The effects of national culture and behavioral pitfalls on investors’ decision-
making: Herding behavior in international stock markets

Chih-Hsiang Chang, Shih-Jia Lin

PII: S1059-0560(15)00003-9
DOI: doi: 10.1016/j.iref.2014.12.010
Reference: REVECO 1027

To appear in: International Review of Economics and Finance

Received date: 19 December 2013

Revised date: 26 October 2014
Accepted date: 30 December 2014

Please cite this article as: Chang, C.-H. & Lin, S.-J., The effects of national cul-
ture and behavioral pitfalls on investors’ decision-making: Herding behavior in in-
ternational stock markets, International Review of Economics and Finance (2015), doi:
10.1016/j.iref.2014.12.010

This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.
 ACCEPTED MANUSCRIPT

The effects of national culture and behavioral pitfalls on investors’

decision-making: Herding behavior in international stock markets

P T
RI
SC
Chih-Hsiang Chang *

NU
Professor and Chairman, Department of Finance, National University of Kaohsiung

No. 700, Kaohsiung University Road, Nanzih District, Kaohsiung 81148, Taiwan (ROC).
MA
E-mail: cch@nuk.edu.tw

Tel: +886-7-5919505
ED

Fax: +886-7-5919329
PT

Shih-Jia Lin
CE

Master, National University of Kaohsiung

AC

No. 700, Kaohsiung University Road, Nanzih District, Kaohsiung 81148, Taiwan (ROC).

E-mail: m0983201@mail.nuk.edu.tw

____________________________________
* Corresponding Author. Chang thanks Ministry of Science and Technology of R.O.C.
(Taiwan) for the financial support (project number: MOST 101-2410-H-390-016-). Helpful
comments from editor and anonymous referee are gratefully acknowledged.
 ACCEPTED MANUSCRIPT

The effects of national culture and behavioral pitfalls on investors’

decision-making: Herding behavior in international stock markets

P T
RI
SC
ABSTRACT

NU
The purpose of this study is to explore the determinants of investor decision-making in
MA
international stock markets. Unlike previous literature, this study provides insight into the

effects of national culture and behavioral pitfalls on investors’ decision-making processes in

ED

international stock markets. Its empirical results provide evidence that herding behaviors
PT

occur in Confucian and less sophisticated equity markets. Additionally, it finds that some
CE

national culture indexes are closely correlated with the exhibition of herding. Finally, it
AC

shows that investors’ behavioral pitfalls dominate their herding tendency, as shown in

cross-sectional absolute deviations of returns.

JEL classification: G02; G11; G15

Keywords: Herding behavior; National culture; Behavioral pitfalls; International stock

markets; Behavioral finance

 ACCEPTED MANUSCRIPT

1. Introduction

With the goal of examining herding behavior in international stock markets in depth, this

T
study looks at the effects of national culture and behavioral pitfalls on irrational

P
decision-making by investors. Since the 1980s, various studies have identified irrational

RI
SC
investment behaviors in markets. Of these, herding behavior has been a subject of particular

concern. Herding behavior is when the same investment strategy is adopted by a majority of

NU
investors during the same period of time, in that they sell or buy the same or similar stocks
MA
during that time. Previous literature has provided evidence of herding behavior among

investors. For example, Choi and Sias (2009) discover that institutional investors sell and buy
ED

stocks in the same industries during the same periods of time, leading to herding behavior.
PT

Venezia, Nashikkar, and Shapira (2011) find strong herding tendencies among investors in
CE

small-cap companies and those facing low systematic risk, and have determined that
AC

professional investors demonstrate weaker herding tendencies than amateur (retail) investors.

Demirer, Kutan, and Zhang (2014) and Yao, Ma, and He (2014) indicate that herding

behavior is more widespread at the industry (sector) level than at the market (country) level.

Lin and Lin (2014) show significant herding tendencies of institutional investors and margin

traders on sharp price movements.

In addition to demonstrating the existence of herding behavior, previous studies have

attempted to examine the causes of this behavior. Scharfstein and Stein (1990) argue that fund
 ACCEPTED MANUSCRIPT

managers ignore private information in order to maintain their reputations, and instead

imitate others’ investment strategies. Banerjee (1992) suggests that investors observe others’

T
behaviors to gain implied information content, and then follow their leads. This imitative

P
herding behavior is like a waterfall, and results in an informational cascade. Hirshleifer,

RI
SC
Subrahmanyam, and Titman (1994) suggest that the employment of the same investment

strategies among investors occurs due to the acquisition of similar or identical information,

NU
based on which investors develop herding behaviors. Dass, Massa, and Patgiri (2008) suggest
MA
that reputation and relative performance among mutual funds results in herding behavior

among fund managers, and contributed to the dot com bubble at the end of the 1990s. Yang
ED

(2011) argues that information externality of the second decision-maker plays a major role in
PT

the efficiency of herding behavior among subsequent decision-makers.

CE

Although investors’ herding behavior has been identified by earlier studies, which have
AC

explored its causes, most have been based on the theoretical approaches of standard

(traditional) finance. Herding is irrational behavior and so can be better examined through the

lens of behavioral finance, which primarily examines investors’ decision-making based on

irrational or long-term market inefficiency. Hirshleifer et al. (1994) indicate that when

investors believe that they are acquiring information quicker than others, they demonstrate

overconfidence and herding behavior. Nofsinger and Sias (1999) find close ties between

herding and positive feedback trading, as well as a positive correlation between the herding
 ACCEPTED MANUSCRIPT

tendency of institutional investors and previous returns. Demarzo, Kaniel, and Kremer (2004)

point out how the major influence of community affects investment decision-making, and

T
explain that the need to “keep up with the Joneses” (compare oneself with one’s neighbors)

P
results in herding behavior among communities, as members follow others’ investment

RI
SC
strategies. Brown, Ivkovic, Smith, and Weisbenner (2008) and Shemesh and Zapatero (2014)

also provide, respectively, evidence of significant community effect for stock market

NU
participation decisions and the decision to buy a car. Moreover, Hong, Kubik, and Stein
MA
(2005) suggest the use of a contagion model to describe the information acquisition of

investors through word of mouth, as well as the influence on fund managers of counterparts
ED

in the same city who buy or sell the same stocks. Venezia et al. (2011) reveal herding
PT

behavior for both amateur and professional investors and the tendency to herd among
CE

amateurs is higher as a result of their financial illiteracy and inexperience.

AC

Factors that influence the levels of irrationality of investors and the degree of inefficiency

of stock markets in different countries mainly involve external environments and inner

psychology. As such, it is beneficial to discuss the causes of herding and the determinants of

investors’ irrational decision-making based on an examination of external national cultures

and internal behavioral pitfalls. Hofstede (2001) addresses national culture in terms of the

five dimensions of power distance, individualism, masculinity, uncertainty avoidance, and

long-term orientation, which he has also quantified. Schmeling (2009) uses individualism and
 ACCEPTED MANUSCRIPT

uncertainty avoidance as proxies for herding behavior and overreaction, and has pointed out

the strong influence of investor sentiment on stock returns in countries that are more likely to

T
demonstrate herding behavior or overreaction. Chui, Titman, and Wei (2010) adopt

P
Hofstede’s (2001) individualism index to conduct an empirical study, and find that a positive

RI
SC
correlation exists between individualism and the profits of momentum strategy.

Differences in national culture are worthy to notice when investigating why investor

NU
herding tendency differs across countries. National culture refers to the behavioral norm and
MA
conventional beliefs of the majority of people in a certain country. It dominates the investors’

behavioral reaction to an information shock and the decision-making for an investment.

ED

Further, it may cause the exhibition of same or similar investment strategies for the majority
PT

of investors in the stock market indicating herding behavior. In other words, differences in
CE

national culture may shed insight into the reasons for herding and cross country culture
AC

comparisons are crucial for the examination of herding behavior in international stock

markets. Specifically, Confucian culture emphasizes ethics, obedience, humanism, and

collectivism, indicating that Confucian society may have a high power distance, low

individualism, high masculinity, low uncertainty avoidance, and low long-term orientation.

On the other hand, western culture is based on science, reality, individualism, and happiness.

Its core ideals is the requirements of human’s fundamental rights. This may imply a

prevention of emergence of herding behavior in western countries. Previous literature has

 ACCEPTED MANUSCRIPT

also found that cross country culture comparisons justify the differences in the human’s

behavior and investor right protection across countries. For example, Stulz and Williamson

T
(2003) show that Protestant countries protect the creditors’ rights better than Catholic

P
countries. Nabar and Boonlert-U-Thai (2007) present that differences in national culture are

RI
SC
one of determinants of manager’s accounting choice across countries.

In addition to national culture, another indicator used to measure the irrationality of

NU
investors and inefficiency of markets is investors’ inevitable tendency toward errors (i.e.,
MA
behavioral pitfalls). Among the behavior pitfalls discussed in existing behavioral finance

literature, excessive optimism, overconfidence, and the disposition effect have been shown to
ED

exist. Excessive optimism refers to the inclination of people to prefer the probability of
PT

occurrence of favorable outcomes to that of unfavorable outcomes. Brown and Cliff (2005)
CE

conclude that lower returns result from investors’ excessive optimism, while Kutsuna, Smith,
AC

and Smith (2009) report that investors’ excessive optimism causes underwriters to adjust IPO

offer prices upwards.

