Advances and Applications in Statistics
Volume 26, Issue 1, Pages 1 - 23
(January 2012)
|
|
A MONTE CARLO SIMULATION STUDY FOR (GLS) AND (GMM) ESTIMATION
El-Houssainy Abd Elbar Rady, Ahmed H. Youssef and Tareq Abdul-Aziz Al-Doub
|
Abstract: Estimation in panel data is a common problem. Many approaches have been developed to face the estimation problems in panel data such as Generalized Least Square (GLS) and Generalized Method of Moment (GMM). The (GLS) technique used by Swamy [9] depends on Ordinary Least Square (OLS), whereas GMM used by Hansen [8] and Verbeek [4] depends on two-stage Least Squares (2-SLS). This study aims to set a comparison between these two techniques using Simple Panel Data (SPD) and Multiple Panel Data (MPD) in linear regression. The purpose of this comparison is to determine which technique is considered a better estimation technique through a simulation study using various sample sizes, models, parameters and standard deviations. Moreover, the following criteria were used to evaluate the two techniques: bias, Mean Square Error (MSE), variances and the negative variances. Results of the study have revealed that (GMM) is more capable and accurate in estimation than (GLS) methodology. |
Keywords and phrases: panel data, generalized least square, generalized method of moment, simple linear regression, multiple linear regression, simulation study, estimation technique, bias, mean square error, negative variances. |
|
Number of Downloads: 587 | Number of Views: 875 |
|