الفهرس 
Title : Some estimation methods for vector autoregression model
ContentsOnly 14 pages are availabe for public view
 |  | from | 125 |  |  |
| | | from | 125 | | |
Abstract
The Vector Autoregressive (VAR) model is considered the most flexible and easiest way to use model in multivariate time series and is considered the most successful one. It is a natural extension of the univariate autoregressive model to a multivariate dynamic time series. Previous studies showed that the VAR model is particularly useful in describing the dynamic behaviors of financial and economic time series and making forecasts as well. It provides better predictions than univariate time series and your theory-based exemplary simultaneous equation models.
This thesis introduces a comparative study of some different estimators in the VAR model. This is achieved by a simulation study and empirical application to evaluate the estimator. The Monte Carlo simulation study results showed that the Akaike information criterion (AIC), Bayesian information criterion (BIC), and mean squared error (MSE) values for the Ridge VAR (BVAR) estimator is smaller than the AIC, BIC, and MSE values of the Maximum Likelihood VAR (MLVAR) and Bayesian VAR (BVAR) estimator for all cases of the simulation. Thus, the estimator of RVAR is better than the other estimators in all different sample sizes. In addition, the results of the empirical application confirmed the simulation results, which indicated that the RVAR estimator is better.