Essays On Macroeconomics And Econometrics PDF Download

The dissertation is composed of three chapters that contribute to the areas of macroeconomics and econometrics. All chapters utilize computer-based simulation methods. Chapter one, "Percentile and Percentile-t Bootstrap Confidence Intervals: A Practical Comparison", employs a Monte Carlo study to compare the performance of bootstrap percentile-t, bootstrap percentile, and standard asymptotic confidence intervals in two distinct heteroscedastic regression models. Results are consistent across models but different among bootstrap algorithms employed, in that for the XY algorithm all three methods perform similarly, but for the wild algorithm the percentile-t significantly outperforms the other methods. The implications are that for the models considered in this study, practitioners who choose to utilize these types of bootstrap confidence intervals should opt for the percentile-t method coupled with the wild bootstrap algorithm. Chapter two, "Asset Pricing with Active Traders and Passive Investors: An Adaptive Learning Approach", constructs a model of active traders and passive investors in production and endowment based stochastic growth asset pricing frameworks. The model replaces rational expectations with an adaptive learning rule that forecasts future equity prices with econometric methods. The paper estimates the model and compares the results to U.S. economic and financial stylized facts. Numerical results show that the model of active traders and passive investors matches several of the stylized facts better than does a model that assumes a fully rational, representative agent. Chapter three, "A Simple Exchange Rate Model with Heterogeneous Agents", employs a simple monetary framework to construct a heterogeneous agent model of flexible exchange rates. The model replaces rational expectations with an adaptive learning rule that forecasts future exchange rates with econometric methods. The paper estimates the model and compares the results to foreign exchange market stylized facts. Numerical results show that the model matches several of the stylized facts better than does a model that assumes a fully rational, representative agent.