Forthcoming in Computational Economics, published on-line.
Extensive exploration of simulation models comes at a high computational cost, all the more when the model involves a lot of parameters. Economists usually rely on random ex- plorations, such as Monte Carlo simulations, and basic econometric modeling to approximate the properties of computational models. This paper aims to provide guidelines for the use of a much more efficient method that combines a parsimonious sampling of the parameter space using a specific design of experiments (DoE), with a well-suited metamodeling method first developed in geostatistics: kriging. We illustrate these guidelines by following them in the analysis of two simple and well known economic models: Nelson and Winter's industrial dynamics model, and Cournot oligopoly with learning firms. In each case, we show that our DoE experiments can catch the main effects of the parameters on the models' dynamics with a much lower number of simulations than the Monte-Carlo sampling (e.g. 85 simulations instead of 2000 in the first case). In the analysis of the second model, we also introduce supplementary numerical tools that may be combined with this method, for characterizing configurations complying with a specific criterion (social optimal, replication of stylized facts, ect.). Our appendix gives an example of the R-project code that can be used to apply this method on other models, in order to encourage other researchers to quickly test this approach on their models.
JEL classification codes: C61; C63; C80; C90.
Keywords: Computational Economics; Exploration of Agent-Based Models; Design of Experiments; Meta-modeling.
Economic Modelling, Volume 34, Pages 114–128, 2013.
This paper investigates the performances of an inflation targeting regime in a learning economy framed as an Agent–Based Model (ABM). We keep our ABM as close as possible to the original New Keynesian (NK) model, but we model the indi- vidual behaviour of the agents under procedural rationality à la Simon. Accordingly, we assume that their behaviour is guided by simple rules of thumb – or heuristics – while a continuous learning process governs the evolution of those rules. Under these assumptions that also allow the emergence of agents heterogeneity, we analyze the dynamics of the economy without assuming rational expectations, and study the role that a central bank, implementing an inflation targeting regime via a monetary policy rule, can play in the orientation of these dynamics. Consequently, our main goal is to analyse the interplay between the learning mechanisms operating at the individual level and the features and performances of the inflation targeting regime. Our results point to the prime importance of the credibility of central bank’s inflation target regarding macroeconomic stabilization, as well as the beneficial role played by that target as an anchoring device for private inflation expectations. We also establish the potential welfare cost of imperfect public information and contribute to the current debate on optimal monetary policy rules under imperfect common knowledge and uncertainty.
JEL classification codes: E52 - E58 - C63 - D83 - D84.
Keywords: inflation targeting, agent-based model, central bank communication, ex- pectations, learning.
Computational Economics, Volume 41 (4), Pages 493-516, 2013.
This article analyses the ability of the learning firms in a Cournot oligopoly to discover market solutions more collusive that the Cournot equilibrium (CE). We start from the results of Vallée and Yildizoglu (2009) and of Alos-Ferrer (2004), and qualify the role of random experimenting, social learning (imitation), and (updated) memory in helping firms to discover such solutions. Our new computational results show, in contradiction with Alos-Ferrer (2004), that memory and its continuous update can indeed allow firms to beat the CE, and benefit from significant periods with higher profits. We show that the results of the literature on evolutionary learning in oligopoly can analytically be characterized through the interaction of three forces, and indicate when these forces can yield more collusive outcomes. We confirm these results with complementary computational experiments that clearly show the role of long memory.
JEL classification codes: L130; L200; D430; C630; C730.
Keywords: Cournot oligopoly; Learning; Evolution; Selection; Evolutionary stability; Nash equilibrium; Collusion.
Forthcoming in Macroeconomic Dynamics
This article questions the rather pessimistic conclusions of Allen and Carroll (2001) about the ability of consumer to learn the optimal buffer-stock based consumption rule. To this aim, we develop an agent based model where alternative learning schemes can be compared in terms of the consumption behaviour that they yield. We show that neither purely adaptive learning, nor social learning based on imitation can ensure satisfactory consumption behaviours. By contrast, if the agents can form adaptive expectations, based on an evolving individual mental model, their behaviour becomes much more interesting in terms of its regularity, and its ability to improve performance (which is as a clear manifestation of learning). Our results indicate that assumptions on bounded rationality, and on adaptive expectations are perfectly compatible with sound and realistic economic behaviour, which, in some cases, can even converge to the optimal solution. This framework may therefore be used to develop macroeconomic models with adaptive dynamics.
