The effects of herding behavior in the Greek and Dutch markets

1

The effects of herding behavior in

the Greek and Dutch markets

Student Name: Kyriakos Skalkos

Student ID: 10828451

Supervisor: Dr. Liang Zou

Master Thesis Finance (ECTS: 15)

MSc Business Economics

Track: Finance

Calendar Year: 2015­2016

2

Contents

The Dutch market……….………12

The Greek market……….………19

4. Discussion………...……….25

Further research………26

3

1. Introduction

As it is known from theoretical background, under normal circumstances, there is a linear relationship between the stock dispersion and the market rate of return. Of course, the term normal is quite subjective. In reality, there is a variety of reasons that formulate share prices and most of them are related to the decisions of human beings. The call to sell or buy a share is based on factors such as the investors’ psychology, the investors’ confidence, the investors’ perception of economic situations and use of available information. All these .calls are made by human beings who have a different temper, different character and have a different perception of how the market works. In essence, the human psychology is one of the main factors that distinguish the share prices. One sort of psychological reaction is the herd behavior, which in practice is the reaction of the masses during specific economic situations. This research discusses the issue of herding for the Dutch and the Greek market for the decade between 2005 and 2015.

Economic psychology

The economic psychology is related to the study of economic behavior, i.e. it tries to explain the behavior of economic substances in cases where there is a distance to the rational economic behavior. Economic behavior is consumer behavior which encompasses financial decision and the consequences that follow them. This sector of study is fairly new (Lea et al. 1987)

Financial acting includes money, time and the effort required to obtain goods, services and rest plus the possibility of choice between alternative products such as resource saving or wasting. In reality, all the calls that include either the choice either the trading of some substitute or of an investment which would bring future benefits are called financial decisions.

The most common factors influencing financial decisions are personal factors, cultural, occasional and in general financial factors that motivate and influence the financial decision receiving process (Lea et al. 1987). What is called rational economic behavior is the behavior during which, every subject takes such decisions in order to maximize its satisfaction taking into account the existing resources. While the classical economic theory initiates with the hypothesis that players have a benefit

4

function which they try to systematically maximize with a rational manner, the economic psychologists consider players as beings that are committed to the principles of rationalism without thinking in particular.

Economic psychology is not based on the existence of the simplified economic subject and intends to propose better models from the behavior of real people which is the outcome of scientific experiments (Lea et al. 1987). One of those examples is the so called ‘herding behaviour’ according to which the subject follows the actions of other subjects.

The effect of financial decision taking is the satisfaction and the prosperity of consumer which comes out of consumption itself. Non satisfaction can lead to complaints towards the administration or the service provider (Lea et al. 1987). Their consequences provide with some kind of experience to the subjects which help them with their future actions.

Most theories of the economic psychology sector come from scientific financing and developing biology while they are also being influenced by psychology. One recent development was the use of neuromonics for the study of decision taking by the human brain.

A number of studies on the other hand have proven that most investors behave in an unorthodox manner. Feelings such as fear, greediness, uncertainty, competition do not let subjects receive sober decisions (Lea et al. 1987). This result has led to the creation of Behavioral Finance which studies the investor behavior taking into account their ‘human side’. Behavioral Finance come in contrast to the efficient market hypothesis which consider the human being as being rational who bases its actions to two factors: risk and market efficiency. According to the decision making models which are based on such theories such as the CAPM (Capital Asset Pricing Models) and the APT model, the investor evaluates every investment taking into account the amount of risk to be taken for a specific rate of return.

Practically, experience shows that decisions are not relying on rationale and the market does not function as a well tuned clock. Behavioral Finance covers all the weaknesses in the interpretation of the Efficient Market Theory by studying and understanding the effect of the subjects’ feelings, for example greediness or

5

confidence during the decision taking process. Thus, knowing the possible mistakes the investors can commit, better forecasts and actions can be held.

One of the reasons why investors do not profit from the market is the unreasonable in some occasions feeling of overconfidence investors have in themselves. The aversion shown by the investors towards losing, combined with their tendency to keep shares that are showing losses for a long time, so that their bad choice is less apparent, leads to liquidation of the profitable rather than the loss positions. It has been calculated that profitable stocks that are liquefied tend to have greater annual return rates by 3,8% compared to the loss stocks in the coming year (Sewell ,2007). Even though Behavioral Finance is quite a new field in Finance, it has been well established and it is taught as a course throughout a number of faculties worldwide.

