Outlier Detection and Identification in Square root Production Function using Robust Regression

Abstract

Outliers can cause inflated error rates as well as serious distortions in parameter and statistic estimates. Outliers in statistical analyses can have both positive and adverse implications. For instance, they increase the error variance and reduce the power of the statistical test. They have the potential to reduce normality and severely bias or affect estimates that may be of tangible significance. For detecting and identifying outliers in large data sets, robust regression estimators can be a very useful tool. Since they are unaffected by outliers, robust regression methods are implemented to solve this problem. In this paper we have used simulation data to examine the behaviour of estimates of the Square root production function in the presence of HLP. We have used MM estimation, S estimation and LTS robust techniques in our study. Robust regression handles outliers in a regression and produces different coefficient estimates than OLS.