Performance of some biased estimators in misspecified model under mahalanobis loss function

In econometric research,misspecification due to omission of relevant variables is a very common phenomena,which leads to biased and inconsistent estimation of parameters. it is needless to mention that the explanatory variables in the misspecified model may be multicollinear and have adverse effects on the least squares estimator. to combat the problem of multicollinearity,several biased estimators have been introduced and widely analyzed under mean squared error criterion when there is no misspecification. furthermore,mahalanobis loss function,which is a weighted quadratic loss function has gained quite an attention as a comparison criterion in recent years. although,not much literature is available on the comparison of the estimators in the presence of multicollinearity under the mahalanobis loss function when there is no misspecification and is almost negligible in case of misspecification due to omission of relevant variables. this article evaluates the performance of some biased estimators in the presence of multicollinearity under the mahalanobis loss function using average loss criterion in the misspecified model. necessary and sufficient conditions for the dominance of one estimator over the others have been derived. further,the findings are supported by a monte carlo simulation and numerical example. © 2017 pakistan journal of statistics.