Overconfidence refers to the perception of people that they are more competent than

average. Gervais and Odean (2001) point out that although overconfident investors are more

likely to have a superior ability to collect information, they fail to make the best use of this

information, and thus receive lower returns. Gervais, Heaton, and Odean (2011) find that

overconfident managers try their best to collect information that will improve their success
 ACCEPTED MANUSCRIPT

rates and the values of projects, and so tend to accept compensation contracts that involve

excessive risk.

T
The disposition effect refers to the tendency of investors to sell gaining investments too

P
soon and hold losing investments too long. Frazzini (2006) shows that the underreaction of

RI
SC
investors to news is due to the disposition effect. Grinblatt, Keloharju, and Linnainmaa (2012)

discover the impact of the disposition effect on capital loss, and have pointed out that

NU
investors with high IQs are less likely to be affected by the disposition effect in their trading
MA
behaviors.

This study aims to investigate the influence of national culture and behavioral pitfalls on
ED

the herding tendency of investors in international stock markets, as well as to examine

PT

herding behavior in international equity markets. This study has four features that distinguish
CE

it from previous studies. First, since investors inevitably imitate public herding behavior, and
AC

this imitating behavior is influenced by national culture, this study examines the influence of

national culture on investor herding behavior in the international stock markets based on

cross country culture comparison. Second, although earlier studies of behavioral finance have

described the various types of behavioral pitfalls made by investors, they have less frequently

examined the influence of behavioral pitfalls on herding behavior. In order to completely

understand the causes of herding behavior, this study investigates the influence of investors’

behavioral pitfalls on their herding tendencies. Third, for the implications of empirical results,
 ACCEPTED MANUSCRIPT

this study discusses the relationship between cultural indices and macro indices, as well as

the effect of Confucian philosophy on investor herding behavior. Finally, this study

T
investigates the herding behavior of investors in more stock markets and for longer intervals

P
than previous studies, covering 50 stock markets over 46 years.

RI
SC
The empirical results of the study first show that a lower percentage of stock markets

exhibit significant herding behavior than was found by Chiang and Zheng (2010) and that a

NU
number of stock markets demonstrating herding behavior are concentrated in Confucian and
MA
less mature countries. Second, our findings demonstrate that national culture is one of the

determinants of investors’ herding behavior and that, among the aspects involved in culture,
ED

power distance, individualism, and masculinity are the most important in this regard. Finally,
PT

we find that behavioral pitfalls not only play a dominant role in investors’ herding tendencies,
CE

in terms of the cross-sectional absolute deviation of returns, but are also the key factors that
AC

lead to investors’ irrational behaviors.

The structure of the rest of this paper is as follows. We introduce our data and

methodologies in Section 2, below, while Section 3 presents our empirical results. Section 4

offers robust analysis including the examination of applying Granger’s (1969) causality test

and the study of using Tang and Koveos’ (2008) national cultural indices, in addition to those

proposed by Hofstede (2001). We discuss further the empirical results in Section 5 and

conclude the paper in Section 6.

 ACCEPTED MANUSCRIPT

2. Data and methodologies

2.1. Data sources and research interval

T
This study uses the returns of market and industrial indexes to calculate cross-sectional

P
RI
absolute deviation of returns, and examines the existence of herding behavior in various

SC
markets. In order to explore the influence of national culture on investors’ herding tendency,

NU
it adopts the national culture indexes proposed by Hofstede (2001) for empirical analysis.

Hofstede (2001) investigates 88,000 IBM employees in 72 countries around the world, using
MA
a questionnaire, and quantified the five dimensions of national nature: Power distance,
ED

individualism, masculinity, uncertainty avoidance, and long-term orientation. In terms of

behavioral pitfall data, the price-to-book ratios of each industry in a stock market (daily data)
PT

were used as proxies for excessive optimism, while the trading volumes of each industry in a
CE

stock market (daily data) were used as proxies for overconfidence and the disposition effect.
AC

The data on price-to-book ratio, trading volume, market indexes, and industrial indexes were

collected from the Datastream database from January 6, 1965 to May 31, 2009, as well as the

Taiwan Economic Journal databank, Mergent Online databank, Yahoo Finance website, and

Business Week website from June 1, 2009 to July 12, 2011.

The research subjects of this study include 50 stock markets. Due to the features of the

data acquired, the research intervals for individual stock markets are inconsistent, with the

longest interval lasting from January 6, 1965 to July 12, 2011. Table 1 shows the research
 ACCEPTED MANUSCRIPT

interval for each stock market.

[INSERT TABLE 1 HERE]

T
2.2. Investigation of investor herding behaviors in individual stock markets

P
RI
This study uses the methods of Chang, Cheng, and Khorana (2000) and Chiang and

SC
Zheng (2010) to examine investors’ herding behaviors in 50 stock markets. Chang et al.

NU
(2000) argue that when herding behavior exists, investors will abandon information that they

possess and follow market trends in making investments. Chang et al. (2000) also use the
MA
interrelationship between cross-sectional absolute deviation of returns and market return
ED

volatility to judge the existence of herding behavior, as shown in Equation (1).

CSADi ,t   i ,0   i ,1Ri ,m,t   i ,2 Ri ,m,t   i ,3Ri2,m,t  i ,t , (1)

PT

where CSADi ,t is cross-sectional absolute deviation (CSAD) in stock market i for day t. As
CE

ni
such, CSADi ,t   Ri , j ,t  Ri ,m,t / n 。 Ri , j ,t is the sector index returns on industry j in stock
j 1
AC

market i for day t. Ri ,m,t is the market index return in stock market i for day t, and ni is the

number of industries in stock market i.  i ,0 , i ,1 , i ,2 , and  i ,3 are regression coefficients,

while  i ,t refers to error term.

Chang et al. (2000) argue that although the return dispersion among stocks based on the

capital asset pricing model (CAPM) increases along with the increase of absolute value of

market returns, this involves a linear correlation. If herding behavior exists in a market during

stock market fluctuation, there is no linear incremental correlation between the return
 ACCEPTED MANUSCRIPT

dispersion among stocks and market returns. Chiang and Zheng (2010), based on the

suggestions of Chang et al. (2000), further conclude that when a stock price change expands

T
and the difference between industrial and market returns narrows, investors demonstrate

P
herding behavior. As a result, in Equation (1) when coefficient  i ,3 has a significant

RI
SC
negative value, it indicates the existence of investor herding behavior.

NU
2.3. Examining the effect of national culture on investor herding behavior

In addition to confirming investor herding behavior in particular stock markets through

MA
the use of coefficient  i ,3 in Equation (1), this study explores the influence of internal
ED

behavioral pitfalls and external national cultures on investors’ decision-making processes.

With regard to the effects of national culture, it aims to investigate whether differences in
PT

national cultures among international stock markets are determinants of investor herding
CE

behavior.
AC

For indexes of national culture, this study refers to Chui et al. (2010) and uses indexes

constructed by Hofstede (2001). Unlike Chui et al. (2010), who only use an individualism

index, we use five indexes of national culture, as suggested by Hofstede (2001), to investigate

the relationship between national culture and investor herding behavior. Hofstede’s (2001)

five indexes of national culture are described in more detail below:

(1) Power distance index (PDI): PDI refers to the expectation and acceptance of unequal

power distribution among the disadvantaged members of national institutions and

 ACCEPTED MANUSCRIPT

organizations. In a high power distance environment, society is unequal and investors

acquire information in unequal conditions. In such cases, investors are more likely to

T
accept implied information of transactions made by others and to follow their leads,

P
which fosters herding behavior.

RI
SC
(2) Individualism index (IDV): IDV refers to the extent to which the people of a society are

loosely connected and care only about themselves and their core families. Thus, in a

NU
society with a lower level of individualism, investors are more likely to care more about
MA
public opinion than their own opinions, and their trading behavior will thus be more

influenced by market consensus and will trend toward herding.

ED

(3) Masculinity index (MAS): MAS concerns the clear distinction between gender roles,
PT

which regards masculinity as arbitrary, strong, and focused on material achievement,

CE

while femininity is modest, tender, and focused on life quality. Thus, in countries with
AC

high masculinity levels, people care about materialism more and are more likely to

engage in opportunistic behaviors of buying high and selling low. This means that when

investors are very focused on earning a greater profit to satisfy their greed, they are more

likely to develop herding behavior.

(4) Uncertainty avoidance index (UAI): UAI refers to the level of threat perceived by

members of a certain culture when they face an uncertain or unknown situation. We

assume that investors with low UAI have lower awareness of risk, and thus, regardless of
 ACCEPTED MANUSCRIPT

the deviation of current share prices from fundamentals will blindly follow the market in

buying or selling stocks, developing herding behavior.

