Robust Regression Estimation Methods and Intercept Bias: A Capital Asset Pricing Model Application

29 Pages Posted: 26 Jun 2015

See all articles by James McDonald

James McDonald

Brigham Young University

Richard A. Michelfelder

Rutgers, The State University of New Jersey - Rutgers University, Camden

Panayiotis Theodossiou

Cyprus University of Technology

Date Written: June 25, 2015

Abstract

Robust estimation techniques based on symmetric probability distributions are often substituted for OLS to obtain efficient regression parameters with thick-tail distributed data. The empirical, simulation and theoretical results in this paper show that with skewed distributed data, symmetric robust estimation techniques produce biased regression intercepts. This paper evaluates robust methods in estimating the capital asset pricing model and shows skewed stock returns data used with symmetric robust estimation techniques produce biased alphas. The results support the recommendation that robust estimation using the skewed generalized T family of distributions may be used to obtain more efficient and unbiased estimates with skewness.

Keywords: CAPM; quasi-maximum likelihood estimator; robust estimator; skewed generalized T

JEL Classification: G12, C13, C14, C15

Suggested Citation

McDonald, James B. and Michelfelder, Richard A. and Theodossiou, Panayiotis, Robust Regression Estimation Methods and Intercept Bias: A Capital Asset Pricing Model Application (June 25, 2015). Multinational Finance Journal, Vol. 13, No. 3/4, p. 293-321, 2009, Available at SSRN: https://ssrn.com/abstract=2623003

James B. McDonald (Contact Author)

Brigham Young University ( email )

130 Faculty Office Bldg.
Provo, UT 84602-2363
United States
801-378-3463 (Phone)

Richard A. Michelfelder

Rutgers, The State University of New Jersey - Rutgers University, Camden ( email )

Camden, NJ 08102
United States

Panayiotis Theodossiou

Cyprus University of Technology ( email )

Limassol, 3603
Cyprus

Here is the Coronavirus
related research on SSRN

Paper statistics

Downloads
65
Abstract Views
741
rank
380,358
PlumX Metrics