Essays on real business cycle modelling under adaptive learning

The thesis consists on three chapters aiming to contribute to a growing literature on adaptive learning, a form of bounded rationality that has been attracting increasing interest both in the theoretical and practical fields, as an alternative to the commonly used rational expectations hypothesis on how expectations are formed among economic agents. The first chapter investigates whether it is possible to improve the ability of the standard real business cycle model to match the main stylised facts of emerging economies, taking the case of Mexico as an illustration, by assuming that agents are not fully rational and instead form expectations according to an adaptive learning rule. Two well-known rules - recursive least squares and its constant gain variant - are considered for this purpose. The degree of difficulty of the learning process is characterised by different starting values of the algorithms as well as different constant gains. The simulations show that the model under learning generally outperforms its rational expectations counterpart. Therefore, policymakers should take into account the fact that the expected welfare gains/losses of a particular policy reform, conceived assuming a fully-rational environment, might be significantly different if, in practice, agents behave as learners. Using a heterogeneous-agent model with three types of agents, namely capitalists, skilled workers and unskilled workers - assuming constant population shares suggesting low social mobility -, and allowing for different degrees of complementarity among these within the productive structure, the second chapter welfare-evaluates tax reforms consistent with a lower long run debt-to-output ratio for the United Kingdom, both under rational expectations and heterogenous learning. This chapter shows that, relative to the other tax reforms, capital tax cuts lead to the highest aggregate welfare but are skill-biased and can thus increase inequality in the long run. That is, depending on the elasticity of substitution between capital and unskilled labour, falls in the capital tax can result in higher levels of welfare inequality, even in the absence of other frictions and increases in other forms of taxation. On the other hand, reductions in labour taxes can hurt the capitalists. This chapter shows too that including the transition period in the welfare evaluation lowers the inequality effects of capital tax reductions since the complementarity between capital and all labour inputs is higher in the short- than in the long-run. Finally, while heterogeneous learning in the shape of differing initial beliefs after the reform can lead to a form of "irrational exuberance" after a tax cut, it can also exacerbate welfare inequality. Finally, the third chapter presents an heterogeneous-agent model with two types of agents, capitalists and workers - with constant population shares given the strong evidence on low social mobility -, calibrated to Bolivia´s data in order to examine the short and long-run effectiveness and distributional effects of various fiscal rules designed to impose restrictions on the evolution of public debt as a share of output, in response to two different sources of exogenous volatility (i.e. productivity and commodity shocks) and under different ways of forming expectations, namely rational expectations and heterogenous learning. The results show that under full rationality the fiscal rules generate a trade-off between debt-stabilisation and higher income inequality while, under some conditions, heterogenous learning can help to break such trade-off so that some of the rules can perform well in both fields. However, given the significantly high levels of income inequality and dependence on commodity revenues experienced by Bolivia, finding the best performing rule in response to all the relevant exogenous shocks this economy might face, appears to be a challenging task