Catálogo de publicaciones - libros

This paper provides an overview of the research activities that have already been undertaken regarding the development and implementation of the idea of the Turing Tournament. This is a two-sided Tournament designed to encourage improvement of the existing as well as creation of new models of human behavior, , that will be capable of replicating the main features that characterize behavior of experimental human subjects in a variety of economic environments. The other side of the Tournament is represented by the algorithms designed to distinguish between machine and human generated behavior. The paper discusses general design questions and its first implementation within the context of repeated games. Finally, the paper describes further stages of the Tournament development which will include its implementation in more complicated economic environments with larger strategy space.

Notwithstanding the recognized importance of traders' expectations in characterizing the observed market dynamics, for instance the formation of speculative bubbles and crashes on financial markets, little attention has been devoted so far by economists to a rigorous study of expectation formation in the laboratory.

In this work we describe a laboratory experiment on the emergence and coordination of expectations in a pure exchange framework. We largely base our study on previous experiments on expectation formation in a controlled laboratory environment by Cars Hommes, Joep Sonnemans, Ian Tuinstra and Henk van de Velden (2002a).

We consider a simple two asset economy with a riskless bond and a risky stock. Each market is composed of six experimental subjects who act as financial advisors of myopic risk-averse utility maximizing investors and are rewarded according to how well their forecasts perform in the market. The participants are asked to predict not only the price of the risky asset at time + 1, as in Hommes et al. (2002a), but also the confidence interval of their prediction, knowing the past realizations of the price until time − 1. The realized asset price is derived from a Walrasian market equilibrium equation, unknown to the subjects, with feedback from individual forecasts. Subjects' earnings are proportional to the increase in their wealth level. With respect to previous experiments that did not include an explicit evaluation of risk by participants, we observe a higher price volatility, a decreased likelihood of bubble dynamics and, in general, a higher heterogeneity of predictions.

Auctions are increasingly being considered as a mechanism for allocating conservation contracts to private landowners. This interest is based on the widely held belief that competitive bidding helps minimize information rents. This study constructs an agent-based model to evaluate the long term performance of conservation auctions under settings where bidders are allowed to learn from previous outcomes. The results clearly indicate that the efficiency benefits of one-shot auctions are quickly eroded under dynamic settings. Furthermore, the auction mechanism is not found to be superior to fixed payment schemes except when the latter involve the use of high prices.

We examine stock market traders' marginal benefits of additional information on the intrinsic value of a stock market security. We use the method of experimental economics to control carefully for the degree of traders' information. Contrary to conventional wisdom we find that it is possible that a marginal unit of additional information does not lead to a marginal increase for a trader's profits. Relatively poorly informed traders can even lose money by using their (limited) available information. However, well informed traders benefit significantly from more information and from using their information when trading.

We discuss a crowd-based theory for describing the collective dynamical behavior of a population of competitive agents in a model market setting. This Crowd-Anticrowd theory, which incorporates the strong correlations between agents' strategies, provides a quantitative description of the fluctuations in the excess demand of this model.

The Minority Games is a good examples of asymmetric coordination problems that are well suited to represent some economic situations. Normally agents play the game with all other agents and this type of game is called Global Minority Game (GMG). If agents play the game only with their neighbors it is called Local Minority Game (LMG). This distinction aims to introduce the limited ability of the agent to receive, decide, and act upon information in the course of interaction. An agent is modeled with its rules and they are updated and selected by natural selection. We propose the rule of give-and-take that is completely different from the conventional rules. With the rule of give-and-take, an agent gives to others when he receives a payoff. On the contrary, a conventional agent only pursues his benefit. We show that the simulation results of give-and-take agents are significantly better than that of selfish agents in both GMG and LMG. We also discuss spatio-temporal patterns and optimality in LMG and whether it can be obtained by evolutionary learning agents.

We use a spatial iterated Prisoner's Dilemma game (IPD) to investigate the spatial-temporal evolution of heterogeneity in agents' strategies. In our model, agents are spatially distributed on a lattice and each agent is assumed to interact with her 4 local neighbors a number of times during each generation. If the agent has a one-step memory for the last action of each individual neighbor, this results in a total of eight different strategies for the game. After each generation, the agent will be replaced by an offspring that adopts the strategy of her most successful neighbor.

The agents are heterogeneous in that they play different strategies dependent on (i) their past experience, (ii) their local neighborhood. The spatial-temporal distribution of these strategies is investigated by means of computer simulations on a cellular automaton. In particular, we study the incluence of on the dynamics of the global frequencies of the different strategies and the conditions for a stationary (frozen) or non-stationary (dynamic) coexistence of particular strategies on a spatial scale.

This paper investigates the long-run properties of chaotic dynamics in a Cournot duopoly market. To this end, it constructs a nonlinear, discrete-time Cournot model in which duopoly firms are heterogeneous and have nonlinear reaction functions with different microeconomic foundations. Due to the nonlinearity, the output adjustment process of the firms generates chaotic fluctuations. This paper demonstrates that for firms, the long-run average profit along a chaotic trajectory can be higher than the profit at all possible equilibria when external effects on production are asymmetric. Further, it is proved that consumer surplus is also higher along the trajectory. The result thus indicates a possibility that from a long-run viewpoint, chaotic fluctuations can be Pareto-improving.

Novelty and heterogeneity are two closely related issues. Heterogeneity is not only a result of the emergence of novelty which creates variety in any evolving system. Heterogeneous elements are also required as inputs for the recombination processes underlying the generation of novelty. However, while heterogeneity figures prominently in computational and agent-based economics and in complex adaptive systems analysis, novelty and its emergence are neglected topics. In order to make progress with the latter the paper starts with a discussion of how novelty is being generated both in the case of genetic novelty and that of mental novelty. For the case of mental novelty it is then shown that the bottleneck in our understanding of novelty is not the generation procedure proper, but rather the procedure by which our mind evaluates or interprets the outcome. On the basis of this distinction it is briefly sketched how, for different forms of novelty, the degree of novelty may be rank-ordered and how the limits to predictability in the context of novelty vary with that degree.

In this paper, we present a model of ‘collective innovation’ building upon the network formation formalism introduced by Jackson and Wolinski (1996) and Jackson and Watts (2002). Agents localized on a circle benefit from knowledge flows from some others with whom they are directly or indirectly connected. They also face costs for direct connections which are linearly increasing with geographical distance separating them. The dynamic process of network formation departs from available literature in that it exhibits preferential meetings for agents close to each other. As our main result, we provide a characterisation of the set of stochastically stable networks selected in the long run. Their architectures are compared to the ones obtained in the simple ‘connections model'. Our main result is to show under what circumstances pairwise stable “small worlds” networks are stochastically selected.