T
(5) Long-term orientation index (LTO): LTO refers to virtues, and especially persistence

P
and frugality, which promote and encourage the search for the orientation of future

RI
SC
returns. A society with a low LTO is concerned with more than current returns, and

investors are easily influenced by market opinion and thus blindly engage in short-term

NU
profitable herding trading rather than holding stocks for a longer term.
MA
In terms of the effects of national culture on investor herding behavior, when coefficient

 i ,3 in Equation (1) has a significant negative value, it indicates the existence of significant
ED

herding behavior in stock market i. This study thus uses D i ,3 as the dependent variable (if
PT

 i ,3 has a significant negative value, then D i ,3 =1; if  i ,3 has a significant positive value,
CE

then D i ,3 =-1; otherwise, D i ,3 =0), and also conducts a multinomial logistic regression
AC

analysis. The description of this is below:

D i ,3  0  1PHIi  2 IDVi  3MASi  4UAIi  5LTOi  i , (2)

3
 h  1 , (3)
h 1

where PHIi , IDVi , MASi , UAIi , and LTOi are respectively the power distance index,

individualism index, masculinity index, uncertainty avoidance index, and long-term

orientation index in stock market i, as proposed by Hofstede (2001), and where

1  P( D i ,3  1 ) ,  2  P( D i ,3  0 ) , and  3  P( D i ,3  1) . { 1 , 2 ,3 } represents

 ACCEPTED MANUSCRIPT

response probability. 1,2 ,3 ,4 , and 5 are regression coefficients, and i is an error

term. According to Equation (2), if coefficient 1,2 ,3 ,4 , or 5 is statistically and

T
significantly different from zero, it indicates that a national cultural index significantly

P
influences investors’ herding behaviors.

RI
SC
2.4. Examining the effect of investors’ behavioral pitfalls on their herding tendencies

NU
Although we are not able to independently judge whether investors significantly

demonstrate herding behavior according to CSAD, CSAD represents return dispersion

MA
between industry sectors and market indexes, as well as the common rising and falling trends
ED

of stocks. A smaller CSAD, therefore, indicates high investor herding tendency. Chiang and

Zheng (2010) also argue that CSAD refers to the investor herding formation. Based on the
PT

aforementioned arguments, this study uses the CSAD as a proxy for investor herding
CE

tendency, and then investigates the effect of investors’ behavioral pitfalls on their herding
AC

tendencies.

Chui et al. (2010) contest the influence of overconfidence and self-attributed bias on

profit from momentum strategy. Because herding behavior may result in abnormal profit

from momentum strategy, behavioral pitfalls will not only affect momentum strategy but also

impact herding tendency. This study examines the influence of investors’ behavioral pitfalls,

such as excessive optimism, overconfidence, and the disposition effect, on their herding

tendencies. Notably, it is helpful to explore the effect of excess optimism on investors’

 ACCEPTED MANUSCRIPT

herding tendency under the condition of having measured overconfidence. Shefrin (2007)

argues that although excess optimism often comes with overconfidence, there are differences

T
between excess optimism and overconfidence. Shefrin (2007) thinks that the degree of excess

P
optimism for a human is not always positively associated with his/her overconfidence. For

RI
SC
example, an investor who is pessimistic about the prospect of the stock market may still

actively buy certain stocks because of his/her overconfidence in selecting stocks. Based on

NU
Shefrin’s (2007) argument, this study measures both overconfidence and excess optimism.
MA
The proxies for excessive optimism, overconfidence, and the disposition effect are described

in more detail below:

ED

(1) Excessive optimism (EO)

PT

EO refers to the tendency of people to overestimate the probability of favorable outcomes

CE

but underestimate that of unfavorable outcomes. Investors with higher EO are more likely to
AC

demonstrate the transaction behavior of buying high and selling low, and to imitate others’

decision-making processes and develop herding behaviors. Yi, El-Badawi, and Lin (2008)

demonstrate the significant positive correlation between investors’ excessive optimism and

the poor performance of stocks after being issued. Based on the definition of EO above, this

study uses EOi ,t to measure the tendency of EO. The calculation method is as follows. First,

to avoid the influence of industrial difference, this study defines PBid, j ,t as the division of the

average daily price-to-book ratio of the industry during the research interval by the
 ACCEPTED MANUSCRIPT

price-to-book ratio of industry j with the declined industrial index in stock market i for day t.

Then, we use the average PBid, j ,t (that is, PBid,t ) of all industries with declined industrial

T
indexes in stock market i for day t as a proxy for EO in stock market i for day t; the larger the

P
value of this proxy, the stronger the demonstrated tendency of investors toward excessive

RI
SC
optimism.

EOi ,t  PBid,t (4)

NU
where EOi ,t is investors’ excessive optimism tendency in stock market i for day t. PBid,t is
MA
the average PBid, j ,t of all industries with decreased industrial indexes in stock market i for
qi ,t
day t. That is, PBi,dt   PBid, j ,t / qi ,t ; qi ,t is the number of industries with decreased
ED

j 1

industrial indexes in stock market i for day t.

PT

(2) Overconfidence (OC)

CE

OC is investors’ tendency to perceive their abilities as above average. Investors’ OC

AC

makes them less likely to imitate others’ decision-making processes, or to develop the

herding behavior of selling low and buying high. Menkhoff, Schmeling, and Schmidt (2010)

suggest that significant differences exist among the OC levels of institutional investors,

investment advisers, and retail investors, and have found significant negative correlations

between investor characteristics (experience and age) and OC level. In accordance with the

above definition of OC, this study uses OCi ,t to measure the tendency of OC. The

calculation method is as follows. First, in order to avoid the influence of industrial difference,
 ACCEPTED MANUSCRIPT

this study defines Vi d, j ,t as the division of the average daily trading volume of an industry

during the research interval by the trading volume of industry j with decreased industrial

T
indexes in stock market i for day t. Then, we use the average Vid, j ,t (that is, Vi d,t ) of all

P
industries with decreased industrial indexes in stock market i for day t as a proxy for OC in

RI
SC
stock market i for day t; the larger the value of this proxy, the stronger the tendency of

investors toward OC.

OCi ,t  Vi d,t
NU (5)
MA
where OCi ,t is investors’ OC tendency in stock market i for day t. Vid,t is the average Vid, j ,t

of all industries with decreased industrial indexes in stock market i for day t. That is,
ED

qi ,t
Vi,dt  Vi d, j ,t / qi ,t .
j 1
PT

(3) Disposition effect (DE)

CE

DE refers to investors’ tendency to hold losing stocks too long and sell gaining stocks too
AC

early. A strong DE prevents investors from using the trading strategy of selling low and

buying high, making it less likely that they will follow the majority when buying and selling.

Barber, Lee, Liu, and Odean (2007) find the existence of DE among investors in the

Taiwanese stock market. Employing the above definition of DE, this study uses DEi ,t to

measure the strength of DE. The calculation method is as follows. First, in order to avoid the

influence of industrial difference, this study defines Vi,uh,t as the division of the average daily

trading volume of the industry during the research interval by the trading volume of industry
 ACCEPTED MANUSCRIPT

h with increased industrial indexes in stock market i for day t. Then, we use the average Viu,h,t

(that is, Vi,ut ) of all industries with increased industrial indexes in stock market i for day t to

T
divide Vi,dt as a proxy for DE in stock market i for day t. When this proxy is larger than 1, it

P
indicates that investors demonstrate DE. The larger the proxy value, the stronger the DE.

RI
Vi,ut

SC
DEi ,t  , (6)
Vi,dt

NU
where DEi ,t is the strength of DE in stock market i for day t. Vi,ut is the average Viu,h ,t of
MA
all industries with increased industrial indexes in stock market i for day t. In other words,
Qi ,t
Vi,ut  Viu,h,t / Qi ,t . Qi ,t refers to the number of industries with increased industrial indexes
h1
ED

in stock market i for day t.

PT

This study uses pooling regression to investigate the influence of investors’ behavioral

pitfalls on their herding tendencies (or CSAD). The pooling regression model is:
CE

CSADi ,t  0  1EOi ,t  2OCi,t  3DEi,t   i ,t , (7)

AC

where 0 ,1 ,2 , and 3 are regression coefficients, and  i,t is an error term. According

to Equation (7), if coefficient 1 ,2 , or 3 is statistically different from zero, it indicates a

significant influence of a certain type of behavioral pitfall on investor herding tendency.

3. Empirical results

3.1. Herding behavior in international stock markets

As shown in Table 2, in the 18 stock markets of Australia, Austria, Chile, China, Hong
 ACCEPTED MANUSCRIPT

Kong, Hungary, Ireland, Japan, Malaysia, Poland, Portugal, Switzerland, Taiwan, Thailand,

Turkey, the UK, the US, and Venezuela, the coefficient  i ,3 has a significant negative value,

T
indicating that among 50 major global stock markets, these 18 involve significant herding

P
behavior. Moreover, Table 2 indicates that among five markets under the influence of

RI
SC
Confucian culture (China, Hong Kong, Japan, South Korea, and Taiwan), four demonstrate

significant herding behavior (Even for South Korea, without demonstrating significant

NU
herding behavior, the coefficient  i ,3 is still a negative value having a t-statistic of -1.61),
MA
with percentages of up to 80%, which indicates a tie between investors from Confucian

cultures and herding behavior. In the end, the stock markets under the influence of Western
ED

culture emerge with obviously a different picture in comparison with those of Confucian
PT

culture. As shown in Table 2, only nine markets (Australia, Austria, Hungary, Ireland, Poland,
CE

Portugal, Switzerland, the UK, and the US) with percentages of 37.5% among twenty-four
AC

markets (Australia, Austria, Belgium, Canada, Czech, Demark, Finland, France, Germany,

Hungary, Ireland, Italy, Luxembourg, Netherlands, New Zealand, Norway, Poland, Portugal,

Slovenia, Spain, Sweden, Switzerland, the UK, and the US) under the influence of Western

culture exhibit evidence of herding behavior.

