Macroeconomic dynamics and adaptive learning

A key feature of modern macroeconomic modelling is the expectations of economic agents. Since expectations play a central role in the analysis of macroeconomic variables, a natural question is how they are formed. For many years, research in macroeconomics has been dominated by the rational expectations hypothesis, i. e. the hypothesis that agents' expectations about the future are correct on average. More recently, there has been a large literature that re-examines the way expectations are formed. In macroeconomics, limited rationality assumes that agents may form rational expectations in the long run, but are uncertain about the path to this equilibrium, and that expectations are revised in each period by taking into account the forecasting error. One way to model limited rationality is adaptive learning. In this framework, economic agents act as statisticians that use econometric rules to forecast the future state of the economy. The present thesis contributes to the literature of analysing macroeconomic dynamics with adaptive learning. The first part of the thesis, titled `Heterogeneous Learning', studies adaptive learning dynamics in a broad class of linear stochastic macroeconomic models, when the agents in the economy are heterogeneous. Agents may differ in (a) their initial perceptions about the evolution of the economy, (b) the degrees of inertia in revising their expectations or (c) the learning rules they use to forecast the future state of the economy. The first chapter provides an introduction, a review of the related literature and the basic framework. The second chapter provides conditions such that the agents' expectations become rational in the long run or, in other words, conditions such that the economy converges to the equilibrium predicted by the rational expectations hypothesis. In the third chapter, the above results are applied to several examples of standard macroeconomic models. The second part of the thesis, titled "Supply-Side Economics and Learning", applies adaptive learning to a macroeconomic policy issue, namely an analysis of the effects of cutting taxes on capital income. The first chapter of this part is an introduction and a review of the literature. The second chapter focuses on the analysis of two important aspects: (a) the identification of the nature and the magnitude of the short run effects of such a reform, and (b) the length of the transition after the reform. The aim of this work is twofold. First, to compare the results from this analysis with the corresponding results predicted by the rational expectations hypothesis; second, to give a policy recommendation about when it is advisable for such tax cuts to be implemented. It turns out that the transition dynamics with adaptive learning are very different from the ones with rational expectations. Furthermore, and contrary to common belief, under the assumption of adaptive learning, cutting capital income taxes before or during a recession may not be an effective means for short-run fiscal stimulus