In order to the banking sector investing advisors to provide with credible advice to their clients who belong to the very specialized target market of private banking, and they have to know in depth the factors that influence the buying behavior of investors of investors, apart from the rationalism imposed by the financial data. As it is known, the factors that affect the market stock prices are mainly:

 Geopolitical changes and the international economic environment

 The particular circumstances of a national economy

 Factors related to the economics and the perspectives of every company participating in the stock market

Given the fact that the share prices show a give or take chaotic behavior that is influenced by the predictable and a number of non predictable variables, people not always have full knowledge of the market and cannot take advantage of all the information available (Sewell ,2007). Thus, they are limited by their personal goals, targets, habits and their cognitive capabilities. Therefore, it is accepted that there is no specific predicting model for the course of shares, but in the best case scenario, there are some general trends. The rules and reasons which the economic science is being deformed during the procedure of taking financial decisions under sentimental reactions that lead investors in taking decisions within short time lapse, contributing to what is called cognitive economy which is based on generalizations and is influenced by spontaneous judgment and intuition (Sewell ,2007).

6

2. Related Literature

Herding Behavior

The phenomenon of herding behavior is described by the tendency of investors to imitate the behavior of their counterparts in the financial market. This phenomenon, when generalized can have an impact on the overall market return (Hwang and Samon 2004). According to Christie and Huang (1995), market stress is described as the market efficiency during extreme conditions (low/high). The result of this herding phenomenon is the decrease of dispersion between the individual stock and the market return. According to Chang et al. (2000) this dispersion is increasing with a decreasing rate and when the herding effect is heavy then it tends to disappear. During times where herding behavior is apparent, investors tend to ignore their private information and follow the path shown by a large group of investors which is driven by their sentiments or instincts. According to Van Campenhout and Verhestraeten (2010) “this outcome may be a conscious or an unconscious decision which can be a cognitive bias or even a strategic move”. Bikhchandani et al. (1992) claims, that a specific moment this behavior might seem profitable for a sole investor, but if generalized it leads to inefficient market returns and non equilibrant situations.

In the general context, herding may not be irrational and can act as a medicine to high stock volatility or to bad quality information (Bikhchandani and Sharma, 2000). Interestingly enough, Henker et al. (2006) provides evidence that herding leads to mixed results when an increased price movement occurs, especially upwards and in the down market when volatility is low or when the investors’ morale is low as shown by Chang et al. (2000).

Radner (1962) and Vives (1995) question the validity of private information at times of massive stock movement and praise the value of ‘precise public information’ which can be utilized by investors to overcome. Apart from the Efficient Market hypothesis developed by Fama (1970), there have been a number of other scholars who intended to explain the impact of human behavior on market return. Shiller (1987); Bikhchandani et al. (1992); Banerjee (1992); Scharfstein & Stein (1990) were the one of the few to study the effect of massive movement towards the market trend while neglecting the value of private information. As herding behavior is not rational, as the

7 value of a share is not accurately depicted, stocks tend to be overrated; creating bubbles that may lead to material welfare loss, market inefficiency but also makes good managers get carried away to market equilibrium (Allen et al., 1992). Van Campenhout (2010) claims that investors disseminate quality information because that they cannot apply them in rational asset pricing models such as the CAPM. According to Truman (1994), correlated trading also leads to non optimal use of resources that would have led to the actual pricing of the assets. Olsen’s study (1996) revealed that for the years 1985-1987, the vast majority of managers was prone to following a correlated forecast behavior due to the uncertainty in predicting returns and as this uncertainty increased, the correlation became even stronger. However, there have been cases where investors would deviate from the optimal choice due to risk miscalculation associated with the assets combined with an optimistic behavior (Hirshleifer & Teoh, 2003). Similar results were indicated by Ciccone (2005) who revealed that 40% of the investors showed a deviant, more optimistic behavior from the optimal choice, leading to 20% higher forecast errors.

