Revisiting Individual Evolutionary Learning in the Cobweb Model - An Illustration of the Virtual Spite-Effect

We examine the Cournot oligopoly model in the context of social and individual learning. In both models of learning, firms update their decisions about how much to produce via variants of the genetic algorithm updating procedure. Arifovic (1994) found that both models of social and individual learning converged to the Walrasian, competitive equilibrium. Vriend (2000) reports that the model of social learning converges to the Walrasian equilibrium outcome, while the model of individual learning converges to the Cournot-Nash equilibrium. We revisit the issue and conduct simulations varying elements of the updating algorithms, as well as of the underlying economic model. In the analysis of the outcomes of our simulations, we conclude that the convergence to the Cournot-Nash equilibrium is due to two things: the specific way in which production rules’ performance is evaluated coupled with a specific cost function specification. JEL classification: B41, C63, C81, D83, H41.