Advanced

Empirical tests of Fama-French three-factor model and Principle Component Analysis on the Chinese stock market

Guo, Jingjing LU and Wang, Kaiwen (2014) BUSN88 20141
Department of Business Administration
Abstract
Date: 2014-06-03
Authors: Kaiwen Wang Jingjing Guo
     fin13kwa@student.lu.se fin13jgu@student.lu.se
Mobile: 0762063660 0762187877
Title: Empirical tests of Fama-French three-factor model and Principle Component Analysis on the Chinese stock market
Tutor: Anders Vilhelmsson, Department of Business Administration, Lund University
Purpose: This paper aim to verify that the Fama-French three factor model (FF) captures more cross-sectional variation in returns for the Chinese stock market than the CAPM, over the period January 2004 to December 2013. Furthermore, we construct statistically optimal factors by using the principal component analysis (PCA) for the Fama-French portfolios and test whether the FF model leaves anything... (More)
Date: 2014-06-03
Authors: Kaiwen Wang Jingjing Guo
     fin13kwa@student.lu.se fin13jgu@student.lu.se
Mobile: 0762063660 0762187877
Title: Empirical tests of Fama-French three-factor model and Principle Component Analysis on the Chinese stock market
Tutor: Anders Vilhelmsson, Department of Business Administration, Lund University
Purpose: This paper aim to verify that the Fama-French three factor model (FF) captures more cross-sectional variation in returns for the Chinese stock market than the CAPM, over the period January 2004 to December 2013. Furthermore, we construct statistically optimal factors by using the principal component analysis (PCA) for the Fama-French portfolios and test whether the FF model leaves anything significant that can be explained by the PCA factors.
Method: Following the procedure in Fama and French (1993), first we construct FF factors and portfolios based on firm size and book-to-market equity, and then compare the performance between CAPM and FF models by applying time-series regressions. For deeper comparison, we continue to explain the return matrix (120*9) with principal component analysis, which produces several PCs for new time-series regressions and study the overall fitness and factor loadings of both FF and PCA models. To see which model captures the most variation, we run cross-sectional regressions with respect to all the three afore-mentioned models.
Conclusion: Our results show that the FF model tends to be more powerful than CAPM for explaining the variations in cross-sectional returns. Yet within the FF model, our data contains one divergence from the US market, we actually find a reversal of book-to-market equity effect. Finally, our results suggest that the PCA model performs better than the FF model. (Less)
Please use this url to cite or link to this publication:
author
Guo, Jingjing LU and Wang, Kaiwen
supervisor
organization
course
BUSN88 20141
year
type
H1 - Master's Degree (One Year)
subject
language
English
id
4462911
date added to LUP
2014-07-02 09:26:05
date last changed
2014-07-02 09:26:05
@misc{4462911,
  abstract     = {Date: 2014-06-03
Authors: Kaiwen Wang Jingjing Guo
     fin13kwa@student.lu.se fin13jgu@student.lu.se
 Mobile: 0762063660 0762187877
Title: Empirical tests of Fama-French three-factor model and Principle Component Analysis on the Chinese stock market
Tutor: Anders Vilhelmsson, Department of Business Administration, Lund University
Purpose: This paper aim to verify that the Fama-French three factor model (FF) captures more cross-sectional variation in returns for the Chinese stock market than the CAPM, over the period January 2004 to December 2013. Furthermore, we construct statistically optimal factors by using the principal component analysis (PCA) for the Fama-French portfolios and test whether the FF model leaves anything significant that can be explained by the PCA factors.
Method: Following the procedure in Fama and French (1993), first we construct FF factors and portfolios based on firm size and book-to-market equity, and then compare the performance between CAPM and FF models by applying time-series regressions. For deeper comparison, we continue to explain the return matrix (120*9) with principal component analysis, which produces several PCs for new time-series regressions and study the overall fitness and factor loadings of both FF and PCA models. To see which model captures the most variation, we run cross-sectional regressions with respect to all the three afore-mentioned models.
Conclusion: Our results show that the FF model tends to be more powerful than CAPM for explaining the variations in cross-sectional returns. Yet within the FF model, our data contains one divergence from the US market, we actually find a reversal of book-to-market equity effect. Finally, our results suggest that the PCA model performs better than the FF model.},
  author       = {Guo, Jingjing and Wang, Kaiwen},
  language     = {eng},
  note         = {Student Paper},
  title        = {Empirical tests of Fama-French three-factor model and Principle Component Analysis on the Chinese stock market},
  year         = {2014},
}