However, there have been cases reported where this deviant behavior has been proven the most rational decision under specific circumstances. There have been cases where investors decided to follow an up market trend which may entail an optimistic view, and have benefited from that choice (Bange,2000). On the other hand, herding behavior has led many investors into trusting over-valued stock while they performed well in the short run, while this choice has been proven erroneous in the long run (Shiller, 1989). The herding phenomenon applies for investors who are modest and do not like to stand out in the crowd (Scharfstein and Stein 1990). Nevertheless a deviant behavior is not the socially rational path to follow, but it can be detrimental to the investor’s career in the labor market, leaving the herding behavior to seem like the only rational choice. Investors with narrow access to information or do not have the cognitive tools to perform sophisticated technical analysis tend to monitor and follow the steps of more experienced investors and to exit the market upon the breaking of bad news (Bikhchandani and Sharma 2001). Chiang et al (2010) explain the negative effect of herding behavior as situations in which only arbitragers profit and, when the share is later priced investors lose out. As the issue compounds on itself, it can create huge inefficiencies which further worsen the situation. If the market fails to correct these mispricings, this can induce considerable shock to the economy. Investors tend

8 to show a herding behavior usually when the market is down compared to high return shares (Fu 2010). Welch (1999) explains the way managers decide to follow the herd in order to protect their reputation, taking into account their reluctance to short losses. The lack of valuable information is also considered as an explanatory factor for this behavior, since standing out is a risk which managers are not willing to take (Welch, 1999). On the other hand, in case market returns are not optimistic then managers tend to share the blame with the economy going bad and other types of excuses. Of course, following the herd is quite the rational thing to do under extreme conditions (Bikhchandani and Sharma 2001). Bikhchandani and Sharma (2001), give the example of a bank run when the rumor spreads about the lack of liquidity in the banking system, where the herding behavior is the rational thing to do. In such cases, private information plays little or no role. On the other hand, when information -is blurry and cannot be verified with certainty, investors assume that the leader’s information is of better quality, mainly because of the prestige of the position and the available information from the source. Once the leader is followed by analysts, especially reliable ones, a trend is created and smaller investors tend to follow this trend. This trend might have an effect on smart investors who have more valuable information, which proves that psychology can outrun valuable information and carry away even the smartest of investors who have benefited from their source of information during the past.

The outcome of Christie and Huang’s study in 1995 provides support to all the above, i.e. there is no general conclusion of herd behavior and the outcome can have a mixed significance. Herding appears much more often in emerging markets and not that often in the already mature markets. Even in emerging markets there are different levels of herding behavior between different stocks and individuals. Furthermore, investors tend to herd more in a down market than in an up market (Gebka et al 2009); extreme markets and rumors can become a driving force of correlated trading. Shiller and Pound (1989) and Christie and Huang (1995) discriminate the cases of herd behavior among institutional investors.

Christie and Huang (1995) have developed a fairly simple model that detects herding behavior, gives rise to the nature of the phenomenon, gives scholars a notion of the cases where herding is going to lead to dispersion in the stock returns while the stocks become more volatile and are driven away from their underlying value. The

9 dispersion of the cross sectional standard deviation (CSSD) is a measure that acts as an alert of possible herding behavior. When there is a correlation between the market returns and the share returns then the difference between the two is going to approach zero. When no herding occurs then the dispersion has a higher mean. This variable was employed by Hwang and Salmon (2004) who, using a dataset containing 10 years of stock returns from Korean and American markets, found evidence of herding behavior in both bull and bear markets. They used the Fama and French (1993) factor model rather than the cross sectional returns of the individual stocks, since according to scholars, the cross sectional average return does not describe the risk factor as well as it is described by the CAPM. The results have shown insignificant herding during the down market, compared to the times when the market was up. Chang Chen and Khorana (2000) employed the same CSAD model, but this time measuring the average dispersion of the shares compared to the market returns. Apart from this fact, the model employed was non linear compared to the linear model proposed by Christie and Huang (1995). The findings propose a linear model when there is no herding behavior in the market. During herding times the dispersion was increasing with a decreasing rhythm and in the existence of herding it was heavily decreasing. Caparelli et al. (2004) repeated the Christie and Huang model for the Italian market during 1988 and 2001, with numbers showing that there was no herding behavior in the Italian market. Economou et al (2010) performed an analysis on 4 different stock markets in the Mediterranean countries (Spain, Portugal, Italy and Greece) from 1998 to 2008 with no evidence of correlated trading in all countries, except Portugal where the phenomenon was present. Caparrelli employed models both the Christie and Huang (1995) and Cheng et al. (2000) model to check for the extreme conditions of the Athens Stock exchange bubble market in 1999. Kallinterakis and Lodetti (2009) repeated the same procedure for the Montenegro stock exchange market and provided no proof of herding for the period between 2003 and 2008. In total, there is no conclusion someone could jump into for the case of herding behavior, since results have been mixed for both normal conditions and extreme conditions.

