Review of Quantitative Finance and Accounting, 24: 65–91, 2005
2005 Springer Science + Business Media, Inc. Manufactured in The Netherlands.
Overreaction after Controlling for Size
and Book-to-Market Effects and its Mimicking
Portfolio in Japan
DAVID C. CHENG
Department of Finance, National Dong Hwa University, Hualien, Taiwan, Tel.: 886-38633135; Fax: 886-38633130
Department of Economics, National Dong Hwa University, Hualien, Taiwan
Abstract. In this paper we observe that ﬁrm size (SZ) and book-to-market (BM) cannot fully explain stock returns
on prior-return- (PR-) based portfolios in the Japanese stock market. The overreaction effect after controlling for
the SZ and BM effects is signiﬁcant and persistent, and accounts for a large part of the zero-investment returns
on the loser to the winner. We therefore propose a new mimicking portfolio whose returns mimic the common
factor in returns related to overreaction. Our evidence shows that the proposed four-factor model captures common
variation in returns on portfolios, based on stocks’ SZ, BM, and PR, better than the well-known three-factor model
Keywords: size, book-to-market, prior return, overreaction
JEL Classiﬁcation: G11, G12
From the early Sharpe-Lintner version of CAPM and APT (Ross, 1976) to the recent three-
factor model (Fama and French (FF), 1993) and the characteristic-based model (Daniel
and Titman (DT), 1997; Daniel, Titman and Wei (DTW), 2001), economists and ﬁnancial
practitioners have devoted great efforts to identify variables that explain stock returns best.
For instance, the ﬁrm size (SZ) and the book-to-market (BM) are the two well-recognized
characteristics in predicting returns.
DeBondt and Thaler (1985, 1987) document the differences in returns between winners
and losers based on prior-returns in the U.S. market. Observing that past losers signiﬁcantly
outperform past winners, the authors assert that the phenomenon is attributed to investors’
overreaction. Gunaratne and Yonesawa (1997) observe a similar pattern in Japan.
In this paper, our tasks are twofold. First, we try to justify the relative importance of
the overreaction effect to the famous characteristic effects, such as the SZ and BM effects,
by the comparisons of return behaviors between characteristic- and PR-based portfolios
in the Japanese market. Second, given the observed importance of overreaction, we shall