[INSERT TABLE 2 HERE]

Confucian culture emphasizes ethics, obedience, humanism, and collectivism indicating a

high power distance, low individualism, high masculinity, low uncertainty avoidance, and
 ACCEPTED MANUSCRIPT

low long-term orientation society. The aforementioned behavioral norm and conventional

beliefs may facilitate the formation of herding. That is, Confucian cultural characteristics may

T
lead to the tendency among investors to follow others’ investment strategies and develop

P
herding behaviors. Unlike Confucian countries, Western culture is based on science, reality,

RI
SC
individualism, and happiness. Its core ideal is the requirements of humans’ fundamental

rights. These personal values under the influence of Western culture may prevent the

NU
exhibition of herding behavior for investors in the stock market.
MA
In addition to cultural factors, investor sophistication is rationally assumed to have a

direct influence on the inevitable tendency toward errors (behavioral pitfalls), and to have an
ED

indirect influence on herding tendency. According to Table 2, in addition to Australia, Hong

PT

Kong, Ireland, Japan, Switzerland, the UK, and the US, 11 out of 18 stock markets with
CE

significant herding behavior belong to countries with low levels of investor sophistication,
AC

indicating that significant herding tendency results in more serious behavioral pitfalls among

investors, such as excessive optimism, overconfidence, and a strong disposition effect.

Table 2 also indicates that in 39 of 50 stock markets, coefficient  i ,1 has a statistically

significant positive value, suggesting that in the majority of stock markets, the CSAD

(investor herding tendency) of market index underperformance is lower (higher) than the

CSAD of market index outperformance. That is to say, when there is a bear market, investors

are more likely to follow the trading behavior of the majority due to panic, and thus to
 ACCEPTED MANUSCRIPT

demonstrate a high level of herding tendency.

Overall, Table 2 shows evidence of significant herding behavior in stock markets under

T
the influence of Confucian culture and those with low sophistication. Like Chiang and Zheng

P
(2010), who investigated 18 stock markets, we discovered more significant herding behavior

RI
SC
in Asian stock markets (most of which are under the influence of Confucian culture) than in

other continents. This study, however, identifies more significant herding behavior in stock

NU
markets with low sophistication, as well as insignificant herding behavior in major stock
MA
markets around the globe with higher percentages (64%=32/50), than that of Chiang and

Zheng (2010). This result is consistent with Venezia et al’s (2011) findings that the herding
ED

tendency is negatively correlated with investors’ financial training. There are two possible
PT

causes for this result. First, this study investigates a longer research interval and includes
CE

more recent data than that of Chiang and Zheng (2010). Another reason is that, compared to
AC

those of Chiang and Zheng (2010), our research subjects include more stock markets with

low sophistication. This study finds that differences vary in investor herding tendencies in

stock markets around the world, in how national cultures are involved, and in investors’

inevitable tendency toward errors in stock markets. As such, in order to gain a more

comprehensive understanding of these differences, we attempt to examine the causes of

herding behaviors, and how they are based on national cultures and behavioral pitfalls.
 ACCEPTED MANUSCRIPT

3.2. The effect of national culture on investor herding behavior

As shown in Table 3, a simple logistic regression from Models (1) to (5) shows that the

regression coefficient of masculinity ( 3 ) is statistically different from zero, indicating more

P T
significant investor herding behavior in countries with high masculinity indexes. Additionally,

RI
SC
in the multiple logistic regressions in Model (7), the regression coefficient of masculinity has

a significant positive value. This again indicates the dominance of the cultural feature of

NU
masculinity in investors’ herding behavior. When the masculinity index in a country is high,
MA
male investors in that country are inclined to be arbitrary, strong, and materialistic (while

female investors are modest, tender, and focused on life quality). Therefore, in stock markets
ED

with a high level of masculinity, investors are more likely to care for material things, to
PT

imitate other investors’ strategy and to engage in buying high and selling low, thus
CE

developing herding behavior.

AC

[INSERT TABLE 3 HERE]

In Section 3.1, a cross country comparison of herding by culture indicates that the

majority of the markets under the influence of Confucian philosophy display evidence of

herding while a lower percentage of Western culture countries exhibit herding behavior. One

of the significant differences between Confucian culture and Western culture is the distinction

between gender roles. Although Confucian culture countries have emphasized gender

equality in recent years, the arguments of Confucian philosophy require a clearer distinction
 ACCEPTED MANUSCRIPT

between gender roles than those of Western culture. Therefore, the markets under the

influence of Confucian culture exhibit a higher masculinity index than Western countries.

T
The markets with a higher masculinity index may demonstrate investor herding.

P
In addition to the regression coefficient of masculinity, Table 3 indicates that the

RI
SC
regression coefficients of the other four national cultural indexes in Models (1) to (7) are all

insignificantly different from zero. However, Hofstede’s (2001) investigation of national

NU
cultural indexes was carried out in 1973, and is now 40 years old. At present, the national
MA
cultures in each country are different from what they were 40 years ago. As a result, we are

not able to exclude the possibility of a stronger influence of national cultural factors than is
ED

shown by the empirical findings of investors’ herding behavior in Table 3.

PT

Overall, Table 3 shows that national cultural factors may play an important role in
CE

investors’ decision-making processes, with a positive relationship between masculinity and

AC

herding. In addition, our empirical results indicate the need for the national cultural indexes

of Hofstede (2001) to be updated. In order to clearly understand whether investor behavior is

dominated by national cultural differences, recent literature on national cultural investigations

is needed for empirical analysis.

3.3. The relationship between investor herding tendency and behavioral pitfalls

As shown in Table 4, the regression coefficients of excessive optimism ( 1 ) in Models (1)

to (4) have significant negative values, while those of overconfidence ( 2 ) and disposition
 ACCEPTED MANUSCRIPT

effect ( 3 ) have significant positive values, indicating that when investors in a stock market

are characterized by excessive optimism, overconfidence, and a weaker disposition effect, the

T
CSAD is smaller (higher investor herding tendency). In other words, when investors

P
incorrectly perceive a higher probability of favorable outcomes than unfavorable outcomes

RI
SC
(excessive optimism) they will demonstrate a higher herding tendency. Moreover, when

investors incorrectly perceive that the probability of them making a mistake is much lower

NU
than the actual occurrence of an error (overconfidence) or they are eager to sell gaining
MA
stocks and unwilling to sell losing stocks (the disposition effect), they will demonstrate lower

herding tendency.
ED

[INSERT TABLE 4 HERE]

PT

A comparison of Tables 3 and 4 shows that behavioral pitfalls have a more significant
CE

influence on investor herding tendency than national cultural factors, there are two possible
AC

causes for this. First, internal psychological factors (investors’ behavioral pitfalls) are more

helpful than external environmental factors (national culture) in explaining differences in

investor herding tendency in each national stock market. Second, it has been a long time

since Hofstede (2001) proposed the national cultural indexes used in this paper, so the

relationship of these indexes with herding behavior is less significant and more up-to-date

national cultural indexes are needed for empirical analysis.

According to Table 4, behavioral pitfalls have a significant, or even dominant, influence

 ACCEPTED MANUSCRIPT

on investor herding tendency. This indicates that the inevitable inclination of humans toward

error is the key factor that influences herding behavior. Because behavioral pitfalls are

T
irrational behaviors influenced by psychological factors, and herding behavior is an irrational

P
phenomenon of human behavior, herding behavior is inevitably influenced by behavioral

RI
SC
pitfalls.

NU
4. Robust Analysis
MA
4.1. Empirical results of applying Tang and Koveos’ (2008) national cultural indices

In Section 3.2, we find that, in addition to the regression coefficient of masculinity, the
ED

other four national culture indexes are insignificantly different from zero, but this may be
PT

because we use the national culture indexes proposed by Hofstede (2001) from 1973. The
CE

current national culture of each country is different from what it was 40 years ago. As a result,

in this section, we use the five dimensions proposed by Tang and Koveos (2008) in 2007 in
AC

response to those of Hofstede (2001), which are power distance, individualism, masculinity,

uncertainty avoidance, and long-term orientation, to update the national cultural indexes for

empirical analysis. The empirical results are shown in Table 5.

[INSERT TABLE 5 HERE]

Table 5 shows that although in the simple logistic regression in Models (1) to (5), only the

regression coefficient of masculinity is significantly greater than zero, in the multiple logistic
 ACCEPTED MANUSCRIPT

regressions in Models (6) and (7), the regression coefficients of power distance ( 1 ) and

masculinity are both significantly larger than zero, while the regression coefficient of

individualism ( 2 ) is significantly below zero. The above results mean that when a country

P T
has a high power distance index, low individualism index, or high masculinity index,

RI
SC
investors in its stock market are more likely to pay attention to implied informational content

in the transaction activities of others than to care for their own thoughts or material

NU
approaches (more likely to imitate the strategies of investors earning profit). These cultural
MA
characteristics may result in investor herding behavior.

