Persistency of the momentum effect

• DOI: http://doi.org/10.1111/eufm.12140
• Date: 01 November 2018
• Published date: 01 November 2018

DOI: 10.1111/eufm.12140

ORIGINAL ARTICLE

Persistency of the momentum effect

Hong-Yi Chen

1

|

Pin-Huang Chou

2

|

Chia-Hsun Hsieh

2

1

Department of Finance, National

Chengchi University, Taiwan, 64, Zhinan

Rd., Taipei 11605, Taiwan

Email: fnhchen@nccu.edu.tw

2

Department of Finance, National Central

University, Taiwan, 300, Zhongda Rd.,

Taoyuan City 32001, Taiwan

Emails: 994408006@cc.ncu.edu.tw;

choup@cc.ncu.edu.tw

Abstract

The intermediate-term momentum persistency is not

universal among all stocks. More than 40% of winners

and losers immediately fall out of their respective groups in

the month following formation, resulting in a monthly loss

of more than 17% for a momentum strategy constructed on

such stocks. By contrast, persistent winners and losers,

defined as those staying in their groups for at least one more

month, exhibit much stronger momentum persistency.

Further analysis indicates that, consistent with the delayed

reaction hypothesis for price momentum, the persistency is

stronger for stocks with greater information asymmetry and

more extensively heterogeneous investor beliefs.

KEYWORDS

delayed reaction hypothesis, duration, heterogeneous beliefs,

information asymmetry, persistent momentum strategy

JEL CLASSIFICATION

G11, G14

1

|

INTRODUCTION

The existence of intermediate-term price momentum, as documented by Jegadeesh and Titman (1993),

remains one of the most robust puzzles in financial research. A variety of theories offer explanations of

The authors are especially grateful to John Doukas (the Editor) and the anonymous ref eree for valuable suggestions.

The authors also appreciate helpful comments from Stephen Brown, Chun-Yang Hwang, Cheng-Few Lee, Ghon

Rhee, as well as seminar participants at National Taiwan University, National Chengchi University, 2014 Financial

Management Association Annual Meeting (Nashville, USA), the 22nd Annual Conference on Pacific Basin Finance,

Economics, Accounting, and Management (Nagoya, Japan), and the 7th Annual Financial Engineering and Risk

Management Conference (Hsinchu, Taiwan). Previous versions of this paper were circulated under the title

‘Profitability of Momentum Strategies: The Role of Consistent Winners and Losers.’

856

|

© 2017 John Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/eufm Eur Financ Manag. 2018;24:856–892.

the momentum effect, ranging from rational accounts, such as risk-based explanations (e.g., Asness,

Moskowitz, & Pedersen, 2013; Barroso & Santa-Clara, 2015; Liu & Zhang, 2008; Sadka, 2006) and

limits to arbitrage (Arena, Haggard, & Yan, 2008), to behavioral explanations, such as overreaction

(e.g., Byun, Lim, & Yun, 2016; Chui, Titman, & Wei, 2010; Cooper, Gutierrez, & Hameed, 2004;

Daniel, Hirshleifer, & Subrahmanyam, 1998; Hillert, Jacobs, & Mueller, 2014) and underreaction

(e.g., Antoniou, Doukas, & Subrahmanyam, 2013; Chen & Zhao, 2012; Da, Gurun, & Warachka, 2014;

Zhang, 2006). Although researchers have extensively investigated the sources of the momentum

profits, what drives intermediate-term momentum persistency remains a problem to be solved.

In addition to the abnormal profits of momentum strategies, the persistency of the momentum

effect has been investigated in empirical asset pricing literatures. For example, Lee and Swaminathan

(2000) propose a momentum life-cycle hypothesis and demonstrate that low-volume winners and

high-volume losers do not attract investors’notice and have a more persistent momentum effect.

Grinblatt and Moskowitz (2004) find that the return consistency can predict future stock returns.

Gutierrez and Kelley (2008) find a long-lasting momentum in weekly returns following the initial

reversal. Chen, Chen, Hsin, and Lee (2014) examine the persistency of profits for earnings, revenue,

and price momentum strategies, and find the earnings momentum strategy exhibits the strongest

persistence, while the revenue momentum strategy is relatively short lived. However, existing studies

only examine the persistency of momentum profitability at the portfolio level, ignoring the fact that

there are variations in performance at the individual firm level. Indeed, we find that, within a 6-month

formation period, approximately 47% of winner (loser) stocks fall out of their respective portfolios

immediately following formation. The return reversals that such nonpersistent winners and losers

exhibit in the month following formation are strong, resulting in a monthly loss of more than 17% for a

momentum strategy constructed on such stocks. Overall, they exhibit weak persistency in

performance. By contrast, persistent winners and losers, defined as those that stayed in their groups

for at least one more month, exhibit much stronger performance persistency. This finding suggests that

intermediate-term momentum persistency is not a universal phenomenon for all stocks with extreme

past performance. In turn, it is of interest to explore the characteristics and driving forces that give rise

to the differences in performance persistency at the individual firm level. To do so, we introduce the

winner (loser) duration as the number of months that a stock consecutively stays in the winner (loser)

portfolio and conduct a comprehensive duration analysis at the firm level to examine various risk-based

and behavioral hypotheses in explaining the persistency of the momentum effect.

