Description
The bulk of research in modern economics has been built on the notion that human beings are rational agents who attempt to maximize wealth while minimizing risk. These agents carefully assess the risk and return of all possible investment options to arrive at an investment portfolio that suits their level of risk aversion.
Electronic copy available at:http://ssrn.com/abstract=1872211
!"# %#"&'()* )+ ,-.('(./&0 ,-'#12)*1
"#$% &' "$#()#
*#$%+$,) -./001 02 &$3$4)5)3,
6378)#97,: 02 ;$1720#37$< =$879
=$879< ;> ?@ABA
C@DEF G@HIE@BH
(5($#()#J+.%$879')%+
KKK'495'+.%$879')%+LM(5($#()#
N)##$3.) O%)$3
P
Q$$9 -./001 02 "+973)99
6378)#97,: 02 ;$1720#37$< ")#R)1):
")#R)1):< ;> ?SGHE
C@BEF ASHIAGAG
0%)$3J/$$9'()#R)1):')%+
KKK'0%)$3'+9
-)T,)5()# HEBB
P
U) ,/$3R V7./01$9 "$#()#79< -7503 *)#8$79< &$#RR+ W$+9,7$< &$,,7 W)10/$#X+<
>3%#)7 -750308< Y$010 -0%737< Z)3) -,+1[< -/)#7%$3 N7,5$3< -,)T/)3 6,R+9< \734 ]$0<
$3% ^+0 _+0 20# .055)3,9 03 ,/79 T$T)#' U) ,/$3R V0$/ -,0225$3 20# T#087%734 +9
K7,/ $3 $3$1:979 02 ,/) %79T097,703 )22)., 20# ,/) `73379/ %$,$9),' ^$3): -57,/
T#087%)% 8$1+$(1) #)9)$#./ $9979,$3.)'
Electronic copy available at:http://ssrn.com/abstract=1872211
Abstract
The Behavior of Individual Investors
We provide an overview of research on the stock trading behavior of individual investors.
This research documents that individual investors (1) underperform standard benchmarks
(e.g., a low cost index fund), (2) sell winning investments while holding losing
investments (the “disposition effect”), (3) are heavily influenced by limited attention and
past return performance in their purchase decisions, (4) engage in naïve reinforcement
learning by repeating past behaviors that coincided with pleasure while avoiding past
behaviors that generated pain, and (5) tend to hold undiversified stock portfolios. These
behaviors deleteriously affect the financial well being of individual investors.
JEL Codes: D12, G11, H31
B
The bulk of research in modern economics has been built on the notion that
human beings are rational agents who attempt to maximize wealth while minimizing risk.
These agents carefully assess the risk and return of all possible investment options to
arrive at an investment portfolio that suits their level of risk aversion. Models based on
these assumptions yield powerful insights into how markets work. For example, in the
Capital Asset Pricing Model—the reigning workhorse of asset pricing models—investors
hold well-diversified portfolios consisting of the market portfolio and riskfree
investments. In Grossman and Stiglitz’s (1980) rational expectations model, some
investors choose to acquire costly information and others choose to invest passively.
Informed, active, investors earn higher pre-cost returns, but, in equilibrium, all investors
have the same expected utility. And in Kyle (1985), an informed insider profits at the
expense of noise traders who buy and sell randomly.
A large body of empirical research indicates that real individual investors behave
differently from investors in these models. Most individual investors hold
underdiversified portfolios. Many apparently uninformed investors trade actively,
speculatively, and to their detriment. And, as a group, individual investors make
systematic, not random, buying and selling decisions.
Transaction costs are an unambiguous drag on the returns earned by individual
investors. More surprisingly, many studies document that individual investors earn poor
returns even before costs. Put another way, many individual investors seem to have a
desire to trade actively coupled with perverse security selection ability!
