TY - JOUR
T1 - Factors or characteristics? That is the question
AU - Gharghori, Philip
AU - Chan, Howard
AU - Faff, Robert
PY - 2006/3/1
Y1 - 2006/3/1
N2 - Daniel and Titman (1997) contend that the Fama-French three-factor model’s ability to explain cross-sectional variation in expected returns is a result of characteristics that firms have in common rather than any risk-based explanation. The primary aim of the current paper is to provide out-of-sample tests of the characteristics versus risk factor argument. The main focus of our tests is to examine the intercept terms in Fama-French regressions, wherein test portfolios are formed by a three-way sorting procedure on book-tomarket, size and factor loadings. Our main test focuses on ‘characteristicbalanced’ portfolio returns of high minus low factor loading portfolios, for different size and book-to-market groups. The Fama-French model predicts that these regression intercepts should be zero while the characteristics model predicts that they should be negative. Generally, despite the short sample period employed, our findings support a risk-factor interpretation as opposed to a characteristics interpretation. This is particularly so for the HML loading-based test portfolios. More specifically, we find that: the majority of test portfolios tend to reveal higher returns for higher loadings (while controlling for book-to-market and size characteristics); the majority of the Fama-French regression intercepts are statistically insignificant; for the characteristic-balanced portfolios, very few of the Fama-French regression intercepts are significant.
AB - Daniel and Titman (1997) contend that the Fama-French three-factor model’s ability to explain cross-sectional variation in expected returns is a result of characteristics that firms have in common rather than any risk-based explanation. The primary aim of the current paper is to provide out-of-sample tests of the characteristics versus risk factor argument. The main focus of our tests is to examine the intercept terms in Fama-French regressions, wherein test portfolios are formed by a three-way sorting procedure on book-tomarket, size and factor loadings. Our main test focuses on ‘characteristicbalanced’ portfolio returns of high minus low factor loading portfolios, for different size and book-to-market groups. The Fama-French model predicts that these regression intercepts should be zero while the characteristics model predicts that they should be negative. Generally, despite the short sample period employed, our findings support a risk-factor interpretation as opposed to a characteristics interpretation. This is particularly so for the HML loading-based test portfolios. More specifically, we find that: the majority of test portfolios tend to reveal higher returns for higher loadings (while controlling for book-to-market and size characteristics); the majority of the Fama-French regression intercepts are statistically insignificant; for the characteristic-balanced portfolios, very few of the Fama-French regression intercepts are significant.
UR - http://www.scopus.com/inward/record.url?scp=69249113692&partnerID=8YFLogxK
U2 - 10.1108/01140580610732769
DO - 10.1108/01140580610732769
M3 - Article
AN - SCOPUS:69249113692
SN - 0114-0582
VL - 18
SP - 21
EP - 46
JO - Pacific Accounting Review
JF - Pacific Accounting Review
IS - 1
ER -