The purpose of this study is to introduce several criteria and to find the best method for estimating dynamic models with the autoregressive moving average residuals. First, Monte Carlo simulations are used to compare the two stage least square method with the ordinary least square estimates of the residuals (2SLS/OLS), the 2SLS method with Box-Jenkins' estimates of the residuals (2SLS/Box-Jenkins), as well as the 2SLS method with the recursive maximum likelihood method on the residuals (2SLS/ML). The results show that the 2SLS/ML method performs best within the stationary constraints. When the residuals are white noise or out of stationary constraints, the 2SLS/OLS method performs best. The results are ranked according to the criteria of mean square errors (MSE), variance, bias squared, mean absolute deviations (MAD), and the percentage of prediction errors (PPE). / Second, the simulations are used to compare the relative effects of the ARMA errors on the estimates of regression coefficients. We compare the ordinary least square (OLS) and the generalized least square method with the recursive maximum likelihood estimates of the residuals (GLS/ML), and the maximum likelihood method with the maximum likelihood estimates of the autoregressive residuals (ML/ML). The results show that within the unit circle, the ML/ML estimator performs best according to four criteria: deviations between the estimates and the true coefficients, their frequency distributions, the standard deviations of the estimates of dependent variables, and the adjusted coefficient of determination. / Source: Dissertation Abstracts International, Volume: 42-06, Section: A, page: 2792. / Thesis (Ph.D.)--The Florida State University, 1981.
Identifer | oai:union.ndltd.org:fsu.edu/oai:fsu.digital.flvc.org:fsu_74530 |
Contributors | HSIEH, HSIH-CHIA., Florida State University |
Source Sets | Florida State University |
Detected Language | English |
Type | Text |
Format | 302 p. |
Rights | On campus use only. |
Relation | Dissertation Abstracts International |
Page generated in 0.0013 seconds