Unlike those in models, real investors tend to sell winning investments while
holding on to their losing investments—a behavior dubbed the "disposition effect." The
disposition effect is among the most widely replicated observations regarding the
behavior of individual investors. While taxes clearly affect the trading of individual
investors, the disposition effect tends to increase, rather than decrease, an investor’s tax
bill since in many markets selling winners generates a tax liability that might be deferred
simply by selling a losing, rather than winning, investment.
H
Real investors are influenced by where they live and work. They tend to hold
stocks of companies close to where they live and invest heavily in the stock of their
employer. These behaviors lead to an investment portfolio far from the market portfolio
proscribed by the CAPM and arguably expose investors to unnecessarily high levels of
idiosyncratic risk.
Real investors are influenced by the media. They tend to buy, rather than sell,
stocks when those stocks are in the news. This attention-based buying can lead investors
to trade too speculatively and has the potential to influence the pricing of stocks.
With this paper, we enter a crowded field of excellent review papers in the field of
behavioral economics and finance (Rabin (1998), Shiller (1999) Hirshleifer (2001),
Daniel, Hirshleifer, and Teoh (2002), Barberis and Thaler (2003), Campbell (2006),
Benartzi and Thaler (2007), Subrahmanyam (2008), and Kaustia (2010a)). We carve out
a specific niche in this field—the behavior of individual investors—and focus on
investments in, and the trading of, individual stocks. We organize the paper around
documented patterns in the investment behavior, as these patterns are generally quite
robust. In contrast, the underlying explanations for these patterns are, to varying degrees,
the subject of continuing debate.
We cover five broad topics: the performance of individual investors, the
disposition effect, buying behavior, reinforcement learning, and diversification. As is the
case with any review paper, we will miss many papers and topics that some deem
relevant. We are human, and all humans err. As is the case for individual investors, so is
the case for those who study them.
D
1. The Performance of Individual Investors
1.1. The Average Individual
In this section, we provide an overview of evidence on the average performance
of individual investors. In Table 1, we provide a brief summary of the articles we discuss.
Collectively, the evidence indicates that the average individual investor underperforms
the marketaboth before and after costs. However, this average (or aggregate)
performance of individual investors masks tremendous variation in performance across
individuals.
In research published through the late 1990s, the study of investor performance
had focused almost exclusively on the performance of institutional investors, in general,
and, more specifically, equity mutual funds.
1
This was partially a result of data
availability (there was relatively abundant data on mutual fund returns and no data on
individual investors). In addition, researchers were searching for evidence of superior
investors to test the central prediction of the efficient markets hypothesis: investors are
unable to earn superior returns (at least after a reasonable accounting for opportunity and
transaction costs).
While the study of institutional investor performance remains an active research
area, several studies provide intriguing evidence that some institutions are able to earn
superior returns. Grinblatt and Titman (1989) and Daniel, Grinblatt, Titman, and
Wermers (DGTW, 1997) study the quarterly holdings of mutual funds. Grinblatt and
Titman conclude (p.415) “superior performance may in fact exist” for some mutual funds.
DGTW (1997) use a much larger sample and time period and document (p.1037) “as a
group, the funds showed some selection ability.” In these studies, the stock selection
1
A notable exception to this generalization is Schlarbaum, Lewellen, and Lease (1978), who analyze the
round-trip trades in 3,000 accounts at a full-service US broker over the period 1964 to 1970. They
document strong returns before trading costs, but after costs returns fail to match a passive index. One
concern with these results is that the authors analyze the internal rate of return on round-trip trades, which
biases their results toward positive performance since investors tend to sell winners and hold losers (the
disposition effect). This dataset is also used in Cohn, Lewellen, Lease, and Schlarbaum, 1975; Lease,
Lewellen, and Scharbaum, 1974; Lewellen, Lease, and Scharbaum, 1977.
S
ability of fund managers generates strong before-fee returns, but is insufficient to cover
the fees funds charge.
2
In financial markets, there is an adding up constraint. For every buy, there is a sell.
