This dissertation examines the forecast model selection problem in economics in both theoretical and empirical settings. The forecast model selection problem is that there often exists a menu of different suitable models to forecast the same economic variable of interest. The theoretical portion of this dissertation considers agents who face this problem in two distinct scenarios. The first scenario considers the case where agents possess a menu of different forecast techniques which includes rational expectations but where the selection of rational expectations is costly. The assumptions that are necessary to include rational expectations as a choice are characterized and the equilibrium dynamics of a model under the appropriate assumptions is studied and shown to exhibit chaotic dynamics. The second scenario considers agents who possess a menu of econometric forecast models and examines the equilibrium outcomes when agents combine the different forecasts using strategies suggested by the forecasting literature. The equilibrium outcomes under these forecasting assumptions are shown to exhibit time-varying volatility and endogenous structural breaks, which are common features of macroeconomic data.
The empirical portion of the dissertation proposes a new dynamic combination strategy for the forecast model selection problem to forecast inflation. The procedure builds on recent research on inflation persistence in the U.S. and on explanations for the efficacy of simple combination strategies, often referred to as the forecast combination puzzle. The new combination strategy is shown to forecast well in real-time out-of-sample forecasting exercises.
| Identifer | oai:union.ndltd.org:uoregon.edu/oai:scholarsbank.uoregon.edu:1794/13279 |
| Date | 03 October 2013 |
| Creators | Gibbs, Christopher |
| Contributors | Evans, George |
| Publisher | University of Oregon |
| Source Sets | University of Oregon |
| Language | en_US |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
| Rights | All Rights Reserved. |
Page generated in 0.0022 seconds