Applying Non-Homogeneity of Variance in The Error to The Investment Function in Iraq

Abstract

The present study aims at studying the problem of variance in the random error in the multiple linear regression sample by using investment function in Iraq from 1980 to 2015 and testing it by using standard methods by comparing the weighted least squares method, which is considered to be one of the ways to treat the problem, with the ordinary least squares method, (WLS) using the efficiency scale. As a result, it has been shown that the weighted least squares method is better than the ordinary least squares method (OLS).The major hypothesis of the present study is that random error does not have the problem of variance in the error. To achieve the aims of the present study, to validate the hypotheses, and to interpret the results, a sophisticated statistical package in this area has been used; (E views 9 program) in which chart plot test, Breusch Pajan, weighted least squares method have been used. The problem is that the dependent variable is limited because of the nature of the phenomenon, which may not lead to the validation of those assumptions. The study has adopted the standard quant itative analysis method with data time series for a random sample whose size is 36 from the Iraqi economy which has shown that there is a variance in the random error in the investment function; i.e. the variance is not equal in the investment function. That has led to some recommendations that may be useful to those concerned with the present study.