If one investor beats the market, someone else must underperform. Collectively, we must
earn the market return before costs. The presence of exceptional investors dictates the
need for subpar investors. With some notable exceptions, which we describe at the end of
this section, the evidence indicates that individual investors are subpar investors.
To preview our conclusions, the aggregate (or average) performance of individual
investors is poor. A big part of the performance penalty borne by individual investors
can be traced to transaction costs (e.g., commissions and bid-ask spread). However,
transaction costs are not the whole story. Individual investors also seem to lose money on
their trades before costs.
The one caveat to this general finding is the intriguing evidence that stocks
heavily bought by individuals over short horizons in the U.S. (e.g., a day or week) go on
to earn strong returns in the subsequent week, while stocks heavily sold earn poor returns.
It should be noted that the short-run return predictability and the poor performance of
individual investors are easily reconciled, as the average holding period for individual
investors is much longer than a few weeks. For example, Barber and Odean (2000)
document that the annual turnover rate at a U.S. discount brokerage is about 75%
annually, which translates into an average holding period of 16 months. (The average
holding period for the stocks in a portfolio is equal to the reciprocal of the portfolios'
turnover rate.) Thus, short-term gains easily could be offset by long-term losses, which is
consistent with much of the evidence we summarize in this section (e.g., Barber, Odean,
and Zhu (2009a)).
2
See also Fama and French (2010), Kosowski, Timmerman, Wermers, and White (2006), and citations
therein. Later in this paper, we discuss evidence from Grinblatt and Keloharju (2000) and Barber, Lee, Liu,
and Odean (2009) that documents strong performance by institutions in Finland and Taiwan, respectively.
@
It should be noted that all of the evidence we discuss in this section focuses on
pre-tax returns. To our knowledge, there is no detailed evidence on the after-tax returns
earned by individual investors because no existing dataset contains the account-level tax
liabilities incurred on dividends and realized capital gains. Nonetheless, we observe that
trading generally hurts performance. With some exceptions (e.g., trading to harvest
capital losses), it is safe to assume that ceteris paribus investors who trade actively in
taxable accounts will earn lower after-tax returns than buy-and-hold investors. Thus,
when trading shortfalls can be traced to high turnover rates, it is likely that taxes will only
exacerbate the performance penalty from trading.
1. 1. 1. Long-Horizon Results
Odean (1999) analyzes the trading records of 10,000 investors at a large discount
broker over the period 1987-1993. Using a calendar-time approach, he finds that the
stocks bought by individuals underperform the stocks sold by 23 basis points per month
in the 12 months after the transaction (with p-values of approximately 0.07) and that this
result persists even when trades more likely to have been made for liquidity, rebalancing,
or tax purposes are excluded from the analysis. These results are provocative on two
dimensions. First, this is the first evidence that there is a group of investors who
systematically earn subpar returns before costs. These investors have perverse security
selection ability. Second, individual investors seem to trade frequently in the face of poor
performance.
Barber and Odean (2000) analyze the now widely used dataset of 78,000 investors
at the same large discount brokerage firm (henceforth referred to as the LDB dataset).
Unlike the earlier dataset, which contained only trading records, this dataset was
augmented with positions and demographic data on the investors, and the analysis here
focuses on positions rather than trades. The analysis of positions, from a larger sample of
investors (78,000 v. 10,000) and a different time period (1991-1996 v. 1987-1993),
provides compelling evidence that individual investors self-managed stock portfolios
underperform the market largely because of trading costs.
A
Barber and Odean (2000) sort households into quintiles based on their monthly
turnover from 1991-1996. The total sample consists of about 65,000 investors, so each
quintile represents about 13,000 households. The 20 % of investors who trade most
actively earn an annual return net of trading costs of 11.4 %. Buy-and-hold investors (i.e.,
the 20 % who trade least actively) earn 18.5 % net of costs. The spread in returns is an
economically large 7 percentage points per year.