In general, Table 5 shows that national cultural factors of power distance, individualism,
ED

and masculinity are closely associated with the exhibition of herding. People with the
PT

influence of Confucian culture focusing on ethics, obedience, humanism, and collectivism

CE

display the acceptance of unequal power distribution among the disadvantaged members of
AC

national institutions and organizations, close connection with the public, and the clear

distinction between gender roles. In other words, investors in the market under the influence

of Confucian culture demonstrate a high power distance, low individualism, and high

masculinity, indicating that there is a greater possibility of exhibiting herding behavior. Hence,

the results in Table 5 are consistent with the explanations and arguments in Section 3.1.

When Table 3 and Table 5 are compared, we find that when we use the more recently

updated national culture indexes of Tang and Koveos (2008), rather than those proposed by
 ACCEPTED MANUSCRIPT

Hofstede (2001) in 1973 for empirical analysis, a more significant influence of national

cultural indexes on investor herding behavior is identified. In addition to internal behavioral

T
pitfalls, external national culture is also one of the important factors affecting investor

P
decision-making processes.

RI
SC
4.2. The causal relationships between investor herding tendency and behavioral pitfalls

NU
In Section 3.3, this study shows that proxies of behavioral pitfalls are significantly

correlated with investor herding tendency and the inevitable inclination of humans toward
MA
error plays an important role in investors’ decision-making. In order to explore whether
ED

behavioral pitfalls have a significant, or even dominant, influence on investor herding

tendency, it is necessary to examine the causal relationships between investor herding

PT

tendency and behavioral pitfalls.

CE

This study uses Granger’s (1969) causality test with panel data to investigate the causal
AC

relationships between CSAD and proxies of behavioral pitfalls. The causality test fitting a

vector autoregressive model is:

5 5
CSADi ,t   0  l CSADi ,t l   l Bk ,i ,t l  1,t , k=1,2,3, (8)
l 1 l 1
5 5
Bk ,i ,t  0  l CSADi ,t l   l Bk ,i ,t l  2,t , k=1,2,3, (9)
l 1 l 1

where  0 ,0 ,l ,l ,l , and  l are regression coefficients; 1,t and 2,t are error terms;

5
B1,i,t  EOi,t ; B2,i,t  OCi,t ; B3,i,t  DEi,t . According to Equations (8) and (9), if  l is
l 1

significantly different from zero, it indicates that proxies of behavioral pitfalls Granger cause
 ACCEPTED MANUSCRIPT

5
investor herding tendency. Additionally, if l is statistically different from zero, it
l 1

indicates that investor herding tendency Granger causes proxies of behavioral pitfalls.

T
The F-statistic in Table 6 indicates that proxies of excess optimism, overconfidence, and

P
disposition effect (i.e., EOi,t , OCi,t , and DEi,t ) Grander cause investor herding tendency

RI
SC
(i.e., CSADi,t ) and investor herding tendency Grander cause proxies of excess optimism,

overconfidence, and disposition effect. This implies that there is a feedback relationship

NU
between proxies of behavioral pitfalls and investor herding tendency. Overall, Table 6 reveals
MA
that the inevitable inclination of humans toward error (i.e., behavioral pitfalls) has a

significant (or even dominant) influence on investor herding tendency and is the key factor
ED

influencing herding behavior.

PT

[INSERT TABLE 6 HERE]

CE

5. Discussions
AC

5.1. The relationship between cultural indices and macro indices

This study shows that some national cultural indexes (i.e., power distance, individualism,

and masculinity) are correlated with the exhibition of herding. The aforementioned finding

implies that culture factors are crucial to investment decision-making. National culture

determines the human perception of a certain event and frames the behavioral reaction to an

information shock. The majority of investors in a stock market inevitably tend toward the
 ACCEPTED MANUSCRIPT

normative belief arising from national culture.

Previous studies have indicated the significant influence of cultural factors on human

T
decision-making behavior. Siegel, Licht, and Schwartz (2011) reveal the negative correlation

P
between egalitarianism distance and international portfolio distribution. Aggarwal, Kearney,

RI
SC
and Lucey (2012) show that the aspects of culture can promote foreign portfolio management.

Ferris, Jayaraman, and Sabherwal (2013) report that individualism and long-term orientation

NU
increases and decreases the CEO’s overconfidence tendency, respectively. Holderness (2014)
MA
finds the interdependence between culture and ownership concentration of public

corporations. According to the findings of previous studies, national culture dominates the
ED

individual’s investment and financing decisions. It should be reasonable to conclude that

PT

national culture may be correlated with the aggregate economic activities in a certain country.
CE

As a result of the possibility of a close connection between national culture and economic
AC

activities, earlier studies were devoted to the examination of the role of national culture in the

economic growth in a certain country. Chow, Shields, and Chan (1991) report the significant

influence of individualism in employees on manufacturing performance. Peters (2001) states

that both enterprise culture and enterprise education are an attempt to blend the public

concept of person. Li, Griffin, Yue, and Zhao (2013) indicate that both individualism and

uncertainty avoidance are significantly positively correlated with country GDP per capita.

Xue and Cheng (2013) argue that national culture and credit market conditions are correlative.
 ACCEPTED MANUSCRIPT

These earlier studies shed some insight on the direct effect of national culture on humans’

behavioral decisions and the indirect effect of national culture on aggregate economic

T
activities.

P
Herding refers to the decision of adopting the investment strategies that are the same or

RI
SC
similar to the majority of investors in the market. It is also the behavioral reaction to

following the trend. Culture is the exhibition of human’s subjective feeling, common habit,

NU
shared beliefs, and consistent perception of a society. The effect of culture on our society is
MA
extensively presented in the various kinds of human’s activities including investment,

economic, education, politics, and religion. Therefore, cultural indices are closely correlated
ED

with some macro indexes such as GDP per capita, education, etc. In other words, the macro
PT

indexes may also be the determinants of herding.

CE

5.2. The effect of Confucian philosophy on investors’ herding behavior

AC

Confucian philosophy attaches very great importance to the morality, ethics, obedience to

elders, and respect for ancestors. The society under the influence of Confucian philosophy

builds the behavioral norm of meeting the requirements of loyalty, filial piety, credibility,

fraternity, courtesy, justice, uprightness, and forbearance. The core ideals of Confucian

philosophy are charity, forgiveness, and sincerity. People embracing the Confucian

philosophy are overwhelmed with longing for their ancestors and hometowns. They are also

proud of their culture and custom. Although people in the Confucian culture society are
 ACCEPTED MANUSCRIPT

willing to absorb the essence of foreign cultures, they do not allow foreign influence to

destroy their culture. In a society under the influence of Confucian philosophy, people not

T
only help each other but also seek a good reputation by abiding to Confucian ethics.

P
The founder of Confucianism is Confucius, who is one of the greatest educators and

RI
SC
philosophers in Chinese history. Confucianism is the mainstay of Chinese culture and the

ethical system of Chinese society. Although Confucian philosophy drives from China, it not

NU
only dominates the core value of global Chinese people but also influences the humans’
MA
behavioral norm in adjacent countries. Overall, countries under the influence of Confucian

philosophy include China, Hong Kong, Japan, Korea, Macau, Taiwan, and Vietnam.
ED

According to the arguments of Confucian philosophy, this study concludes that investors
PT

in the stock markets under the influence of Confucian philosophy are living in a society with
CE

a high power distance, low individualism, high masculinity, low uncertainty avoidance, and
AC

low long-term orientation. The aforementioned culture characteristics of Confucian countries

may cause the investors in these countries to exhibit visible herding behavior. In Section 3.1,

our empirical results show that the majority of Confucian countries display significant

herding behavior (Note: Macau and Vietnam under the influences of Confucian culture

were not the research subjects of this study). Moreover, this study indicates that power

distance, individualism, and masculinity are significantly associated with the formation of

investors’ herding behavior. These results are consistent with the aforementioned conclusions.
 ACCEPTED MANUSCRIPT

According to the above discussion and our empirical results, this study argues that

investors in stock markets dominated by Confucian philosophy emphasize public morality

T
and follow the behavior of the majority of people. This not only impacts investors’

P
decision-making but also frames their reaction to a certain informational shock. In other

RI
SC
words, this affects how investors in the stock markets invest. Therefore, Confucian

philosophy may trigger investors’ herding behavior.

