The use of the weighted least squares method when the error variance is heteroscedastic

Abstract

Homoscedastic property of a variance of the errors in a linear regression is among the assumptions of the Ordinary Least Square (OLS) method. When this assumption of homoscedasticity is violated, it causes the regression coefficients to be biased and inconsistent. Various methods have been used in the literature to detect the presence of heteroscedasticity. This study compares two of the existing methods of detecting the presence of heteroscedasticity. The two methods are; Goldfeld-Quandt (GQ) and Breusch Pagan-Gofrey (BP) test. Results show that the GQ test is better than the BPG test in terms of their P-values. In the presence of heteroscedasticity, the study adopts the method of Weighted Least Squares (WLS) to circumvent the problems of associated with heteroscedasticity.