Description
This study investigates the relation between trading patterns and performance in the TAIEX futures
market. The research shows that individual investors are poor market timers and earn negative returns;
institutional investors have success in timing the market and their trades make positive returns. Individual
trading activity is more aggressive in terms of a higher proportion of the market order and a
shorter holding period for a round-trip trade. Individual trading is also more motivated by behavioral
bias, like overconfidence and disposition effect. Institutional investors exhibit significant overconfidencebased
trading when opening extremely small or relatively large positions.
Trading patterns in the TAIEX futures markets: Information- or behavioral-based
trades?
Mei-Chen Lin
a, *
, Ming-Ti Chiang
a, b
a
National Taipei University, Taiwan
b
Hsing Wu University, Taiwan
a r t i c l e i n f o
Article history:
Received 3 August 2012
Accepted 20 October 2014
Keywords:
Individual investors
Institutional investors
Overcon?dence
Disposition effect
a b s t r a c t
This study investigates the relation between trading patterns and performance in the TAIEX futures
market. The research shows that individual investors are poor market timers and earn negative returns;
institutional investors have success in timing the market and their trades make positive returns. Indi-
vidual trading activity is more aggressive in terms of a higher proportion of the market order and a
shorter holding period for a round-trip trade. Individual trading is also more motivated by behavioral
bias, like overcon?dence and disposition effect. Institutional investors exhibit signi?cant overcon?dence-
based trading when opening extremely small or relatively large positions.
© 2015 College of Management, National Cheng Kung University. Production and hosting by Elsevier
Taiwan LLC. All rights reserved.
1. Introduction
Institutional and individual investors are two major players who
compete to obtain limited pro?tability in ?nancial markets. In-
stitutions generally differ from individuals due to their size and
sophistication (Kaniel, Saar, & Titman, 2005). In particular, indi-
vidual investors are generally less well informed and prone to
misinterpret available information or trade for non-informational
reasons. In comparison, institutional investors have better re-
sources and training than do individual investors. Although in-
stitutions cannot be immune from the same cognitive biases as
individuals, the impacts from behavioral biases may be alleviated
since they may overcome these biases through better information
and analytical skill. Thus, behavioral biases may have different ef-
fects on the trading patterns of institutional and individual traders.
Consistent with this, institutions are found to be informed investors
(e.g., Chakravarty, 2001; Jones & Lipson, 2004), by contrast, indi-
vidual investors are irrational noise traders and frequently succumb
to their cognitive biases (Bange, 2000; Frazzini & Lamont, 2008).
However, some papers posit that individuals make excess returns
through providing liquidity for institutional trading demands
(Campbell, Ramadorai, & Vuolteenaho, 2005; Kaniel et al., 2005).
Therefore, the evidence so far is mixed regarding the roles of in-
dividual and institutional investors.
In view of these con?icting ?ndings, this study uses a new data
set with detailed transaction information to explore whether
trading decisions are in?uenced more by knowledge about value
(information-based trading) or by psychological biases (behavioral-
based trading). The sample used in this study contains 38,684,525
trades of the Taiwan Stock Exchange Capitalization Weighted Stock
Index (TAIEX) futures executed by individual investors and
13,057,657 trades of TAIEX futures executed by institutional in-
vestment accounts during the period January 2004 through
December 2008.
Some previous studies have examined the TAIEX futures mar-
kets. The issues include the information conveyed by trade types of
different categories of investors (Lin, 2011), the impact of a tax rate
reduction on the market quality (Chou & Wang, 2006), the
expiration-day effects (Chou, Chen, & Chen, 2006; Chueh & Yang,
2005), the costeminimization combination of margins, spot price
limits, and futures price limits (Chou, Lin, &Yu, 2006), and the daily
dynamic relation between returns and trades by institutional and
individual investors (Lin, 2011). Other authors compare the infor-
mation transmission between TAIEX and Mini-TAIEX Stock Index
Futures (Lin, Hsu, & Chiang, 2004) and price discovery between
TAIFEX TAIEX index futures and SGX MSCI Taiwan index futures
(Chen, Lin, Chou, & Hwang, 2002).
The research by Cheng, Lin, and Chuang (2007), Kuo and Lin
(2013, 2011) is associated with behavioral-based trades in the
TAIEX futures market. Kuo and Lin (2013) investigate the
* Corresponding author. National Taipei University, 151, University Rd., San Shia
District, New Taipei City 23741, Taiwan.
E-mail address: [email protected] (M.-C. Lin).
Peer review under responsibility of College of Management, National Cheng
Kung University.
HOSTED BY
Contents lists available at ScienceDirect
Asia Paci?c Management Review
j ournal homepage: www. el sevi er. com/ l ocat e/ apmrv
http://dx.doi.org/10.1016/j.apmrv.2014.10.002
1029-3132/© 2015 College of Management, National Cheng Kung University. Production and hosting by Elsevier Taiwan LLC. All rights reserved.
Asia Paci?c Management Review 20 (2015) 165e176
performance of individual day traders in the TAIEX market and ?nd
that individual day traders incur a signi?cant loss. Cheng et al.
(2007) examine the trading behavior and performance of traders
in the TAIEX market. They ?nd that the individual traders are
positive feedback traders while foreign investors tend to engage in
negative feedback trading. Lin (2011) shows that open trading by
foreign institutional investors conveys more information regarding
the underlying index, and open selling of individual investors is
more likely to introduce noise signals to the spot market.
Different from Kuo and Lin (2013), this paper not only in-
vestigates the performance of individual traders, but also compares
the performance and trading behavior of individual and institu-
tional traders in the TAIEX market. Besides, although Cheng et al.
(2007) have examined the trading behavior and performance of
traders in the TAIEX market, they do not discuss whether the dif-
ference in performance between individual and institutional
traders arises from their different behavioral biases. Lin (2011)
shows that individuals are more irrational and more prone to
misinterpret available information. Nevertheless, she does not
address the relationship between trading decisions and behavioral
biases. Additionally, to my best knowledge, no research has yet
investigated whether behavioral biases, like overcon?dence and
disposition bias, will affect individual and institutional investors'
tendency to open a new contract or close an existing contract. This
study intends to ?ll in this gap. By separating trades into open
volume and close volume, it has been possible to examine whether
decisions to open a new contract or close an existing contract are
more affected by information motives or by behavioral biases.
I ?rst compare the return performance and trading behavior of
institutions and individual investors. I ?nd that the average trading
return for individual contracts is negative, whereas the average
return obtained by institutional investors is positive. Furthermore,
institutional investors appear to have some success in market
timing. In particular, the TAIEX futures market experiences positive
returns after institutional buying and negative returns after insti-
tutional selling. By contrast, individual investors are poor market
timers; market return is negative after their buying and positive
after their selling. Nofsinger and Sias (1999) and Kamesaka,
Nofsinger, and Kawakita (2003) posit that a strategy earning posi-
tive returns indicates that it is motivated more by information,
whereas trading which results in a negative return indicates a
higher probability of behavioral-based motivation. When com-
bined with the preliminary ?ndings of this study, this research
indicates that the decisions of individuals are more behavioral-
based, whereas those of institutional investors are more informa-
tion-based.
The behavioral-based model argues that investor trading de-
cisions are in?uenced by behavioral biases, like overcon?dence and
disposition effect (Daniel, Hirshleifer, & Subrahmanyam, 1998;
Gervais & Odean, 2001). Both the overcon?dence and disposition
biases may have effects in the decisions to close a position. But only
overcon?dence bias may affect the decisions to open a new con-
tract. The results show that, the round-trip trade of individuals has
a shorter period than that of institutional investors; individuals
exhibit a tendency toward the disposition effect, but institutional
investors display a reverse disposition effect. Individuals are also
more aggressive in terms of a higher proportion of the market order
and a shorter holding period for a round-trip trade. A further
regression test con?rms the positive relationship between trading
behavior (including both open and close trading) and over-
con?dence among individual investors, and the disposition effect of
individuals occurs when they close an extremely low amount of
positions or relatively large positions. Similarly, overcon?dence
induces institutional investors to open an extremely small amount
of positions or relatively large positions, and institutional investors'
desire to realize gains soon and ride on losses contributes a high
closing volume. Thus, I conclude that the trading activity of both
types of investors shows evidence of behavioral-bias motivation.
However, a further comparison of trades by individual and insti-
tutional investors in the TAIEX futures markets shows that indi-
vidual investors are less-informed and their trading decisions are
more motivated by behavioral biases.
The remainder of this paper is organized as follows. Section 2
introduces the data used in this study; Section 3 compares the
trading performance of individual and institutional investors;
Section 4 explores the motivation behind trading and compares the
trading behavior for institutions and individual traders; with the
conclusions being provided in the ?nal section.
2. Data
The data for this study consist of all of the trades of the spot-
month Taiwan Stock Exchange Capitalization Weighted Stock In-
dex (TAIEX) futures contracts from the Taiwan Futures Exchange
(TAIFEX) during the period January 2004 through December 2008.
The TAIEX is a market capitalization weighted index composed of
all stocks listed in the Taiwan Stock Exchange. The contract size per
contract is the TAIEX index point multiplied by 200 New Taiwan
Dollars (NTD). Contract months of TAIEX index futures are spot
month, the next calendar month, and the next three quarterly
months. The last trading day for each contract month is the third
Wednesday of the delivery month for each contract. Since launched
on 21 July 1998, TAIEX futures contracts have been growing fast and
have made up the largest part of futures contracts in the TAIFEX.
According to Futures Industry Association (FIA), the Taiwan Stock
Index Futures contract (TX) is the sixth largest one of Asian index
futures contracts in 2004 and TAIFEX's global ranking on trading
volume rose to rank 17th in 2008 from 57th in 1998 (Lin, 2011).
The data include trader's ID codes, trading directions (buy/sell),
transaction prices and volume (in number of contracts), and the
time of each transaction. This unique dataset allow me to correctly
identify the trade type for each transaction, including open buy,
open sell, close buy, and close sell. This helps me to reduce the error
likely to occur in studies using the Lee and Ready (1991) algorithm
to speculate on buyer-initiated and seller-initiated volume (see e.g.,
Chan et al., 2002). In addition, different from prior research with
mature futures markets where the institutional investors are the
major participants, individual investors are the major participants
in the TAIEX futures markets. Because of these features and the
availability of data with trade- and trader-type classi?cations, the
TAIFX futures market is an appropriate environment to test the
relation between trading decisions and behavioral bias.
To obtain an indication of whether their trading is based more
on information or behavioral bias, I ?rst provide the net returns per
contract for each investor type. Appendix illustrates the calculation
of major relevant variables: the net realized returns, net unrealized
returns, duration, numbers of realized gains, numbers of realized
losses, numbers of unrealized gains, numbers of unrealized losses,
open buy volume, open sell volume, close buy volume, and close
sell volume. The procedure is as follows. For each trader and each
contract, I ?rst sort trades based on transaction time. Once the ?rst
trade is located, I track each subsequent trade. I mark to market
after each trade and calculate statistics such as the weighted
average costs, open interests (OIs), trading volume, realized gains/
losses and unrealized gains/losses. The weighted average cost for
trade j at time t þ1 is de?ned as: AVC
j;tþ1
¼
Pj$t Vj;t þPj;tþ1Vj;tþ1
Vj;t þVj;tþ1
, where P
jt
(P
jt þ 1
) is the futures price for trade j at time t (t þ1) and V
jt
(V
jt þ 1
)
is the numbers of contracts of the trade for trade j at time t (t þ 1).
For a long (short) position being closed, the sales (purchase) price is
M.-C. Lin, M.-T. Chiang / Asia Paci?c Management Review 20 (2015) 165e176 166
compared to its weighted average costs to determine if it is closed
at a gain or loss. For a position not closed at the end of the day, its
paper gain/loss is calculated based on the daily settlement price.
For example, on December 16, 2004, Trader A starts his trade
by longing (i.e., open buy) seven TX futures contracts with
expiration day on January 25 (TXFA5) at a price of 6005. I then
track his subsequent trades until the maturity of TXFA5. At this
time, his record shows an average cost of 6005, an open interest
of 7 and trading volume of 7. In reality, the Clearing Department
of TAIFEX monitors intra-day price ?uctuations throughout the
trading period in order to evaluate the impact of such changes on
the clearing members. Therefore, I follow Locke and Mann (2005)
to mark to market after each trade, and calculate realized gains/
losses and unrealized gains/losses. Because this trade is not
closed on December 16, 2004, I calculate its corresponding gross
paper gains (losses) based on its concurrent daily settlement
price. The contract size for the TAIEX futures is the index value of
TAIEX Â 200 New Taiwan Dollars (NT$). Thus, with the daily
settlement price being 6020, an average cost of 6005, and open
interest of 7, the trader has a gross unrealized gain of NT$21,000
((6020 e 6005) Â 200 Â 7 ¼ 21,000).
To calculate the net pro?t, I subtract the commission and
transaction tax, which is 1/100th of one percent of the transaction
value before 1, January, 2006. Afterwards, it is 0.4/100th of one
percent of the transaction value. The commission varies among the
brokerage houses and the average is about 150 New Taiwan Dollars
(NTD) for each contract. Then, the net realized gains are NT$ 6519
(7600 À 150 Â 2 Â 2 À (6024 þ 6005)*0.01% Â 200).
On the next day, the trader ?rst closes 2 contracts by selling (i.e.,
close sell) at a price of 6024, and its corresponding gross realized
gains, NT$7600, are determined by comparing its costs
((6024e6005) Â 200 Â 2). Meanwhile, the remaining 5 long con-
tracts are marked to market, and now their unrealized gains are
NT$16,297. His record now thus shows an open interest of 5 and an
average cost of 6005. The duration is calculated by comparing the
time to open and to close a contract. Note that, in reality, TAIFEX is
operated from 8:45 am to 1:45 pm, Monday through Friday
(excluding public holidays). Therefore, I calculate the duration
based on the ?ve trading hours a day. For example, from Table 1, at
8:58 on December 17 of 2004, trader A closes two contracts which
are longing at 9:00 on December 16 of 2004. Thus, the duration is
298 min (285 min on December 16 and 13 min on December 17) for
these two contracts. Because its corresponding net realized returns
are positive, the duration is referred to as duration of the pro?table
round trip (DG). The same procedure is used to identify the dura-
tion for a losing contract (DL).
The same calculations are repeated for the following trades.
Notice that, on December 22, after longing one contract at 6042, the
trader buys seven more contracts in his second trade at a price of
6064. At this time, his record is updated to showan average cost of
6061.25 ((6042 Â1 þ6064 Â 7) ÷ (1 þ7)) and an open interest of 8.
Note that for contracts that are held until maturity and closed by
the exchange, I calculate their net realized gains/losses based on the
?nal settlement price of the contract. Note that, Locke and Mann
(2005) assume that open interest is zero at the end of each
trading day and determine realized and unrealized gains/losses by
?rst accumulating a sequence of buy (sell) trades. In comparison,
my calculation of realized gains/losses provides a more accurate
measure of realized gains/losses.
Table 1 reports summary statistics with respect to individual
and institutional trades. Over the sample period, the average daily
trading accounts for individuals and institutional investors are
6373.52 and 12.89, respectively. The percentage of individual in-
vestors is approximately 98.14%, which is strikingly higher than
that of institutional investors (1.86%). However, in terms of trading
volume (i.e., numbers of contracts), the percentage of individual
investors is not so high. Individual investors contribute 59.56%
(93,182.73/(93,182.73 þ 63,263.3)) of the gross volume of trade. In
contrast, 40.44% of the gross volume of trade is by institutional
investors. This indicates that individual investors are the major
participants in the TAIEX futures markets. This ?nding differs from
prior research with mature futures markets where the institutional
investors are the major participants.
The average daily buy-sell imbalances for individual and insti-
tutional investors are À0.121 and 0.886, respectively. It appears that
individual investors were net sellers and institutional investors
were net buyers during the sample period. The trade size per order
of individual investors (8.579) is signi?cantly lower than that of
institutional investors (23.987). The ratio of limit order to total
order, including limit order and market order, of individual in-
vestors (0.856) is lower than that of institutional investors (0.973).
Meanwhile, individuals leave fewer open interests (3.419 < 26.640)
and hold the contract for a shorter time (327.0 < 860.7) than in-
stitutions do. By contrast, individuals have higher turnover ratios
(59.84 > 19.68). This indicates that individual investors are more
aggressive in terms of a higher proportion of the market order and a
shorter holding period for a round-trip trade.
3. The trading performance of individual and institutional
investors
An abundant literature has theoretically argued that individuals
are overcon?dent and exhibit the disposition effect. They trade
frequently because they overestimate the precision of their
knowledge and underestimate the riskiness of the expected return
(see Benos, 1998; Kyle & Wang, 1997; Odean, 1998b; Wang, 1998,
Table 1
Summary statistics for trades of individual and institutional investors.
Individuals Institutions Difference t-value
Average numbers of traders 6373.52 120.98 6252.55 96.303***
Numbers of contracts 93,182.73 63,263.30 29,919.43 11.867***
Trade imbalance À0.0121 0.0886 À0.1007 À10.781***
Trade-size per order 8.579 23.987 À15.407 À38.398***
Limit-order ratio 0.856 0.973 À0.117 À14.403***
Open interests 3.419 26.640 À23.221 À29.08***
Turnover ratio 59.84 19.68 40.16 130.45***
Duration 327.0 860.7 À533.7 À155.2***
This table reports the numbers of traders, numbers of contracts, trade imbalance, trade size per order and limit order ratio for individual and institutional investors over the
sample period from 1, January 2004 to 31, December 2008.
Open interest refers to the total number of contracts that have not been settled on the trading day.
Turnover ratio is represented in percent.
Duration is the time to hold a round-trip contract, which is represented in minutes.
*, **, and *** denote signi?cant at the 1%, 5%, and 10% signi?cance level, respectively.
“Difference” denotes difference between individuals and institutions.
M.-C. Lin, M.-T. Chiang / Asia Paci?c Management Review 20 (2015) 165e176 167
2001). In addition, they also under-react to more relevant infor-
mation; this leads to positive feedback trading (see De Long,
Shleifer, Summers, & Waldmann, 1990; Daniel et al., 1998;
Hirshleifer, Subrahmanyam, & Titman, 1994; Odean, 1998b). By
contrast, though institutional investors are subject to the same
cognitive biases as individual investors, they are on average better
trained and have better resources than individual investors. The
better information and analysis skills may allow institutions to
overcome these biases. As a consequence, most academics argue
that institutional investors are more likely to behave rationally and
less likely to trade on noise than individual investors. Institutional
investors are also characterized as smart money in the sense of
being informed (e.g., Campbell & Kyle, 1993).
To investigate whether individual investors and institutional
investors in the TAIEX futures markets are smart money or dumb
money investors, I evaluate their performance using two different
methods. First, I examine the market timing performance after
those days when investors conduct heavy buying or selling
(Kamesaka et al., 2003). Second, I depict the return per contract for
each investor group.
3.1. Market timing
To study the market performance after individual and institu-
tional heavy buying and selling days, I need to calculate the trade
imbalance (TI) measure. My de?nition of trade imbalance follows
Chan and Fong (2000), Chordia, Roll, and Subrahmanyam (2002)
and Kamesaka et al. (2003) calculate the trading imbalance of eq-
uities on the Tokyo Stock Exchange (TSE). They de?ne trading
imbalance as (buy Àsell)/(buy þsell). Different fromKamesaka et al.
(2003), I examine the trading imbalance of TAIEX futures contract.
Because the buy volume of futures contracts includes the numbers
of open buy and close buy, and the sell volume of futures contracts
includes the numbers of open sell and close sell, I revise the trading
imbalance equation of Kamesaka et al. (2003) and de?ne it as
follows:
TI
it
¼
OB
it
ÀOS
it
þCB
it
ÀCS
it
OB
it
þOS
it
þCB
it
þCS
it
; (1)
where OB
it
is the ‘open-buy’ volume for investor group i on day t,
OS
it
is the ‘open-sell’ volume for investor group i on day t, CB
it
is the
‘close-buy’ volume for investor group i on day t, and CS
it
is the
‘close-sell’ volume for investor group i on day t. Following Chang,
Hsieh, and Lai (2009) and Chang, Hsieh, and Wang (2010), ‘open-
buy’ is denoted as opening new long contracts, ‘open-sell’ is
denoted as opening new short contracts, ‘close-buy’ is denoted as
closing existing long contracts, and ‘close-sell’ is denoted as closing
existing short contracts. The trade type is ‘open-buy’ if the trader
?rst opens a new long position for which I can ?nd no corre-
sponding ‘open-sell’ before the time and day that the buy order is
initiated. Conversely, it's ‘close-buy’ if the trader opens a new sell
position prior to initiating the buy order. With similar logic, I can
determine both ‘open-sell’ and ‘close-sell’, as well as ‘open-interest’
by collecting those positions which exist on each trading day.
