Macroeconometrics with Bayesian estimation of dynamic stochastic general equilibrium models

This dissertation presents three essays on macroeconometrics. Their common denominator is the use of Bayesian techniques to estimate and analyze dynamic stochastic general equilibrium (DSGE) models of the US economy.; The first chapter evaluates the performance of simple monetary policy rules in an estimated DSGE model of the US economy. The principal welfare criterion is derived from a second-order approximation of the household's utility function and a proper solution method allows us to work with the distorted steady-state of the economy. The estimation of the DSGE model allows the consideration of parameter uncertainty. Our findings show (i) that wage inflation is a better anchor than price inflation, and (ii) that parameter uncertainty matters. While the effect of the latter on the level of welfare is small, perturbations of the parameters within the bounds of the posterior distribution, for a given policy, can significantly increase the variance of the nominal interest rate to levels that are at odds with the conventional views.; The second chapter estimates and compares, within unified framework, three popular pricing models for the US economy: the New Keynesian model, the Wolman model and versions of the sticky-information framework developed by Mankiw and Reis. We interpret our results as favoring the Wolman pricing mechanism presented over the New Keynesian (NK) model with indexation and the sticky information model of Mankiw and Reis. The key factor that explains the performance of the Wolman model is that the data reject the key assumption of the NK model that the firm's probability of price change is constant over time and independent of the contract's vintage.; The nature of the third chapter is more methodological than those of the other chapters. It investigates the implications of accounting for the heteroskedastic nature of the economic shocks on the estimation and study of the model. We find that even though modeling heteroskedasticity improves the model's fit, it does not have much affect on the posterior distribution; the most significant difference appears to be that the persistence of the exogenous shocks are in general more tightly estimated under heteroskedasticity.