Reversal and momentum patterns in weekly stock returns: European evidence
Date | 01 April 2019 |
Author | Hannah Lea Hühn,Hendrik Scholz |
Published date | 01 April 2019 |
DOI | http://doi.org/10.1002/rfe.1037 |
ORIGINAL ARTICLE
Reversal and momentum patterns in weekly stock returns:
European evidence
Hannah Lea Hühn
|
Hendrik Scholz
Friedrich-Alexander-Universität (FAU)
Erlangen-Nürnberg, Chair of Finance and
Banking, Nürnberg, Germany
Correspondence
Hendrik Scholz, Friedrich-Alexander-
Universität (FAU) Erlangen-Nürnberg,
Chair of Finance and Banking, Nürnberg,
Germany.
Email: hendrik.scholz@fau.de.
Abstract
We analyze short‐term reversal and medium‐term momentum patterns in weekly
stock returns in Europe. Focusing on raw and stock‐specific returns, our empirical
results show for both return specifications (a) a negative relation between weekly
past returns and future returns in the short run and (b) a positive relation in the
medium run. However, returns from reversal and momentum strategies based on
stock‐specific returns are less volatile. In further analyses, we find short‐term
reversal and medium‐term momentum patterns to be connected to stock character-
istics. Looking at the potential causes of these effects, our results do not support
the idea that short‐term reversal in weekly stock returns is due to an over‐or
underreaction to firm‐specific news nor that it is mainly driven by illiquidity.
Medium‐term momentum in weekly stock returns, on the other hand, can be con-
nected to behavioral biases. Our concluding tests confirm that our findings are
robust among industries, in subperiods, for the January effect and in varying mar-
ket states. Finally, while medium‐term momentum strategies remain profitable
after accounting for transaction costs, short‐term reversal strategies can be mainly
explained by transaction costs due to their high turnover.
JEL CLASSIFICATION
G11, G12, G14
KEYWORDS
conventional strategy, medium-term momentum, short-term reversal, stock characteristics, stock-
specific strategy
1
|
INTRODUCTION
This paper contributes to the literature by studying the relation between reversal and momentum patterns in weekly Euro-
pean stock returns. So far, most researchers have focused on reversal patterns
1
in short‐term stock returns and on momen-
tum patterns
2
in medium‐term stock returns. However, Gutierrez and Kelley (2008) find that return reversals in the first
weeks are offset by an even stronger momentum pattern in the medium run. They thus conclude that not reversal but
momentum seems to be the dominant factor of weekly returns. Gutierrez (2016), studying NASDAQ stocks, confirms these
findings. Chai, Limkriangkrai, and Ji (2015) finds a relation between reversal and momentum in Australian stocks. Bianchi,
Drew, and Fan (2015) connect momentum and reversal patterns in commodity futures. However, to the best of our knowl-
edge, no study has so far analyzed the coexistence and interaction of reversal and momentum for one of the most liquid
markets, the European stock market. Our study gives an out‐of‐sample test to the existing literature and corroborates that
the coexistence of reversal and momentum is not driven by a data snooping bias in the sense of Lo and MacKinlay
Received: 19 March 2018
|
Revised: 15 May 2018
|
Accepted: 29 May 2018
DOI: 10.1002/rfe.1037
272
|
© 2018 The University of New Orleans wileyonlinelibrary.com/journal/rfe Rev Financ Econ. 2019;37:272–296.
(1990b). We follow, among others, Rouwenhorst (1998), Annaert, De Ceuster, and Verstegen (2013) and Walkshäusl
(2014), who also use European stock markets as a new dataset to test the robustness of market anomalies like momentum
and the MAX effect.
Following Gutierrez and Kelley (2008), we examine the performance of strategies not only in the short run but also in
the medium run. We analyze these return patterns for strategies based on raw returns and for strategies based on stock‐spe-
cific returns.
3
For both return specifications and applying various weekly formation periods, the performance of these port-
folios shows reversal patterns in the short run and momentum patterns in the medium run. However, for the stock‐specific
strategy, reversal seems to be more pronounced and is still present in the second week after portfolio formation. Moreover,
ranking on stock‐specific returns leads to less volatile returns from these strategies. Implementing asset pricing models, we
find these results to be robust on an individual stock level.
