Estimation of Linear Regression Model with Correlated Regressors in the Presence of Autocorrelation

When using the linear statistical model, researchers face variety of problems due to non experimental nature i.e uncertainity about the nature of the error process, model mis- specifications, dependent regressors etc. The phenomenon of correlated errors in linear regression models involving time series data is called autocorrelation. Violation of the assumption of independent regressors leads to multicollinearity. Hence, Ordinary ridge estimates are imprecise to be of much use in case of autocorrelated regression model with the multicollinearity problem. Objective: To develop a new estimator for the regression parameter in the presence of multicollinearity and autocorrelation. To choose an appropriate ridge parameter for the proposed estimator using Monte Carlo simulation. Materials and Methods: Monte Carlo simulation study is carried out using the Statistical programming language MATLAB version 7.0 to evaluate the performance of the proposed estimator based on the Mean squared error (MSE) criterion. Findings: Determined the regions where a particular method for estimating ridge parameter performs better among different existing methods. This estimate of ridge parameter is used in the proposed estimator. The proposed estimator performs better than the existing estimator under the MSE criterion.