NU
MA
6. Conclusions

This study aims to investigate investor herding behavior in international stock markets, as
ED

well as examine the influence of internal behavioral pitfalls and external national culture on
PT

herding behaviors and tendencies. Unlike previous literature, this study has four unique

features. First, in lieu of addressing the cause of imitating public behavior, this study
CE

examines the causes of herding behavior from the perspective of a cross country culture
AC

comparison. Second, while herding behavior among investors is irrational and behavioral

pitfalls have been identified in existing behavioral finance literature, the causes of herding

tendency have seldom been addressed from the perspective of behavioral pitfalls. With this in

mind, this study examines the causes of herding tendency from that perspective. Third, this

study discusses whether macro indexes are the determinants of herding and why Confucian

philosophy may lead to herding. Finally, this study explores more subjects and uses a longer

interval than the majority of earlier studies, covering 50 stock markets over 46 years.
 ACCEPTED MANUSCRIPT

Our empirical results show that first, among 50 stock markets, only 18 exhibit significant

herding behavior, but that among Confucian stock markets and those with low sophistication,

T
herding behavior is significantly more common. Second, our findings show that national

P
cultural factors influence herding behavior to some degree, and that aspects such as power

RI
SC
distance, individualism, and masculinity have more dominant influences than others. Finally,

we find that behavioral pitfalls, including excessive optimism, overconfidence, and the

NU
disposition effect, significantly dominate investors’ herding tendencies.
MA
Based on the empirical results of this study, we find that investor herding tendency varies

among stock markets. The causes of differences in investors’ herding tendencies, in addition
ED

to those described in previous literature, include internal behavioral pitfalls and external
PT

national cultures, and especially investors’ inevitable tendency toward error (behavioral
CE

pitfalls) in decision-making processes. As a result, when engaging in international portfolio

AC

management, mutual fund managers should consider differences in national cultures and the

sophistication of investors in particular stock markets.

References

Aggarwal, R., Kearney, C., & Lucey, B. (2012). Gravity and culture in foreign portfolio

investment. Journal of Banking and Finance, 36, 525-538.

Banerjee, A. V. (1992). A simple model of herd behavior. Quarterly Journal of Economics,

107, 797-817.
 ACCEPTED MANUSCRIPT

Barber, B. M., Lee, Y. T., Liu, Y. J., & Odean, T. (2007). Is the aggregate investor reluctant

to realise losses? Evidence from Taiwan. European Financial Management, 13, 423-447.

T
Brown, G. W., & Cliff, M. T. (2005). Investor sentiment and asset valuation. Journal of

P
Business, 78, 405-440.

RI
SC
Brown, J. R., Ivkovic, Z., Smith, P. A., & Weisbenner, S. (2008). Neighbors matter: Casual

community effects and stock market participation. Journal of Finance, 63, 1509-1531.

NU
Chang, E. C., Cheng, J. W., & Khorana, A. (2000). An examination of herd behavior in
MA
equity markets: An international perspective. Journal of Banking and Finance, 24,

1651-1679.
ED

Chiang, T. C., & Zheng, D. (2010). An empirical analysis of herd behavior in global stock
PT

markets. Journal of Banking and Finance, 34, 1911-1921.

CE

Choi, N., & Sias, R. W. (2009). Institutional industry herding. Journal of Financial
AC

Economics, 94, 469-491.

Chow, C. W., Shields, M. D., & Chan, Y. K. (1991). The effects of management controls and

national culture on manufacturing performance: An experimental investigation.

Accounting, Organizations and Society, 16, 209-226.

Chui, A. C. W., Titman, S., & Wei, K. C. J. (2010). Individualism and momentum around the

world. Journal of Finance, 65, 361-392.

Dass, N., Massa, M., & Patgiri, R. (2008). Mutual funds and bubbles: The surprising role of
 ACCEPTED MANUSCRIPT

contractual incentives. Review of Financial Studies, 21, 51-99.

Demarzo, P. M., Kaniel, R., & Kremer, I. (2004). Diversification as a public good:

T
Community effects in portfolio choice. Journal of Finance, 59, 1677-1716.

P
Demirer, R., Kutan, A. M., & Zhang, H. (2014). Do ADR investors herd? Evidence from

RI
SC
advanced and emerging markets. International Review of Economics and Finance, 30,

138-148.

NU
Ferris, S. P., Jayaraman, N., & Sabherwal, S. (2013). CEO overconfidence and international
MA
merger and acquisition activity. Journal of Financial and Quantitative Analysis, 48,

137-164.
ED

Frazzini, A. (2006). The disposition effect and underreaction to news. Journal of Finance, 61,
PT

2017-2046.
CE

Gervais, S., & Odean, T. (2001). Learning to be overconfident. Review of Financial Studies,
AC

14, 1-27.

Gervais, S., Heaton, J. B., & Odean, T. (2011). Overconfidence, compensation contracts, and

capital budgeting. Journal of Finance, 66, 1735-1777.

Granger, C. W. J. (1969). Investigating causal relations by econometric models and

cross-spectral methods. Econometrica, 37, 424-438.

Grinblatt, M., Keloharju, M., & Linnainmaa, J. T. (2012). IQ, trading behavior, and

performance. Journal of Financial Economics, 104, 339-362.

 ACCEPTED MANUSCRIPT

Hirshleifer, D., Subrahmanyam, A., & Titman, S. (1994). Security analysis and trading

patterns when some investors receive information before others. Journal of Finance, 49,

T
1665-1698.

P
Hofstede, G. (2001). Culture's consequences: Comparing values, behaviors, institutions, and

RI
SC
organizations across nations. (2nd ed.). California: Sage.

Holderness, C. G. (2014). Culture and the ownership concentration of public corporations

NU
around the world. Journal of Corporate Finance, Article in Press, DOI:
MA
10.1016/j.jcorpfin.2014.07.002.

Hong, H., Kubik, J. D., & Stein, J. C. (2005). Thy neighbor's portfolio: Word-of-mouth
ED

effects in the holdings and trades of money managers. Journal of Finance, 60,
PT

2801-2824.
CE

Kutsuna, K., Smith, J. K., & Smith, R. L. (2009). Public information, IPO price formation,
AC

and long-run returns: Japanese evidence. Journal of Finance, 64, 505-546.

Li, K., Griffin, D., Yue, H., & Zhao, L. (2013). How does culture influence corporate

risk-taking? Journal of Corporate Finance, 23, 1-22.

Lin, A. Y., & Lin, Y.-N. (2014). Herding of institutional investors and margin traders on

extreme market movements. International Review of Economics and Finance, 33,

186-198.

Menkhoff, L., Schmeling, M., & Schmidt, U. (2010). Overconfidence, experience, and
 ACCEPTED MANUSCRIPT

professionalism: An experimental study. Working Paper, Leibniz Universität Hannover.

Nabar, S., & Boonlert-U-Thai, K. K. (2007). Earnings management, investor protection, and

T
national culture. Journal of International Accounting Research, 6, 35-54.

P
Nofsinger, J. R., & Sias, R. W. (1999). Herding and feedback trading by institutional and

RI
SC
individual investors. Journal of Finance, 54, 2263-2295.

Peters, M. (2001). Education, enterprise culture and the entrepreneurial self: A foucauldian

NU
perspective. Journal of Educational Enquiry, 2, 58-71.
MA
Scharfstein, D. S., & Stein, J. C. (1990). Herd behavior and investment. American Economic

Review, 80, 465-479.

ED

Schmeling, M. (2009). Investor sentiment and stock returns: Some international evidence.
PT

Journal of Empirical Finance, 16, 394-408.

CE

Shefrin, H. (2007). Behavioral corporation finance: Decisions that create value. New York:
AC

McGraw-Hill.

Shemesh, J. and Zapatero, F. (2014). Thou shalt not covet thy (suburban) neighbor’s car.

Working paper, University of Melbourne, SSRN: http://ssrn.com/abstract=1805206 or

http://dx.doi.org/10.2139/ssrn.1805206.

Siegel, J. I., Licht, A. N., & Schwartz, S. H. (2011). Egalitarianism and international

investment. Journal of Financial Economics, 102, 621-642.

Stulz, R. M., & Williamson, R. (2003). Culture, openness, and finance. Journal of Financial
 ACCEPTED MANUSCRIPT

Economics, 70, 313-349.

Tang, L., & Koveos, P. E. (2008). A framework to update Hofstede's cultural value indices:

T
Economic dynamics and institutional stability. Journal of International Business Studies,

P
39, 1045-1063.

RI
SC
Venezia, I., Nashikkar, A., & Shapira, Z. (2011). Firm specific and macro herding by

professional and amateur investors and their effects on market volatility. Journal of

Banking and Finance, 35, 1599-1609.

NU
MA
Xue, M., & Cheng, W. (2013). National culture, market condition and market share of

foreign bank. Economic Modelling, 33, 991-997.

ED

Yang, W.-R. (2011). Herding with costly information and signal extraction. International
PT

Review of Economics and Finance, 20, 624-632.

CE

Yao, J., Ma, C., & He, W. P. (2014). Investor herding behavior of Chinese stock market,
AC

International Review of Economics and Finance, 29, 12-19.

Yi, B., El-Badawi, M. H., & Lin, B. (2008). Pre-issue investor optimism and post-issue

underperformance. Financial Analysts Journal, 64, 77-87.

 ACCEPTED MANUSCRIPT

Table 1
The research intervals for individual stock markets. This table reports the research intervals
for each stock market. The research subjects include 50 stock markets. Due to the features of
the data acquired, the research intervals for individual stock markets are inconsistent, with the
longest interval lasting from January 6, 1965 to July 12, 2011.

