The first part of this paper presents the background information necessary for determining the direction of the study. The general forms of univariate and multivariate ARIMA models are presented, and time series and econometric models are compared and contrasted. Past work in both univariate and multivariate time series approaches to macroeconomic forecasting are reviewed. The definition of Granger causality is presented, along with Pierce's approach to measuring causality. / The second part of the study chooses a set of thirteen variables which describe the workings of the Florida economy. Using an analysis of causal flows among the variables, the variables are divided into clusters for purposes of estimating multivariate ARIMA models. The models are then estimated, described statistically and economically, and are compared to equations contained in an econometric model of the state's economy. Forecasts are derived from both types of models, and are compared using graphical analysis and four- and eight-quarter ahead root mean squared errors. / Multivariate ARIMA modeling is found to be successful when applied to a regional macroeconomy. The forecasts outperform those of the econometric model in eight of thirteen cases according to the four-quarter ahead root mean squared error and in nine of thirteen cases according to the eight-quarter ahead root mean squared error. / Source: Dissertation Abstracts International, Volume: 45-08, Section: A, page: 2595. / Thesis (Ph.D.)--The Florida State University, 1984.
| Identifer | oai:union.ndltd.org:fsu.edu/oai:fsu.digital.flvc.org:fsu_75358 |
| Contributors | JOHNSON, PAMELA HENDERSON., Florida State University |
| Source Sets | Florida State University |
| Detected Language | English |
| Type | Text |
| Format | 206 p. |
| Rights | On campus use only. |
| Relation | Dissertation Abstracts International |
Page generated in 0.002 seconds