Duration Dependence, Behavioral Restrictions, and the Market Timing Ability of Commodity Trading Advisors
Author | Michael Frömmel,Alexander Mende,Gert Elaut |
Date | 01 September 2017 |
DOI | http://doi.org/10.1111/irfi.12114 |
Published date | 01 September 2017 |
Duration Dependence, Behavioral
Restrictions, and the Market Timing
Ability of Commodity Trading
Advisors
*
GERT ELAUT
†
,MICHAEL FRÖMMEL
†
AND ALEXANDER MENDE
‡
†
Department of Financial Economics, Ghent University, Ghent, Belgium and
‡
RPM Risk and Portfolio Management AB, Stockholm, Sweden
ABSTRACT
This paper addresses a potential shortcoming in the work on the market
timing ability of fund managers. We adapt the Henriksson-Merton (1981) test
for market timing by relaxing a behavioral assumption that is implicit in the
use of daily data. To this end, we relax the assumption that managers base
their market timing decisions on daily excess returns. Instead, we use results
from the literature on bull and bear markets and test whether fund managers
can successfully time such trends in financial markets. We make use of a pro-
prietary dataset of daily Commodity Trading Advisors (CTAs) returns to show
that CTAs, on average, are able to time the bull and bear markets we identify.
JEL Codes: G10; G11; G15
INTRODUCTION
In general, the value potentially added through active management can stem
from one or two sources. First, there is the traditional security selection, that is,
the ability to add value by selecting securities that subsequently outperform.
Second, managers could also add value by successfully anticipating market trends
and reacting to these trends by entering or exiting the market accordingly. This is
referred to as market timing ability and has received considerable attention over
the last two decades.
However, empirical evidence on whether managers do in fact add value
through one or both approaches is mixed. One of the first prominent studies
on mutual fund performance is Sharpe (1966). He finds no evidence of excess per-
formance for funds compared to the DJIA over the period 1954–1965. This result
* Funding by the European Commission under the 7
th
Framework Programme is gratefully acknowl-
edged (IAPP 324440 “Futures”). We would like to thank seminar participants at the HEC Liège Man-
agement School—University of Liège, the University of Duisburg-Essen, participants of the 2015
Joint Conference on “Institutional Investors/Hedge Funds and Emerging Market Finance”at Ghent
University, and an anonymous reviewer.
© 2017 International Review of Finance Ltd. 2017
International Review of Finance, 17:3, 2017: pp. 427–450
DOI: 10.1111/irfi.12114
is confirmed by Jensen (1968), who shows that the average “alpha”of mutual
funds in his dataset is not significantly different from zero. Subsequent evidence
is more mixed, but seems to gravitate to the null hypothesis of no significant
outperformance by mutual funds. While “alpha”captures security selection,
other studies focus on fund managers’market timing ability, that is, the ability
to adjust one’s market exposure in anticipation of future (stock) market move-
ments. The majority of these studies finds no (or sometimes even negative) mar-
ket timing ability for mutual funds (see e.g. Jensen 1972; Henriksson and Merton
1981; Merton 1981; Admati et al. 1986; Lehmann and Modest 1987; Ferson and
Schadt 1996; Kao et al. 1998; Becker et al. 1999).
As such, the consensus for mutual funds seems to emerge that mutual fund
managers, on average, add little value for investors. To some extent, fees charged
by these funds seem to explain most of the lack of performance: Many studies,
most recently Fama and French (2010), find that funds’gross returns outperform
the market, while the net-of-fee returns do not. This suggests that fund managers
are capturing the outperformance through fees.
Evidence for market timing among hedge funds is also mixed, although more
recent work indicates some market timing skill for these managers. Whereas
Fung et al. (2002) do not find evidence for market timing ability among hedge
funds Chen and Liang (2007) study a sample of self-declared market timing
hedge funds and find evidence of market timing ability. Chen (2007), who exam-
ines the timing ability of hedge funds with regard to their focus markets, also
finds evidence that a number of categories of hedge funds (CTAs and Global
Macro) can successfully time certain asset markets. Finally, Kazemi and Li’s
(2009) findings suggest that CTAs generate their returns mostly from successful
market timing.
However, whereas early studies use monthly returns to test for timing ability,
more recent studies such as Bollen and Busse (2001) and Jiang et al. (2007)
who use daily data come to more encouraging conclusions about managers’mar-
ket timing abilities. These results provide evidence that confirm the findings by
Goetzmann et al. (2000) that the use of daily data appears to increase the power
of the market timing models to detect market timing ability. Chance and Hemler
(2001) analyze daily explicit recommendations by market participants and also
find evidence of market timing ability.Results in both papers further suggest that,
when monthly data is used, the evidence of positive market timing ability
disappears.
One major drawback in applying existing market timing models to monthly
data is that the researcher implicitly assumes that the trading frequency is also
monthly.Goetzmann et al. (2000) are the first to point out this behavioral restric-
tion. The authors propose an adjustment that assumes daily timing but that does
not require collecting daily returns. Nevertheless, they note that applying market
timing models directly to daily data is preferable. However, the application to
daily data creates a potential conflict: Standard tests for market timing (Treynor
and Mazuy 1966; Henriksson and Merton 1981) use the market’s excess return
as benchmark for market timing. While this might be a reasonable assumption
International Review of Finance
© 2017 International Review of Finance Ltd. 2017428
To continue reading
Request your trial