Campinas, Faculdade de Engenharia Elétrica e de Computação / Made available in DSpace on 2018-08-16T03:39:08Z (GMT). No. of bitstreams: 1 Walter_IgorAlexandre_D.pdf: 1762436 bytes, checksum: 257485271a6580f86b0b466799ceff14 (MD5) Previous issue date: 2010 / Resumo: Sugerimos uma abordagem evolutiva para o projeto de estratégias de interação em sistemas multiagentes, especialmente estratégias de oferta modeladas como sistemas baseados em regras nebulosas. O objetivo é a aprendizagem das estratégias de oferta em leilões em modelos em que a base de conhecimento sofre evolução para melhorar o desempenho dos agentes atuando em um ambiente competitivo. Dados para aprendizagem e otimização das estratégias são raros em ambientes competitivos como os leilões. Introduzimos um modelo de sistema genético fuzzy (GFS) cujos operadores genéticos utilizam uma representação de tamanho variável do cromossomo e uma relação hierárquica estabelecida através do fitness dos indivíduos, em um esquema que explora e explota o espaço de busca ao longo das gerações. A evolução de estratégias de interação permite a descoberta de comportamentos dos agentes previamente desconhecidos e inesperados, permitindouma análise mais rica dos mecanismos de negociação e seu papel como protocolo de coordenação. A aplicação da abordagem proposta no mercado de energia elétrica permite a simulação destes mercados através da evolução de estratégias de oferta (bidding) em leilões de energia. A reestruturação destes mercados nas economias contemporâneas apresenta novos desafios e oportunidades, uma vez que não há consenso sobre qual seria sua melhor organização. A evolução da estrutura organizacional destes mercados salienta a falta de discernimento sobre as principais questões a serem analisadas e levadas em consideração. Argumenta-se que a abordagem econômica neoclássica se mostra limitada na análise dos efeitos da reestruturação e no estudo do comportamento dos agentes econômicos competindo nos mercados de energia elétrica reestruturados. Apresentamos uma arquitetura computacional inspirada na Economia Computacional baseada em Agentes que permite a modelagem, estudo e simulação destes mercados. Aplicamos ferramentas de inteligência computacional adequadas à concepção dos agentes participantes nos mercados de energia e que podem ser estendidas a outros mecanismos de mercado e negociação. Os mercados de energia elétrica são sistemas complexos habitados por agentes econômicos com interesse próprio que interagem entre si. Concluímos que é natural modelar e simular estes mercados como sistemas multiagentes. A evolução de estratégias de oferta permite a descoberta de comportamentos que auxiliam na tomada de decisão de um participante e na avaliação do mecanismo de negociação por parte de seus projetistas / Abstract: We suggest an evolutionary approach to design interaction strategies for multiagent systems, focusing on strategies modeled as fuzzy rule-based systems. The aim is to learn models represented by evolving knowledge bases to achieve agents' performance improvement when playing in a competitive environment. In competitive situations data for learning and tuning are rare and rule bases must jointly evolve with the databases. We introduce an evolutionary algorithm whose operators use variable length chromosome, a hierarchical relationship among individuals through fitness, and a scheme that successively explores and exploits the search space along generations. Evolution of interaction strategies uncovers unknown and unexpected agent behaviors and allows a richer analysis of negotiation mechanisms and their role as a coordination protocol. An application concerning an electricity market illustrates the effectiveness of the approach and allows to simulate the market through evolutionary bidding strategies. The restructuring process of power markets raises new challenges and opportunities, since there is no consensual market architecture. The evolution of the power industry organization shows a lack of insight about the issues to be addressed and taken into account. Several authors have considered the available tools based on the neoclassical economics theory a limited approach to analyze the effects of the industry restructuring and to study economical agents behavior participating in a restructured electricity market. We present Artificial Economy Multiagent System (AEMAS), a computational architecture inspired on Agent-based Computational Economics (ACE) that allows to model, study and simulate a power market. We apply Computational Intelligence tools to conceive the market agents that we expect could be extended to other negotiation environments. A power market is a complex system populated by self interested economical agents that interact. We conclude that it is feasible to model and simulate these markets on a multiagent system based approach. The evolution of bidding strategies allows to uncover new and unexpected behaviors that help to address the negotiation mechanism analysis by its designers and to support a market player decision process / Doutorado / Engenharia de Computação / Doutor em Engenharia Elétrica