These raw return results are confirmed with typical asset-pricing tests. Consider
results based on the Fama-French three-factor model. After costs, the stock portfolio of
the average individual investors earns a three-factor alpha of -31.1 basis points (bps) per
month (-3.7 percentage points (pps) annually). Individuals who trade more perform even
worse. The quintile of investors who trade most actively averages annual turnover of
258 %; these active investors churn their portfolios more than twice per year! They earn
monthly three-factor alphas of -86.4 bps (-10.4 pps annually) after costs.
Grinblatt and Keloharju (2000) analyze two years of trading in Finland and
provide supportive evidence regarding the poor gross returns earned by individual
investors. The focus of their investigation is whether certain investors follow momentum
or contrarian behavior with respect to past returns. In addition, they examine the
performance of different categories of investors. Hampered by a short time-series of
returns, they do not calculate the returns on portfolios that mimic the buying and selling
behavior of investors. Instead, they calculate the buy ratio for a particular stock and
investor category on day t, conditional on its future performance from day t+1 to day
t+120, and test the null hypothesis that the buy ratio is equal for the top and bottom
quartile of future performers. For households, the buy ratio for the top quartile is greater
than the buy ratio for the bottom quartile on only 44.8% of days in the two-year sample
period (p=0.08). For Finnish financial firms and foreigners, the difference in the ratios is
positive on more than 55% of days. Individual investors are net buyers of stocks with
weak future performance, while financial firms and foreigners are net buyers of stocks
with strong future performance.
G
Further confirmation regarding the perverse trading ability of individual investors
comes from Taiwan. Barber, Lee, Liu, and Odean (2009) analyze the trading records of
Taiwanese investors over the period 1995 to 1999. They construct portfolios that mimic
the trading of individuals and institutions, respectively. When portfolios are constructed
assuming holding periods that range from one day to six months, the stocks bought by
institutions (sold by individuals) earn strong returns, while stocks bought by individuals
(sold by institutions) perform poorly. A long-short strategy that mimics the buying and
selling of individual investors and assumes a holding period of 140 trading days earns a
negative abnormal return of 75 basis points per month before accounting for transaction
costs (p
The bulk of research in modern economics has been built on the notion that human beings are rational agents who attempt to maximize wealth while minimizing risk. These agents carefully assess the risk and return of all possible investment options to arrive at an investment portfolio that suits their level of risk aversion.
Electronic copy available at:http://ssrn.com/abstract=1872211
!"# %#"&'()* )+ ,-.('(./&0 ,-'#12)*1
"#$% &' "$#()#
*#$%+$,) -./001 02 &$3$4)5)3,
6378)#97,: 02 ;$1720#37$< =$879
=$879< ;> ?@ABA
C@DEF G@HIE@BH
(5($#()#J+.%$879')%+
KKK'495'+.%$879')%+LM(5($#()#
N)##$3.) O%)$3
P
Q$$9 -./001 02 "+973)99
6378)#97,: 02 ;$1720#37$< ")#R)1):
")#R)1):< ;> ?SGHE
C@BEF ASHIAGAG
0%)$3J/$$9'()#R)1):')%+
KKK'0%)$3'+9
-)T,)5()# HEBB
P
U) ,/$3R V7./01$9 "$#()#79< -7503 *)#8$79< &$#RR+ W$+9,7$< &$,,7 W)10/$#X+<
>3%#)7 -750308< Y$010 -0%737< Z)3) -,+1[< -/)#7%$3 N7,5$3< -,)T/)3 6,R+9< \734 ]$0<
$3% ^+0 _+0 20# .055)3,9 03 ,/79 T$T)#' U) ,/$3R V0$/ -,0225$3 20# T#087%734 +9
K7,/ $3 $3$1:979 02 ,/) %79T097,703 )22)., 20# ,/) `73379/ %$,$9),' ^$3): -57,/
T#087%)% 8$1+$(1) #)9)$#./ $9979,$3.)'
Electronic copy available at:http://ssrn.com/abstract=1872211
Abstract
The Behavior of Individual Investors
We provide an overview of research on the stock trading behavior of individual investors.