TI is positive (negative) when the investor group buys more
(less) than sells contracts during the day. A large trade imbalance in
either direction is an indication of market timing (Kamesaka et al.,
2003). Informed traders with positive information are more likely
to be net buyers. Conversely, informed traders are net sellers when
they possess negative information. The trade imbalance (TI) for
each investor type is sorted onto ?ve equal sets. The quintile with
the highest positive trade imbalance is designated as the buying
day for that investor type. The quintile with the largest negative
trade imbalance is the selling day. To examine the market timing
ability of individual and institutional investors, I compute the one-
day, two-day, and three-day TAIEX futures return following the
trading day.
Table 2 reports the results. Individual investors have a trade
imbalance of 0.325 on buying days and À0.0733 on selling days.
One day after individual buying, the market decreases À0.0977%,
on average. One day after individual selling, the market increases
0.0186%. The post one-day return difference between buy days and
sell days is not signi?cantly different for individual investors. The
two-day period following the buy and sell days also experiences a
market return of À0.1572% and 0.1066%, respectively. The
difference, À0.2638%, is also not signi?cant. The 3-day return
following individual investor trading is À0.3679% after buying days
and 0.2660% after selling days. The difference, À0.6339%, becomes
signi?cant at the 5% level. Overall, individual investors appear to
exhibit poor market timing ability in the TAIEX futures markets.
The trade imbalance for the institutional buy and sell days are
0.1685 and À0.1550, respectively. The difference in trade imbalance
is 0.3235, which is relatively lower than that of individual investors.
On day one, the market experiences a 0.8187% return following
institutional buying and a À0.8582% return following institutional
selling. The difference, 1.6769%, is signi?cant. A market return of
1.6369% occurs two days after institutional buying. This is signi?-
cantly larger than the two-day return of À1.7010% following insti-
tutional selling. The three-day market return following
institutional buying is also signi?cantly larger than the return
following institutional selling. This indicates that institutional in-
vestors appear to have some success in market timing.
3.2. Trading performance
The average net realized gains of the winning contract (RG
i,t
) and
the average net realized losses of the losing contract (RL
i,t
) for
investor type i on date t are denoted as follows:
RG
i;t
¼
PN
i
RG;t
j¼1
RG
i
j;t
N
i
RG;t
; RL
i;t
¼
PN
i
RL;t
j¼1
RL
i
j;t
N
i
RL;t
; (2)
N
i
RG;t
¼ number of trades by investor type i where a gain is
realized on day t
N
i
RL;t
¼ number of trades by investor type i where a loss is
realized on day t
RG
i
j;t
¼ the net realized gains of the j winning contract for
investor type i on day t
RL
i
j;t
¼ the net realized losses of the j losing contract for investor
type i on day t
Because some contracts may not be closed by the end of a day, I
calculate total performance which includes paper gains or paper
losses. The average total net gains of the winning contract (TG
i,t
)
and the average total net losses of the losing contract (TL
i,t
) for
investor type i on date t are denoted as follows:
TG
i;t
¼
PN
i
RG;t
j¼1
RG
i
j;t
þ
PN
i
PG;t
j¼1
PG
i
j;t
N
i
RG;t
þN
i
PG;t
; TL
i;t
¼
PN
i
RL;t
j¼1
RL
i
j;t
þ
PN
i
PL;t
j¼1
PL
i
j;t
N
i
RL;t
þN
i
PL;t
;
(3)
N
i
PG;t
¼ number of trades by investor type i where a gain is
unrealized on day t
M.-C. Lin, M.-T. Chiang / Asia Paci?c Management Review 20 (2015) 165e176 168
N
i
PL;t
¼ number of trades by investor type i where a loss is un-
realized on day t
PG
i
j;t
¼ the net unrealized gains of the j winning contract for
investor type i on day t
PL
i
j;t
¼ the net unrealized losses of the j losing contract for
investor type i on day t
Other variables are de?ned as those in Eq. (2).
Table 3 reports the average total net gains (losses), net realized
pro?ts, and net unrealized performance per contract. As shown,
individuals have negative total returns of NT$-1307.8 per contract
and institutional investors earn positive total return of NT$1832.6
per contract. This better performance for institutions holds even
when it is measured in terms of either realized or unrealized
returns. This is consistent with Barber, Lee, Liu, and Odean (2005;
2009) and Barber, Odean and Zhu (2009) whose study sample fo-
cuses on the Taiwan equity market. A closer look reveals that in-
stitutions have greater magnitude of gains and losses than
individuals. For example, conditional on winning (losing) contracts,
the average total gains (losses) per contract for institutions, NT$
25,001.8 (NT$ À22,666.1), are larger than those of individuals,
NT$9278.5 (NT$ À8666.4). Taking gains and losses together, it is
apparent that the total net returns are higher for institutional in-
vestors than for individuals.
Nofsinger and Sias (1999) and Kamesaka et al. (2003) posit that
trading with high returns indicates that the trading is motivated by
information; in contrast, trading with low returns reveals that it is
motivated by behavioral-based biases. In accordance with this
insight, Barber and Odean (2001) and Odean (1999) show that
overcon?dence is associated with poor investment performance;
Locke and Mann (2005) ?nd that the least successful traders of
professional futures hold losers the longest, while the most suc-
cessful traders hold losers for the shortest time; Wermers (2003)
reports that mutual funds with poor performance have a greater
tendency to hold losing stocks. Above evidence indicates that, if
individual investors, compared with institutional investors, are
more inclined toward behavioral biases, on average, their perfor-
mance will be smaller. Therefore, following Nofsinger and Sias
(1999) and Kamesaka et al. (2003), I ?rst use the past trading
returns to get a preliminary result. I then take a further step to
examine the relationship between open (close) volume and
behavioral biases, like overcon?dence and disposition effect, to
identify whether trading decisions are in?uenced more by
information-based or behavioral-based motivation.
The preliminary results in Tables 2 and 3 show that there is an
signi?cant difference in performance and trading behavior among
these two investor-types, as has previously been suggested (Odean,
1999). Institutional investors are more likely to be informed traders
Table 2
Market performance after buying and selling by investor type.
Trade imbalance 1-day Returns (%) 2-day Returns (%) 3-day Returns (%)
Panel A: Individual investors
Buy 0.3025 À0.0977 À0.1572 À0.3679
t-value 39.2*** À0.20 À1.84* À2.27**
Sell À0.0733 0.0186 0.1066 0.2660
t-value À1.15 0.11 À1.19 2.86**
Difference 0.3759 À0.1162 À0.2638 À0.6339
t-value 60.20*** À0.83 À1.43 À2.85***
Panel B: Institutional investors
Buy 0.1685 0.8187 1.6369 2.4492
t-value 19.3*** 4.9*** 7.6*** 12.9***
Sell À0.1550 À0.8582 À1.7010 À2.5579
t-value À16.75*** À5.13*** À8.98*** À14.25***
Difference 0.3235 1.6769 3.3379 5.0071
t-value 47.71*** 11.63*** 16.61*** 2.33***
Each investor-type's trade imbalance is sorted into ?ve equal sets. The quintile of the days within the highest positive trade imbalance is designated as the buying days for that
investor type. The quintile of the largest negative trade imbalance is the selling days. The buy and sell days represent the largest trade imbalance by the type of investors. Each
quintile of buy and sell days has 241 observations.
I compute the 1-day, 2-day, and 3-day returns following the buy or sell trading days.
“Difference” denotes difference in trade imbalance or returns between buy and sell.
The return difference between buy days and sell days is tested using a difference in means t-statistics.
*, **, and *** denote signi?cant at the 1%, 5%, and 10% signi?cance level, respectively.
Table 3
Performance for individual and institutional traders.
Individuals t-value Institutions t-value Difference t-value
Panel A: Total contracts
Total À1307.8 À358.9*** 1832.6 182.6*** À3140.4 À275.3***
Realized À1122.5 À260.7*** 1310.0 87.6*** À2432.5 À213.5***
Unrealized À998.9 À400.2*** 1113.4 143.0*** À2112.8 À158.6***
Panel B: Winning contracts
Total 9278.5 1980.4*** 25,001.8 1517.6*** À15,723.3 À1997.5***
Realized 7778.7 1986.1*** 20,064.4 1239.4*** À12,285.7 À1066.7***
Unrealized 6252.7 1750.8*** 15,686.1 1767.8*** À9433.4 À973.2***
Panel C: Losing contracts
Total À8666.4 À1822.7*** À22,666.1 À1480.5*** 18,999.7 758.9***
Realized À11,670.4 À1277.2*** À17,803.8 À1254.0*** 6133.4 439.7***
Unrealized À5191.1 À1736.0*** À14,077.3 À1666.5*** 8996.2 169.5***
This table reports the returns per contract for individual and institutional investors over the sample period from 1, January 2004 to 31, December 2008.
“Difference” denotes the difference between individuals and institutions, which is tested using a difference in means t-statistic.
*, **, and *** denote signi?cant at the 1%, 5%, and 10% signi?cance level, respectively.
M.-C. Lin, M.-T. Chiang / Asia Paci?c Management Review 20 (2015) 165e176 169
and individual investors are prone to be noisy traders in the TAIEX
futures markets.
4. Trading due to information or behavioral bias?
Because the only assets examined are the TAIEX futures con-
tracts, unlike equity stocks, they do not permit an examination of a
wide range of behavioral biases like familiarity, limited attention,
and representativeness. Thus, the study focuses on only two biases:
overcon?dence and disposition effect.
4.1. Overcon?dence
Overcon?dence is considered to be the most robust ?nding in
the psychology of judgment (De Bondt & Thaler, 1995). Over-
con?dence is induced by two cognitive biases: biased self-
attribution and con?rmatory bias (Daniel et al., 1998; Gervais &
Odean, 2001; Hirshleifer, 2001; Odean, 1998b). Biased self-
attribution is that people tend to attribute successes to their own
abilities and failure to bad luck or external factors. Con?rmatory
bias is that people are prone to interpret evidence as consistent
with their prior beliefs. Einhorn (1980) claims that individuals have
a higher tendency to be overcon?dent in a setting where more
judgment is required to evaluate information, and where the
feedback on the judgment is ambiguous in the short run. Obviously,
the futures markets are the circumstances where investment de-
cisions require professional knowledge and the feedback is slow
and noisy. Therefore, an abundance of overcon?dent investors exist
in futures markets.
To gain insights into whether the poor performance of indi-
vidual trades is motivated by psychological biases, I use turnover
ratio as a proxy for overcon?dence (Barber & Odean, 2001; Odean,
1999). Turnover ratio measures the total number of contracts
traded in a period relative to the number of open positions at the
end of the period. The turnover ratio for investors within investor
type i on date t is de?ned as:
TURN
i;t
¼
P
N
i
t
j¼1
TV
i
j;t
P
N
i
t
j¼1
OP
i
j;t
; (4)
where N
i
t
is number of investors for investor type i on day t, TV
i
j;t
is
the trading volume of the j trader within investor type i on day t,
and OP
i
j;t
is the number of open interests of the j trader within
investor type i on day t.
From Table 4, it is evident that individuals have higher turnover
ratio than institutional investors do. On average, the turnover ratio
is 59.84% for individuals and 19.68% for institutions. No matter
whether they hold a winning or losing contract, individuals have a
higher turnover in the positions than institutions do. As indicated
by Barber and Odean (2001), the overcon?dence-induced trading
by individual investors is associated with poor investment
performance. Combined with their increased aggressiveness in
placing orders (see Table 1), the trading decisions of individual
investors appear to be motivated more by cognitive biases, such as
overcon?dence, than by information.
4.2. Disposition effect
In addition to overcon?dence, disposition effect is one of the
most well-known behavioral biases of investors. Behavioral re-
searchers attribute this phenomenon to loss aversion (e.g.,
Kahneman & Tversky, 1979; Odean, 1998a; Kyle, Hui, & Xiong,
2006; Hens & Vlcek, 2011; Barberis & Xiong, 2009). Loss aversion
is proposed by Kahneman and Tversky (1979) and as part of pros-
pect theory. According to the prospect theory, the decision-making
under risk is associated with gains and losses, not ?nal wealth
levels; investors are more sensitive to losses than to gains, and are
risk averse for gains and risk seeking for losses. This risk-averse
behavior for gains and risk-seeking behavior for losses then lead
to the disposition effect. In addition to loss aversion, Shefrin and
Statman (1985) suggest regret and pride, which has recently been
supported with experimental evidence (O'Curry Fogel & Berry,
2006), as another explanation for the disposition effect. Wanting
to feel pride by realizing gains and avoiding regret by delaying
realizing losses is what causes investors to realize gains more
quickly than losses.
I use two methodologies to measure the disposition effect. First
of all, following Odean's (1998a) methodology, I measure the
disposition effect by calculating and comparing the difference be-
tween investors' propensity to realize gains (PGR) and their pro-
pensity to realize losses (PLR). The proportion of gain realized
(PGR
i,t
) and the proportion of loss realized (PLR
i,t
) for investor type i
on date t are de?ned as:
PGR
i;t
¼
N
i
RG;t
N
i
RG;t
þN
i
PG;t
; PLG
i;t
¼
N
i
RL;t
N
i
RL;t
þN
i
PL;t
; (5)
with N
i
RG;t
and N
i
RL;t
having the same de?nition as found in equation
(2), and N
i
PG;t
and N
i
PL;t
being de?ned as follows:
N
i
PG;t
¼ number of trades by investor type i where there is a
paper gain on day t
N
i
PL;t
¼ number of trades by investor type i where there is a
paper loss on day t
The Odean's disposition effect (DE1) for investor type i on day t
is computed as:
DE1
i;t
¼ PGR
i;t
ÀPLR
i;t
: (6)
A positive DE1
i,t
indicates that investor type i tends to realize
gains more than losses on day t. Table 5 reports the disposition
effect. The results show that individuals have higher PGR than PLR,
Table 4
Turnover ratio for individual and institutional traders.
Individuals t-value Institutions t-value Difference t-value
Total 59.84 255.04*** 19.68 98.76*** 40.16 130.45***
Winners 78.09 197.16*** 26.75 74.80*** 51.34 96.21***
Losers 33.72 116.85*** 15.62 74.93*** 18.11 50.86***
Winners-Losers 44.37 108.10*** 11.13 76.02*** 33.23 40.39***
This table reports the turnover ratio for individual and institutional investors over the sample period from 1, January 2004 to 31, December 2008.
Turnover ratio measures the total number of contracts traded in a period relative to the size of open positions at the end of the period.
The turnover ratio is represented in percent.
“Difference” denotes the difference between individuals and institutions, which is tested using a difference in means t-statistic.
*, **, and *** denote signi?cant at the 1%, 5%, and 10% signi?cance level, respectively.
M.-C. Lin, M.-T. Chiang / Asia Paci?c Management Review 20 (2015) 165e176 170
but institutions have lower PGR than PLR. In other words, indi-
vidual investors exhibit the disposition effect, but institutional in-
vestors show evidence for the reverse-disposition effect. A closer
look reveals that individuals have a higher tendency to realize both
winning and losing contracts than institutions do. This is in line
with the higher turnover ratio for individual investors in Table 4.
The other method to measure the disposition effect is proposed
by Shapira and Venezia (2001), who compare the duration of losing
round trips to those of winning round trips. I de?ne the average
duration of the pro?table round trip (DG
i,t
) and the average dura-
tion of the losing round trip (DL
i,t
) for investor type i on date t as:
DG
i;t
¼
PN
i
RG;t
j¼1
DG
i
j;t
N
i
RG;t
; DL
i;t
¼
PN
i
RL;t
j¼1
DL
i
j;t
N
i
RL;t
; (7)
with N
i
RG;t
and N
i
RL;t
having the same de?nition as found in equation
(2), and DG
i
j;t
and DL
i
j;t
being de?ned as follows:
DL
i
j;t
¼the duration of the j winning round trip for investor type i
on day t
DL
i
j;t
¼ the duration of the j losing round trip for investor type i
on day t
The Shapira and Venezia's disposition effect (DE2) for investor
type i on day t is then computed as:
DE2
i;t
¼ DG
i;t
ÀDL
i;t
: (8)
A negative DE2
i,t
indicates that investor type i holds a winning
contract shorter than a losing contract. That is, they have a higher
tendency to realize gains than losses on day t. Table 6 reports the
duration of a round-trip trade, which is represented in minutes. It is
evident that individuals hold the futures contract for a shorter time
frame than institutional investors do. On average, the duration of a
round trip is 5.45 (327/60) hours for individuals, and 14.345 (860.7/
60) hours for institutions. Regardless of whether they hold a win-
ning or losing contract, institutions hold contracts longer than
individuals do. That is, individual investors have a higher pro-
pensity to realize both gains and losses than institutional investors
do. In addition, the holding time per winning contract is shorter
(longer) than the holding time per losing contract for individual
(institutional) investors. This is consistent with the results based on
comparing the proportion of gain realized (PGR
i,t
) and the propor-
tion of loss realized (PLR
i,t
). In other words, individuals, instead of
institutions, suffer from disposition biases.
4.3. Quantile regression
To performa robust check, we run a regression test of predicting
open or close trades. To know well about the information about the
tail behaviors of open or close trades' distribution, I adopt a quantile
regression approach proposed by Koenker and Bassett (1978). The
quantile regression permits the estimation of various quantile
functions of a conditional distribution, with the median (0.5th
quantile) function being a special case.
Given the data (y
t
, x
t
) for t ¼ 1, …, T, where x
t
is k  1, the linear
speci?cation for the conditional quantiles of y can be considered as
follows:
yt ¼ x
t
b þe
t
: (9)
The qth quantile regression estimator of b is obtained by mini-
mizing the average of asymmetrically weighted absolute errors
with weight q on positive errors and weight (q À 1) on negative
errors:
V
T
V
T
ðb; qÞ ¼
1
T
2
4
q
X
t:yt x
0
t
b
y
t
Àx
0
t
b
þð1 ÀqÞ
X
t:yt x
0
t
b
y
t
Àx
0
t
b
3
5
:
(10)
Each quantile regression describes a particular (center or tail)
point of the conditional distribution. This approach is particularly
useful when the conditional distribution of ?nancial variables is
heterogenous and does not have a “standard” shape, such as an
asymmetric, fat-tailed or truncated.
4.4. Regression model
To assess the relation between individual (institutional) trading
behavior and behavioral bias, I test for the ratios of open trading
volume to total volume (OP) and close trading volume to total
volume (CL) as follows:
OP
it
¼ a
11
þa
12
OC
itÀ1
þ
X
K
k¼1
b
1k
R
tÀk
þ
X
K
k¼1
l
1k
logðV
tÀk
Þ
þ
X
K
k¼1
g
1k
s
2
tÀk
þ 3
OPt
; (11)
Table 5
Disposition effect individual and institutional traders.
Total Individuals Institution Difference
PGR 0.4848 0.6918 0.3320 0.3598
t-value 56.43*** 76.58*** 30.97*** 75.73***
PLR 0.4434 0.5760 0.3652 0.2108
t-value 54.65*** 85.09*** 18.71*** 82.81***
DE 0.0414 0.1158 À0.0332 0.1490
t-value 15.04*** 13.85*** À7.08*** 14.29***
This table reports the mean of PGR, PLR, and DE for individual and institutional
investors over the sample period from 1, January 2004 to 31, December 2008.
PGR is the number of realized gains divided by the number of realized gains plus the
number of paper gains, and PLR is the number of realized losses divided by the
number of realized losses plus the number of paper losses.
DE is the difference of PGR and PLR.
*, **, and *** denote signi?cant at the 1%, 5%, and 10% signi?cance level, respectively.
Table 6
Duration for individual and institutional traders.
Individuals t-value Institutions t-value Difference t-value
Total 327.0 1556.6*** 860.7 1413.5*** À533.7 À155.2***
Winners 294.5 111.7*** 871.0 131.2*** À576.4 À861.2***
Losers 365.7 193.6*** 850.3 968.9*** À484.6 À632.4***
Winners-Losers À71.2 À168.9*** 20.7 16.99*** À91.9 À107.3***
This table reports the duration measure for individual and institutional investors over the sample period from 1, January 2004 to 31, December 2008.
Duration is the time to hold a round-trip contract, which is represented in minutes.
“Difference” denotes the difference between individuals and institutions, which is tested using a difference in means t-statistic.