Besides research on the profitability of reversal and momentum strategies, several studies find evidence that these effects
are stronger for stocks with certain firm characteristics in the United States (e.g., Gutierrez and Kelley (2008), Da et al. (2014)
for reversal strategies and Jegadeesh and Titman (1993), Hong, Lim, and Stein (2000) and Zhang (2006), Asness (1997), and
Daniel and Titman (1999) for momentum strategies). Our paper contributes to this literature by analyzing, for the first time in
Europe, whether short‐term reversal and medium‐term momentum in weekly stock returns are related to firm characteristics.
We follow Bali, Cakici, and Whitelaw (2011) and include size, book‐to‐market equity, beta, skewness, idiosyncratic volatility,
illiquidity, and past 1‐year return as stock characteristics. Our results indicate that reversal and momentum patterns are still
present after accounting for these stock characteristics and are thus not solely driven by sorts in characteristics. However, they
are different within characteristic‐based quintiles. Connecting these results to potential causes, we find no evidence that short‐
term reversal might be due to an over‐or underreaction to firm‐specific news nor that it is mainly driven by illiquidity (e.g.,
Campbell, Grossman, & Wang, 1994). On the other hand, medium‐term momentum can be connected to behavioral biases
(e.g., Barberis, Shleifer, & Vishny, 1998; Chen & Lu, 2017; Daniel, Hirshleifer, & Subrahmanyam, 1998; Hong & Stein,
1999) as size, beta, idiosyncratic volatility, and illiquidity represent proxies for (information) uncertainty and limits to arbi-
trage. We test the strength of these results by implementing cross‐sectional regressions of future returns on past weekly stock‐
specific returns, including our stock characteristics and interaction variables which relate weekly stock‐specific returns to stock
characteristics. This cross‐sectional analysis confirms most results of quintile analyses.
To complete our empirical analysis by testing the robustness of our findings, we analyze our stock‐specific strategy
within industries, subperiods, for calendar month and within different market states. We find short‐term reversal and med-
ium‐term momentum patterns to be robust within industries and during both of our 12‐year subperiods. Moreover, while
short‐term reversal is present during January and non‐January weeks, we only find significant medium‐term momentum
returns in non‐January weeks. There is no clear relation between both phenomena and market states. Finally, when imple-
menting transaction costs, high turnovers partly remove the profitability of short‐term reversal strategies in Europe. Thus,
short‐term reversal strategies can be mainly explained by transaction costs. On the other hand, medium‐term momentum
patterns seem to remain profitable even after accounting for transaction costs.
Overall, our out‐of‐sample study confirms the results of Gutierrez and Kelley (2008). Thus, the interaction between
reversal and momentum does not seem to be driven by a data snooping bias in the sense of Lo and MacKinlay (1990b).
Our study proceeds as follows: Section 2 provides the methodology for constructing strategies based on past raw returns
and based on past stock‐specific returns. Moreover, it presents the performance evaluation models applied. Section 3
describes the data and reports our main empirical findings. Section 4 examines whether short‐term reversal and medium‐
term momentum are related to stock characteristics. Section 5 contains robustness tests including transaction costs, and Sec-
tion 6 concludes.
2
|
PORTFOLIO CONSTRUCTION AND PERFORMANCE EVALUATION
We implement the following two strategies to determine whether weekly stock returns exhibit reversal or momentum pat-
terns. First, following conventional strategy (e.g., Gutierrez & Kelley, 2008; Jegadeesh, 1990), we calculate the past raw
return for each stock iby cumulating weekly returns during the respective J‐week formation period (J= 1, 2, 3, 4). For
each week t, we then sort stocks into deciles based on these past raw returns, after which we form equally weighted decile
portfolios and construct a zero‐investment strategy that buys stocks in the winner decile (highest returns) and sells stocks in
the loser decile (lowest returns). This conventional strategy is rebalanced every week.
Our second strategy is based on the Fama and French (1993) (henceforth FF) three‐factor model which divides the
return r
i,t
of stock iin excess of the risk‐free return in week tinto its factor‐related return contribution (β
i
RMRF
t
+
HÜHN AND SCHOLZ
|
273
To continue reading
Request your trial