This research documents that individual investors (1) underperform standard benchmarks
(e.g., a low cost index fund), (2) sell winning investments while holding losing
investments (the “disposition effect”), (3) are heavily influenced by limited attention and
past return performance in their purchase decisions, (4) engage in naïve reinforcement
learning by repeating past behaviors that coincided with pleasure while avoiding past
behaviors that generated pain, and (5) tend to hold undiversified stock portfolios. These
behaviors deleteriously affect the financial well being of individual investors.
JEL Codes: D12, G11, H31
B
The bulk of research in modern economics has been built on the notion that
human beings are rational agents who attempt to maximize wealth while minimizing risk.
These agents carefully assess the risk and return of all possible investment options to
arrive at an investment portfolio that suits their level of risk aversion. Models based on
these assumptions yield powerful insights into how markets work. For example, in the
Capital Asset Pricing Model—the reigning workhorse of asset pricing models—investors
hold well-diversified portfolios consisting of the market portfolio and riskfree
investments. In Grossman and Stiglitz’s (1980) rational expectations model, some
investors choose to acquire costly information and others choose to invest passively.
Informed, active, investors earn higher pre-cost returns, but, in equilibrium, all investors
have the same expected utility. And in Kyle (1985), an informed insider profits at the
expense of noise traders who buy and sell randomly.
A large body of empirical research indicates that real individual investors behave
differently from investors in these models. Most individual investors hold
underdiversified portfolios. Many apparently uninformed investors trade actively,
speculatively, and to their detriment. And, as a group, individual investors make
systematic, not random, buying and selling decisions.
Transaction costs are an unambiguous drag on the returns earned by individual
investors. More surprisingly, many studies document that individual investors earn poor
returns even before costs. Put another way, many individual investors seem to have a
desire to trade actively coupled with perverse security selection ability!
Unlike those in models, real investors tend to sell winning investments while
holding on to their losing investments—a behavior dubbed the "disposition effect." The
disposition effect is among the most widely replicated observations regarding the
behavior of individual investors. While taxes clearly affect the trading of individual
investors, the disposition effect tends to increase, rather than decrease, an investor’s tax
bill since in many markets selling winners generates a tax liability that might be deferred
simply by selling a losing, rather than winning, investment.
H
Real investors are influenced by where they live and work. They tend to hold
stocks of companies close to where they live and invest heavily in the stock of their
employer. These behaviors lead to an investment portfolio far from the market portfolio
proscribed by the CAPM and arguably expose investors to unnecessarily high levels of
idiosyncratic risk.
Real investors are influenced by the media. They tend to buy, rather than sell,
stocks when those stocks are in the news. This attention-based buying can lead investors
to trade too speculatively and has the potential to influence the pricing of stocks.
With this paper, we enter a crowded field of excellent review papers in the field of
behavioral economics and finance (Rabin (1998), Shiller (1999) Hirshleifer (2001),
Daniel, Hirshleifer, and Teoh (2002), Barberis and Thaler (2003), Campbell (2006),
Benartzi and Thaler (2007), Subrahmanyam (2008), and Kaustia (2010a)). We carve out
a specific niche in this field—the behavior of individual investors—and focus on
investments in, and the trading of, individual stocks. We organize the paper around
documented patterns in the investment behavior, as these patterns are generally quite
robust. In contrast, the underlying explanations for these patterns are, to varying degrees,
the subject of continuing debate.
We cover five broad topics: the performance of individual investors, the
disposition effect, buying behavior, reinforcement learning, and diversification. As is the
case with any review paper, we will miss many papers and topics that some deem
relevant. We are human, and all humans err. As is the case for individual investors, so is
the case for those who study them.
D
1. The Performance of Individual Investors
1.1. The Average Individual
In this section, we provide an overview of evidence on the average performance
of individual investors. In Table 1, we provide a brief summary of the articles we discuss.
Collectively, the evidence indicates that the average individual investor underperforms
the marketaboth before and after costs. However, this average (or aggregate)
performance of individual investors masks tremendous variation in performance across
individuals.