*, **, and *** denote signi?cant at the 1%, 5%, and 10% signi?cance level, respectively.
M.-C. Lin, M.-T. Chiang / Asia Paci?c Management Review 20 (2015) 165e176 171
Table 7
Regression test: Open trading.
OLS 0.05 0.1 0.2 0.3 0.4 0.5(LAD) 0.6 0.7 0.8 0.9 0.95
Panel A: Individual investors
Intercept 6.228 0.716 0.757 0.797 0.823 0.860 0.881 0.916 0.969 1.042 1.606 16.505
(4.78)*** (7.60)*** (7.69)*** (7.66)*** (5.10)*** (8.79)*** (6.86)*** (7.03)*** (7.69)*** (5.04)*** (3.46)*** (2.77)**
OC
t À 1
12.629 0.036 0.091 0.183 0.182 0.238 0.220 0.252 0.283 0.259 2.341 54.891
(5.15)*** (0.44) (1.57) (3.59)*** (3.67)*** (4.70)*** (4.10)*** (3.88)*** (3.70)*** (2.39)** (2.25)** (3.04)***
r
t À 1
0.631 0.143 0.334 0.258 0.278 0.169 0.112 0.081 À0.304 À0.690 À5.378 À6.858
(1.04) (0.68) (1.91)* (1.72) (1.64) (0.91) (0.51) (0.35) (À1.36) (À1.99)** (À1.15) (À0.66)
r
t À 2
À0.805 0.355 0.293 0.136 0.147 0.059 0.151 0.015 À0.248 À0.245 À0.395 3.947
(À0.84) (0.93) (1.60) (0.86) (0.94) (0.33) (0.79) (0.07) (À1.21) (À0.64) (À0.15) (0.41)
V
t À 1
À0.177 À0.001 À0.001 À0.001 À0.001 À0.001 À0.001 À0.001 À0.002 À0.003 À0.013 À0.171
(À6.85)*** (À3.78)*** (À2.98)*** (À3.77)*** (À4.46)*** (À5.74)*** (À5.90)*** (À5.62)*** (À6.28)*** (À7.36)*** (À1.54) (À1.83)*
V
t À 2
À0.117 0.000 À0.001 À0.001 À0.001 À0.001 À0.002 À0.002 À0.002 À0.003 À0.018 À0.326
(À4.11)*** (À0.37) (À3.78)*** (À5.19)*** (À4.72)*** (À4.94)*** (À6.31)*** (À5.04)*** (À4.71)*** (À4.87)*** (À1.02) (À1.77)
s
2
tÀ1
1.603 0.004 0.008 0.007 0.012 0.013 0.015 0.018 0.029 0.061 0.426 7.408
(5.37)*** (0.66) (1.67) (1.93)* (3.76)*** (4.24)*** (5.59)*** (3.51)*** (2.47)** (3.39)*** (0.85) (1.18)
s
2
tÀ2
0.743 À0.001 À0.002 0.000 0.000 0.002 0.003 0.008 0.007 0.017 0.232 1.723
(1.75) (À0.21) (À0.50) (0.02) (À0.08) (0.99) (1.20) (2.85)*** (1.70) (1.54) (0.67) (0.47)
Adj. R
2
0.134 0.105 0.105 0.106 0.105 0.105 0.105 0.105 0.105 0.106 0.111 0.196
Panel B: Institutional investors
Intercept 1.128 0.162 0.178 0.211 0.240 0.273 0.305 0.336 0.396 0.511 0.856 1.165
(7.63)*** (5.64)*** (2.75)** (8.04)*** (7.51)*** (8.92)*** (3.40)*** (8.68)*** (6.21)*** (5.70)*** (9.22)*** (5.13)***
OC
t À 1
16.928 À5.138 À4.829 4.426 5.593 6.636 8.361 8.180 11.729 17.614 37.036 52.263
(1.67) (À2.71)*** (À1.98)** (1.55) (1.29) (1.65) (1.57) (1.89)* (5.09)*** (4.42)*** (4.36)*** (4.10)***
r
t À 1
0.769 0.266 0.067 0.276 0.394 0.449 0.417 0.402 0.567 0.929 1.805 4.433
(1.05) (1.31) (0.27) (1.26) (2.01)** (2.78)** (2.89)*** (2.78)** (2.97)*** (2.48)** (1.43) (1.60)
r
t À 2
0.680 0.031 À0.037 À0.217 À0.391 À0.318 À0.283 À0.288 À0.575 À0.408 0.482 0.976
(0.20) (0.14) (À0.16) (À1.06) (À2.03)** (À1.89)* (À1.73) (À1.59) (À2.20)** (À0.99) (0.34) (0.18)
V
t À 1
À0.019 0.000 0.000 0.000 0.000 0.000 0.000 0.000 À0.001 À0.002 À0.006 À0.010
(À2.03)** (0.73) (0.88) (0.69) (À0.73) (À0.55) (À0.66) (À2.57)** (À3.87)*** (À4.47)*** (À4.46)*** (À3.95)***
V
t À 2
À0.001 0.001 0.001 0.001 0.001 0.000 0.000 0.000 0.000 À0.001 À0.003 À0.007
(À0.26) (2.95)*** (3.04)*** (2.92)*** (2.52)** (0.86) (À0.39) (À0.22) (À0.84) (À1.16) (À0.74) (À1.43)
s
2
tÀ1
0.116 À0.017 À0.009 À0.009 À0.006 0.000 0.001 0.003 0.008 0.029 0.083 0.213
(1.73) (À2.71)** (À3.02)*** (À2.80)** (À1.16) (0.12) (0.62) (1.41) (1.58) (1.41) (1.24) (0.93)
s
2
tÀ2
0.059 À0.003 À0.006 À0.001 0.000 0.000 0.001 0.000 0.000 0.011 0.048 0.190
(1.56) (À1.59) (À2.41)** (À0.44) (À0.10) (À0.20) (0.52) (À0.22) (À0.15) (1.02) (2.41)** (1.76)
Adj. R
2
0.138 0.134 0.124 0.112 0.107 0.105 0.105 0.105 0.107 0.112 0.128 0.151
This table reports the relationship between individual and institutional trading and the overcon?dence level. The individual and institutional open trading volume is normalized by total trading volume. Above variables are
regressed on the proxy of overcon?dence level (OC
t
), with lagged index futures return (r
t À 1
and r
t À 2
), lagged index futures trading volume (V
t À 1
and V
t À 2
) and lagged daily 5-min volatility (s
2
tÀ1
and s
2
tÀ2
) being included as
control variables. T-values are estimated using heteroscedasticity-consistent standard errors.
Values in parenthesis are T-values.
*** denotes signi?cant at the 1% level.
** denotes signi?cant at the 5% level.
* denotes signi?cant at the 10% level, respectively.
M
.
-
C
.
L
i
n
,
M
.
-
T
.
C
h
i
a
n
g
/
A
s
i
a
P
a
c
i
?
c
M
a
n
a
g
e
m
e
n
t
R
e
v
i
e
w
2
0
(
2
0
1
5
)
1
6
5
e
1
7
6
1
7
2
Table 8
Regression test: Close trading.
OLS 0.05 0.1 0.2 0.3 0.4 0.5(LAD) 0.6 0.7 0.8 0.9 0.95
Panel A: Individual investors
Intercept 5.917 0.655 0.697 0.748 0.774 0.813 0.840 0.868 0.938 1.017 1.466 17.780
(4.63)*** (3.26)*** (4.28)*** (7.86)*** (6.12)*** (6.02)*** (8.24)*** (7.54)*** (4.55)*** (4.67)*** (2.62)** (3.58)***
OC
t
1.44 1.06 1.05 1.14 0.17 0.18 0.15 0.14 0.16 0.02 1.99 4.95
(3.39)*** (2.81)*** (2.71)*** (3.03)*** (1.73) (1.18) (1.44) (1.24) (1.64) (1.980)** (2.78)*** (2.19)**
DE
t
À1.805 À0.315 À0.248 À0.181 À0.102 À0.148 À0.138 À0.130 0.315 0.347 0.681 1.500
(À1.93)* (À2.36)** (À2.93)*** (À2.24)** (À1.20) (À1.86)* (À1.60) (À1.29) (2.35)** (2.52)** (2.47)** (2.75)**
r
t À 1
0.576 0.014 0.138 0.014 0.127 0.380 0.419 0.445 1.078 1.675 0.854 0.321
(2.33)** (0.05) (0.75) (0.12) (0.71) (1.76) (2.23)** (2.16)** (2.81)** (4.83)*** (0.95) (0.72)
r
t À 2
À0.636 À0.182 À0.166 0.154 À0.023 À0.244 À0.230 À0.385 À0.520 À0.794 À0.217 0.253
(À0.98) (À0.54) (À0.95) (1.07) (À0.15) (À1.17) (À1.08) (À1.75) (À2.08)** (À2.30)** (À0.08) (0.44)
V
t À 1
À0.172 À0.001 À0.001 À0.001 À0.001 À0.001 À0.002 À0.002 À0.002 À0.003 À0.010 À0.156
(À6.84)*** (À1.45) (À2.10)** (À4.83)*** (À4.82)*** (À3.98)*** (À5.25)*** (À5.93)*** (À5.49)*** (À7.17)*** (À1.23) (À3.32)***
V
t À 2
À0.113 0.000 0.000 0.000 0.000 À0.001 À0.001 À0.001 À0.002 À0.003 À0.016 À0.345
(À4.07)*** (0.07) (À0.14) (À1.44) (À1.77) (À2.26)** (À3.03)*** (À3.29)*** (À3.06)*** (À4.59)*** (À0.74) (À1.86)*
s
2
tÀ1
1.554 0.003 0.002 0.006 0.006 0.011 0.014 0.016 0.034 0.057 0.381 7.634
(5.34)*** (0.65) (0.77) (3.12)*** (2.41)** (2.63)** (4.30)*** (3.43)*** (2.08)** (3.39)*** (0.67) (1.17)
s
2
tÀ2
0.754 0.000 0.000 0.000 0.001 0.002 0.006 0.006 0.010 0.018 0.167 1.278
(1.87)* (À0.07) (0.00) (À0.18) (0.55) (0.85) (1.36) (1.75) (1.73) (1.95)* (0.48) (0.30)
Adj. R
2
0.138 0.102 0.102 0.103 0.103 0.103 0.104 0.104 0.105 0.106 0.111 0.198
Panel B: Institutional investors
Intercept 1.086 0.123 0.147 0.178 0.203 0.232 0.265 0.298 0.340 0.426 0.708 1.188
(7.26)*** (4.26)*** (7.34)*** (5.07)*** (3.61)*** (3.51)*** (5.52)*** (5.99)*** (5.10)*** (9.87)*** (2.27)** (9.42)***
OC
t
1.941 1.409 1.002 0.324 0.060 0.439 1.188 2.819 3.242 2.472 6.449 14.312
(1.17) (0.63) (1.17) (0.21) (0.04) (0.26) (1.25) (1.43) (1.75) (1.86)* (4.51)*** (2.97)***
DE
t
1.800 0.141 0.447 0.218 0.138 0.215 1.161 1.446 1.946 3.582 11.167 19.386
(0.62) (0.24) (0.65) (0.32) (0.24) (0.31) (1.71) (1.68) (188)* (3.53)*** (9.76)*** (3.63)***
r
t À 1
À6.702 À0.001 À0.065 À0.379 À0.269 À0.325 À0.349 À0.433 À0.616 À0.961 À2.291 À2.902
(À2.01)** (À0.01) (À0.34) (À2.33)** (À1.55) (À1.95)* (À2.26)** (À2.52)** (À3.60)*** (À3.57)*** (À2.43)** (À0.39)
r
t À 2
0.178 À0.230 À0.290 À0.333 À0.317 À0.326 À0.122 À0.148 À0.169 À0.484 À0.073 1.138
(0.05) (À1.29) (À1.48) (À1.87) (À1.93)* (À2.00)** (À0.70) (À0.78) (À0.80) (À1.46) (À0.06) (0.37)
V
t À 1
À0.019 0.000 0.000 0.000 0.000 0.000 0.000 0.000 À0.001 À0.002 À0.005 À0.007
(À1.95)* (2.07)** (1.84)* (1.40) (1.72) (1.76) (1.59) (10.67) (À3.39)*** (À3.97)*** (À4.07)*** (À1.94)*
V
t À 2
À0.001 0.001 0.001 0.001 0.001 0.000 0.000 0.000 0.000 0.000 À0.001 À0.011
(À0.25) (5.17)*** (3.86)*** (3.86)*** (1.67) (0.71) (À0.43) (À0.09) (0.20) (À0.34) (À0.61) (À1.33)
s
2
tÀ1
0.119 À0.007 À0.008 À0.008 À0.004 À0.001 0.002 0.006 0.011 0.018 0.068 0.229
(2.76)** (À3.23)*** (À3.17)*** (À2.36)** (À1.06) (À0.28) (0.49) (1.61) (4.33)*** (2.18)** (1.42) (0.99)
s
2
tÀ2
0.057 À0.005 À0.004 À0.001 À0.001 À0.001 À0.002 À0.003 À0.005 0.009 0.026 0.175
(1.50) (2.56)** (À1.29) (À0.40) (À0.33) (À0.66) (À0.83) (À1.51) (À2.26)** (0.95) (1.14) (0.57)
Adj. R
2
0.137 0.136 0.123 0.115 0.108 0.105 0.103 0.103 0.104 0.108 0.126 0.150
This table reports the relationship between individual and institutional trading and the overcon?dence level. The individual and institutional close trading volume is normalized by total trading volume. Above variables are
regressed on the proxy of overcon?dence level (OC
t
) and the proxy of disposition effect (DE
t
), with lagged index futures return (r
t À 1
and r
t À 2
), lagged index futures trading volume (V
t À 1
and V
t À 2
) and lagged daily 5-
min volatility (s
2
tÀ1
and s
2
tÀ2
) being included as control variables. T-values are estimated using heteroscedasticity-consistent standard errors.
Values in parenthesis are T-values.
*** denotes signi?cant at the 1% level.
** denotes signi?cant at the 5% level.
* denotes signi?cant at the 10% level, respectively.
M
.
-
C
.
L
i
n
,
M
.
-
T
.
C
h
i
a
n
g
/
A
s
i
a
P
a
c
i
?
c
M
a
n
a
g
e
m
e
n
t
R
e
v
i
e
w
2
0
(
2
0
1
5
)
1
6
5
e
1
7
6
1
7
3
CL
it
¼ a
21
þa
22
OC
it
þa
23
DE1
it
þ
X
K
k¼1
b
2k
R
tÀk
þ
X
K
k¼1
l
2k
logðV
tÀk
Þ þ
X
K
k¼1
g
2k
s
2
tÀk
þ?
CLt
: (12)
For the open trading regression, the proxy of overcon?dence
(OC), which is the daily turnover, is one of the independent vari-
ables. This variable is included since one of the most robust facts
about the trading of investors is overcon?dence, and an over-
con?dent investor is thought to trade excessively. Additionally,
investors have an overall tendency to exhibit the disposition effect,
which is a propensity to sell winners and hold on to losers. Chou
and Wang (2011) argue that, if overcon?dent investors have pre-
viously held a long (short) position, then trading gains from that
position would induce them to buy (sell) more in the subsequent
period, and to do so more aggressively. They also argue that, if
disposition-biased investors have previously held a pro?table long
(short) position, then in order to quickly realize their gains, they
will hastily sell off their long (short) position in the subsequent
period. Because the disposition bias affects an investor's decision
to close a position, not to open a position, we consider it only
when investors try to close a position. But, overcon?dence bias
may affect the decisions to both open and close a new contract.
Therefore, I use the lagged daily turnover (OC
i,t À 1
) to predict the
contemporaneous opening trading volume (OP
i,t
) in equation (11)
to examine the effect of overcon?dence on open decisions. I also
use the concurrent daily turnover (OC
i,t
) and Odean's (1998a)
disposition measure (
[fx4]
) to predict the closing trading volume
(CL
i,t
) in equation (12) to examine whether the overcon?dence and
disposition effect will affect close decisions. Because trading
behavior can be in?uenced by some factors beyond behavioral
biases, I add certain control variables into the regression. They
include lagged TAIEX futures return (R
t À k
), lagged log trading
volume (log(V
t À k
)) and lagged daily volatility as a proxy risk
factor (s
2
tÀk
). The Schwartz Bayesian Criterion (SBC) is used to
determine lagged terms.
As shown in Panel A of Table 7, after controlling for other effects,
a high overcon?dence level among individual investors is associ-
ated with a high open trading volume. Because the OLS results
represent the “averaging” behavior, they yield little information
about the tail behaviors of the given distribution. Therefore, I take a
further look at the results of the quantile regression. It's found that
the overcon?dence tendency is prevalent among individual in-
vestors, with the exception in the lower “tail” of the open trading
distribution. This indicates that overcon?dent characteristics
inspire individual investors to increase their transactions on the
basis of the erroneous belief that they have valuable fundamental
information. By contrast, the OLS results in Panel B indicate that
this does not occur among institutional investors. Interestingly, the
negative coef?cient in the quintiles of q 0.1 reveals that over-
con?dence is correlated with the extremely low open volume of
institutions. But, the relationship within the extreme right tails is
statistically signi?cant in the quantiles of q S 0.6 at the 10% sig-
ni?cance level and q S0.7 at the 1% signi?cance level. This reveals
that overcon?dence also plays a role for institutional investors to
open large positions.
To know whether the decision to close a position is also
motivated by overcon?dence and disposition effect, I have re-
ported the regression of close trading volume in Table 8. The re-
sults are similar to those that were obtained with respect to open
trading. Individual investors tend to close a position due to
overcon?dence and disposition effect, but institutional investors
do not. In particular, a higher overcon?dence level is associated
with both high and low close trading volume among individual
investors. As demonstrated by the quantile regression in Panel A,
the signi?cantly negative coef?cient of disposition effect for q
0.2 indicates that, under the lower bound of close trading volume,
if individuals have higher tendency to realize their gains than to
realize losses, this would contribute to lower individual close
trading volume. Under the upper bound of close trading volume,
individuals' close trading volume increases with tendency to
realize their gains. This leads to insigni?cant coef?cients within
the middle range of close volume and a signi?cant and positive
coef?cient at the upper bound (q S 0.7). However, institutional
investors have the tendency to close an extremely high volume of
positions (qS0.8) only when they have a higher tendency of
disposition bias. This ?nding indicates that a desire to realize
gains soon and ride on losses contributes to a high closing
volume.
In sum, both open and close trading decisions of individual
investors are motivated to some degree by overcon?dence.
There is also a correlation between the close trading decisions
and the disposition effect for individual investors. Institutional
trading is less subject to behavioral biases, such as over-
con?dence and the disposition effect. The exception occurs
when the open and close volume is extremely large. Over-
con?dence induces institutional investors to open large posi-
tions and the desire to realize gains and ride on losses leads
them to close large positions.
5. Conclusion
This paper employs a unique data set to examine the trading
behaviors of individual and institutional investors in the TAIEX
futures markets. The average return per contract for individual
investors is negative, but positive for institutional investors. The
market returns are positive following the heavy buying by insti-
tutional investors, and negative after their heavy selling. By
contrast, the returns are negative after individual heavy buying and
positive after individual heavy selling. These results suggest that
individual investors are poor market timers, whereas institutional
investors have success in timing the market.
In addition, individual trading activity is more aggressive for
their higher proportion of market order and shorter holding time of
a round trip trade. Further testing indicates that both individual
and institutional trading activity is motivated by behavioral biases.
In particular, individuals are more overcon?dent than individuals;
individuals exhibit a tendency toward the disposition effect, but
institutional investors display a reverse disposition effect. When
institutions are overcon?dent, their open volume is either
extremely low or high. For the lower bounds of quantiles, when
individuals have lower tendency to realize their gains than to
realize losses, close trading volume is relatively low. For the upper
bounds of quantiles, their close trading volume increases with the
increasing tendency to realize their gains. Under the upper bounds
of quantiles, similar results are revealed for institutional investors.
That is, disposition bias leads to a higher close volume. This result is
consistent with Lin (2011), who shows that trades of all categories
of investors are motivated by both behavioral- and market-related
factors, and individuals tend to be more irrational and prone to
misinterpret available information or trade for non-informational
reasons.
Con?icts of interest
All contributing authors declare no con?icts of interest.
M.-C. Lin, M.-T. Chiang / Asia Paci?c Management Review 20 (2015) 165e176 174
A
p
p
e
n
d
i
x
.