We begin by providing a clearer understanding of the distribution of post-formation returns and the

turnover rates for stocks in winner and loser portfolios. We find that at least one-quarter of winner

(loser) stocks experience negative (positive) post-formation returns; thus, a momentum strategy may

fail if an investor concurrently goes long for winner stocks and shorts loser stocks. We also find that, in

terms of the 6-month formation period, more than 40% of winner and loser stocks are nonpersistent

winners and losers. Among persistent winners (losers), 18.6% (19.0%) of those stocks have only a

1-month duration, 10.8% (10.9%) have a 2-month duration, and approximately 97% persist for less than

6 months. This large dispersion of post-formation performance and high turnover rates for winner and

loser stocks inspires us to explore the difference in post-formation performance for persistent and

nonpersistent winners (losers) further. To this end, we separate winner (loser) stocks into persistent

and nonpersistent subsamples and examine their post-formation returns. The empirical results show

that during the post-formation period, persistent winners exhibit greater price appreciation than

nonpersistent winners, and persistent losers tend to lose more than nonpersistent losers.

Moreover, we demonstrate that a persistent momentum strategy, that is, buying persistent winners

and selling persistent losers, can generate monthly returns as high as 1.25%, which amounts to an

annual return of 15%. Such a persistent momentum strategy outperforms a price momentum strategy by

CHEN ET AL.

|

857

19 basis points in monthly returns (an 18% improvement). The persistent momentum effect remains

robust after adjusting for risk factors including the market factor, five risk factors proposed by Fama

and French (2015), Carhart’s (1997) momentum factor, and Pastor and Stambaugh’s (2003) liquidity

factor.

Among various explanations for the momentum effect, a recent and growing body of literature

demonstrates that the delayed reaction hypothesis can explain momentum profits. For example,

Barberis, Shleifer, and Vishny (1998) develop a theoretical model suggesting that conservatism bias

can lead investors to underreact to new information, and O’Hara (2003), Chen and Zhao (2012), and

Chen, Lee, and Shih (2016) show that greater information asymmetry will make investors react to a

firm’s good or bad news more conservatively, leading to a delayed price adjustment from new

information. Moreover, Dische (2002), Doukas and McKnight (2005), Hong and Stein (2007),

Banerjee, Kaniel, and Kremer (2009), Verardo (2009), Makarov and Rytchkov (2012), and

Ben-Rephael, Da, and Israelsen (2017) argue that gradual information diffusion results in greater

heterogeneity among investors (hereinafter, ‘heterogeneous beliefs’) and leads to a stronger

momentum effect. The current study therefore takes advantage of the duration of persistent winners

and losers to examine whether the information asymmetry and heterogeneous beliefs are associated

with momentum persistency at the individual firm level. We find that, consistent with the delayed

reaction hypothesis, winner and loser stocks with a higher level of information asymmetry or higher

degree of heterogeneous beliefs tend to stay in their respective portfolios in the next formation period.

In addition, those persistent portfolios with higher levels of information asymmetry or higher degrees

of heterogeneous beliefs can generate more momentum profits in the subsequent 1-year holding period.

We further conduct a duration model to investigate whether characteristics associated with

information asymmetry and heterogeneous beliefs can affect winner and loser durations. We show that

persistent winners and losers of relatively small size, with highly idiosyncratic volatility and low

institutional ownership, will have a longer duration as winners (losers). The finding implies that

investors tend to underreact to information about firms with greater information asymmetry, leading to

a longer momentum effect. In addition, the positive association among the duration, trading volume,

and dispersion of analyst forecasts indicates that momentum stocks endure longer if investor opinions

differ about their fundamental value. We also observe a strong momentum persistency for

high-information asymmetry and high-heterogeneous belief stocks that exhibit a large price

appreciation (depreciation). That is, the momentum effect is more persistent when stocks have more

incremental information content during the formation period and those prices thus do not reflect this

information efficiently. In addition, the duration model shows an asymmetric effect on momentum

duration: the formation-period return plays a relatively important role for persistent losers, but the post-

formation return is more important for persistent winners.

Overall, this study contributesto the finance literature in several ways. First, we provideevidence of

the implementationcost of the momentum strategy.Knez and Ready (1996), Shleiferand Vishny (1997),

Grundy andMartin (2001), Korajczyk and Sadka(2004), and Israel and Moskowitz (2013)argue that the

trading costmay deter investors from applying the momentumstrategy, though the effect of tradingcosts

cannot fully explain momentum profits. The current study demonstrates a large dispersion of post-

formation returns and a highturnover rate for winner and loser portfolios, indicating that not all winner

and loser stocks follow the midterm return continuation. The momentum strategy contains an

implementationrisk: investors could face a potential loss if they cannot go long for all winner stocksand

go short for all loserstocks. Such an indirect implementation cost for the momentumstrategy is new to the

literature and presents a possible avenue to track the sources of the momentum effect.

Second, we find a stronger post-formation performance for persistent momentum portfolios and

demonstrate that the persistent momentum effect cannot be explained by risk factors. Prior literature

858

|

CHEN ET AL.