JEL classification: E21, D91, D83, D84,
Keywords: Consumption decisions; Learning; Expectations; Adaptive behaviour, Computational economics
Structural Change and Economic Dynamics, 21: 3, 181-196, 2010.
This article is dedicated to the analysis of the first innovation survey of the Tunisian firms. Starting from basic mechanisms of innovation processes and existing results on other developing countries, we test a set of assumptions on the characteristics of innovating firms in a developing country like Tunisia. The analysis of product and process innovations shows the essential role played by external technical knowledge sources: firms must be able to benefit at least from one such a source in order to attain a significant innovation propensity. We also show that the profile of Tunisian firms can be contrasted with other developing countries. The main contrast concerns the limited role of internal R&D and the insignificant role played by foreign participation. For both types of innovations, another important contrasting result concerns the role of sectoral membership. In Tunisia, this dimension does not seem to structure enough systematically the innovative capabilities of firms. That could indicate an immaturity of sectoral systems of innovations in this country.
JEL classification codes: O120; O300
Keywords: Innovation, development, absorptive capacity, learning, technology policy
Journal of Economic Behaviour and Organization, 72, 670-690, 2009.
Convergence to the Nash equilibrium in a Cournot oligopoly is a question that recurrently arises as a subject of controversy in economics. The development of evolutionary game theory has provided an equilibrium concept more directly connected with adjustment dynamics, and the evolutionary stability of the equilibria of the Cournot game have been extensively studied in the literature. Several articles show that the Walrasian equilibrium is the stable ESS of the Cournot game. But no general result has been established for the difficult case of simultaneous heterogenous mutations. Authors propose specific selection dynamics to analyze this case. Vriend (2000) proposes using a genetic algorithm for studying learning dynamics in this game and obtains convergence to Cournot equilibrium with individual learning. The resulting convergence has been questioned by Arifovic and Maschek (2006). The aim of this article is to clarify this controversy. It analyzes the mechanisms that are behind these contradictory results and underlines the specific role of the spite effect. We show why social learning gives rise to the Walrasian equilibrium and why, in a general setup, individual learning can effectively yield convergence to the Cournot equilibrium. We also illustrate these general results by systematic computational experiments.
JEL classification codes: L130; L200; D430; C630; C730;
Keywords: Cournot oligopoly; Learning; Evolution; Selection; Evolutionary stability; Nash equilibrium; Genetic algorithms
We analyze in this article main determinants of technology dynamics in Tunisian manufacturing sectors. The data from the industrial survey provided by Ministry of Scientific Research, Technology and Competency Development (MSRTCD) for the period 2002-2004 is explored using regression trees and Probit models in order to discover main factors that favor the innovative capacity of Tunisian firms. Our results show that we must distinguish process and product innovations because they are driven by different mechanisms. Moreover, we observe that sectoral heterogeneity should not be neglected and we study more in detail fours sectors that are particularly well represented in our sample. This analysis allows us to suggest some differentiated policy indications for fostering innovative capacity in these sectors.
Review of Economic Design, 11, 339-359, 2008.
Searching for efficient networks can prove a very difficult analytical and even computational task. In this paper, we explore the possibility of using the genetic algorithms (GA) technique to identify efficient network structures in the case of non-trivial payoff functions. The robustness of this method in predicting optimal networks is tested on the two simple stylized models introduced by Jackson and Wolinski (1996), for which the efficient networks are known over the whole state space of the parameters' values. This approach allows us to obtain new exploratory results in the case of the linear-spatialized connections model proposed by Johnson and Gilles (2000), for which the efficient allocation of bilateral connections is driven by contradictory forces that push either for a centralized structure around a coordinating agent, or for only locally and evenly distributed connections.
in Cantner, U., J.-L. Gaffard and L. Nesta (eds), "Schumpeterian Perspectives on Innovation, Competition and Growth", Springer, 2009.
This article develops an evolutionary model of industry dynamics in order to carry out a richer theoretical analysis of the consequences of a stronger patent system. This model explicitly takes into account the potentially positive effects of the patents : Publication of patents participates to the building of a collective knowledge stock on which the innovations can rely, and dropped patents can provide a source of technological progress for firms that are lagging behind the leaders of the industry. These dimensions of the patent system are used to question the negative results of Vallée & Yildizoglu (2006). The main results of the new model show that these positive effects do not counterbalance the negative effects of a stronger patent system on social welfare and global technological progress, even if it is a source of better protection and higher profits for the firms.