A
n
e
x
a
m
p
l
e
o
f
t
h
e
c
a
l
c
u
l
a
t
i
o
n
o
f
t
h
e
v
a
r
i
a
b
l
e
s
o
f
g
a
i
n
s
/
l
o
s
s
e
s
D
a
t
e
B
u
y
/
S
e
l
l
P
r
i
c
e
N
u
m
b
e
r
s
o
f
c
o
n
t
r
a
c
t
s
T
i
m
e
S
e
t
t
l
e
m
e
n
t
p
r
i
c
e
O
p
e
n
i
n
t
e
r
e
s
t
s
A
v
e
r
a
g
e
c
o
s
t
G
r
o
s
s
r
e
a
l
i
z
e
d
g
a
i
n
/
l
o
s
s
C
o
m
m
i
s
s
i
o
n
T
a
x
N
e
t
r
e
a
l
i
z
e
d
g
a
i
n
/
l
o
s
s
D
u
r
a
t
i
o
n
G
r
o
s
s
u
n
r
e
a
l
i
z
e
d
g
a
i
n
/
l
o
s
s
C
o
m
m
i
s
s
i
o
n
_
e
s
t
i
m
a
t
e
T
a
x
_
e
s
t
i
m
a
t
e
N
e
t
u
n
r
e
a
l
i
z
e
d
g
a
i
n
/
l
o
s
s
N
u
m
b
e
r
s
o
f
r
e
a
l
i
z
e
d
g
a
i
n
N
u
m
b
e
r
s
o
f
r
e
a
l
i
z
e
d
l
o
s
s
N
u
m
b
e
r
s
o
f
p
a
p
e
r
g
a
i
n
N
u
m
b
e
r
s
o
f
p
a
p
e
r
l
o
s
s
C
l
o
s
e
b
u
y
v
o
l
u
m
e
C
l
o
s
e
s
e
l
l
v
o
l
u
m
e
O
p
e
n
b
u
y
v
o
l
u
m
e
O
p
e
n
s
e
l
l
v
o
l
u
m
e
2
0
0
4
1
2
1
6
B
6
0
0
5
7
0
9
:
0
0
:
3
6
6
0
2
0
7
6
0
0
5
.
0
0
2
1
.
0
0
0
2
.
1
1
.
6
8
4
1
7
.
2
1
7
7
7
2
0
0
4
1
2
1
7
S
6
0
2
4
2
0
8
:
5
8
:
2
5
6
0
1
9
5
6
0
0
5
.
0
0
7
.
6
0
0
0
.
6
0
.
4
8
1
6
.
5
1
9
2
9
8
1
9
.
0
0
0
1
.
5
1
.
2
0
3
1
6
.
2
9
7
2
5
2
2
0
0
4
1
2
1
7
S
6
0
2
4
2
0
8
:
5
8
:
3
0
6
0
1
9
3
6
0
0
5
.
0
0
7
.
6
0
0
0
.
6
0
.
4
8
1
6
.
5
1
9
2
9
8
1
1
.
4
0
0
0
.
9
0
.
7
2
2
9
.
7
7
8
2
3
2
2
0
0
4
1
2
1
7
S
6
0
2
4
1
0
8
:
5
8
:
3
0
6
0
1
9
2
6
0
0
5
.
0
0
3
.
8
0
0
0
.
3
0
.
2
4
1
3
.
2
5
9
2
9
8
7
.
6
0
0
0
.
6
0
.
4
8
1
6
.
5
1
9
1
2
1
2
0
0
4
1
2
1
7
S
6
0
2
4
2
0
8
:
5
8
:
3
5
6
0
1
9
0
7
.
6
0
0
0
.
6
0
.
4
8
1
6
.
5
1
9
2
9
8
2
2
2
0
0
4
1
2
1
7
B
6
0
0
6
5
0
9
:
2
5
:
0
6
6
0
1
9
5
6
0
0
6
.
0
0
0
.
0
0
0
1
.
5
1
.
2
0
1
À
2
.
7
0
1
5
5
2
0
0
4
1
2
1
7
B
6
0
0
6
1
0
9
:
2
5
:
0
6
6
0
1
9
6
6
0
0
6
.
0
0
0
.
0
0
0
1
.
8
1
.
4
4
1
À
3
.
2
4
1
6
1
2
0
0
4
1
2
1
7
B
6
0
0
6
1
0
9
:
2
5
:
0
7
6
0
1
9
7
6
0
0
6
.
0
0
1
8
.
2
0
0
2
.
1
1
.
6
8
4
1
4
.
4
1
7
7
1
2
0
0
4
1
2
2
1
S
5
9
7
6
7
1
1
:
0
2
:
5
5
5
9
7
8
0
À
4
2
.
0
0
0
2
.
1
1
.
6
7
7
À
4
5
.
7
7
7
6
9
7
7
7
2
0
0
4
1
2
2
2
B
6
0
4
2
1
0
9
:
2
4
:
1
5
6
0
2
5
1
6
0
4
2
.
0
0
0
.
0
0
0
0
.
3
0
.
2
4
2
À
0
.
5
4
2
1
1
2
0
0
4
1
2
2
2
B
6
0
6
4
7
0
9
:
3
0
:
1
5
6
0
2
5
8
6
0
6
1
.
2
5
4
.
4
0
0
2
.
4
1
.
9
4
0
0
.
0
6
0
8
7
2
0
0
4
1
2
2
2
S
6
0
6
2
1
0
9
:
3
9
:
1
1
6
0
2
5
7
6
0
6
1
.
2
5
0
.
1
5
0
0
.
3
0
.
2
4
2
À
0
.
3
9
2
9
.
7
5
1
.
0
5
0
2
.
1
1
.
6
9
7
À
2
.
7
4
7
1
7
1
2
0
0
4
1
2
2
2
S
6
0
4
6
1
1
0
:
1
8
:
3
3
6
0
2
5
6
6
0
6
1
.
2
5
À
3
.
0
5
0
0
.
3
0
.
2
4
2
À
3
.
5
9
2
4
8
.
7
5
À
1
8
.
3
0
0
1
.
8
1
.
4
5
3
À
2
1
.
5
5
3
1
6
1
2
0
0
4
1
2
2
2
S
6
0
4
6
6
1
0
:
1
8
:
3
5
6
0
2
5
0
À
1
8
.
3
0
0
1
.
8
1
.
4
5
3
À
2
1
.
5
5
3
4
8
.
7
5
6
6
2
0
0
4
1
2
2
3
B
5
9
9
1
7
0
9
:
4
9
:
1
2
6
0
0
2
7
5
9
9
1
.
0
0
0
.
0
0
0
2
.
1
1
.
6
7
7
À
3
.
7
7
7
7
7
2
0
0
4
1
2
2
3
S
5
9
7
1
7
1
0
:
0
3
:
4
2
6
0
0
2
0
À
2
8
.
0
0
0
2
.
1
1
.
6
7
5
À
3
1
.
7
7
5
1
4
7
7
2
0
0
4
1
2
2
8
B
6
0
4
1
7
0
9
:
1
4
:
5
6
6
0
2
0
7
6
0
4
1
.
0
0
À
2
9
.
4
0
0
2
.
1
1
.
6
8
9
À
3
3
.
1
8
9
7
7
2
0
0
4
1
2
2
9
B
6
0
5
6
1
0
8
:
4
5
:
3
6
6
1
1
8
8
6
0
4
2
.
8
8
2
1
.
0
0
0
2
.
4
1
.
9
3
6
1
6
.
6
6
4
8
1
2
0
0
4
1
2
2
9
B
6
0
5
7
2
0
8
:
4
6
:
0
4
6
1
1
8
1
0
6
0
4
5
.
7
0
2
2
.
6
0
0
3
2
.
4
2
1
1
7
.
1
7
9
1
2
2
0
0
4
1
2
2
9
B
6
0
5
7
1
0
8
:
4
6
:
0
4
6
1
1
8
1
1
6
0
4
6
.
7
3
2
2
.
6
0
0
3
.
3
2
.
6
6
3
1
6
.
6
3
7
1
1
1
2
0
0
4
1
2
2
9
B
6
0
5
7
1
0
8
:
4
6
:
0
5
6
1
1
8
1
2
6
0
4
7
.
5
8
2
2
.
6
0
0
3
.
6
2
.
9
0
5
1
6
.
0
9
5
1
2
1
2
0
0
4
1
2
2
9
B
6
0
5
7
1
0
8
:
4
6
:
0
6
6
1
1
8
1
3
6
0
4
8
.
3
1
2
2
.
6
0
0
3
.
9
3
.
1
4
7
1
5
.
5
5
3
1
3
1
2
0
0
4
1
2
2
9
B
6
0
5
7
2
0
8
:
4
6
:
0
6
6
1
1
8
1
5
6
0
4
9
.
4
7
2
2
.
6
0
0
4
.
5
3
.
6
3
2
1
4
.
4
6
8
1
5
2
2
0
0
4
1
2
2
9
S
6
0
9
2
3
0
9
:
3
2
:
0
9
6
1
1
8
1
2
6
0
4
9
.
4
7
2
5
.
5
2
0
0
.
9
0
.
7
2
8
2
3
.
8
9
2
1
7
3
1
0
2
.
0
8
0
3
.
6
2
.
9
1
4
9
5
.
5
6
6
3
1
2
3
2
0
0
4
1
2
2
9
S
6
0
9
2
5
0
9
:
3
2
:
1
0
6
1
1
8
7
6
0
4
9
.
4
7
4
2
.
5
3
3
1
.
5
1
.
2
1
4
3
9
.
8
1
9
1
7
3
5
9
.
5
4
7
2
.
1
1
.
7
0
0
5
5
.
7
4
7
5
7
5
2
0
0
4
1
2
2
9
B
6
1
0
1
7
1
1
:
0
4
:
4
4
6
1
1
8
1
4
6
0
7
5
.
2
3
7
2
.
1
4
7
4
.
2
3
.
4
0
9
6
4
.
5
3
7
1
4
7
2
0
0
4
1
2
2
9
S
6
1
1
3
2
1
2
:
0
4
:
5
5
6
1
1
8
1
2
6
0
7
5
.
2
3
1
5
.
1
0
7
0
.
6
0
.
4
8
8
1
4
.
0
1
9
1
9
2
.
5
9
0
.
6
4
0
3
.
6
2
.
9
2
5
8
4
.
1
1
5
2
1
2
2
2
0
0
4
1
2
2
9
S
6
1
1
3
1
1
2
:
0
5
:
0
0
6
1
1
8
1
1
6
0
7
5
.
2
3
7
.
5
5
3
0
.
3
0
.
2
4
4
7
.
0
1
0
1
9
3
.
5
8
3
.
0
8
7
3
.
3
2
.
6
8
1
7
7
.
1
0
5
1
1
1
1
2
0
0
4
1
2
2
9
S
6
1
1
3
2
1
2
:
0
5
:
0
4
6
1
1
8
9
6
0
7
5
.
2
3
1
5
.
1
0
7
0
.
6
0
.
4
8
8
1
4
.
0
1
9
1
9
3
.
5
6
7
.
9
8
0
2
.
7
2
.
1
9
4
6
3
.
0
8
6
2
9
2
2
0
0
4
1
2
2
9
S
6
1
1
3
2
1
2
:
0
5
:
0
7
6
1
1
8
7
6
0
7
5
.
2
3
1
5
.
1
0
7
0
.
6
0
.
4
8
8
1
4
.
0
1
9
1
9
3
.
5
5
9
.
8
7
3
2
.
1
1
.
7
0
7
5
6
.
0
6
6
2
7
2
2
0
0
4
1
2
3
1
S
6
1
2
4
7
0
9
:
3
9
:
2
4
6
1
8
8
0
6
8
.
2
7
3
2
.
1
1
.
7
0
8
6
4
.
4
6
5
6
4
7
.
5
7
7
M.-C. Lin, M.-T. Chiang / Asia Paci?c Management Review 20 (2015) 165e176 175
References
Bange, M. M. (2000). Do the portfolios of small investors re?ect positive feedback
trading? Journal of Financial and Quantitative Analysis, 35, 239e255.
Barber, B. M., Lee, Y. T., Liu, Y. J., & Odean, T. (2005). Who loses from trade? Evidence
from Taiwan. In EFA 2005 Moscow Meetings Paper (vol. 529062). http://ssrn.
com/abstract.
Barber, B. M., Lee, Y., Liu, Y., & Odean, T. (2009). Just how much do individual in-
vestors lose by trading? Review of Financial Studies, 22, 609e632.
Barber, B. M., & Odean, T. (2001). Boys will be boys: gender, overcon?dence, and
common stock investment. Quarterly Journal of Economics, 116, 261e292.
Barber, B. M., Odean, T., & Zhu, N. (2009). Systematic noise. Journal of Financial
Markets, 12, 547e569.
Barberis, N., & Xiong, W. (2009). What drives the disposition effect? An analysis of a
long-standing preference-based explanation. Journal of Finance, 64, 751e784.
Benos, A. (1998). Aggressiveness and survival of overcon?dent traders. Journal of
Financial Markets, 1, 353e383.
Campbell, J. Y., & Kyle, A. S. (1993). Smart money, noise trading and stock price
behavior. Review of Economic Studies, 6, 1e34.
Campbell, J. Y., Ramadorai, T., & Vuolteenaho, T. (2005). Caught on tape: Institutional
order ?ow and stock returns (Working paper 11439). Cambridge, MA: National
Bureau of Economic Research.
Chakravarty, S. (2001). Stealth-trading: which trader's trades move stock prices?
Journal of Financial Economics, 61, 289e307.
Chan, K., Chung, Y. P., & Fong, W. (2002). The information role of stock and option
volume. Review of Financial Studies, 15, 1049e1075.
Chan, K., & Fong, W. (2000). Trade size, order imbalance, and the volatility-volume
relation. Journal of Financial Economics, 57, 247e273.
Chang, C. C., Hsieh, P. F., & Lai, H. N. (2009). Do informed option investors predict
stock return? Evidence from the Taiwan stock exchange. Journal of Banking and
Finance, 33, 757e764.
Chang, C. C., Hsieh, P. F., & Wang, Y. H. (2010). Information content of options trading
volume for future volatility: evidence from Taiwan options market. Journal of
Banking and Finance, 34, 174e183.
Cheng, T. Y., Lin, C. H., & Chuang, S. S. (2007). Who is the winner? trading behavior
and performance for major types of tradersdEvidence from Taiwan's futures
market. Asia Paci?c Management Review, 12(1), 13e21.
Chen, S. Y., Lin, C. C., Chou, P. H., & Hwang, D. Y. (2002). Price discovery between
TAIFEX TAIEX index futures and SGX MSCI Taiwan index futures. Journal of
Futures Markets, 22(3), 219e240.
Chordia, T., Roll, R., & Subrahmanyam, A. (2002). Order imbalance, liquidity, and
market returns. Journal of Financial Economics, 65, 111e130.
Chou, H. C., Chen, W. N., & Chen, D. H. (2006). The expiration effects of stock-index
derivatives: empirical evidence from the Taiwan futures exchange. Emerging
Markets Finance and Trade, 42(5), 81e102.
Chou, P. H., Lin, M. C., & Yu, M. T. (2006). Margins and price limits in Taiwan's stock
index futures market. Emerging Markets Finance and Trade, 42(1), 62e88.
Chou, R. K., & Wang, G. H. K. (2006). Transaction tax and market quality of the
Taiwan stock index futures. Journal of Futures Markets, 25(2), 1195e1216.
Chou, R. K., & Wang, Y. Y. (2011). A test of the different implications of the over-
con?dence and disposition hypotheses. Journal of Banking and Finance, 35,
2037e2046.
Chueh, H., & Yang, D. Y. (2005). Expiration-day effects of index futures: some
empirical evidence from Taiwan stock market. Journal of Financial Studies, 13(2),
71e95.
Daniel, K., Hirshleifer, D., & Subrahmanyam, A. (1998). Investor psychology and
security market under- and overreactions. Journal of Finance, 53, 1839e1885.
De Bondt, W., & Thaler, R. H. (1995). Financial decision-making in markets and
?rms: a behavioral perspectives. In R. A. Jarrow, V. Maksimovic, & W. T. Ziemba
(Eds.), Finance handbook in operation research and management science 9 (pp.
385e410). Amsterdam: North Holland.
De Long, B. J., Shleifer, A., Summers, L. H., & Waldmann, R. J. (1990). Noise trader risk
in ?nancial markets. Journal of Political Economy, 99, 73e738.
Einhorn, H. J. (1980). Overcon?dence in judgment. New Directions for Methodology
of Social and Behavioral Science, 4, 1e16.
Frazzini, A., & Lamont, O. A. (2008). Dumb money: mutual fund ?ows and the cross-
section of stock returns. Journal of Financial Economics, 88(2), 299e322.
Gervais, S., & Odean, T. (2001). Learning to be overcon?dent. Review of Financial
Studies, 14, 1e27.
Hens, T., & Vlcek, M. (2011). Does prospect theory explain the disposition effect?
Journal of Behavioral Finance, 12(3), 141e157.
Hirshleifer, D. (2001). Investor psychology and asset pricing. Journal of Finance, 64,
1533e1597.
Hirshleifer, D., Subrahmanyam, A., & Titman, S. (1994). Security analysis and trading
patterns when some investors receive information before others. Journal of
Finance, 49, 1665e1698.
Jones, C. M., & Lipson, M. (2004). Are retail orders different? (Working paper).
Columbia University, New York.
Kahneman, D., & Tversky, A. (1979). Prospect theory: an analysis of decision under
risk. Econometrica, 46, 171e185.
Kamesaka, A., Nofsinger, J., & Kawakita, H. (2003). Investment patterns and per-
formance of investor groups in Japan. Paci?c-Basin Finance Journal, 11, 1e22.
Kaniel, R., Saar, G., & Titman, S. (2005). Individual investor sentiment and stock
returns (Working paper). Duke University, Durham.
Koenker, R., & Bassett, G. (1978). Regression quantiles. Econometrica, 46, 33e35.
Kuo, W. Y., & Lin, T. C. (2013). Overcon?dent individual day traders: evidence from
Taiwan futures market. Journal of Banking and Finance, 37(9), 3548e3561.
Kyle, A. S., Hui, O. Y., & Xiong, W. (2006). Prospect theory and liquidation decisions.
Journal of Economic Theory, 129, 273e288.
Kyle, A. S., & Wang, F. A. (1997). Speculation duopoly with agreement to disagree:
can overcon?dence survive the market test? Journal of Finance, 52, 273e279.
Lee, C. M. C., & Ready, M. J. (1991). Inferring trade direction from intraday data.
Journal of Finance, 46, 733e746.
Lin, M. C. (2011). Information content for investor groups in TAIEX futures trading.
Asia-Paci?c Journal of Financial Studies, 40, 433e466.
Lin, C. H., Hsu, H. N., & Chiang, C. Y. (2004). The information transmission between
two substitutes of index futures: the case of TAIEX and Mini-TAIEX stock index
futures. Asia Paci?c Management Review, 9(4), 689e707.
Locke, P., & Mann, S. (2005). Professional trader discipline and trade disposition.
Journal of Financial Economics, 76, 401e444.
Nofsinger, J., & Sias, R. (1999). Herding and feedback trading by institutional and
individual investors. Journal of Finance, 54, 2263e2295.
Odean, T. (1998a). Are investors reluctant to realize their losses? Journal of Finance,
55, 1775e1798.
Odean, T. (1998b). Volume, volatility, price, and pro?t when all traders are above
average. Journal of Finance, 53, 1887e1934.
Odean, T. (1999). Do investors trade too much? American Economic Review, 89,
1279e1298.
O'Curry Fogel, S., & Berry, T. (2006). The disposition effect and individual investor
decisions: the roles of regret and counterfactual alternatives. Journal of
Behavioral Finance, 7(2), 107e116.
Shapira, Z., & Venezia, I. (2001). Patterns of behavior of professionally managed and
independent investors. Journal of Banking and Finance, 25(8), 1573e1587.
Shefrin, H. M., & Statman, M. (1985). The disposition to sell winners too early and
ride losers too long. Journal of Finance, 4, 777e779.
Wang, Z. (1998). Ef?ciency loss and constraints on portfolio holdings. Journal of
Financial Economics, 48, 359e375.
Wang, C. (2001). Investor sentiment and return predictability in agricultural futures
markets. Journal of Futures Markets, 21, 929e952.
Wermers, R. (2003). Is money really smart? New evidence on the relation between
mutual fund ?ows, manager behavior, and performance persistence (Working
paper). University of Maryland, Maryland.
