Feasible Generalized Least Squares Using Machine Learning
20 Pages Posted: 10 May 2017 Last revised: 11 May 2017
Date Written: May 10, 2017
In the presence of heteroskedastic errors, regression using Feasible Generalized Least Squares (FGLS) offers potential efficiency gains over Ordinary Least Squares (OLS). However, FGLS adoption remains limited, in part because the form of heteroskedasticity may be misspecified. We investigate machine learning methods to address this concern, focusing on Support Vector Regression. Monte Carlo results indicate the resulting estimator and an accompanying standard error correction offer substantially improved precision, nominal coverage rates, and shorter confidence intervals than OLS with heteroskedasticity-consistent (HC3) standard errors. Reductions in root mean squared error are 87-98% of those achievable when the form of heteroskedasticity is known.
Keywords: Heteroskedasticity, Feasible Generalized Least Squares, Machine Learning
JEL Classification: C13, C21
Suggested Citation: Suggested Citation