T
Countries Research interval (yyyy/mm/dd) Countries Research interval (yyyy/mm/dd)

P
Argentina 1988/01/05 to 2011/07/01 Malaysia 1986/01/03 to 2011/07/07

RI
Australia 1973/01/02 to 2011/05/27 Mexico 1988/01/05 to 2011/07/07
Austria 1973/01/02 to 2011/05/27 Netherlands 1973/01/02 to 2011/05/27

SC
Belgium 1973/01/02 to 2011/05/27 New Zealand 1988/01/06 to 2011/05/27
Brazil 1994/07/05 to 2011/07/07 Norway 1980/01/03 to 2011/05/27

NU
Bulgaria 2000/10/04 to 2011/07/11 Pakistan 1992/07/17 to 2011/07/08
Canada 1973/01/03 to 2011/05/27 Peru 1994/01/04 to 2011/07/08
Chile 1989/07/04 to 2011/07/07 Philippines 1987/09/10 to 2011/07/07
MA
China 1991/08/27 to 2011/07/12 Poland 1994/03/03 to 2011/07/08
Colombia 1992/01/03 to 2011/07/08 Portugal 1990/01/03 to 2011/07/07
Czech 1993/11/10 to 2011/07/08 Romania 1996/12/10 to 2011/07/11
ED

Denmark 1973/01/02 to 2011/05/27 Russian 1998/01/28 to 2011/07/11

Finland 1988/03/28 to 2011/05/27 Singapore 1973/01/02 to 2011/07/07
France 1973/01/02 to 2011/05/27 Slovenia 1999/01/04 to 2011/07/11
PT

Germany 1973/01/02 to 2011/05/27 South Africa 1973/01/02 to 2011/05/27

Greece 1988/01/05 to 2011/07/07 South Korea 1987/09/10 to 2011/07/07
CE

Hong Kong 1973/01/02 to 2011/07/07 Spain 1987/03/03 to 2011/05/27

Hungary 1991/06/24 to 2011/07/08 Sweden 1982/01/05 to 2011/05/27
India 1990/01/02 to 2011/05/27 Switzerland 1973/01/03 to 2011/05/27
AC

Indonesia 1990/04/03 to 2011/07/08 Taiwan 1987/09/10 to 2011/05/10

Ireland 1973/01/02 to 2011/05/27 Thailand 1987/01/05 to 2011/07/07
Israel 1993/01/04 to 2011/05/27 Turkey 1988/01/05 to 2011/07/07
Italy 1973/01/02 to 2011/05/27 UK 1965/01/06 to 2011/05/27
Japan 1973/01/03 to 2011/07/01 US 1973/01/03 to 2011/05/27
Luxembourg 1992/01/03 to 2011/07/08 Venezuela 1990/01/03 to 2011/07/08
 ACCEPTED MANUSCRIPT

Table 2
Investor herding behavior in international stock markets. This table reports the investor
herding behavior in international stock markets. This study uses the methods of Chang et al.
(2000) and Chiang and Zheng (2010) to examine investors’ herding behaviors in 50 major
stock markets. When coefficient  i,3 has a significant negative value, it indicates the

T
existence of investor herding behavior. Figures in parentheses are the t-statistic. Adj R2
refers to the adjusted coefficient of determination. *, **, and *** indicate significance at the

P
10%, 5%, and 1% levels, respectively.

RI
Countries  i ,0  i,1  i,2  i,3 Adj  R2 F-statistic
Argentina 0.0113*** -0.0421*** 0.2394*** -0.0381 0.3002 871.81***

SC
(80.75) (-10.96) (35.42) (-1.44)
Australia 0.0073*** 0.014*** 0.4271*** -0.4514*** 0.4303 2445.12***

NU
(133.64) (3.98) (72.53) (-8.67)
Austria 0.0044*** 0.0133*** 0.7041*** -3.8388*** 0.5082 3029.93***
(66.26) (2.91) (63.96) (-16.96)
MA
Belgium 0.0071*** 0.0154*** 0.4683*** 0.5941*** 0.4831 3004.16***
(123.44) (3.96) (52.21) (3.08)
Brazil 0.0099*** 0.0248*** 0.3571*** 0.9078*** 0.3675 817.98***
ED

(58.49) (4.02) (24.27) (4.91)

Bulgaria 0.0085*** 0.0129 0.9066*** 0.9676*** 0.5292 978.36***
(19.38) (0.80) (29.72) (5.90)
PT

Canada 0.0086*** 0.0188*** 0.387*** 0.7038*** 0.4193 2337.28***

(159.88) (5.04) (45.54) (4.08)
CE

Chile 0.0062*** 0.04*** 0.5322*** -2.3054*** 0.3045 806.92***

(58.41) (5.91) (29.80) (-4.65)
China 0.0081*** 0.0264*** 0.2632*** -0.7159*** 0.1952 396.15***
AC

(51.37) (5.64) (27.07) (-10.05)

Colombia 0.0052*** 0.0312*** 0.6768*** -0.2701 0.4979 1570.81***
(44.16) (4.15) (39.17) (-0.86)
Czech 0.0082*** -0.0046 0.6264*** 0.4553*** 0.5517 1767.25***
(52.75) (-0.67) (43.19) (3.07)
Denmark 0.0056*** 0.0452*** 0.5609*** 0.5102*** 0.54 3476.89***
(74.81) (9.60) (59.63) (4.49)
Finland 0.0093*** 0.0144*** 0.4263*** 2.5936*** 0.7051 4634.24***
(86.65) (3.83) (46.66) (22.01)
France 0.0083*** 0.0228*** 0.308*** 1.0066*** 0.4237 2371.77***
(139.62) (7.84) (38.87) (5.88)
 ACCEPTED MANUSCRIPT

Table 2 (continued)
Countries  i ,0  i,1  i,2  i,3 Adj  R2 F-statistic
Germany 0.0064*** 0.0382*** 0.3896*** 2.0814*** 0.517 3447.09***
(112.63) (11.04) (52.47) (15.75)
Greece 0.0107*** 0.0216*** 0.2936*** 1.9482*** 0.2715 730.37***

T
(57.89) (3.31) (16.77) (7.69)

P
Hong Kong 0.0087*** 0.0334*** 0.3459*** -0.3123*** 0.3793 1952.49***

RI
(108.31) (10.30) (62.35) (-8.32)
Hungary 0.0104*** 0.0315*** 0.5948*** -1.1785*** 0.3781 1011.61***

SC
(55.31) (4.45) (34.78) (-5.59)
India 0.0108*** 0.0057 0.2923*** 0.9604*** 0.4143 1196.00***

NU
(84.18) (1.23) (29.97) (10.43)
Indonesia 0.0109*** 0.0186*** 0.4555*** 1.2607*** 0.7614 5501.84***
(68.19) (3.44) (44.59) (42.42)
MA
Ireland 0.0082*** 0.0058 0.7397*** -0.4096* 0.4715 2858.35***
(75.81) (0.99) (56.38) (-1.93)
Israel 0.0093*** 0.0005 0.1937*** 0.2478 0.1379 241.78***
ED

(77.67) (0.10) (13.30) (0.80)

Italy 0.0078*** 0.029*** 0.2564*** 1.1494*** 0.3693 1890.12***
(115.49) (9.95) (31.13) (7.09)
PT

Japan 0.0067*** 0.0349*** 0.3074*** -1.0226*** 0.3279 1560.38***

(135.22) (12.06) (48.35) (-9.09)
CE

Luxembourg 0.006*** 0.016** 0.7055*** 1.6528*** 0.6165 2630.02***

(56.67) (2.52) (42.50) (4.19)
AC

Malaysia 0.0076*** 0.0241*** 0.4166*** -0.5943*** 0.4633 1817.42***

(94.26) (6.03) (51.01) (-8.01)
Mexico 0.0065*** 0.0179*** 0.4058*** 2.7572*** 0.6237 3325.59***
(64.07) (4.22) (35.89) (15.44)
Netherlands 0.0073*** 0.0195*** 0.3863*** 0.9516*** 0.4705 2886.08***
(120.77) (5.83) (44.36) (5.21)
New Zealand 0.0078*** 0.0218*** 0.505*** 0.5446** 0.4184 1409.78***
(85.83) (3.65) (37.95) (2.22)
Norway 0.009*** 0.0219*** 0.3872*** 0.0713 0.3714 1556.17***
(92.15) (5.20) (41.13) (0.56)
Pakistan 0.0086*** 0.0208*** 0.4021*** 2.2166*** 0.6903 3363.04***
(70.80) (4.79) (43.75) (23.77)
Peru 0.0059*** 0.0098 0.7144*** 0.0537 0.5086 1521.04***
(40.18) (1.21) (34.74) (0.14)
 ACCEPTED MANUSCRIPT

Table 2 (continued)
Countries  i ,0  i,1  i,2  i,3 Adj  R2 F-statistic
Philippines 0.0101*** 0.0211*** 0.3636*** 0.4523** 0.3181 920.96***
(82.08) (3.86) (27.93) (2.29)
Poland 0.01*** 0.0124** 0.3805*** -1.8875*** 0.2630 510.68***

T
(64.62) (2.46) (27.09) (-9.43)

P
Portugal 0.0091*** 0.0209** 0.6351*** -2.1152*** 0.3026 779.80***

RI
(68.13) (2.43) (31.18) (-5.03)
Romania 0.0141*** 0.0083 0.5133*** 0.6164*** 0.5639 1554.91***

SC
(58.89) (1.17) (39.28) (9.42)
Russian 0.0084*** 0.0045 0.4847*** -0.2442 0.0990 126.02***