M.-C. Lin, M.-T. Chiang / Asia Paci?c Management Review 20 (2015) 165e176 176
doc_667388190.pdf
This study investigates the relation between trading patterns and performance in the TAIEX futures
market. The research shows that individual investors are poor market timers and earn negative returns;
institutional investors have success in timing the market and their trades make positive returns. Individual
trading activity is more aggressive in terms of a higher proportion of the market order and a
shorter holding period for a round-trip trade. Individual trading is also more motivated by behavioral
bias, like overconfidence and disposition effect. Institutional investors exhibit significant overconfidencebased
trading when opening extremely small or relatively large positions.
Trading patterns in the TAIEX futures markets: Information- or behavioral-based
trades?
Mei-Chen Lin
a, *
, Ming-Ti Chiang
a, b
a
National Taipei University, Taiwan
b
Hsing Wu University, Taiwan
a r t i c l e i n f o
Article history:
Received 3 August 2012
Accepted 20 October 2014
Keywords:
Individual investors
Institutional investors
Overcon?dence
Disposition effect
a b s t r a c t
This study investigates the relation between trading patterns and performance in the TAIEX futures
market. The research shows that individual investors are poor market timers and earn negative returns;
institutional investors have success in timing the market and their trades make positive returns. Indi-
vidual trading activity is more aggressive in terms of a higher proportion of the market order and a
shorter holding period for a round-trip trade. Individual trading is also more motivated by behavioral
bias, like overcon?dence and disposition effect. Institutional investors exhibit signi?cant overcon?dence-
based trading when opening extremely small or relatively large positions.
© 2015 College of Management, National Cheng Kung University. Production and hosting by Elsevier
Taiwan LLC. All rights reserved.
1. Introduction
Institutional and individual investors are two major players who
compete to obtain limited pro?tability in ?nancial markets. In-
stitutions generally differ from individuals due to their size and
sophistication (Kaniel, Saar, & Titman, 2005). In particular, indi-
vidual investors are generally less well informed and prone to
misinterpret available information or trade for non-informational
reasons. In comparison, institutional investors have better re-
sources and training than do individual investors. Although in-
stitutions cannot be immune from the same cognitive biases as
individuals, the impacts from behavioral biases may be alleviated
since they may overcome these biases through better information
and analytical skill. Thus, behavioral biases may have different ef-
fects on the trading patterns of institutional and individual traders.
Consistent with this, institutions are found to be informed investors
(e.g., Chakravarty, 2001; Jones & Lipson, 2004), by contrast, indi-
vidual investors are irrational noise traders and frequently succumb
to their cognitive biases (Bange, 2000; Frazzini & Lamont, 2008).
However, some papers posit that individuals make excess returns
through providing liquidity for institutional trading demands
(Campbell, Ramadorai, & Vuolteenaho, 2005; Kaniel et al., 2005).
Therefore, the evidence so far is mixed regarding the roles of in-
dividual and institutional investors.
In view of these con?icting ?ndings, this study uses a new data
set with detailed transaction information to explore whether
trading decisions are in?uenced more by knowledge about value
(information-based trading) or by psychological biases (behavioral-
based trading). The sample used in this study contains 38,684,525
trades of the Taiwan Stock Exchange Capitalization Weighted Stock
Index (TAIEX) futures executed by individual investors and
13,057,657 trades of TAIEX futures executed by institutional in-
vestment accounts during the period January 2004 through
December 2008.
Some previous studies have examined the TAIEX futures mar-
kets. The issues include the information conveyed by trade types of
different categories of investors (Lin, 2011), the impact of a tax rate
reduction on the market quality (Chou & Wang, 2006), the
expiration-day effects (Chou, Chen, & Chen, 2006; Chueh & Yang,
2005), the costeminimization combination of margins, spot price
limits, and futures price limits (Chou, Lin, &Yu, 2006), and the daily
dynamic relation between returns and trades by institutional and
individual investors (Lin, 2011). Other authors compare the infor-
mation transmission between TAIEX and Mini-TAIEX Stock Index
Futures (Lin, Hsu, & Chiang, 2004) and price discovery between
TAIFEX TAIEX index futures and SGX MSCI Taiwan index futures
(Chen, Lin, Chou, & Hwang, 2002).
The research by Cheng, Lin, and Chuang (2007), Kuo and Lin
(2013, 2011) is associated with behavioral-based trades in the
TAIEX futures market. Kuo and Lin (2013) investigate the
* Corresponding author. National Taipei University, 151, University Rd., San Shia
District, New Taipei City 23741, Taiwan.
E-mail address: [email protected] (M.-C. Lin).
Peer review under responsibility of College of Management, National Cheng
Kung University.
HOSTED BY
Contents lists available at ScienceDirect
Asia Paci?c Management Review
j ournal homepage: www. el sevi er. com/ l ocat e/ apmrv
http://dx.doi.org/10.1016/j.apmrv.2014.10.002
1029-3132/© 2015 College of Management, National Cheng Kung University. Production and hosting by Elsevier Taiwan LLC. All rights reserved.
Asia Paci?c Management Review 20 (2015) 165e176
performance of individual day traders in the TAIEX market and ?nd
that individual day traders incur a signi?cant loss. Cheng et al.
(2007) examine the trading behavior and performance of traders
in the TAIEX market. They ?nd that the individual traders are
positive feedback traders while foreign investors tend to engage in
negative feedback trading. Lin (2011) shows that open trading by
foreign institutional investors conveys more information regarding
the underlying index, and open selling of individual investors is
more likely to introduce noise signals to the spot market.
Different from Kuo and Lin (2013), this paper not only in-
vestigates the performance of individual traders, but also compares
the performance and trading behavior of individual and institu-
tional traders in the TAIEX market. Besides, although Cheng et al.
(2007) have examined the trading behavior and performance of
traders in the TAIEX market, they do not discuss whether the dif-
ference in performance between individual and institutional
traders arises from their different behavioral biases. Lin (2011)
shows that individuals are more irrational and more prone to
misinterpret available information. Nevertheless, she does not
address the relationship between trading decisions and behavioral
biases. Additionally, to my best knowledge, no research has yet
investigated whether behavioral biases, like overcon?dence and
disposition bias, will affect individual and institutional investors'
tendency to open a new contract or close an existing contract. This
study intends to ?ll in this gap. By separating trades into open
volume and close volume, it has been possible to examine whether
decisions to open a new contract or close an existing contract are
more affected by information motives or by behavioral biases.
I ?rst compare the return performance and trading behavior of
institutions and individual investors. I ?nd that the average trading
return for individual contracts is negative, whereas the average
return obtained by institutional investors is positive. Furthermore,
institutional investors appear to have some success in market
timing. In particular, the TAIEX futures market experiences positive
returns after institutional buying and negative returns after insti-
tutional selling. By contrast, individual investors are poor market
timers; market return is negative after their buying and positive
after their selling. Nofsinger and Sias (1999) and Kamesaka,
Nofsinger, and Kawakita (2003) posit that a strategy earning posi-
tive returns indicates that it is motivated more by information,
whereas trading which results in a negative return indicates a
higher probability of behavioral-based motivation. When com-
bined with the preliminary ?ndings of this study, this research
indicates that the decisions of individuals are more behavioral-
based, whereas those of institutional investors are more informa-
tion-based.
The behavioral-based model argues that investor trading de-
cisions are in?uenced by behavioral biases, like overcon?dence and
disposition effect (Daniel, Hirshleifer, & Subrahmanyam, 1998;
Gervais & Odean, 2001). Both the overcon?dence and disposition
biases may have effects in the decisions to close a position. But only
overcon?dence bias may affect the decisions to open a new con-
tract. The results show that, the round-trip trade of individuals has
a shorter period than that of institutional investors; individuals
exhibit a tendency toward the disposition effect, but institutional
investors display a reverse disposition effect. Individuals are also
more aggressive in terms of a higher proportion of the market order
and a shorter holding period for a round-trip trade. A further
regression test con?rms the positive relationship between trading
behavior (including both open and close trading) and over-
con?dence among individual investors, and the disposition effect of
individuals occurs when they close an extremely low amount of
positions or relatively large positions. Similarly, overcon?dence
induces institutional investors to open an extremely small amount
of positions or relatively large positions, and institutional investors'
desire to realize gains soon and ride on losses contributes a high
closing volume. Thus, I conclude that the trading activity of both
types of investors shows evidence of behavioral-bias motivation.
However, a further comparison of trades by individual and insti-
tutional investors in the TAIEX futures markets shows that indi-
vidual investors are less-informed and their trading decisions are
more motivated by behavioral biases.
The remainder of this paper is organized as follows. Section 2
introduces the data used in this study; Section 3 compares the
trading performance of individual and institutional investors;
Section 4 explores the motivation behind trading and compares the
trading behavior for institutions and individual traders; with the
conclusions being provided in the ?nal section.
2. Data
The data for this study consist of all of the trades of the spot-
month Taiwan Stock Exchange Capitalization Weighted Stock In-
dex (TAIEX) futures contracts from the Taiwan Futures Exchange
(TAIFEX) during the period January 2004 through December 2008.
The TAIEX is a market capitalization weighted index composed of
all stocks listed in the Taiwan Stock Exchange. The contract size per
contract is the TAIEX index point multiplied by 200 New Taiwan
Dollars (NTD). Contract months of TAIEX index futures are spot
month, the next calendar month, and the next three quarterly
months. The last trading day for each contract month is the third
Wednesday of the delivery month for each contract. Since launched
on 21 July 1998, TAIEX futures contracts have been growing fast and
have made up the largest part of futures contracts in the TAIFEX.
According to Futures Industry Association (FIA), the Taiwan Stock
Index Futures contract (TX) is the sixth largest one of Asian index
futures contracts in 2004 and TAIFEX's global ranking on trading
volume rose to rank 17th in 2008 from 57th in 1998 (Lin, 2011).
The data include trader's ID codes, trading directions (buy/sell),
transaction prices and volume (in number of contracts), and the
time of each transaction. This unique dataset allow me to correctly
identify the trade type for each transaction, including open buy,
open sell, close buy, and close sell. This helps me to reduce the error
likely to occur in studies using the Lee and Ready (1991) algorithm
to speculate on buyer-initiated and seller-initiated volume (see e.g.,
Chan et al., 2002). In addition, different from prior research with
mature futures markets where the institutional investors are the
major participants, individual investors are the major participants
in the TAIEX futures markets. Because of these features and the
availability of data with trade- and trader-type classi?cations, the
TAIFX futures market is an appropriate environment to test the
relation between trading decisions and behavioral bias.
To obtain an indication of whether their trading is based more
on information or behavioral bias, I ?rst provide the net returns per
contract for each investor type. Appendix illustrates the calculation
of major relevant variables: the net realized returns, net unrealized
returns, duration, numbers of realized gains, numbers of realized
losses, numbers of unrealized gains, numbers of unrealized losses,
open buy volume, open sell volume, close buy volume, and close
sell volume. The procedure is as follows. For each trader and each
contract, I ?rst sort trades based on transaction time. Once the ?rst
trade is located, I track each subsequent trade. I mark to market
after each trade and calculate statistics such as the weighted
average costs, open interests (OIs), trading volume, realized gains/
losses and unrealized gains/losses. The weighted average cost for
trade j at time t þ1 is de?ned as: AVC
j;tþ1
¼
Pj$t Vj;t þPj;tþ1Vj;tþ1
Vj;t þVj;tþ1
, where P
jt
(P
jt þ 1
) is the futures price for trade j at time t (t þ1) and V
jt
(V
jt þ 1
)
is the numbers of contracts of the trade for trade j at time t (t þ 1).
For a long (short) position being closed, the sales (purchase) price is
M.-C. Lin, M.-T. Chiang / Asia Paci?c Management Review 20 (2015) 165e176 166
compared to its weighted average costs to determine if it is closed
at a gain or loss. For a position not closed at the end of the day, its
paper gain/loss is calculated based on the daily settlement price.
For example, on December 16, 2004, Trader A starts his trade
by longing (i.e., open buy) seven TX futures contracts with
expiration day on January 25 (TXFA5) at a price of 6005. I then
track his subsequent trades until the maturity of TXFA5. At this
time, his record shows an average cost of 6005, an open interest
of 7 and trading volume of 7. In reality, the Clearing Department
of TAIFEX monitors intra-day price ?uctuations throughout the
trading period in order to evaluate the impact of such changes on
the clearing members. Therefore, I follow Locke and Mann (2005)
to mark to market after each trade, and calculate realized gains/
losses and unrealized gains/losses. Because this trade is not
closed on December 16, 2004, I calculate its corresponding gross
paper gains (losses) based on its concurrent daily settlement
price. The contract size for the TAIEX futures is the index value of
TAIEX Â 200 New Taiwan Dollars (NT$). Thus, with the daily
settlement price being 6020, an average cost of 6005, and open
interest of 7, the trader has a gross unrealized gain of NT$21,000
((6020 e 6005) Â 200 Â 7 ¼ 21,000).
To calculate the net pro?t, I subtract the commission and
transaction tax, which is 1/100th of one percent of the transaction
value before 1, January, 2006. Afterwards, it is 0.4/100th of one
percent of the transaction value. The commission varies among the
brokerage houses and the average is about 150 New Taiwan Dollars
(NTD) for each contract. Then, the net realized gains are NT$ 6519
(7600 À 150 Â 2 Â 2 À (6024 þ 6005)*0.01% Â 200).
On the next day, the trader ?rst closes 2 contracts by selling (i.e.,
close sell) at a price of 6024, and its corresponding gross realized
gains, NT$7600, are determined by comparing its costs
((6024e6005) Â 200 Â 2). Meanwhile, the remaining 5 long con-
tracts are marked to market, and now their unrealized gains are
NT$16,297. His record now thus shows an open interest of 5 and an
average cost of 6005. The duration is calculated by comparing the
time to open and to close a contract. Note that, in reality, TAIFEX is
operated from 8:45 am to 1:45 pm, Monday through Friday
(excluding public holidays). Therefore, I calculate the duration
based on the ?ve trading hours a day. For example, from Table 1, at
8:58 on December 17 of 2004, trader A closes two contracts which
are longing at 9:00 on December 16 of 2004. Thus, the duration is
298 min (285 min on December 16 and 13 min on December 17) for
these two contracts. Because its corresponding net realized returns
are positive, the duration is referred to as duration of the pro?table
round trip (DG). The same procedure is used to identify the dura-
tion for a losing contract (DL).
The same calculations are repeated for the following trades.
Notice that, on December 22, after longing one contract at 6042, the
trader buys seven more contracts in his second trade at a price of
6064. At this time, his record is updated to showan average cost of
6061.25 ((6042 Â1 þ6064 Â 7) ÷ (1 þ7)) and an open interest of 8.
Note that for contracts that are held until maturity and closed by
the exchange, I calculate their net realized gains/losses based on the
?nal settlement price of the contract. Note that, Locke and Mann
(2005) assume that open interest is zero at the end of each
trading day and determine realized and unrealized gains/losses by
?rst accumulating a sequence of buy (sell) trades. In comparison,
my calculation of realized gains/losses provides a more accurate
measure of realized gains/losses.
Table 1 reports summary statistics with respect to individual
and institutional trades. Over the sample period, the average daily
trading accounts for individuals and institutional investors are
6373.52 and 12.89, respectively. The percentage of individual in-
vestors is approximately 98.14%, which is strikingly higher than
that of institutional investors (1.86%). However, in terms of trading
volume (i.e., numbers of contracts), the percentage of individual
investors is not so high. Individual investors contribute 59.56%
(93,182.73/(93,182.73 þ 63,263.3)) of the gross volume of trade. In
contrast, 40.44% of the gross volume of trade is by institutional
investors. This indicates that individual investors are the major
participants in the TAIEX futures markets. This ?nding differs from
prior research with mature futures markets where the institutional
investors are the major participants.
The average daily buy-sell imbalances for individual and insti-
tutional investors are À0.121 and 0.886, respectively. It appears that
individual investors were net sellers and institutional investors
were net buyers during the sample period. The trade size per order
of individual investors (8.579) is signi?cantly lower than that of
institutional investors (23.987). The ratio of limit order to total
order, including limit order and market order, of individual in-
vestors (0.856) is lower than that of institutional investors (0.973).
Meanwhile, individuals leave fewer open interests (3.419 < 26.640)
and hold the contract for a shorter time (327.0 < 860.7) than in-
stitutions do. By contrast, individuals have higher turnover ratios
(59.84 > 19.68). This indicates that individual investors are more
aggressive in terms of a higher proportion of the market order and a
shorter holding period for a round-trip trade.
3. The trading performance of individual and institutional
investors
An abundant literature has theoretically argued that individuals
are overcon?dent and exhibit the disposition effect. They trade
frequently because they overestimate the precision of their
knowledge and underestimate the riskiness of the expected return
(see Benos, 1998; Kyle & Wang, 1997; Odean, 1998b; Wang, 1998,
Table 1
Summary statistics for trades of individual and institutional investors.
Individuals Institutions Difference t-value
Average numbers of traders 6373.52 120.98 6252.55 96.303***
Numbers of contracts 93,182.73 63,263.30 29,919.43 11.867***
Trade imbalance À0.0121 0.0886 À0.1007 À10.781***
Trade-size per order 8.579 23.987 À15.407 À38.398***
Limit-order ratio 0.856 0.973 À0.117 À14.403***
Open interests 3.419 26.640 À23.221 À29.08***
Turnover ratio 59.84 19.68 40.16 130.45***
Duration 327.0 860.7 À533.7 À155.2***
This table reports the numbers of traders, numbers of contracts, trade imbalance, trade size per order and limit order ratio for individual and institutional investors over the
sample period from 1, January 2004 to 31, December 2008.
Open interest refers to the total number of contracts that have not been settled on the trading day.
Turnover ratio is represented in percent.
Duration is the time to hold a round-trip contract, which is represented in minutes.
*, **, and *** denote signi?cant at the 1%, 5%, and 10% signi?cance level, respectively.
“Difference” denotes difference between individuals and institutions.
M.-C. Lin, M.-T. Chiang / Asia Paci?c Management Review 20 (2015) 165e176 167
2001). In addition, they also under-react to more relevant infor-
mation; this leads to positive feedback trading (see De Long,
Shleifer, Summers, & Waldmann, 1990; Daniel et al., 1998;
Hirshleifer, Subrahmanyam, & Titman, 1994; Odean, 1998b). By
contrast, though institutional investors are subject to the same
cognitive biases as individual investors, they are on average better
trained and have better resources than individual investors. The
better information and analysis skills may allow institutions to
overcome these biases. As a consequence, most academics argue
that institutional investors are more likely to behave rationally and
less likely to trade on noise than individual investors. Institutional
investors are also characterized as smart money in the sense of
being informed (e.g., Campbell & Kyle, 1993).
To investigate whether individual investors and institutional
investors in the TAIEX futures markets are smart money or dumb
money investors, I evaluate their performance using two different
methods. First, I examine the market timing performance after
those days when investors conduct heavy buying or selling
(Kamesaka et al., 2003). Second, I depict the return per contract for
each investor group.
3.1. Market timing
To study the market performance after individual and institu-
tional heavy buying and selling days, I need to calculate the trade
imbalance (TI) measure. My de?nition of trade imbalance follows
Chan and Fong (2000), Chordia, Roll, and Subrahmanyam (2002)
and Kamesaka et al. (2003) calculate the trading imbalance of eq-
uities on the Tokyo Stock Exchange (TSE). They de?ne trading
imbalance as (buy Àsell)/(buy þsell). Different fromKamesaka et al.
(2003), I examine the trading imbalance of TAIEX futures contract.
Because the buy volume of futures contracts includes the numbers
of open buy and close buy, and the sell volume of futures contracts
includes the numbers of open sell and close sell, I revise the trading
imbalance equation of Kamesaka et al. (2003) and de?ne it as
follows:
TI
it
¼
OB
it
ÀOS
it
þCB
it
ÀCS
it
OB
it
þOS
it
þCB
it
þCS
it
; (1)
where OB
it
is the ‘open-buy’ volume for investor group i on day t,
OS
it
is the ‘open-sell’ volume for investor group i on day t, CB
it
is the
‘close-buy’ volume for investor group i on day t, and CS
it
is the
‘close-sell’ volume for investor group i on day t. Following Chang,
Hsieh, and Lai (2009) and Chang, Hsieh, and Wang (2010), ‘open-
buy’ is denoted as opening new long contracts, ‘open-sell’ is
denoted as opening new short contracts, ‘close-buy’ is denoted as
closing existing long contracts, and ‘close-sell’ is denoted as closing
existing short contracts. The trade type is ‘open-buy’ if the trader
?rst opens a new long position for which I can ?nd no corre-
sponding ‘open-sell’ before the time and day that the buy order is
initiated. Conversely, it's ‘close-buy’ if the trader opens a new sell
position prior to initiating the buy order. With similar logic, I can
determine both ‘open-sell’ and ‘close-sell’, as well as ‘open-interest’
by collecting those positions which exist on each trading day.
