Learning agents in a monopolistic competition framework

We propose an agent-based computational model in order to study a general equilibrium macro-economic system within a monopolistic competition setting. We address the framework of monopolistic competition introduced by a seminal contribution of Blanchard-Kiyotaki. We model a number of price-setting firms producing differentiated goods characterized by a constant elasticity of substitution. A representative buyer participates in the goods´ markets, supplies labor to firms and sets the level of wage. We assume a bounded rationality framework where economic agents learn to optimize their own utility in a strategic economic context. The model is suited to be represented by a normal-form game in order to perform a convergence analysis of learning dynamics with respect to game-theoretical solutions. Results point out that the general equilibrium solution la Blanchard-Kiyotaki results to be both a Nash equilibrium and Pareto optima allocation, but it is not the most frequently played solution. Indeed, the most played solution is characterized by the highest competition on the side of producers, resembling the Bertrand-like solution on price competition, and by the highest representative buyer´s welfare. A possible explanation of this finding is that in our setting the highest competition occurs in the price-setting behavior of firms who face residual demands, whereas the single representative buyer has not any opponent on wage setting thus being able to reach her optimal utility.