Are disposition effect and skew preference correlated? Evidence from account‐level ELW transactions
Date | 01 February 2020 |
Author | Eunji Kwon,Woojin Kim,Youngsoo Choi |
DOI | http://doi.org/10.1002/fut.22055 |
Published date | 01 February 2020 |
J Futures Markets. 2020;40:228–246.wileyonlinelibrary.com/journal/fut228
|
© 2019 Wiley Periodicals, Inc.
Received: 15 August 2019
|
Accepted: 24 August 2019
DOI: 10.1002/fut.22055
RESEARCH ARTICLE
Are disposition effect and skew preference correlated?
Evidence from account‐level ELW transactions
Youngsoo Choi
1
|
Woojin Kim
2
|
Eunji Kwon
1
1
Hankuk University of Foreign Studies,
Mohyeon‐myeon, Cheoin‐gu, Yongin‐si,
Gyeonggi‐do, Korea
2
Seoul National University Business
School, Gwanak‐gu, Seoul, Korea
Correspondence
Youngsoo Choi, Hankuk University of
Foreign Studies, 81 Oedae‐ro,
Mohyeon‐myeon, Cheoin‐gu,
Yongin‐si, Gyeonggi‐do 449‐791, Korea.
Email: choiys@hufs.ac.kr
Funding information
Hankuk University of Foreign Studies
Research Fund; Institute of Management
Research at Seoul National University,
Institute of Finance and Banking of Seoul
National University
Abstract
This paper examines whether two well‐known cognitive biases, disposition
effect and skew preference, may reflect a common feature of certain investors.
On the basis of a unique proprietary dataset that provides the details of all
transactions in the Korean equity‐linked warrant (ELW) market between 2009
and 2011, we find that investors who realize gains faster than losses are also
more likely to trade out‐of‐the‐money ELWs. Investors who are less (more)
subject to both biases exhibit the best (worst) risk‐adjusted trading performance.
Our findings suggest that disposition effect and skew preference occur
simultaneously, which could adversely affect trading performance.
KEYWORDS
disposition effect, equity‐linked warrant, Korea, lottery, option, skew preference
JEL CLASSIFICATION
G13; G23; G41
1
|
INTRODUCTION
A growing literature in behavioral finance suggests that various forms of cognitive biases exist among investors,
especially among individual investors. Two of the most widely discussed behavioral biases are the disposition effect, a
tendency to sell winners and hold onto losers, and the preference for skewed return or lottery‐type payoffs.
Recent studies suggest that the degree of disposition effect or skew preference may vary across different investor
clientele. For example, Bailey, Kumar, and Ng (2011) find that individuals with socioeconomic characteristics that are
likely to induce gambling—namely poor, young, and less‐educated investors—prefer low‐priced stocks with high
volatility and high skewness. Dhar and Zhu (2006) report that wealthy, professional individuals, as well as investors
who trade more frequently, exhibit less disposition effect. Fischbacher, Hoffmann, and Schudy (2017) document that
investors who commit to realize a loss in advance by voluntarily signing up for automatic selling mechanisms are also
less subject to disposition effect.
Another stream of literature examines how prior financial performance may influence subsequent risk‐taking.
Although many studies report that experiencing financial losses may lead to more risk aversion, some studies find that
losers are more risk‐seeking than winners.
1
A recent experiment by Imas (2016) formally distinguishes between realized
and paper losses and finds that it is the latter that leads to more subsequent risk‐taking, but not the former. These
studies jointly suggest that disposition effect and skew preference may not be independent behavioral biases but rather
reflect a correlated behavior, potentially more distinct among less sophisticated investors.
1
For example, Andersen, Hanspal, and Nielsen (2019) and Guiso, Sapienza, and Zingales (2018) report a positive relationship between past performance and risk‐taking, whereas Chatrath, Christie‐
David, Miao, and Ramchander (2019) find a negative relationship.
Cumulative prospect theory in Tversky and Kahneman (1992) and Barberis (2012) lays the groundwork for a
theoretical prediction that losers may engage in more risk‐taking than winners due to a dynamic inconsistency in
preference. Nevertheless, empirical study that explicitly considers both disposition effect and skew preference as
common characteristics of a certain class of investors is extremely scant.