TI is positive (negative) when the investor group buys more
(less) than sells contracts during the day. A large trade imbalance in
either direction is an indication of market timing (Kamesaka et al.,
2003). Informed traders with positive information are more likely
to be net buyers. Conversely, informed traders are net sellers when
they possess negative information. The trade imbalance (TI) for
each investor type is sorted onto ?ve equal sets. The quintile with
the highest positive trade imbalance is designated as the buying
day for that investor type. The quintile with the largest negative
trade imbalance is the selling day. To examine the market timing
ability of individual and institutional investors, I compute the one-
day, two-day, and three-day TAIEX futures return following the
trading day.
Table 2 reports the results. Individual investors have a trade
imbalance of 0.325 on buying days and À0.0733 on selling days.
One day after individual buying, the market decreases À0.0977%,
on average. One day after individual selling, the market increases
0.0186%. The post one-day return difference between buy days and
sell days is not signi?cantly different for individual investors. The
two-day period following the buy and sell days also experiences a
market return of À0.1572% and 0.1066%, respectively. The
difference, À0.2638%, is also not signi?cant. The 3-day return
following individual investor trading is À0.3679% after buying days
and 0.2660% after selling days. The difference, À0.6339%, becomes
signi?cant at the 5% level. Overall, individual investors appear to
exhibit poor market timing ability in the TAIEX futures markets.
The trade imbalance for the institutional buy and sell days are
0.1685 and À0.1550, respectively. The difference in trade imbalance
is 0.3235, which is relatively lower than that of individual investors.
On day one, the market experiences a 0.8187% return following
institutional buying and a À0.8582% return following institutional
selling. The difference, 1.6769%, is signi?cant. A market return of
1.6369% occurs two days after institutional buying. This is signi?-
cantly larger than the two-day return of À1.7010% following insti-
tutional selling. The three-day market return following
institutional buying is also signi?cantly larger than the return
following institutional selling. This indicates that institutional in-
vestors appear to have some success in market timing.
3.2. Trading performance
The average net realized gains of the winning contract (RG
i,t
) and
the average net realized losses of the losing contract (RL
i,t
) for
investor type i on date t are denoted as follows:
RG
i;t
¼
PN
i
RG;t
j¼1
RG
i
j;t
N
i
RG;t
; RL
i;t
¼
PN
i
RL;t
j¼1
RL
i
j;t
N
i
RL;t
; (2)
N
i
RG;t
¼ number of trades by investor type i where a gain is
realized on day t
N
i
RL;t
¼ number of trades by investor type i where a loss is
realized on day t
RG
i
j;t
¼ the net realized gains of the j winning contract for
investor type i on day t
RL
i
j;t
¼ the net realized losses of the j losing contract for investor
type i on day t
Because some contracts may not be closed by the end of a day, I
calculate total performance which includes paper gains or paper
losses. The average total net gains of the winning contract (TG
i,t
)
and the average total net losses of the losing contract (TL
i,t
) for
investor type i on date t are denoted as follows:
TG
i;t
¼
PN
i
RG;t
j¼1
RG
i
j;t
þ
PN
i
PG;t
j¼1
PG
i
j;t
N
i
RG;t
þN
i
PG;t
; TL
i;t
¼
PN
i
RL;t
j¼1
RL
i
j;t
þ
PN
i
PL;t
j¼1
PL
i
j;t
N
i
RL;t
þN
i
PL;t
;
(3)
N
i
PG;t
¼ number of trades by investor type i where a gain is
unrealized on day t
M.-C. Lin, M.-T. Chiang / Asia Paci?c Management Review 20 (2015) 165e176 168
N
i
PL;t
¼ number of trades by investor type i where a loss is un-
realized on day t
PG
i
j;t
¼ the net unrealized gains of the j winning contract for
investor type i on day t
PL
i
j;t
¼ the net unrealized losses of the j losing contract for
investor type i on day t
Other variables are de?ned as those in Eq. (2).
Table 3 reports the average total net gains (losses), net realized
pro?ts, and net unrealized performance per contract. As shown,
individuals have negative total returns of NT$-1307.8 per contract
and institutional investors earn positive total return of NT$1832.6
per contract. This better performance for institutions holds even
when it is measured in terms of either realized or unrealized
returns. This is consistent with Barber, Lee, Liu, and Odean (2005;
2009) and Barber, Odean and Zhu (2009) whose study sample fo-
cuses on the Taiwan equity market. A closer look reveals that in-
stitutions have greater magnitude of gains and losses than
individuals. For example, conditional on winning (losing) contracts,
the average total gains (losses) per contract for institutions, NT$
25,001.8 (NT$ À22,666.1), are larger than those of individuals,
NT$9278.5 (NT$ À8666.4). Taking gains and losses together, it is
apparent that the total net returns are higher for institutional in-
vestors than for individuals.
Nofsinger and Sias (1999) and Kamesaka et al. (2003) posit that
trading with high returns indicates that the trading is motivated by
information; in contrast, trading with low returns reveals that it is
motivated by behavioral-based biases. In accordance with this
insight, Barber and Odean (2001) and Odean (1999) show that
overcon?dence is associated with poor investment performance;
Locke and Mann (2005) ?nd that the least successful traders of
professional futures hold losers the longest, while the most suc-
cessful traders hold losers for the shortest time; Wermers (2003)
reports that mutual funds with poor performance have a greater
tendency to hold losing stocks. Above evidence indicates that, if
individual investors, compared with institutional investors, are
more inclined toward behavioral biases, on average, their perfor-
mance will be smaller. Therefore, following Nofsinger and Sias
(1999) and Kamesaka et al. (2003), I ?rst use the past trading
returns to get a preliminary result. I then take a further step to
examine the relationship between open (close) volume and
behavioral biases, like overcon?dence and disposition effect, to
identify whether trading decisions are in?uenced more by
information-based or behavioral-based motivation.
The preliminary results in Tables 2 and 3 show that there is an
signi?cant difference in performance and trading behavior among
these two investor-types, as has previously been suggested (Odean,
1999). Institutional investors are more likely to be informed traders
Table 2
Market performance after buying and selling by investor type.
Trade imbalance 1-day Returns (%) 2-day Returns (%) 3-day Returns (%)
Panel A: Individual investors
Buy 0.3025 À0.0977 À0.1572 À0.3679
t-value 39.2*** À0.20 À1.84* À2.27**
Sell À0.0733 0.0186 0.1066 0.2660
t-value À1.15 0.11 À1.19 2.86**
Difference 0.3759 À0.1162 À0.2638 À0.6339
t-value 60.20*** À0.83 À1.43 À2.85***
Panel B: Institutional investors
Buy 0.1685 0.8187 1.6369 2.4492
t-value 19.3*** 4.9*** 7.6*** 12.9***
Sell À0.1550 À0.8582 À1.7010 À2.5579
t-value À16.75*** À5.13*** À8.98*** À14.25***
Difference 0.3235 1.6769 3.3379 5.0071
t-value 47.71*** 11.63*** 16.61*** 2.33***
Each investor-type's trade imbalance is sorted into ?ve equal sets. The quintile of the days within the highest positive trade imbalance is designated as the buying days for that
investor type. The quintile of the largest negative trade imbalance is the selling days. The buy and sell days represent the largest trade imbalance by the type of investors. Each
quintile of buy and sell days has 241 observations.
I compute the 1-day, 2-day, and 3-day returns following the buy or sell trading days.
“Difference” denotes difference in trade imbalance or returns between buy and sell.
The return difference between buy days and sell days is tested using a difference in means t-statistics.
*, **, and *** denote signi?cant at the 1%, 5%, and 10% signi?cance level, respectively.
Table 3
Performance for individual and institutional traders.
Individuals t-value Institutions t-value Difference t-value
Panel A: Total contracts
Total À1307.8 À358.9*** 1832.6 182.6*** À3140.4 À275.3***
Realized À1122.5 À260.7*** 1310.0 87.6*** À2432.5 À213.5***
Unrealized À998.9 À400.2*** 1113.4 143.0*** À2112.8 À158.6***
Panel B: Winning contracts
Total 9278.5 1980.4*** 25,001.8 1517.6*** À15,723.3 À1997.5***
Realized 7778.7 1986.1*** 20,064.4 1239.4*** À12,285.7 À1066.7***
Unrealized 6252.7 1750.8*** 15,686.1 1767.8*** À9433.4 À973.2***
Panel C: Losing contracts
Total À8666.4 À1822.7*** À22,666.1 À1480.5*** 18,999.7 758.9***
Realized À11,670.4 À1277.2*** À17,803.8 À1254.0*** 6133.4 439.7***
Unrealized À5191.1 À1736.0*** À14,077.3 À1666.5*** 8996.2 169.5***
This table reports the returns per contract for individual and institutional investors over the sample period from 1, January 2004 to 31, December 2008.
“Difference” denotes the difference between individuals and institutions, which is tested using a difference in means t-statistic.
*, **, and *** denote signi?cant at the 1%, 5%, and 10% signi?cance level, respectively.
M.-C. Lin, M.-T. Chiang / Asia Paci?c Management Review 20 (2015) 165e176 169
and individual investors are prone to be noisy traders in the TAIEX
futures markets.
4. Trading due to information or behavioral bias?
Because the only assets examined are the TAIEX futures con-
tracts, unlike equity stocks, they do not permit an examination of a
wide range of behavioral biases like familiarity, limited attention,
and representativeness. Thus, the study focuses on only two biases:
overcon?dence and disposition effect.
4.1. Overcon?dence
Overcon?dence is considered to be the most robust ?nding in
the psychology of judgment (De Bondt & Thaler, 1995). Over-
con?dence is induced by two cognitive biases: biased self-
attribution and con?rmatory bias (Daniel et al., 1998; Gervais &
Odean, 2001; Hirshleifer, 2001; Odean, 1998b). Biased self-
attribution is that people tend to attribute successes to their own
abilities and failure to bad luck or external factors. Con?rmatory
bias is that people are prone to interpret evidence as consistent
with their prior beliefs. Einhorn (1980) claims that individuals have
a higher tendency to be overcon?dent in a setting where more
judgment is required to evaluate information, and where the
feedback on the judgment is ambiguous in the short run. Obviously,
the futures markets are the circumstances where investment de-
cisions require professional knowledge and the feedback is slow
and noisy. Therefore, an abundance of overcon?dent investors exist
in futures markets.
To gain insights into whether the poor performance of indi-
vidual trades is motivated by psychological biases, I use turnover
ratio as a proxy for overcon?dence (Barber & Odean, 2001; Odean,
1999). Turnover ratio measures the total number of contracts
traded in a period relative to the number of open positions at the
end of the period. The turnover ratio for investors within investor
type i on date t is de?ned as:
TURN
i;t
¼
P
N
i
t
j¼1
TV
i
j;t
P
N
i
t
j¼1
OP
i
j;t
; (4)
where N
i
t
is number of investors for investor type i on day t, TV
i
j;t
is
the trading volume of the j trader within investor type i on day t,
and OP
i
j;t
is the number of open interests of the j trader within
investor type i on day t.
From Table 4, it is evident that individuals have higher turnover
ratio than institutional investors do. On average, the turnover ratio
is 59.84% for individuals and 19.68% for institutions. No matter
whether they hold a winning or losing contract, individuals have a
higher turnover in the positions than institutions do. As indicated
by Barber and Odean (2001), the overcon?dence-induced trading
by individual investors is associated with poor investment
performance. Combined with their increased aggressiveness in
placing orders (see Table 1), the trading decisions of individual
investors appear to be motivated more by cognitive biases, such as
overcon?dence, than by information.
4.2. Disposition effect
In addition to overcon?dence, disposition effect is one of the
most well-known behavioral biases of investors. Behavioral re-
searchers attribute this phenomenon to loss aversion (e.g.,
Kahneman & Tversky, 1979; Odean, 1998a; Kyle, Hui, & Xiong,
2006; Hens & Vlcek, 2011; Barberis & Xiong, 2009). Loss aversion
is proposed by Kahneman and Tversky (1979) and as part of pros-
pect theory. According to the prospect theory, the decision-making
under risk is associated with gains and losses, not ?nal wealth
levels; investors are more sensitive to losses than to gains, and are
risk averse for gains and risk seeking for losses. This risk-averse
behavior for gains and risk-seeking behavior for losses then lead
to the disposition effect. In addition to loss aversion, Shefrin and
Statman (1985) suggest regret and pride, which has recently been
supported with experimental evidence (O'Curry Fogel & Berry,
2006), as another explanation for the disposition effect. Wanting
to feel pride by realizing gains and avoiding regret by delaying
realizing losses is what causes investors to realize gains more
quickly than losses.
I use two methodologies to measure the disposition effect. First
of all, following Odean's (1998a) methodology, I measure the
disposition effect by calculating and comparing the difference be-
tween investors' propensity to realize gains (PGR) and their pro-
pensity to realize losses (PLR). The proportion of gain realized
(PGR
i,t
) and the proportion of loss realized (PLR
i,t
) for investor type i
on date t are de?ned as:
PGR
i;t
¼
N
i
RG;t
N
i
RG;t
þN
i
PG;t
; PLG
i;t
¼
N
i
RL;t
N
i
RL;t
þN
i
PL;t
; (5)
with N
i
RG;t
and N
i
RL;t
having the same de?nition as found in equation
(2), and N
i
PG;t
and N
i
PL;t
being de?ned as follows:
N
i
PG;t
¼ number of trades by investor type i where there is a
paper gain on day t
N
i
PL;t
¼ number of trades by investor type i where there is a
paper loss on day t
The Odean's disposition effect (DE1) for investor type i on day t
is computed as:
DE1
i;t
¼ PGR
i;t
ÀPLR
i;t
: (6)
A positive DE1
i,t
indicates that investor type i tends to realize
gains more than losses on day t. Table 5 reports the disposition
effect. The results show that individuals have higher PGR than PLR,
Table 4
Turnover ratio for individual and institutional traders.
Individuals t-value Institutions t-value Difference t-value
Total 59.84 255.04*** 19.68 98.76*** 40.16 130.45***
Winners 78.09 197.16*** 26.75 74.80*** 51.34 96.21***
Losers 33.72 116.85*** 15.62 74.93*** 18.11 50.86***
Winners-Losers 44.37 108.10*** 11.13 76.02*** 33.23 40.39***
This table reports the turnover ratio for individual and institutional investors over the sample period from 1, January 2004 to 31, December 2008.
Turnover ratio measures the total number of contracts traded in a period relative to the size of open positions at the end of the period.
The turnover ratio is represented in percent.
“Difference” denotes the difference between individuals and institutions, which is tested using a difference in means t-statistic.
*, **, and *** denote signi?cant at the 1%, 5%, and 10% signi?cance level, respectively.
M.-C. Lin, M.-T. Chiang / Asia Paci?c Management Review 20 (2015) 165e176 170
but institutions have lower PGR than PLR. In other words, indi-
vidual investors exhibit the disposition effect, but institutional in-
vestors show evidence for the reverse-disposition effect. A closer
look reveals that individuals have a higher tendency to realize both
winning and losing contracts than institutions do. This is in line
with the higher turnover ratio for individual investors in Table 4.
The other method to measure the disposition effect is proposed
by Shapira and Venezia (2001), who compare the duration of losing
round trips to those of winning round trips. I de?ne the average
duration of the pro?table round trip (DG
i,t
) and the average dura-
tion of the losing round trip (DL
i,t
) for investor type i on date t as:
DG
i;t
¼
PN
i
RG;t
j¼1
DG
i
j;t
N
i
RG;t
; DL
i;t
¼
PN
i
RL;t
j¼1
DL
i
j;t
N
i
RL;t
; (7)
with N
i
RG;t
and N
i
RL;t
having the same de?nition as found in equation
(2), and DG
i
j;t
and DL
i
j;t
being de?ned as follows:
DL
i
j;t
¼the duration of the j winning round trip for investor type i
on day t
DL
i
j;t
¼ the duration of the j losing round trip for investor type i
on day t
The Shapira and Venezia's disposition effect (DE2) for investor
type i on day t is then computed as:
DE2
i;t
¼ DG
i;t
ÀDL
i;t
: (8)
A negative DE2
i,t
indicates that investor type i holds a winning
contract shorter than a losing contract. That is, they have a higher
tendency to realize gains than losses on day t. Table 6 reports the
duration of a round-trip trade, which is represented in minutes. It is
evident that individuals hold the futures contract for a shorter time
frame than institutional investors do. On average, the duration of a
round trip is 5.45 (327/60) hours for individuals, and 14.345 (860.7/
60) hours for institutions. Regardless of whether they hold a win-
ning or losing contract, institutions hold contracts longer than
individuals do. That is, individual investors have a higher pro-
pensity to realize both gains and losses than institutional investors
do. In addition, the holding time per winning contract is shorter
(longer) than the holding time per losing contract for individual
(institutional) investors. This is consistent with the results based on
comparing the proportion of gain realized (PGR
i,t
) and the propor-
tion of loss realized (PLR
i,t
). In other words, individuals, instead of
institutions, suffer from disposition biases.
4.3. Quantile regression
To performa robust check, we run a regression test of predicting
open or close trades. To know well about the information about the
tail behaviors of open or close trades' distribution, I adopt a quantile
regression approach proposed by Koenker and Bassett (1978). The
quantile regression permits the estimation of various quantile
functions of a conditional distribution, with the median (0.5th
quantile) function being a special case.
Given the data (y
t
, x
t
) for t ¼ 1, …, T, where x
t
is k  1, the linear
speci?cation for the conditional quantiles of y can be considered as
follows:
yt ¼ x
t
b þe
t
: (9)
The qth quantile regression estimator of b is obtained by mini-
mizing the average of asymmetrically weighted absolute errors
with weight q on positive errors and weight (q À 1) on negative
errors:
V
T
V
T
ðb; qÞ ¼
1
T
2
4
q
X
t:yt x
0
t
b
y
t
Àx
0
t
b
þð1 ÀqÞ
X
t:yt x
0
t
b
y
t
Àx
0
t
b
3
5
:
(10)
Each quantile regression describes a particular (center or tail)
point of the conditional distribution. This approach is particularly
useful when the conditional distribution of ?nancial variables is
heterogenous and does not have a “standard” shape, such as an
asymmetric, fat-tailed or truncated.
4.4. Regression model
To assess the relation between individual (institutional) trading
behavior and behavioral bias, I test for the ratios of open trading
volume to total volume (OP) and close trading volume to total
volume (CL) as follows:
OP
it
¼ a
11
þa
12
OC
itÀ1
þ
X
K
k¼1
b
1k
R
tÀk
þ
X
K
k¼1
l
1k
logðV
tÀk
Þ
þ
X
K
k¼1
g
1k
s
2
tÀk
þ 3
OPt
; (11)
Table 5
Disposition effect individual and institutional traders.
Total Individuals Institution Difference
PGR 0.4848 0.6918 0.3320 0.3598
t-value 56.43*** 76.58*** 30.97*** 75.73***
PLR 0.4434 0.5760 0.3652 0.2108
t-value 54.65*** 85.09*** 18.71*** 82.81***
DE 0.0414 0.1158 À0.0332 0.1490
t-value 15.04*** 13.85*** À7.08*** 14.29***
This table reports the mean of PGR, PLR, and DE for individual and institutional
investors over the sample period from 1, January 2004 to 31, December 2008.
PGR is the number of realized gains divided by the number of realized gains plus the
number of paper gains, and PLR is the number of realized losses divided by the
number of realized losses plus the number of paper losses.
DE is the difference of PGR and PLR.
*, **, and *** denote signi?cant at the 1%, 5%, and 10% signi?cance level, respectively.
Table 6
Duration for individual and institutional traders.
Individuals t-value Institutions t-value Difference t-value
Total 327.0 1556.6*** 860.7 1413.5*** À533.7 À155.2***
Winners 294.5 111.7*** 871.0 131.2*** À576.4 À861.2***
Losers 365.7 193.6*** 850.3 968.9*** À484.6 À632.4***
Winners-Losers À71.2 À168.9*** 20.7 16.99*** À91.9 À107.3***
This table reports the duration measure for individual and institutional investors over the sample period from 1, January 2004 to 31, December 2008.
Duration is the time to hold a round-trip contract, which is represented in minutes.
“Difference” denotes the difference between individuals and institutions, which is tested using a difference in means t-statistic.