In research published through the late 1990s, the study of investor performance
had focused almost exclusively on the performance of institutional investors, in general,
and, more specifically, equity mutual funds.
1
This was partially a result of data
availability (there was relatively abundant data on mutual fund returns and no data on
individual investors). In addition, researchers were searching for evidence of superior
investors to test the central prediction of the efficient markets hypothesis: investors are
unable to earn superior returns (at least after a reasonable accounting for opportunity and
transaction costs).
While the study of institutional investor performance remains an active research
area, several studies provide intriguing evidence that some institutions are able to earn
superior returns. Grinblatt and Titman (1989) and Daniel, Grinblatt, Titman, and
Wermers (DGTW, 1997) study the quarterly holdings of mutual funds. Grinblatt and
Titman conclude (p.415) “superior performance may in fact exist” for some mutual funds.
DGTW (1997) use a much larger sample and time period and document (p.1037) “as a
group, the funds showed some selection ability.” In these studies, the stock selection
1
A notable exception to this generalization is Schlarbaum, Lewellen, and Lease (1978), who analyze the
round-trip trades in 3,000 accounts at a full-service US broker over the period 1964 to 1970. They
document strong returns before trading costs, but after costs returns fail to match a passive index. One
concern with these results is that the authors analyze the internal rate of return on round-trip trades, which
biases their results toward positive performance since investors tend to sell winners and hold losers (the
disposition effect). This dataset is also used in Cohn, Lewellen, Lease, and Schlarbaum, 1975; Lease,
Lewellen, and Scharbaum, 1974; Lewellen, Lease, and Scharbaum, 1977.
S
ability of fund managers generates strong before-fee returns, but is insufficient to cover
the fees funds charge.
2
In financial markets, there is an adding up constraint. For every buy, there is a sell.
If one investor beats the market, someone else must underperform. Collectively, we must
earn the market return before costs. The presence of exceptional investors dictates the
need for subpar investors. With some notable exceptions, which we describe at the end of
this section, the evidence indicates that individual investors are subpar investors.
To preview our conclusions, the aggregate (or average) performance of individual
investors is poor. A big part of the performance penalty borne by individual investors
can be traced to transaction costs (e.g., commissions and bid-ask spread). However,
transaction costs are not the whole story. Individual investors also seem to lose money on
their trades before costs.
The one caveat to this general finding is the intriguing evidence that stocks
heavily bought by individuals over short horizons in the U.S. (e.g., a day or week) go on
to earn strong returns in the subsequent week, while stocks heavily sold earn poor returns.
It should be noted that the short-run return predictability and the poor performance of
individual investors are easily reconciled, as the average holding period for individual
investors is much longer than a few weeks. For example, Barber and Odean (2000)
document that the annual turnover rate at a U.S. discount brokerage is about 75%
annually, which translates into an average holding period of 16 months. (The average
holding period for the stocks in a portfolio is equal to the reciprocal of the portfolios'
turnover rate.) Thus, short-term gains easily could be offset by long-term losses, which is
consistent with much of the evidence we summarize in this section (e.g., Barber, Odean,
and Zhu (2009a)).
2
See also Fama and French (2010), Kosowski, Timmerman, Wermers, and White (2006), and citations
therein. Later in this paper, we discuss evidence from Grinblatt and Keloharju (2000) and Barber, Lee, Liu,
and Odean (2009) that documents strong performance by institutions in Finland and Taiwan, respectively.
@
It should be noted that all of the evidence we discuss in this section focuses on
pre-tax returns. To our knowledge, there is no detailed evidence on the after-tax returns
earned by individual investors because no existing dataset contains the account-level tax
liabilities incurred on dividends and realized capital gains. Nonetheless, we observe that
trading generally hurts performance. With some exceptions (e.g., trading to harvest
capital losses), it is safe to assume that ceteris paribus investors who trade actively in
taxable accounts will earn lower after-tax returns than buy-and-hold investors. Thus,
when trading shortfalls can be traced to high turnover rates, it is likely that taxes will only
exacerbate the performance penalty from trading.