Journal of Economic Behaviour and Organization, 67: 2, 495-511,2008.
In this paper, we make an exploratory use of numerical techniques (genetic algorithms and Monte Carlo simulations) to compute efficient and emergent networks in a spatialized version of the connections model of Jackson and Wolinski (1996). This approach allows us to observe and discuss the coordination failures that arise in a strategic network formation context with link-mediated positive externalities to connections and geographically based connection costs. Our results highlight that, depending on the strength of the externalities, emergent and efficient networks may share several structural properties. Nevertheless, emergent networks have too few local and distant connections and are also too less "coordinated" around some central agents than they should.
Review of Economic Design, 11, 339-359, 2008.
The modelling of networks formation has recently became the object of an increasing interest in economics. One of the important issues raised in this literature is the one of networks efficiency. Nevertheless, for non trivial payoff functions, searching for efficient network structures turns out to be a very difficult analytical problem as well as a huge computational task, even for a relatively small number of agents. In this paper, we explore the possibility of using genetic algorithms (GA) techniques for identifying efficient network structures, because the GA have proved their power as a tool for solving complex optimization problems. The robustness of this method in predicting optimal network structures is tested on two simple stylized models introduced by Jackson and Wolinski (1996), for which the efficient networks are known over the whole state space of parameter values.
Journal of Evolutionary Economics, 16,189-206, 2006
This article develops an evolutionary model of industry dynamics in order to carry out a richer theoretical analysis of the consequences of a stronger patent system. The first results obtained in our article are rather consistent with the anti-patent arguments and they do not favour the case for a stronger patent system: higher social welfare and technical progress are observed in our model in industries with milder patent systems (lower patent height and patent life).
Journal of Economic Behaviour and Organization, 61, 668-690, 2006.
In a preceding article we have studied the Communities of Practice and their conditions of emergence using an Agent based model of a set of agents facing a continuous flow of problems. We center now our analysis on the performance of this organizational structure in comparison with a two-level hierarchical delegation structure. Our results show the crucial role played by the communication and the specialisation of the agents.
Revue d'Economie Industrielle, 103, p.91-110, 2003.
This contribution focuses on the simulation of the emergence and evolution of communities of practice, a concept brought forwards originally in sociology in the early 90's by Lave and Wenger (1990) and Brown and Duguid (1991). In a first part, the main theoretical principles on which relies the notion of community of practice are recalled. Then, the contribution exposes the characteristics of the simulation, and discusses the main results that have been obtained.
Revue d'Economie Politique, 114, 711-745, 2004
A quick analysis of the most recent literature in economics shows a tremendous increase of the number of the articles using Genetic Algorithms. This tendency can be observed nearly in all areas of economics (from econometrics to finance). The aim of this article is to give a presentation of the main mechanisms of this approach and a panorama of its applications in economics. The article is completed by a significative bibliography.
European Journal of Economics and Social Sciences, 15(3), 2001
In this methodological work I explore the possibility of explicitly modelling expectations conditioning the R&D decisions of firms. In order to isolate this problem from the controversies of cognitive science, I propose a black box strategy through the concept of ``internal model''. The last part of the article uses artificial neural networks to model the expectations of firms in a model of industry dynamics based on Nelson & Winter (1982).
This article aims to test the relevance of learning through Genetic Algorithms (GA) and Learning Classifier Systems (LCS), in opposition with fixed R&D rules, in a simplified version of the evolutionary industry model of Nelson and Winter. These three R&D strategies are compared from the points of view of industry performance (welfare): the results of simulations clearly show that learning is a source of technological and social efficiency.
Computational Economics, 2002, 19:51-65, 2002
This article aims to test the relevance of learning through Genetic Algorithms, in opposition with fixed R&D rules, in a simplified version of the evolutionary industry model of Nelson and Winter. These two R&D strategies are compared from the points of view of industry performance (welfare) and firms' relative performance (competitive edge): the results of simulations clearly show that learning is a source of technological and social efficiency as well as a mean for market domination.