*, **, and *** denote signi?cant at the 1%, 5%, and 10% signi?cance level, respectively.
M.-C. Lin, M.-T. Chiang / Asia Paci?c Management Review 20 (2015) 165e176 171
Table 7
Regression test: Open trading.
OLS 0.05 0.1 0.2 0.3 0.4 0.5(LAD) 0.6 0.7 0.8 0.9 0.95
Panel A: Individual investors
Intercept 6.228 0.716 0.757 0.797 0.823 0.860 0.881 0.916 0.969 1.042 1.606 16.505
(4.78)*** (7.60)*** (7.69)*** (7.66)*** (5.10)*** (8.79)*** (6.86)*** (7.03)*** (7.69)*** (5.04)*** (3.46)*** (2.77)**
OC
t À 1
12.629 0.036 0.091 0.183 0.182 0.238 0.220 0.252 0.283 0.259 2.341 54.891
(5.15)*** (0.44) (1.57) (3.59)*** (3.67)*** (4.70)*** (4.10)*** (3.88)*** (3.70)*** (2.39)** (2.25)** (3.04)***
r
t À 1
0.631 0.143 0.334 0.258 0.278 0.169 0.112 0.081 À0.304 À0.690 À5.378 À6.858
(1.04) (0.68) (1.91)* (1.72) (1.64) (0.91) (0.51) (0.35) (À1.36) (À1.99)** (À1.15) (À0.66)
r
t À 2
À0.805 0.355 0.293 0.136 0.147 0.059 0.151 0.015 À0.248 À0.245 À0.395 3.947
(À0.84) (0.93) (1.60) (0.86) (0.94) (0.33) (0.79) (0.07) (À1.21) (À0.64) (À0.15) (0.41)
V
t À 1
À0.177 À0.001 À0.001 À0.001 À0.001 À0.001 À0.001 À0.001 À0.002 À0.003 À0.013 À0.171
(À6.85)*** (À3.78)*** (À2.98)*** (À3.77)*** (À4.46)*** (À5.74)*** (À5.90)*** (À5.62)*** (À6.28)*** (À7.36)*** (À1.54) (À1.83)*
V
t À 2
À0.117 0.000 À0.001 À0.001 À0.001 À0.001 À0.002 À0.002 À0.002 À0.003 À0.018 À0.326
(À4.11)*** (À0.37) (À3.78)*** (À5.19)*** (À4.72)*** (À4.94)*** (À6.31)*** (À5.04)*** (À4.71)*** (À4.87)*** (À1.02) (À1.77)
s
2
tÀ1
1.603 0.004 0.008 0.007 0.012 0.013 0.015 0.018 0.029 0.061 0.426 7.408
(5.37)*** (0.66) (1.67) (1.93)* (3.76)*** (4.24)*** (5.59)*** (3.51)*** (2.47)** (3.39)*** (0.85) (1.18)
s
2
tÀ2
0.743 À0.001 À0.002 0.000 0.000 0.002 0.003 0.008 0.007 0.017 0.232 1.723
(1.75) (À0.21) (À0.50) (0.02) (À0.08) (0.99) (1.20) (2.85)*** (1.70) (1.54) (0.67) (0.47)
Adj. R
2
0.134 0.105 0.105 0.106 0.105 0.105 0.105 0.105 0.105 0.106 0.111 0.196
Panel B: Institutional investors
Intercept 1.128 0.162 0.178 0.211 0.240 0.273 0.305 0.336 0.396 0.511 0.856 1.165
(7.63)*** (5.64)*** (2.75)** (8.04)*** (7.51)*** (8.92)*** (3.40)*** (8.68)*** (6.21)*** (5.70)*** (9.22)*** (5.13)***
OC
t À 1
16.928 À5.138 À4.829 4.426 5.593 6.636 8.361 8.180 11.729 17.614 37.036 52.263
(1.67) (À2.71)*** (À1.98)** (1.55) (1.29) (1.65) (1.57) (1.89)* (5.09)*** (4.42)*** (4.36)*** (4.10)***
r
t À 1
0.769 0.266 0.067 0.276 0.394 0.449 0.417 0.402 0.567 0.929 1.805 4.433
(1.05) (1.31) (0.27) (1.26) (2.01)** (2.78)** (2.89)*** (2.78)** (2.97)*** (2.48)** (1.43) (1.60)
r
t À 2
0.680 0.031 À0.037 À0.217 À0.391 À0.318 À0.283 À0.288 À0.575 À0.408 0.482 0.976
(0.20) (0.14) (À0.16) (À1.06) (À2.03)** (À1.89)* (À1.73) (À1.59) (À2.20)** (À0.99) (0.34) (0.18)
V
t À 1
À0.019 0.000 0.000 0.000 0.000 0.000 0.000 0.000 À0.001 À0.002 À0.006 À0.010
(À2.03)** (0.73) (0.88) (0.69) (À0.73) (À0.55) (À0.66) (À2.57)** (À3.87)*** (À4.47)*** (À4.46)*** (À3.95)***
V
t À 2
À0.001 0.001 0.001 0.001 0.001 0.000 0.000 0.000 0.000 À0.001 À0.003 À0.007
(À0.26) (2.95)*** (3.04)*** (2.92)*** (2.52)** (0.86) (À0.39) (À0.22) (À0.84) (À1.16) (À0.74) (À1.43)
s
2
tÀ1
0.116 À0.017 À0.009 À0.009 À0.006 0.000 0.001 0.003 0.008 0.029 0.083 0.213
(1.73) (À2.71)** (À3.02)*** (À2.80)** (À1.16) (0.12) (0.62) (1.41) (1.58) (1.41) (1.24) (0.93)
s
2
tÀ2
0.059 À0.003 À0.006 À0.001 0.000 0.000 0.001 0.000 0.000 0.011 0.048 0.190
(1.56) (À1.59) (À2.41)** (À0.44) (À0.10) (À0.20) (0.52) (À0.22) (À0.15) (1.02) (2.41)** (1.76)
Adj. R
2
0.138 0.134 0.124 0.112 0.107 0.105 0.105 0.105 0.107 0.112 0.128 0.151
This table reports the relationship between individual and institutional trading and the overcon?dence level. The individual and institutional open trading volume is normalized by total trading volume. Above variables are
regressed on the proxy of overcon?dence level (OC
t
), with lagged index futures return (r
t À 1
and r
t À 2
), lagged index futures trading volume (V
t À 1
and V
t À 2
) and lagged daily 5-min volatility (s
2
tÀ1
and s
2
tÀ2
) being included as
control variables. T-values are estimated using heteroscedasticity-consistent standard errors.
Values in parenthesis are T-values.
*** denotes signi?cant at the 1% level.
** denotes signi?cant at the 5% level.
* denotes signi?cant at the 10% level, respectively.
M
.
-
C
.
L
i
n
,
M
.
-
T
.
C
h
i
a
n
g
/
A
s
i
a
P
a
c
i
?
c
M
a
n
a
g
e
m
e
n
t
R
e
v
i
e
w
2
0
(
2
0
1
5
)
1
6
5
e
1
7
6
1
7
2
Table 8
Regression test: Close trading.
OLS 0.05 0.1 0.2 0.3 0.4 0.5(LAD) 0.6 0.7 0.8 0.9 0.95
Panel A: Individual investors
Intercept 5.917 0.655 0.697 0.748 0.774 0.813 0.840 0.868 0.938 1.017 1.466 17.780
(4.63)*** (3.26)*** (4.28)*** (7.86)*** (6.12)*** (6.02)*** (8.24)*** (7.54)*** (4.55)*** (4.67)*** (2.62)** (3.58)***
OC
t
1.44 1.06 1.05 1.14 0.17 0.18 0.15 0.14 0.16 0.02 1.99 4.95
(3.39)*** (2.81)*** (2.71)*** (3.03)*** (1.73) (1.18) (1.44) (1.24) (1.64) (1.980)** (2.78)*** (2.19)**
DE
t
À1.805 À0.315 À0.248 À0.181 À0.102 À0.148 À0.138 À0.130 0.315 0.347 0.681 1.500
(À1.93)* (À2.36)** (À2.93)*** (À2.24)** (À1.20) (À1.86)* (À1.60) (À1.29) (2.35)** (2.52)** (2.47)** (2.75)**
r
t À 1
0.576 0.014 0.138 0.014 0.127 0.380 0.419 0.445 1.078 1.675 0.854 0.321
(2.33)** (0.05) (0.75) (0.12) (0.71) (1.76) (2.23)** (2.16)** (2.81)** (4.83)*** (0.95) (0.72)
r
t À 2
À0.636 À0.182 À0.166 0.154 À0.023 À0.244 À0.230 À0.385 À0.520 À0.794 À0.217 0.253
(À0.98) (À0.54) (À0.95) (1.07) (À0.15) (À1.17) (À1.08) (À1.75) (À2.08)** (À2.30)** (À0.08) (0.44)
V
t À 1
À0.172 À0.001 À0.001 À0.001 À0.001 À0.001 À0.002 À0.002 À0.002 À0.003 À0.010 À0.156
(À6.84)*** (À1.45) (À2.10)** (À4.83)*** (À4.82)*** (À3.98)*** (À5.25)*** (À5.93)*** (À5.49)*** (À7.17)*** (À1.23) (À3.32)***
V
t À 2
À0.113 0.000 0.000 0.000 0.000 À0.001 À0.001 À0.001 À0.002 À0.003 À0.016 À0.345
(À4.07)*** (0.07) (À0.14) (À1.44) (À1.77) (À2.26)** (À3.03)*** (À3.29)*** (À3.06)*** (À4.59)*** (À0.74) (À1.86)*
s
2
tÀ1
1.554 0.003 0.002 0.006 0.006 0.011 0.014 0.016 0.034 0.057 0.381 7.634
(5.34)*** (0.65) (0.77) (3.12)*** (2.41)** (2.63)** (4.30)*** (3.43)*** (2.08)** (3.39)*** (0.67) (1.17)
s
2
tÀ2
0.754 0.000 0.000 0.000 0.001 0.002 0.006 0.006 0.010 0.018 0.167 1.278
(1.87)* (À0.07) (0.00) (À0.18) (0.55) (0.85) (1.36) (1.75) (1.73) (1.95)* (0.48) (0.30)
Adj. R
2
0.138 0.102 0.102 0.103 0.103 0.103 0.104 0.104 0.105 0.106 0.111 0.198
Panel B: Institutional investors
Intercept 1.086 0.123 0.147 0.178 0.203 0.232 0.265 0.298 0.340 0.426 0.708 1.188
(7.26)*** (4.26)*** (7.34)*** (5.07)*** (3.61)*** (3.51)*** (5.52)*** (5.99)*** (5.10)*** (9.87)*** (2.27)** (9.42)***
OC
t
1.941 1.409 1.002 0.324 0.060 0.439 1.188 2.819 3.242 2.472 6.449 14.312
(1.17) (0.63) (1.17) (0.21) (0.04) (0.26) (1.25) (1.43) (1.75) (1.86)* (4.51)*** (2.97)***
DE
t
1.800 0.141 0.447 0.218 0.138 0.215 1.161 1.446 1.946 3.582 11.167 19.386
(0.62) (0.24) (0.65) (0.32) (0.24) (0.31) (1.71) (1.68) (188)* (3.53)*** (9.76)*** (3.63)***
r
t À 1
À6.702 À0.001 À0.065 À0.379 À0.269 À0.325 À0.349 À0.433 À0.616 À0.961 À2.291 À2.902
(À2.01)** (À0.01) (À0.34) (À2.33)** (À1.55) (À1.95)* (À2.26)** (À2.52)** (À3.60)*** (À3.57)*** (À2.43)** (À0.39)
r
t À 2
0.178 À0.230 À0.290 À0.333 À0.317 À0.326 À0.122 À0.148 À0.169 À0.484 À0.073 1.138
(0.05) (À1.29) (À1.48) (À1.87) (À1.93)* (À2.00)** (À0.70) (À0.78) (À0.80) (À1.46) (À0.06) (0.37)
V
t À 1
À0.019 0.000 0.000 0.000 0.000 0.000 0.000 0.000 À0.001 À0.002 À0.005 À0.007
(À1.95)* (2.07)** (1.84)* (1.40) (1.72) (1.76) (1.59) (10.67) (À3.39)*** (À3.97)*** (À4.07)*** (À1.94)*
V
t À 2
À0.001 0.001 0.001 0.001 0.001 0.000 0.000 0.000 0.000 0.000 À0.001 À0.011
(À0.25) (5.17)*** (3.86)*** (3.86)*** (1.67) (0.71) (À0.43) (À0.09) (0.20) (À0.34) (À0.61) (À1.33)
s
2
tÀ1
0.119 À0.007 À0.008 À0.008 À0.004 À0.001 0.002 0.006 0.011 0.018 0.068 0.229
(2.76)** (À3.23)*** (À3.17)*** (À2.36)** (À1.06) (À0.28) (0.49) (1.61) (4.33)*** (2.18)** (1.42) (0.99)
s
2
tÀ2
0.057 À0.005 À0.004 À0.001 À0.001 À0.001 À0.002 À0.003 À0.005 0.009 0.026 0.175
(1.50) (2.56)** (À1.29) (À0.40) (À0.33) (À0.66) (À0.83) (À1.51) (À2.26)** (0.95) (1.14) (0.57)
Adj. R
2
0.137 0.136 0.123 0.115 0.108 0.105 0.103 0.103 0.104 0.108 0.126 0.150
This table reports the relationship between individual and institutional trading and the overcon?dence level. The individual and institutional close trading volume is normalized by total trading volume. Above variables are
regressed on the proxy of overcon?dence level (OC
t
) and the proxy of disposition effect (DE
t
), with lagged index futures return (r
t À 1
and r
t À 2
), lagged index futures trading volume (V
t À 1
and V
t À 2
) and lagged daily 5-
min volatility (s
2
tÀ1
and s
2
tÀ2
) being included as control variables. T-values are estimated using heteroscedasticity-consistent standard errors.
Values in parenthesis are T-values.
*** denotes signi?cant at the 1% level.
** denotes signi?cant at the 5% level.
* denotes signi?cant at the 10% level, respectively.
M
.
-
C
.
L
i
n
,
M
.
-
T
.
C
h
i
a
n
g
/
A
s
i
a
P
a
c
i
?
c
M
a
n
a
g
e
m
e
n
t
R
e
v
i
e
w
2
0
(
2
0
1
5
)
1
6
5
e
1
7
6
1
7
3
CL
it
¼ a
21
þa
22
OC
it
þa
23
DE1
it
þ
X
K
k¼1
b
2k
R
tÀk
þ
X
K
k¼1
l
2k
logðV
tÀk
Þ þ
X
K
k¼1
g
2k
s
2
tÀk
þ?
CLt
: (12)
For the open trading regression, the proxy of overcon?dence
(OC), which is the daily turnover, is one of the independent vari-
ables. This variable is included since one of the most robust facts
about the trading of investors is overcon?dence, and an over-
con?dent investor is thought to trade excessively. Additionally,
investors have an overall tendency to exhibit the disposition effect,
which is a propensity to sell winners and hold on to losers. Chou
and Wang (2011) argue that, if overcon?dent investors have pre-
viously held a long (short) position, then trading gains from that
position would induce them to buy (sell) more in the subsequent
period, and to do so more aggressively. They also argue that, if
disposition-biased investors have previously held a pro?table long
(short) position, then in order to quickly realize their gains, they
will hastily sell off their long (short) position in the subsequent
period. Because the disposition bias affects an investor's decision
to close a position, not to open a position, we consider it only
when investors try to close a position. But, overcon?dence bias
may affect the decisions to both open and close a new contract.
Therefore, I use the lagged daily turnover (OC
i,t À 1
) to predict the
contemporaneous opening trading volume (OP
i,t
) in equation (11)
to examine the effect of overcon?dence on open decisions. I also
use the concurrent daily turnover (OC
i,t
) and Odean's (1998a)
disposition measure (
[fx4]
) to predict the closing trading volume
(CL
i,t
) in equation (12) to examine whether the overcon?dence and
disposition effect will affect close decisions. Because trading
behavior can be in?uenced by some factors beyond behavioral
biases, I add certain control variables into the regression. They
include lagged TAIEX futures return (R
t À k
), lagged log trading
volume (log(V
t À k
)) and lagged daily volatility as a proxy risk
factor (s
2
tÀk
). The Schwartz Bayesian Criterion (SBC) is used to
determine lagged terms.
As shown in Panel A of Table 7, after controlling for other effects,
a high overcon?dence level among individual investors is associ-
ated with a high open trading volume. Because the OLS results
represent the “averaging” behavior, they yield little information
about the tail behaviors of the given distribution. Therefore, I take a
further look at the results of the quantile regression. It's found that
the overcon?dence tendency is prevalent among individual in-
vestors, with the exception in the lower “tail” of the open trading
distribution. This indicates that overcon?dent characteristics
inspire individual investors to increase their transactions on the
basis of the erroneous belief that they have valuable fundamental
information. By contrast, the OLS results in Panel B indicate that
this does not occur among institutional investors. Interestingly, the
negative coef?cient in the quintiles of q 0.1 reveals that over-
con?dence is correlated with the extremely low open volume of
institutions. But, the relationship within the extreme right tails is
statistically signi?cant in the quantiles of q S 0.6 at the 10% sig-
ni?cance level and q S0.7 at the 1% signi?cance level. This reveals
that overcon?dence also plays a role for institutional investors to
open large positions.
To know whether the decision to close a position is also
motivated by overcon?dence and disposition effect, I have re-
ported the regression of close trading volume in Table 8. The re-
sults are similar to those that were obtained with respect to open
trading. Individual investors tend to close a position due to
overcon?dence and disposition effect, but institutional investors
do not. In particular, a higher overcon?dence level is associated
with both high and low close trading volume among individual
investors. As demonstrated by the quantile regression in Panel A,
the signi?cantly negative coef?cient of disposition effect for q
0.2 indicates that, under the lower bound of close trading volume,
if individuals have higher tendency to realize their gains than to
realize losses, this would contribute to lower individual close
trading volume. Under the upper bound of close trading volume,
individuals' close trading volume increases with tendency to
realize their gains. This leads to insigni?cant coef?cients within
the middle range of close volume and a signi?cant and positive
coef?cient at the upper bound (q S 0.7). However, institutional
investors have the tendency to close an extremely high volume of
positions (qS0.8) only when they have a higher tendency of
disposition bias. This ?nding indicates that a desire to realize
gains soon and ride on losses contributes to a high closing
volume.
In sum, both open and close trading decisions of individual
investors are motivated to some degree by overcon?dence.
There is also a correlation between the close trading decisions
and the disposition effect for individual investors. Institutional
trading is less subject to behavioral biases, such as over-
con?dence and the disposition effect. The exception occurs
when the open and close volume is extremely large. Over-
con?dence induces institutional investors to open large posi-
tions and the desire to realize gains and ride on losses leads
them to close large positions.
5. Conclusion
This paper employs a unique data set to examine the trading
behaviors of individual and institutional investors in the TAIEX
futures markets. The average return per contract for individual
investors is negative, but positive for institutional investors. The
market returns are positive following the heavy buying by insti-
tutional investors, and negative after their heavy selling. By
contrast, the returns are negative after individual heavy buying and
positive after individual heavy selling. These results suggest that
individual investors are poor market timers, whereas institutional
investors have success in timing the market.
In addition, individual trading activity is more aggressive for
their higher proportion of market order and shorter holding time of
a round trip trade. Further testing indicates that both individual
and institutional trading activity is motivated by behavioral biases.
In particular, individuals are more overcon?dent than individuals;
individuals exhibit a tendency toward the disposition effect, but
institutional investors display a reverse disposition effect. When
institutions are overcon?dent, their open volume is either
extremely low or high. For the lower bounds of quantiles, when
individuals have lower tendency to realize their gains than to
realize losses, close trading volume is relatively low. For the upper
bounds of quantiles, their close trading volume increases with the
increasing tendency to realize their gains. Under the upper bounds
of quantiles, similar results are revealed for institutional investors.
That is, disposition bias leads to a higher close volume. This result is
consistent with Lin (2011), who shows that trades of all categories
of investors are motivated by both behavioral- and market-related
factors, and individuals tend to be more irrational and prone to
misinterpret available information or trade for non-informational
reasons.
Con?icts of interest
All contributing authors declare no con?icts of interest.
M.-C. Lin, M.-T. Chiang / Asia Paci?c Management Review 20 (2015) 165e176 174
A
p
p
e
n
d
i
x
.