1. 1. 1. Long-Horizon Results
Odean (1999) analyzes the trading records of 10,000 investors at a large discount
broker over the period 1987-1993. Using a calendar-time approach, he finds that the
stocks bought by individuals underperform the stocks sold by 23 basis points per month
in the 12 months after the transaction (with p-values of approximately 0.07) and that this
result persists even when trades more likely to have been made for liquidity, rebalancing,
or tax purposes are excluded from the analysis. These results are provocative on two
dimensions. First, this is the first evidence that there is a group of investors who
systematically earn subpar returns before costs. These investors have perverse security
selection ability. Second, individual investors seem to trade frequently in the face of poor
performance.
Barber and Odean (2000) analyze the now widely used dataset of 78,000 investors
at the same large discount brokerage firm (henceforth referred to as the LDB dataset).
Unlike the earlier dataset, which contained only trading records, this dataset was
augmented with positions and demographic data on the investors, and the analysis here
focuses on positions rather than trades. The analysis of positions, from a larger sample of
investors (78,000 v. 10,000) and a different time period (1991-1996 v. 1987-1993),
provides compelling evidence that individual investors self-managed stock portfolios
underperform the market largely because of trading costs.
A
Barber and Odean (2000) sort households into quintiles based on their monthly
turnover from 1991-1996. The total sample consists of about 65,000 investors, so each
quintile represents about 13,000 households. The 20 % of investors who trade most
actively earn an annual return net of trading costs of 11.4 %. Buy-and-hold investors (i.e.,
the 20 % who trade least actively) earn 18.5 % net of costs. The spread in returns is an
economically large 7 percentage points per year.
These raw return results are confirmed with typical asset-pricing tests. Consider
results based on the Fama-French three-factor model. After costs, the stock portfolio of
the average individual investors earns a three-factor alpha of -31.1 basis points (bps) per
month (-3.7 percentage points (pps) annually). Individuals who trade more perform even
worse. The quintile of investors who trade most actively averages annual turnover of
258 %; these active investors churn their portfolios more than twice per year! They earn
monthly three-factor alphas of -86.4 bps (-10.4 pps annually) after costs.
Grinblatt and Keloharju (2000) analyze two years of trading in Finland and
provide supportive evidence regarding the poor gross returns earned by individual
investors. The focus of their investigation is whether certain investors follow momentum
or contrarian behavior with respect to past returns. In addition, they examine the
performance of different categories of investors. Hampered by a short time-series of
returns, they do not calculate the returns on portfolios that mimic the buying and selling
behavior of investors. Instead, they calculate the buy ratio for a particular stock and
investor category on day t, conditional on its future performance from day t+1 to day
t+120, and test the null hypothesis that the buy ratio is equal for the top and bottom
quartile of future performers. For households, the buy ratio for the top quartile is greater
than the buy ratio for the bottom quartile on only 44.8% of days in the two-year sample
period (p=0.08). For Finnish financial firms and foreigners, the difference in the ratios is
positive on more than 55% of days. Individual investors are net buyers of stocks with
weak future performance, while financial firms and foreigners are net buyers of stocks
with strong future performance.
G
Further confirmation regarding the perverse trading ability of individual investors
comes from Taiwan. Barber, Lee, Liu, and Odean (2009) analyze the trading records of
Taiwanese investors over the period 1995 to 1999. They construct portfolios that mimic
the trading of individuals and institutions, respectively. When portfolios are constructed
assuming holding periods that range from one day to six months, the stocks bought by
institutions (sold by individuals) earn strong returns, while stocks bought by individuals
(sold by institutions) perform poorly. A long-short strategy that mimics the buying and
selling of individual investors and assumes a holding period of 140 trading days earns a
negative abnormal return of 75 basis points per month before accounting for transaction
costs (p