An Applied Study to Compare Some Estimators for Panel Data Model Under The Two Problems of Heteroscedasticity  and Autocorrelation

Jawad, Hussin Adnan; Karosa, Ahmed; Elatar, Labiba

doi:10.21608/acj.2023.311407

An Applied Study to Compare Some Estimators for Panel Data Model Under The Two Problems of Heteroscedasticity and Autocorrelation

Document Type : Original Article

Authors

¹ Master researcher Department of Statistics, Mathematical and Insurance Faculty of Commerce, University of Alexandria

² Lecturer in the Department of Statistics, Sports and Insurance Faculty of Commerce, University of Alexandria The Egyptian Arabic Republic

³ Department of Statistics, Sports and Insurance The Egyptian Arabic Republic

10.21608/acj.2023.311407

Abstract

The use of panel data models is one of the modern methods used in quantitative standard analysis. This study dealt with some methods of estimating panel data models in the event that the assumptions of heteroscedasticity and autocorrelation between those errors. The problem of the study is to answer the following question: What is the best estimator in light of the two problems of heteroscedasticity and autocorrelation among several estimators? Where a applied studywas designed on the volume of electric energy production in the General Directorate of Electric Power Production in the Republic of Iraq for the period (2013-2021), which represents the dependent variable, and three independent variables (the work component, the investment component, and the volume of electrical energy produced in the previous period). Using a set of tests on this data, it was found that there are two problems of heteroscedasticity and autocorrelation in the residual series. The estimate was made using several estimators, including: combined ordinary least squares, possible generalized least squares, robust cluster standard error estimators, and Newey-west standard errors. And the fixed effects model, and the random effects model, and the results of the statistical tests indicated that the fixed effects model is the appropriate model, as the FEM fixed effects model was able to solve the problems of heteroscedasticity and autocorrelation in the residuals, in addition to the significance of the model.

Keywords