Comparison of Estimate Methods of Multiple Linear Regression Model with Auto-Correlated Errors when the Error Distributed with General Logistic

Abstract

In this research, we studied the multiple linear regression models for two variables in the presence of the autocorrelation problem for the error term observations and when the error is distributed with general logistic distribution. The auto regression model is involved in the studying and analyzing of the relationship between the variables and through this relationship, the forecasting is completed with the variables as values. A simulation technique is used for comparison methods depending on the mean square error criteria in where the estimation methods that were used are (generalized least squares, M robust and Laplace) and for different sizes of samples (20, 40, 60, 80, 100, 120). The M robust method is demonstrated the best method for all values of correlation coefficients as (ϕ = -0.9, -0.5, 0.5, 0.9). So, we applied it to the data that was obtained from the Ministry of Planning in Iraq/Central Organization for Statistics which represents the consumer price index for the years 2004-2016. So, we confirmed that the dollar exchange rate is directly affected by the increase in annual inflation rates and the ratio of currency to the money supply.