Expectations and Adaptive Behaviours: the Missing Trade-off in Models of Innovation
We explore the modelling of the determination of the level of R&D investment of firms. This means that we do not tackle the decision of being an innovator or not, nor the adoption of a new technology. We exclude these decisions and focus on the situations where firms invest in internal R&D in order to produce an innovation. In that case the problem is to determine the level of R&D investment. Our interest is to analyse how expectation and adaptation can be combined in the modelling of R&D investment rules. In the literature both dimensions are generally split up: rational expectations are assumed in neoclassical models whereas alternative approaches (institutional and/or evolutionary) generally adopt a purely adaptive representation.
Sources of Technological Diversity
Economic theory has recently recognized that local interaction is a source of aggregate diversity. In this paper, we examine possible sources of intra-industry technological diversity. We study two distinct representations of firms interactions, which respectively take place in the space of firms and along a technological trajectory. Technological change is explicitly localized and firms productivities are altered by technological externalities. Moreover, the network of inter-firm relations constrains the diffusion of new technologies. We study the resultant effects and give qualitative results concerning their respective importance. We show that localization of learning can be a source of diversity but technological externalities do not favor technological diversity.
Formalizations of Technological Flexibility à la Stigler
After having exposed the technological flexibility concept which appears in the seminal work of Georges Stigler, we analyse different formalizations proposed in the literature. At the end, we propose a general formalization independent of the structure of cost functions. (in French)
Technological Diversity in an Evolutionary Industry Model with Localized Learning and Network Externalities
Structural Change and Economic Dynamics, vol 9(1), pp.33-51, 1998
We extend the model of Nelson and Winter with a lattice--based spatial structure in order to study the effects of localized learning and increasing returns to adoption on the emergence and persistence of technological diversity. We show that an intermediate level of localization of learning can effectively be a source of diversity. Network effects give rise to the lock-in effect even if the technology set is endogenously modified by innovations. Combination of localized learning and network externalities can nevertheless imply a greater diversity compared to Nelson and Winter's model.
Market Occupation in a Circular Market: Product Innovation and/or Patenting
We study two strategies that an incumbent can use in order to occupy his market and to blockade entry: variety proliferation (by product innovation) and patenting. We use Salop's Circular City model in the study of this differentiated good market. We show that the incumbent will generally choose variety proliferation even if he can protect his market by many patents.
Asymmetric Information and Welfare: The retail market
The agency problem between a producer and a distributor has two facets. At the one hand, the existence of an incomplete information implies some efficiency loss because of the necessity to incite the distributor to reveal his private information. At the other hand, the market power induced by this exclusive relationship conduces to a monopoly position. We show that the public authority can resolve the latter problem by introducing a tax which will establish a competitive position. The first problem is quite tricky and it can only be resolved if the public authority is better informed than the producer. (in French)
Technology Choice and Tip-toe Entry under Incomplete Information
This model extends the analysis of the barriers to entry in a context where the potential entrant can choose his technology before the entry. When studied under complete information, this model is very close to the models of ''strategic investment". But we show that the entrant can always put his toe on the market: the variable scale entry is always possible. Under incomplete information our model extends the ''limit--pricing'' models to a setting where the incumbent's signal can be given by any relevant economic decision.
Technological Flexibility vs Efficiency and Variable Scale Entry
This model extends the analysis of the barriers to entry in a context where the potential entrant is given the opportunity to choose his technology before the entry. This choice is characterized as a particular trade--off between technological flexibility and technological efficiency. This model is very close to the models of ''strategic investment" . But we show that the technology of the incumbent can never preclude entry and the entrant can always put his toe in the market: the variable scale entry is always possible.
Investissement stratégique, incertitude et effet d'irréversibilité
Annales de l'Économie et de Statistique, n°35, 1994
In a simple model of two stage competition, we show that the strategic effect of an irreversible commitment in ex ante capacity can an insufficient motivation to blockade the entry for the incumbent if the ex post demand on the market is uncertain. In this context, we show that this strategic effect must be strong enough to counterbalance the option value of the irreversible commitment.
Return to the
Top of this page
Note importante : Tous les documents disponibles sur ce site sont fournis tels quels. Quand de nouveaux documents deviendront disponibles, ils seront placés sur ce site. Veuillez ne pas me contacter pour me demander des documents supplémentaires (exercices, polycopiés...).