2
The lack of such study is presumably due to
unavailability of an appropriate dataset that can simultaneously measure an investor’s susceptibility to both disposition
effect and skew preference. For example, detailed information on account‐level transactions is required to construct a
measure of the disposition effect. In addition, we need a measure of ex ante skewness of a return distribution to proxy
for skew preference.
3
In this study, we take advantage of a unique dataset that allows one to test the potential coexistence of both
disposition effect and skew preference, and how they affect trading performance. Our tests are based on a
proprietary dataset from the Korean equity‐linked warrant (ELW) market, which provides detailed trading records
at each account level. Our dataset includes account identifiers for both the buyer and the seller, which allows us to
directly track the trade history and the dollar amount gains and losses to each and every account in our sample.
4
Specifically, our dataset includes all trades and quotes of ELWs listed in the Korea Exchange from January 2, 2009
to June 30, 2011. There are a total of more than 94,000 accounts available, out of which 53,620 accounts trade ELWs
based on individual stocks.
ELWs are very similar to standard equity options in terms of the strike price, maturity, underlying assets, and payoff
structure, except that they are created and sold by securities companies. Once sold, they are listed on the stock
exchange, rather than on options exchange, and they trade just like any other stock so that no margin is required as in
standard options trading. However, retail investors can only initially buy ELWs and cannot take short positions. As
such, conventional option trading strategies, like writing covered calls or creating various types of spreads, are not
feasible.
According to a recent model developed by Boyer and Vorkink (2014), ex ante skewness of an option can be proxied
by its (out‐of‐the) moneyness. Since ELWs are essentially options that allow only long positions, we can measure an
investor’s skew preference by the average moneyness of the ELWs purchased.
And since we can identify gains and losses in our transaction‐level dataset, we can also construct a measure of the
disposition effect. We construct a measure of the disposition effect, similar to those developed in Shefrin and Statman
(1985) and Shapira and Venezia (2001) as follows. We classify all sell transactions in our dataset into a gain, loss, or a
draw, and conditional on each outcome, calculate the length of the round‐trip holding period from the initial purchase.
Then, for each account in our sample, we obtain the average holding periods for losses and gains, and take the ratio of
average time until losses over average time until gains as a measure of the disposition effect. This measure precisely
reflects how losers are held longer compared with winners.
Consistent with the previous literature, we find that our ELW investors indeed exhibit a disposition
effect on average. For example, the mean (median) round‐trip holding period for a transaction incurring a
loss is more than 5 (close to 2) times longer than the corresponding length for a transaction with a gain. In
roughly 80% of our sample accounts, average time held until a loss is longer than average time held until
again.
In addition, we find that disposition effect is more pronounced when a given account purchases more out‐of‐
the‐money ELWs, which is a measure of ex ante skew preference based on Boyer and Vorkink (2014). This is
broadly consistent with experimental evidence provided in Imas (2016), where individuals engage in more risk‐
taking following a paper loss. Both disposition effect and skew preference are strongeramonginvestorswhotradea
smallernumberofELWsorunderlyingassets,aproxyforinvestor sophistication. The positive correlation between
disposition effect and skew preference holds after controlling for additional account‐level characteristics in a
multivariate framework.
More importantly, both disposition effect and ex ante skew preference of an account adversely affect its
risk‐adjusted trading performance. There have been some studies that examine how skew preferences may affect
performance (e.g., Bailey, Kumar, & Ng, 2011), but very few studies examine how disposition effect may affect
2
Two related studies are Imas (2016), who examines risk‐taking behavior subsequent to paper losses in an experimental setting, and Chatrath et al. (2019), who examine the link between option
traders’performance, pricing of their holdings, and potential disposition effect.
3
Ex post skewness obtained from realized returns of an investor’s portfolio is positively correlated with realized performance of that portfolio by construction, since it contains at least one large realized
gain. As such, it cannot be used to proxy for investors’ex ante behavioral biases.
4
This approach is similar to Barber and Odean (2000) and Baron, Brogaard, and Kirilenko (2012).
CHOI ET AL.
|
229
To continue reading
Request your trial