Please use this identifier to cite or link to this item:
Title: Estimating Multi-Beta Pricing Models : With or Without an Intercept: Further Results from Simulations
Authors : Akermann, Florin Ernst
Advisors / Reviewers : Bänziger-Aiba, Armin
Extent : 45
Publisher / Ed. Institution : ZHAW Zürcher Hochschule für Angewandte Wissenschaften
Issue Date: 2016
License (according to publishing contract) : CC BY 4.0: Attribution 4.0 International
Language : English
Subject (DDC) : 658.8: Marketing management
Abstract: The two-pass method is a common approach for estimating risk premia and examining factor pricing models. It consists of a time-series regression (first-pass) and a cross-sectional regression (second-pass). Two common problems of this approach are the downward bias of the estimates and the large standard error of the estimates. A previous study showed through a simulation approach that the problem of the bias can be mitigated by running at least one of the two regressions without an intercept, whereas the problem of the large standard error can be mitigated by running the second regression without an intercept. The previous study used a single factor pricing model as the underlying model for its simulation. The objective of this thesis was to provide further evidence for this mitigation method (leaving out the intercepts) by analyzing the mitigating effects in the case of the Fama and French three-factor model. For this purpose, the simulation was based on the simulation of the previous study and was extended to suit the properties of the threefactor model. The simulation comprised two main parts: first, the test data was generated artificially, then the two-pass method was applied on each set of this generated data. Similar to the findings of the underlying study, it was found that omitting the intercept in at least one of the two regressions decreases the bias of the estimated market premium. Furthermore, omitting the intercept in the cross-sectional regression decreases the standard deviation of the market premium estimates. However, for the two additionally estimated risk premiums (size and value), the mitigating effect on the biases is hardly observable, as the biases of these estimates are already small without omitting the intercept in either of the two regressions. Moreover, the standard errors of the estimates for the size and value premiums did not decrease when the intercept was omitted in at least one of the regressions. In all of the applied variants, the standard error of the estimates for the three premiums are persistently large. Therefore, even with this partially effective mitigation method, it remains difficult to draw statistical conclusions from the two-pass method.
Departement: School of Management and Law
Publication type: Bachelor Thesis
DOI : 10.21256/zhaw-1241
Appears in Collections:BSc Betriebsökonomie

Files in This Item:
File Description SizeFormat 
Akermann_Florin_W.BA.BO.BF (PiE).pdf996.77 kBAdobe PDFThumbnail

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.