الفهرس 
Title : Handling Outliers in Seemingly Unrelated Regression Equations Model
ContantsOnly 14 pages are availabe for public view
 |  | from | 158 |  |  |
| | | from | 158 | | |
Abstract
The concept of outliers for seemingly unrelated regressions equations (SURE) model is discussed and the consequences of outliers. Several approaches for detecting outliers are presented and shown to lead to the same diagnostic procedure.The problem of outliers is one of the oldest in econometrics and statistics, during the last century; interest in it has waxed and waned several times, one of the main problems in regression estimation methods. The SURE model is one of the econometric developments that have found considerable use in applied statistics.SURE model is one regression multivariate case, which have especial assumption, i.e., correlation between errors on the multivariate linear models (MLMs), by considering multiple regression equations that are linked by contemporaneously correlated disturbances. The assumptions underlying most SURE estimators give little consideration to influential observations, which may be present in the dataset. To overcome this problem, robust estimation is commonly applied to solve the problem caused by outliers.This thesis introduces a comparative study for some different robust estimators in SURE model.This is achieved by simulation study and empirical application to evaluate the robust estimator.The Monte Carlo simulation and application results indicate that the (non-robust) Ordinary Least Squares (OLS), Maximum Likelihood (ML) and Feasible Generalized Least Squares (FGLS) estimators are very sensitive to outliers, while robust (M-estimate, S-estimate and MM-estimate) estimators are more effective. In addition, MM-estimation method is more efficient, and the MM-estimator outperforms the other estimators in terms of relative absolute bias (RAB), total mean squared error (TMSE) and total mean absolute error (TMAE)