Economou (2011) et al performed a study on the markets of Portugal, Italy, Greece and Spain, 4 markets, the economies of which have been subjected to severe crisis during the past years. Results show more intense herding phenomena mainly at the economies of Greece and Portugal for the period between 1999 and 2008. According

10 to the study, there is no clear evidence of herding for the case of the Spanish market while the results are mixed for the case of Portugal. Another crucial finding of this study is that portfolio diversification for these markets are not offering the expected benefits compared to other markets.

3. Data ­ Methodology

The setting

The purpose of this study is to compare possible herding behaviors between the Dutch and the Greek stock markets. The model employed is the one by Chang, Cheng and Khorana (2000). The model suggests that the cross sectional average deviation (CSAD) is in line with the Capital Asset Pricing Model, i.e. it follows a linear relationship with respect to the return rates. This linearity is tested by the regression model

The CSAD term is calculated as the average deviation of every stock in the market from the total market return.

Where is the return rate of stock i at time t, and is the market return at time t.

Under rational circumstances the coefficient is positive, reflecting the association between return rate and standard deviation (risk), while the term should be 0. When herding occurs is expected to be negative indicating that the average deviation is either accelerating with a decreasing velocity or decelerating in total. A positive indicates that market movements cause more dispersion in stock returns than expected under rational pricing. A positive indicates a divergence in investment behavior.

Chang, Cheng and Khorana (2000) also examined for possible asymmetric herding for the two cases when the market had profit and when it encountered loss. Thus, the initial model becomes

11

for the case of the profitable market and

For the case of the market with losses where and are the market returns of a profitable and a lossy market respectively.

Initially, the model of the cross sectional standard deviation employed by Christie and Huang (1995) is employed. According to this model, investors tend to ignore information about the market when large market price movements occur and base their positions on the market. If herding occurs the individual stock returns tend to follow the returns of the market. Christie and Huang (1995) calculate the cross sectional standard deviation using the formula

Finally, the regression analysis is run with respect to two dummy variables and

which take the value 1 whenever one stock belongs to the upper/ lower

extreme of the market rate of return for the 10th, 5th and 1st percentile. In essence the regression analysis formula will be of the form

Rational asset pricing suggests that an increase in market returns will be associated with an increase in the cross sectional standard deviation of stock returns given the exposure of individual stock returns to the market portfolio. In contrast, in the presence of herding, CSSDt is expected to increase at a decreasing rate or it might even fall if herding is severe. In the absence of herding, during extreme market movements, individual stock returns exhibit high volatility and the estimated coefficients bL and bU are positive. In contrast, negative estimates of coefficients and are consistent with the presence of herding behavior.

This study also focuses on two periods, before and after 2007 in order to investigate the effect for the two markets before and after the 2008 economic crisis.

12

The data available lie between 1/08/2005 and 31/07/2015

The Dutch market

Table 1

Descriptive statistics for the case of the Dutch market

CSAD Rm CSADup CSADdown CSSD

Max 0,18 0,29 0,29 -0,22 0,75

Min 0,00 -0,22 0 -0,01 0,05

mean 0,08 0,00 0,01 -0,02 0,09

σ 0,01 0,03 0,03 0,03 0,03

Figure 1

The Dutch market return

It can be deduced that the market return has encountered great profits and great losses, i.e., the value of the market return is a volatile and fluctuating market, especially during the first years of the sample.

