* Corresponding author: Dr. Shinn-Juh Lin at 64, SEC. 2, Tz-Nan Rd., Wenshan, Taipei 116, Taiwan. Tel.: + 886 2 29393091x81106; fax: + 886 2 29387699, email: [email protected].
2
1 Introduction
On security market, it is typical to observe stock prices that have been increasing in the past keep increasing in the short to intermediate future, and hence produce the so-called
momentum effect. Jegadeesh and Titman (1993) document that firms with high returns over the past three months to one year continue to outperform firms with low past returns over the same period. Since then, researchers have attempted to examine such momentum effect from different perspectives. Moskowitz and Grinblatt (1999) show that individual stock momentum is largely driven by industry momentum, and that stocks within the same industry tend to be more highly correlated than stocks across industries. George and Hwang (2004) find that the 52-week high price explains a large portion of the profits from momentum investing. Park (2010) proposes that investors’ anchoring bias by using the moving averages or the 52-week high as reference points for estimating fundamental values is the primary source of
momentum effects. He finds that the moving average ratio (MAR, ratio of short-term moving average to the long-term moving average) combined with nearness to the 52-week high explains most of the intermediate-term momentum profits. Blitz, Huij and Martens (2011) argue that conventional momentum strategies exhibit substantial time-varying exposures to the Fama and French factors. They propose a residual momentum strategy which is based on residual returns estimated using the Fama and French three-factor model. By examining the profitability of residual momentum strategy using the all U.S. domestic data that covers the period from January 1926 to December 2009, they find that the residual momentum strategy succeeds in improving upon a total return momentum strategy of Jegadeesh and Titman (1993, 2001). Chan, Jegadeesh and Lakonishok (1996) find that medium-term return continuation can be explained in part by under-reaction to earnings information. They show that earnings
3
momentum strategies are profitable even among larger stocks and that the profitability cannot be explained by the Fama-French three-factor model. To summarize, the price-, the industry-, the 52-week high-, the moving average ratio-, the residual- and the earnings- momentum are the six most often examined momentum strategies.
In Taiwan’s stock market, profitability of the price-, the earnings-, the industry- and the 52-week high moment strategies have been separately examined, while the MAR- and the residual momentum strategies have not received due attention in the literature. Intriguingly, these studies have not demonstrated unambiguous results regarding profitability of each momentum strategy. More importantly, there has not been a comprehensive study nor
comparison of profitability of popular moment strategies in Taiwan’s stock market. Therefore, by employing data of common stocks listed on the Taiwan Stock Exchange from January 1981 to December 2010, this paper proposes to thoroughly examine profitability of all those six popular momentum strategies with different holding periods. With pairwise comparison, we compare profitability of momentum strategies in pairs. Furthermore, we follow regression model of George and Hwang (2004) to to ascertain whether profitability of a particular momentum strategy is significantly better than others. As robustness checks, we also examine whether our empirical results are affected by the January effect, different lengths of the portfolio formation period, and different market states (bull or bear markets).
Overall, our empirical results show that the 52-week high-, the residual-, and the earnings-momentum strategies generate significantly positive profit, with the 52-week high-momentum strategy stands out as the most profitable momentum indicator in Taiwan’s stock market. With different lengths of portfolio formation period, we find that a price momentum portfolio formed according to the past 6 month returns can produce significant
4
positive profit, which is consistent with findings in Jegadeesh and Titman (1993). However, the industry momentum portfolios show no significant profit. In addition, the 52-week high- and earnings-momentum strategies both provide additional explanatory power to returns conditional on the other four momentum strategies. On the contrary, the other four momentum strategies do not offer additional explanatory power to returns conditional on the 52-week high- or the earnings-momentum strategies. Furthermore, following George and Hwang’s (2004) regression model, we also conduct a composite analysis of the six momentum strategies, and find that the profitability of the 52-week high momentum strategy is
significantly better than other strategies. In robustness tests, our empirical results show that returns to losers portfolio become much smaller after January months are excluded, and result in even higher momentum profit. This indicates that the January effect does have a
significant effect on momentum profitability. Furthermore, the profitability of the 52-week high-momentum strategy is not affected by different lengths of portfolio formation period, and different market states.
The rest of this paper is organized as follows. In Section 2, we describe the data used in this paper. Construction of various residual momentum strategies and our research
methodologies are introduced in Section 3. Section 4 presents empirical results with robustness checks. The last section offers concluding remarks.
2 Data Description
Our sample consists of returns and firm characteristics of all common stocks listed on the Taiwan Stock Exchange (TWSE) from January 1981 to December 2010. The data are
retrieved from the Taiwan Economic Journal (TEJ), which is a local data vendor in Taiwan.
Financial firms and firms with negative book values are excluded from our sample.
5
Continuously compound returns,
R , are calculated as
tP , ln P
= R
1
t-t t (1)
where
P is the closing price at time t. Industry portfolios are constructed according to the 28
t industry groups designed by the TWSE.2 Within each industry, returns are weighted by market values of component stocks. For residual momentum strategy, the Fama and French (1993) three-factor model is employed to computed residual returns. Following Fama and French (1992), a firm's book-to-market equity for July of year t to June of year t + 1 is calculated as the book value of fiscal year t − 1, divided by market equity at the end of calendar year t − 1.3 Methodology
3.1 Construction of Momentum Strategy Indicators
Momentum strategies examined in this paper include the price momentum strategy of Jegadeesh and Titman (1993) (JT hereafter), the industry momentum strategy of Moskowitz and Grinblatt (1999) (MG hereafter), the 52-week high momentum strategy of George and Hwang (2004) (H52 hereafter), the moving average ratio momentum strategy of Park (2010) (MAR hereafter), the residual momentum strategy of Blitz, Huij and Martens (2011) (RS hereafter), and the earnings momentum strategy of Chan, Jegadeesh and Lakonishok (1996) (Earn hereafter). In the following, we briefly outline the construction of the indicator for each momentum strategy in sequence.
2 According to the TWSE, individual stocks are grouped into the following categories: Cement, Foods, Plastics, Textiles, Electric Machinery, Electrical and Cable, Glass & Ceramics, Paper & Pulp, Iron and Steel, Rubber, Automobile, Building Material and Construction, Shipping and Transportation, Tourism, Finance and Insurance, Trading and Consumers' Goods, Chemical, Biotechnology and Medical Care, Oil, Gas and Electricity, Semiconductor, Computer and Peripheral Equipment, Optoelectronics, Communications and Internet, Electronic Parts & Components, Electronic Products Distribution, Information Service, Other Electronics, and Others,
6
1. The JT momentum strategy：
Following Jegadeesh and Titman (1993), the JT momentum strategy indicator is constructed based on average returns of the past twelve months as follows
12 , 1
, 1: 12 ,
12
i t j j i t t
R R
(2)where
R
i,t represents return of stock i at month t; whileR
i,t1:t-12 represents average return of stock i during the past twelve months.2. The MG momentum strategy：
Following Moskowitz and Grinblatt (1999), the MG momentum strategy indicator is constructed based on average returns of each industrial group during the past twelve months as follows,
12 , 1
, 1: -12 ,
12
i t j j
i t t
IR IR
(3)where