A
n
e
x
a
m
p
l
e
o
f
t
h
e
c
a
l
c
u
l
a
t
i
o
n
o
f
t
h
e
v
a
r
i
a
b
l
e
s
o
f
g
a
i
n
s
/
l
o
s
s
e
s
D
a
t
e
B
u
y
/
S
e
l
l
P
r
i
c
e
N
u
m
b
e
r
s
o
f
c
o
n
t
r
a
c
t
s
T
i
m
e
S
e
t
t
l
e
m
e
n
t
p
r
i
c
e
O
p
e
n
i
n
t
e
r
e
s
t
s
A
v
e
r
a
g
e
c
o
s
t
G
r
o
s
s
r
e
a
l
i
z
e
d
g
a
i
n
/
l
o
s
s
C
o
m
m
i
s
s
i
o
n
T
a
x
N
e
t
r
e
a
l
i
z
e
d
g
a
i
n
/
l
o
s
s
D
u
r
a
t
i
o
n
G
r
o
s
s
u
n
r
e
a
l
i
z
e
d
g
a
i
n
/
l
o
s
s
C
o
m
m
i
s
s
i
o
n
_
e
s
t
i
m
a
t
e
T
a
x
_
e
s
t
i
m
a
t
e
N
e
t
u
n
r
e
a
l
i
z
e
d
g
a
i
n
/
l
o
s
s
N
u
m
b
e
r
s
o
f
r
e
a
l
i
z
e
d
g
a
i
n
N
u
m
b
e
r
s
o
f
r
e
a
l
i
z
e
d
l
o
s
s
N
u
m
b
e
r
s
o
f
p
a
p
e
r
g
a
i
n
N
u
m
b
e
r
s
o
f
p
a
p
e
r
l
o
s
s
C
l
o
s
e
b
u
y
v
o
l
u
m
e
C
l
o
s
e
s
e
l
l
v
o
l
u
m
e
O
p
e
n
b
u
y
v
o
l
u
m
e
O
p
e
n
s
e
l
l
v
o
l
u
m
e
2
0
0
4
1
2
1
6
B
6
0
0
5
7
0
9
:
0
0
:
3
6
6
0
2
0
7
6
0
0
5
.
0
0
2
1
.
0
0
0
2
.
1
1
.
6
8
4
1
7
.
2
1
7
7
7
2
0
0
4
1
2
1
7
S
6
0
2
4
2
0
8
:
5
8
:
2
5
6
0
1
9
5
6
0
0
5
.
0
0
7
.
6
0
0
0
.
6
0
.
4
8
1
6
.
5
1
9
2
9
8
1
9
.
0
0
0
1
.
5
1
.
2
0
3
1
6
.
2
9
7
2
5
2
2
0
0
4
1
2
1
7
S
6
0
2
4
2
0
8
:
5
8
:
3
0
6
0
1
9
3
6
0
0
5
.
0
0
7
.
6
0
0
0
.
6
0
.
4
8
1
6
.
5
1
9
2
9
8
1
1
.
4
0
0
0
.
9
0
.
7
2
2
9
.
7
7
8
2
3
2
2
0
0
4
1
2
1
7
S
6
0
2
4
1
0
8
:
5
8
:
3
0
6
0
1
9
2
6
0
0
5
.
0
0
3
.
8
0
0
0
.
3
0
.
2
4
1
3
.
2
5
9
2
9
8
7
.
6
0
0
0
.
6
0
.
4
8
1
6
.
5
1
9
1
2
1
2
0
0
4
1
2
1
7
S
6
0
2
4
2
0
8
:
5
8
:
3
5
6
0
1
9
0
7
.
6
0
0
0
.
6
0
.
4
8
1
6
.
5
1
9
2
9
8
2
2
2
0
0
4
1
2
1
7
B
6
0
0
6
5
0
9
:
2
5
:
0
6
6
0
1
9
5
6
0
0
6
.
0
0
0
.
0
0
0
1
.
5
1
.
2
0
1
À
2
.
7
0
1
5
5
2
0
0
4
1
2
1
7
B
6
0
0
6
1
0
9
:
2
5
:
0
6
6
0
1
9
6
6
0
0
6
.
0
0
0
.
0
0
0
1
.
8
1
.
4
4
1
À
3
.
2
4
1
6
1
2
0
0
4
1
2
1
7
B
6
0
0
6
1
0
9
:
2
5
:
0
7
6
0
1
9
7
6
0
0
6
.
0
0
1
8
.
2
0
0
2
.
1
1
.
6
8
4
1
4
.
4
1
7
7
1
2
0
0
4
1
2
2
1
S
5
9
7
6
7
1
1
:
0
2
:
5
5
5
9
7
8
0
À
4
2
.
0
0
0
2
.
1
1
.
6
7
7
À
4
5
.
7
7
7
6
9
7
7
7
2
0
0
4
1
2
2
2
B
6
0
4
2
1
0
9
:
2
4
:
1
5
6
0
2
5
1
6
0
4
2
.
0
0
0
.
0
0
0
0
.
3
0
.
2
4
2
À
0
.
5
4
2
1
1
2
0
0
4
1
2
2
2
B
6
0
6
4
7
0
9
:
3
0
:
1
5
6
0
2
5
8
6
0
6
1
.
2
5
4
.
4
0
0
2
.
4
1
.
9
4
0
0
.
0
6
0
8
7
2
0
0
4
1
2
2
2
S
6
0
6
2
1
0
9
:
3
9
:
1
1
6
0
2
5
7
6
0
6
1
.
2
5
0
.
1
5
0
0
.
3
0
.
2
4
2
À
0
.
3
9
2
9
.
7
5
1
.
0
5
0
2
.
1
1
.
6
9
7
À
2
.
7
4
7
1
7
1
2
0
0
4
1
2
2
2
S
6
0
4
6
1
1
0
:
1
8
:
3
3
6
0
2
5
6
6
0
6
1
.
2
5
À
3
.
0
5
0
0
.
3
0
.
2
4
2
À
3
.
5
9
2
4
8
.
7
5
À
1
8
.
3
0
0
1
.
8
1
.
4
5
3
À
2
1
.
5
5
3
1
6
1
2
0
0
4
1
2
2
2
S
6
0
4
6
6
1
0
:
1
8
:
3
5
6
0
2
5
0
À
1
8
.
3
0
0
1
.
8
1
.
4
5
3
À
2
1
.
5
5
3
4
8
.
7
5
6
6
2
0
0
4
1
2
2
3
B
5
9
9
1
7
0
9
:
4
9
:
1
2
6
0
0
2
7
5
9
9
1
.
0
0
0
.
0
0
0
2
.
1
1
.
6
7
7
À
3
.
7
7
7
7
7
2
0
0
4
1
2
2
3
S
5
9
7
1
7
1
0
:
0
3
:
4
2
6
0
0
2
0
À
2
8
.
0
0
0
2
.
1
1
.
6
7
5
À
3
1
.
7
7
5
1
4
7
7
2
0
0
4
1
2
2
8
B
6
0
4
1
7
0
9
:
1
4
:
5
6
6
0
2
0
7
6
0
4
1
.
0
0
À
2
9
.
4
0
0
2
.
1
1
.
6
8
9
À
3
3
.
1
8
9
7
7
2
0
0
4
1
2
2
9
B
6
0
5
6
1
0
8
:
4
5
:
3
6
6
1
1
8
8
6
0
4
2
.
8
8
2
1
.
0
0
0
2
.
4
1
.
9
3
6
1
6
.
6
6
4
8
1
2
0
0
4
1
2
2
9
B
6
0
5
7
2
0
8
:
4
6
:
0
4
6
1
1
8
1
0
6
0
4
5
.
7
0
2
2
.
6
0
0
3
2
.
4
2
1
1
7
.
1
7
9
1
2
2
0
0
4
1
2
2
9
B
6
0
5
7
1
0
8
:
4
6
:
0
4
6
1
1
8
1
1
6
0
4
6
.
7
3
2
2
.
6
0
0
3
.
3
2
.
6
6
3
1
6
.
6
3
7
1
1
1
2
0
0
4
1
2
2
9
B
6
0
5
7
1
0
8
:
4
6
:
0
5
6
1
1
8
1
2
6
0
4
7
.
5
8
2
2
.
6
0
0
3
.
6
2
.
9
0
5
1
6
.
0
9
5
1
2
1
2
0
0
4
1
2
2
9
B
6
0
5
7
1
0
8
:
4
6
:
0
6
6
1
1
8
1
3
6
0
4
8
.
3
1
2
2
.
6
0
0
3
.
9
3
.
1
4
7
1
5
.
5
5
3
1
3
1
2
0
0
4
1
2
2
9
B
6
0
5
7
2
0
8
:
4
6
:
0
6
6
1
1
8
1
5
6
0
4
9
.
4
7
2
2
.
6
0
0
4
.
5
3
.
6
3
2
1
4
.
4
6
8
1
5
2
2
0
0
4
1
2
2
9
S
6
0
9
2
3
0
9
:
3
2
:
0
9
6
1
1
8
1
2
6
0
4
9
.
4
7
2
5
.
5
2
0
0
.
9
0
.
7
2
8
2
3
.
8
9
2
1
7
3
1
0
2
.
0
8
0
3
.
6
2
.
9
1
4
9
5
.
5
6
6
3
1
2
3
2
0
0
4
1
2
2
9
S
6
0
9
2
5
0
9
:
3
2
:
1
0
6
1
1
8
7
6
0
4
9
.
4
7
4
2
.
5
3
3
1
.
5
1
.
2
1
4
3
9
.
8
1
9
1
7
3
5
9
.
5
4
7
2
.
1
1
.
7
0
0
5
5
.
7
4
7
5
7
5
2
0
0
4
1
2
2
9
B
6
1
0
1
7
1
1
:
0
4
:
4
4
6
1
1
8
1
4
6
0
7
5
.
2
3
7
2
.
1
4
7
4
.
2
3
.
4
0
9
6
4
.
5
3
7
1
4
7
2
0
0
4
1
2
2
9
S
6
1
1
3
2
1
2
:
0
4
:
5
5
6
1
1
8
1
2
6
0
7
5
.
2
3
1
5
.
1
0
7
0
.
6
0
.
4
8
8
1
4
.
0
1
9
1
9
2
.
5
9
0
.
6
4
0
3
.
6
2
.
9
2
5
8
4
.
1
1
5
2
1
2
2
2
0
0
4
1
2
2
9
S
6
1
1
3
1
1
2
:
0
5
:
0
0
6
1
1
8
1
1
6
0
7
5
.
2
3
7
.
5
5
3
0
.
3
0
.
2
4
4
7
.
0
1
0
1
9
3
.
5
8
3
.
0
8
7
3
.
3
2
.
6
8
1
7
7
.
1
0
5
1
1
1
1
2
0
0
4
1
2
2
9
S
6
1
1
3
2
1
2
:
0
5
:
0
4
6
1
1
8
9
6
0
7
5
.
2
3
1
5
.
1
0
7
0
.
6
0
.
4
8
8
1
4
.
0
1
9
1
9
3
.
5
6
7
.
9
8
0
2
.
7
2
.
1
9
4
6
3
.
0
8
6
2
9
2
2
0
0
4
1
2
2
9
S
6
1
1
3
2
1
2
:
0
5
:
0
7
6
1
1
8
7
6
0
7
5
.
2
3
1
5
.
1
0
7
0
.
6
0
.
4
8
8
1
4
.
0
1
9
1
9
3
.
5
5
9
.
8
7
3
2
.
1
1
.
7
0
7
5
6
.
0
6
6
2
7
2
2
0
0
4
1
2
3
1
S
6
1
2
4
7
0
9
:
3
9
:
2
4
6
1
8
8
0
6
8
.
2
7
3
2
.
1
1
.
7
0
8
6
4
.
4
6
5
6
4
7
.
5
7
7
M.-C. Lin, M.-T. Chiang / Asia Paci?c Management Review 20 (2015) 165e176 175
References
Bange, M. M. (2000). Do the portfolios of small investors re?ect positive feedback
trading? Journal of Financial and Quantitative Analysis, 35, 239e255.
Barber, B. M., Lee, Y. T., Liu, Y. J., & Odean, T. (2005). Who loses from trade? Evidence
from Taiwan. In EFA 2005 Moscow Meetings Paper (vol. 529062). http://ssrn.
com/abstract.
Barber, B. M., Lee, Y., Liu, Y., & Odean, T. (2009). Just how much do individual in-
vestors lose by trading? Review of Financial Studies, 22, 609e632.
Barber, B. M., & Odean, T. (2001). Boys will be boys: gender, overcon?dence, and
common stock investment. Quarterly Journal of Economics, 116, 261e292.
Barber, B. M., Odean, T., & Zhu, N. (2009). Systematic noise. Journal of Financial
Markets, 12, 547e569.
Barberis, N., & Xiong, W. (2009). What drives the disposition effect? An analysis of a
long-standing preference-based explanation. Journal of Finance, 64, 751e784.
Benos, A. (1998). Aggressiveness and survival of overcon?dent traders. Journal of
Financial Markets, 1, 353e383.
Campbell, J. Y., & Kyle, A. S. (1993). Smart money, noise trading and stock price
behavior. Review of Economic Studies, 6, 1e34.
Campbell, J. Y., Ramadorai, T., & Vuolteenaho, T. (2005). Caught on tape: Institutional
order ?ow and stock returns (Working paper 11439). Cambridge, MA: National
Bureau of Economic Research.
Chakravarty, S. (2001). Stealth-trading: which trader's trades move stock prices?
Journal of Financial Economics, 61, 289e307.
Chan, K., Chung, Y. P., & Fong, W. (2002). The information role of stock and option
volume. Review of Financial Studies, 15, 1049e1075.
Chan, K., & Fong, W. (2000). Trade size, order imbalance, and the volatility-volume
relation. Journal of Financial Economics, 57, 247e273.
Chang, C. C., Hsieh, P. F., & Lai, H. N. (2009). Do informed option investors predict
stock return? Evidence from the Taiwan stock exchange. Journal of Banking and
Finance, 33, 757e764.
Chang, C. C., Hsieh, P. F., & Wang, Y. H. (2010). Information content of options trading
volume for future volatility: evidence from Taiwan options market. Journal of
Banking and Finance, 34, 174e183.
Cheng, T. Y., Lin, C. H., & Chuang, S. S. (2007). Who is the winner? trading behavior
and performance for major types of tradersdEvidence from Taiwan's futures
market. Asia Paci?c Management Review, 12(1), 13e21.
Chen, S. Y., Lin, C. C., Chou, P. H., & Hwang, D. Y. (2002). Price discovery between
TAIFEX TAIEX index futures and SGX MSCI Taiwan index futures. Journal of
Futures Markets, 22(3), 219e240.
Chordia, T., Roll, R., & Subrahmanyam, A. (2002). Order imbalance, liquidity, and
market returns. Journal of Financial Economics, 65, 111e130.
Chou, H. C., Chen, W. N., & Chen, D. H. (2006). The expiration effects of stock-index
derivatives: empirical evidence from the Taiwan futures exchange. Emerging
Markets Finance and Trade, 42(5), 81e102.
Chou, P. H., Lin, M. C., & Yu, M. T. (2006). Margins and price limits in Taiwan's stock
index futures market. Emerging Markets Finance and Trade, 42(1), 62e88.
Chou, R. K., & Wang, G. H. K. (2006). Transaction tax and market quality of the
Taiwan stock index futures. Journal of Futures Markets, 25(2), 1195e1216.
Chou, R. K., & Wang, Y. Y. (2011). A test of the different implications of the over-
con?dence and disposition hypotheses. Journal of Banking and Finance, 35,
2037e2046.
Chueh, H., & Yang, D. Y. (2005). Expiration-day effects of index futures: some
empirical evidence from Taiwan stock market. Journal of Financial Studies, 13(2),
71e95.
Daniel, K., Hirshleifer, D., & Subrahmanyam, A. (1998). Investor psychology and
security market under- and overreactions. Journal of Finance, 53, 1839e1885.
De Bondt, W., & Thaler, R. H. (1995). Financial decision-making in markets and
?rms: a behavioral perspectives. In R. A. Jarrow, V. Maksimovic, & W. T. Ziemba
(Eds.), Finance handbook in operation research and management science 9 (pp.
385e410). Amsterdam: North Holland.
De Long, B. J., Shleifer, A., Summers, L. H., & Waldmann, R. J. (1990). Noise trader risk
in ?nancial markets. Journal of Political Economy, 99, 73e738.
Einhorn, H. J. (1980). Overcon?dence in judgment. New Directions for Methodology
of Social and Behavioral Science, 4, 1e16.
Frazzini, A., & Lamont, O. A. (2008). Dumb money: mutual fund ?ows and the cross-
section of stock returns. Journal of Financial Economics, 88(2), 299e322.
Gervais, S., & Odean, T. (2001). Learning to be overcon?dent. Review of Financial
Studies, 14, 1e27.
Hens, T., & Vlcek, M. (2011). Does prospect theory explain the disposition effect?
Journal of Behavioral Finance, 12(3), 141e157.
Hirshleifer, D. (2001). Investor psychology and asset pricing. Journal of Finance, 64,
1533e1597.
Hirshleifer, D., Subrahmanyam, A., & Titman, S. (1994). Security analysis and trading
patterns when some investors receive information before others. Journal of
Finance, 49, 1665e1698.
Jones, C. M., & Lipson, M. (2004). Are retail orders different? (Working paper).
Columbia University, New York.
Kahneman, D., & Tversky, A. (1979). Prospect theory: an analysis of decision under
risk. Econometrica, 46, 171e185.
Kamesaka, A., Nofsinger, J., & Kawakita, H. (2003). Investment patterns and per-
formance of investor groups in Japan. Paci?c-Basin Finance Journal, 11, 1e22.
Kaniel, R., Saar, G., & Titman, S. (2005). Individual investor sentiment and stock
returns (Working paper). Duke University, Durham.
Koenker, R., & Bassett, G. (1978). Regression quantiles. Econometrica, 46, 33e35.
Kuo, W. Y., & Lin, T. C. (2013). Overcon?dent individual day traders: evidence from
Taiwan futures market. Journal of Banking and Finance, 37(9), 3548e3561.
Kyle, A. S., Hui, O. Y., & Xiong, W. (2006). Prospect theory and liquidation decisions.
Journal of Economic Theory, 129, 273e288.
Kyle, A. S., & Wang, F. A. (1997). Speculation duopoly with agreement to disagree:
can overcon?dence survive the market test? Journal of Finance, 52, 273e279.
Lee, C. M. C., & Ready, M. J. (1991). Inferring trade direction from intraday data.
Journal of Finance, 46, 733e746.
Lin, M. C. (2011). Information content for investor groups in TAIEX futures trading.
Asia-Paci?c Journal of Financial Studies, 40, 433e466.
Lin, C. H., Hsu, H. N., & Chiang, C. Y. (2004). The information transmission between
two substitutes of index futures: the case of TAIEX and Mini-TAIEX stock index
futures. Asia Paci?c Management Review, 9(4), 689e707.
Locke, P., & Mann, S. (2005). Professional trader discipline and trade disposition.
Journal of Financial Economics, 76, 401e444.
Nofsinger, J., & Sias, R. (1999). Herding and feedback trading by institutional and
individual investors. Journal of Finance, 54, 2263e2295.
Odean, T. (1998a). Are investors reluctant to realize their losses? Journal of Finance,
55, 1775e1798.
Odean, T. (1998b). Volume, volatility, price, and pro?t when all traders are above
average. Journal of Finance, 53, 1887e1934.
Odean, T. (1999). Do investors trade too much? American Economic Review, 89,
1279e1298.
O'Curry Fogel, S., & Berry, T. (2006). The disposition effect and individual investor
decisions: the roles of regret and counterfactual alternatives. Journal of
Behavioral Finance, 7(2), 107e116.
Shapira, Z., & Venezia, I. (2001). Patterns of behavior of professionally managed and
independent investors. Journal of Banking and Finance, 25(8), 1573e1587.
Shefrin, H. M., & Statman, M. (1985). The disposition to sell winners too early and
ride losers too long. Journal of Finance, 4, 777e779.
Wang, Z. (1998). Ef?ciency loss and constraints on portfolio holdings. Journal of
Financial Economics, 48, 359e375.
Wang, C. (2001). Investor sentiment and return predictability in agricultural futures
markets. Journal of Futures Markets, 21, 929e952.
Wermers, R. (2003). Is money really smart? New evidence on the relation between
mutual fund ?ows, manager behavior, and performance persistence (Working
paper). University of Maryland, Maryland.
M.-C. Lin, M.-T. Chiang / Asia Paci?c Management Review 20 (2015) 165e176 176
doc_667388190.pdf