Three essays on economic forecasting and theory examination

Abstract

In the first chapter, Monte Carlo simulation and bootstrap methods are used to compare the actual and nominal coverage probabilities of prediction intervals constructed using the Prais-Winsten modified weighted symmetric least squares (PW-MWSLS) estimation method. The evidence suggests that the PW-MWSLS estimator, the best point predictor, for the linear trend model with first-order autoregressive errors also leads to prediction intervals with the most accurate coverage rates for the linear trend model with first-order autoregressive errors.;The second chapter employs an innovative methodology to construct inflation expectations by incorporating information in the commodity futures market. The empirical results from the vector dynamic system show that the constructed expected rate of inflation series provides the best in-sample and out-of-sample forecasts over the sample period under investigation.;Chapter three applies the constructed time series of inflation expectations in the second chapter to examine two broadly debated topics in the field of economics, the Fisher effect and the Phillips curve. The findings provide support for the existence of the short-run Fisher effect; and for the examination of the two main alternative specifications of the Phillips curve, the New Keynesian Phillips curve and the expectations-augmented Phillips curve, the empirical evidence is in favor of the former.