NU
(10.27) (0.25) (10.54) (-0.72)
Singapore 0.0074*** 0.0244*** 0.3779*** 0.3141*** 0.3969 2119.33***
(96.58) (6.35) (51.10) (3.92)
MA
Slovenia 0.0066*** 0.0080 0.4837*** -0.6050 0.4134 734.13***
(57.29) (1.09) (25.54) (-1.45)
South Africa 0.0074*** 0.0123*** 0.4569*** -0.1834 0.4526 2677.72***
ED

(97.27) (3.38) (53.21) (-1.26)

South Korea 0.0111*** 0.0157*** 0.2684*** -0.3102 0.2416 623.18***
(80.69) (3.79) (21.10) (-1.61)
PT

Spain 0.0082*** 0.0313*** 0.3716*** 0.5698** 0.4075 1401.37***

(91.39) (7.18) (32.17) (2.48)
CE

Sweden 0.0082*** 0.0210*** 0.3317*** 0.1374 0.3393 1265.84***

(85.72) (5.40) (28.88) (0.62)
AC

Switzerland 0.0057*** 0.0328*** 0.5134*** -0.6684*** 0.5018 3244.16***

(106.41) (8.71) (60.88) (-3.86)
Taiwan 0.0097*** 0.0096** 0.2349*** -2.0630*** 0.0913 194.80***
(71.52) (2.47) (17.59) (-9.12)
Thailand 0.0104*** 0.0197*** 0.3813*** -0.6524*** 0.3256 971.23***
(74.17) (4.26) (30.77) (-3.95)
Turkey 0.0122*** 0.0141*** 0.3503*** -1.5817*** 0.2158 556.58***
(66.59) (3.41) (28.95) (-11.86)
UK 0.0064*** 0.0194*** 0.3640*** -0.3714*** 0.4275 2744.5***
(129.57) (7.11) (53.84) (-2.65)
US 0.0066*** 0.0173*** 0.2891*** -0.1424* 0.3671 1873.86***
(146.99) (6.18) (53.97) (-1.87)
Venezuela 0.0077*** -0.0031 0.7165*** -1.2915*** 0.5072 1839.76***
(41.74) (-0.46) (46.50) (-8.44)
 ACCEPTED MANUSCRIPT

Table 3
The effect of Hofstede’s (2001) national culture indexes on investor herding behavior. This
table reports the effect of Hofstede’s (2001) national culture indexes on investor herding
behavior. This study thus uses D i,3 as the dependent variable (if  i,3 has a significant
negative value, then D i,3 =1; if  i,3 has a significant positive value, then D i,3 =-1;

T
otherwise, D i,3 =0), and also conducts a multinomial logistic regression analysis. The

P
description of this is below:
D i,3  0  1PHIi  2 IDVi  3MASi  4UAIi  5LTOi  i , (2)

RI
3
 h  1 . (3)

SC
h 1
If coefficient 1, 2 , 3 , 4 , or 5 is statistically and significantly different from zero, it
indicates that a national cultural index significantly influences investors’ herding behaviors.

NU
Figures in parentheses are the t-statistic. Pseudo R2 refers to the pseudo coefficient of
determination. *, **, and *** indicate significance at the 10%, 5%, and 1% levels,
respectively.
MA
Coefficients Model Model Model Model Model Model Model
(1) (2) (3) (4) (5) (6) (7)
0 0.915 0.773 -0.722 2.522** 0.692 -0.922 -0.443
(0.96) (1.04) (-0.76) (2.07) (0.79) (-0.38) (-0.12)
ED

1 0.008 0.036 0.012

(0.52) (1.51) (0.36)
2
PT

-0 .013 -0.031 -0.032

(-0.89) (-1.44) (-1.20)
3 0.045** 0.067**
CE

(2.22) (2.31)
4 -0.016 -0.016 -0.028
AC

(-1.01) (-0.90) (-1.16)

5 0.014 0.004
(0.84) (0.19)
Pseudo R2 0.005 0.016 0.112 0.022 0.014 0.081 0.247
2 0.27 0.82 5.61** 1.10 0.73 4.06 11.45*
 ACCEPTED MANUSCRIPT

Table 4
The effect of investors’ behavioral pitfalls on their herding tendencies. This table reports the
effect of investors’ behavioral pitfalls on their herding tendencies (or CSAD). This study uses
pooling regression to investigate the influence of investors’ behavioral pitfalls on their
herding tendencies. The pooling regression is:

T
CSADi,t  0  1EOi,t  2OCi,t  3DEi,t  i,t . (7)

P
If coefficient 1, 2 , or 3 is statistically different from zero, it indicates a significant

RI
influence of a certain type of behavioral pitfall on investor herding tendency. Figures in
parentheses are the t-statistic. *, **, and *** indicate significance at the 10%, 5%, and 1%

SC
levels, respectively.
Coefficients Model Model Model Model
(1) (2) (3) (4)

NU
0 0.012251*** 0.012195*** 0.012315*** 0.012125***
(618.43) (673.69) (740.35) (576.72)
1 -0.000071*** -0.000073***
MA
(-6.37) (-6.51)
2 0.000142*** 0.000150***
(16.68) (17.61)
ED

3 6.13×10-10** 6.25×10-10***
(2.49) (2.62)
F-statistic 40.60*** 278.30*** 6.19** 119.19***
PT
CE
AC
 ACCEPTED MANUSCRIPT

Table 5
The effect of Tang and Koveos’ (2008) national culture indexes on investor herding behavior.
This table reports the effect of Tang and Koveos’ (2008) national culture indexes on investor
herding behavior. This study thus uses D i,3 as the dependent variable (if  i,3 has a
significant negative value, then D i,3 =1; if  i,3 has a significant positive value, then

T
D i,3 =-1; otherwise, D i,3 =0), and also conducts a multinomial logistic regression analysis.

P
The description of this is below:
D i ,3  0  1PHIi  2 IDVi  3MASi  4UAIi  5LTOi  i . (2)

RI
3
 h  1 . (3)

SC
h 1
If coefficient 1, 2 , 3 , 4 , or 5 is statistically and significantly different from zero, it
indicates that a national cultural index significantly influences investors’ herding behaviors.
Figures in parentheses are the t-statistic. Pseudo R2 refers to the pseudo coefficient of

NU
determination. *, **, and *** indicate significance at the 10%, 5%, and 1% levels,
respectively.
Coefficients Model Model Model Model Model Model Model
MA
(1) (2) (3) (4) (5) (6) (7)
0 1.372 0.565 -1.192 1.858 1.534* -7.286 -25.197**
(1.40) (0.67) (-0.81) (1.23) (1.84) (-1.54) (-2.26)
ED

1 2.08×10 -4
0.086* 0.215**
(0.02) (1.83) (2.16)
2
PT

-0.015 -0.078** -0.189**

(-1.03) (-1.98) (-2.24)
3 0.055* 0.132**
CE

(1.72) (2.22)
4 -0.007 0.006 0.023
(-0.33) (0.24) (0.73)
AC

5 -0.002 -0.030
(-0.12) (-0.91)
Pseudo R2 0.000 0.027 0.083 0.002 0.000 0.128 0.363
 2
0.000 1.1 3.34* 0.11 0.01 5.13* 14.57**
 ACCEPTED MANUSCRIPT

Table 6
The causal relationships between investor herding tendency and behavioral pitfalls. This table
reports the results of applying Granger’s (1969) causality test with panel data to investigate
the causal relationships between CSAD and proxies of behavioral pitfalls. The causality test
fitting a vector autoregressive model is:
5 5
CSADi ,t   0  l CSADi ,t l   l Bk ,i ,t l  1,t , k=1,2,3, (8)

T
l 1 l 1

P
5 5
Bk ,i ,t  0  l CSADi ,t l   l Bk ,i ,t l  2,t , k=1,2,3. (9)
l 1 l 1

RI
5
If  l is significantly different from zero, it indicates that proxies of behavioral pitfalls
l 1

SC
5
Granger cause investor herding tendency. Additionally, if l is statistically different from
l 1
zero, it indicates that investor herding tendency Granger causes proxies of behavioral pitfalls.

NU
*, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively.
Dependent variables Independent variables F-statistic
MA
3293.02***
CSADi,t CSADi,t l , B1,i,t l (EOi,t l )

3183.57***
B1,i,t (EOi,t ) CSADi,t l , B1,i,t l (EOi,t l )
ED

3293.55***
CSADi,t CSADi,t l , B2,i,t l (OCi,t l )
PT

51.42***
B2,i,t (OCi,t ) CSADi,t l , B2,i,t l (OCi,t l )

3292.52***
CE

CSADi,t CSADi,t l , B3,i,t l (DEi,t l )

2.64***
B3,i,t (DEi,t ) CSADi,t l , B3,i,t l (DEi,t l )
AC
 ACCEPTED MANUSCRIPT

Research highlights

 This study provides insight into the effects of national culture and behavioral pitfalls on

T
investors’ decision-making processes in international stock markets.

P
RI
 Empirical results show that herding behaviors occur in Confucian and less sophisticated

SC
equity markets

 National culture has an impact on investor herding behavior

 NU
Investors’ behavioral pitfalls dominate their herding tendency.
MA
ED
PT
CE
AC