Détermination, apprentissage adaptatif et règles de politique monétaire dans

This paper presents the concepts, modelling methods and analytical tools used by the literature on adaptive learning in the new canonical macroeconomic model. The paper?s aim is to provide all elements needed to appreciate the issues and insights which are related to the introduction of learning into this model. Rational expectations which are a building block of the original model can thus be replaced by expectations based on estimation algorithms. The permanent updating of these estimations can be interpreted as a learning process of the model?s agents. This learning process induces additional dynamics into the model. The literature in question uses two criteria in order to analyse the ability of monetary policy rules to stabilise the economy: (i) uniqueness of the (non-explosive) rational expectations equilibrium in the original model and (ii) stability in regards to learning in the modified model. Taking learning into account enables to detect shortcomings of a rule according to (ii) that would not been seen in a rational expectations model. The paper ends with a discussion on the specific problems met in the introduction of a learning process in the new canonical model.