-0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4

Market return

13

Table 2

The results of herding behavior according to Christie and Huang (1995)

Beta t sig Adj R^2 D-W

10% .093 -4.803 .000 .03 1.93 -.139 7.179 .000 5% .081 -4.139 .000 .012 1.937 -.076 3.919 .000 1% -.016 -.800 .424 0 1.92 .013 .675 .500

As we can see the results failed to reject the null hypothesis that for 1% extreme values, there exists some kind of herding behavior. As we notice, the variable Du is significantly different for both the 10% and the 5% sample and has a negative sign which means that for the 10% and the 5% percentiles. The value of CSSD is significantly greater than for the rest of the sample while its value is significantly lower than for the rest of the sample which in essence. Therefore, identifies some sort of herding behavior for the upper 10% and 5%.

Figure 2 indicates the average deviation over the years of the study for the Dutch market. There are noticeable fluctuations between different years which indicate significant differences between the values of CSAD for different years.

14

Figure 2

The course of the cross sectional average deviation over time in the Dutch market

As we can see, during the whole 10 years, there is only clear evidence of herding behavior during the 2008 in cases where the value of the average deviation is converging to the value of 0.

The very first model that is run encompasses the total Dutch market for all the period between 2005 and 2014. The results of this analysis indicate that the relationship between the market return and the average deviation is not linear as prpoposed by theory. The existance of a relationship between CSAD and market return is verified by information provided by the ANOVA statistics

F(2. 2608)= 157.45. p=0.000<0.05

The p value of the hypothesis that there is no relationship between the dependent variable (CSAD) and the two independent variables is less than the upper limit of 0.05 and thus, this null hypothesis is rejected. Table 3 indicates the coefficients of the relationships betweed CSAD and the two market rate variables:

0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2

CSAD

15

Table 3

Coefficients of the regression analysis for the total Dutch market

Coefficientsa

Model Beta t Sig.

1 (Constant) 421.558 0.000

.337 17.743 .000 -.069 -3.632 .000

A very first glimpse indicates that the relationship between the average deviation and the market return is not linear since we fail to reject the null hypothesis that there is no relationship between CSAD and the squared market return. Thus. the CAPM model does not apply. The negative sign in the relationship indicates that there is some kind of herding involved in the relationship since returns are either growing with a decelerated pace or they encounter losses.. On the other hand the positive sign of the coefficient verifies the positive relationship between deviation and risk. What can be deduced for the table above is that as the CSAD measure increases, the market return increases, but with a delayed pace, as it is denoted by the sign of the relationship between the term and the CSAD term.

For the case of the up and down market. results are similar. Table 4 shows the coefficients of the analysis for the up market:

Table 4

Herding behavior on the Dutch up market

Coefficientsa

Model Beta t Sig.

1 (Constant) 111.898 0.000

.771 7.036 .000 .773 7.056 .000

16 Once again it is proven that there is no validation of the CAPM model since we failed to reject the null hypothesis that there is no relationship between the average deviation and the squared market return. The positive sign shows that the market encounters greater dispersion and no herding behavior is encountered. In practice, this means that following the up market trend offers a great opportunity of arbitraging, i.e. taking advantage of stock that may be overrated. The dispersion grows with an accelerating rhythm, offering good opportunities of profit for the people who decided to follow this trend

For the case of the down market we employed the absolute values of the rates of return and run the regression again.

Table 5

Herding behavior on the Dutch down market

Coefficientsa

Model Beta t Sig.

1 (Constant) 105.083 0.000

.687 6.173 .000 -.680 -8.105 .000

Results indicate a more intensified herding behavior in absolute numbers which means that investors tend to liquefy stocks much more easily when the market is down compared to buying stocks when the market is up. The greater the risk in a stock (expressed by the CSAD variable) the faster the market rate drops and the faster investors liquefy their positions. In practice, as the market return grows in slower pace, the distance between the share prices and the stock market is closing down and in essence, herding behavior exists. Taking into account the fact that is negative, it can be concluded that the ones who decided to invest in the down market hoping for some underrated stock for a long time, encountered severe losses. The procedure took place by examining two different periods, one before 2007 and one 2007 onwards. With this way, the effect of economic crisis on investor behavior is going to be examined. Theory advises that during times of recession, herding

17 towards liquidating positions is much more intense - which heavily impacts the market return. For the period 2005 and 2006 the results do not indicate some herding behavior since the rational rule of CAPM applies:

Table 6

Herding behavior test before the financial crisis

Coefficientsa

Model Beta t Sig.

1 (Constant) 215.381 0.000

-.014 -.263 .793 .235 4.465 .000

As the statistics show, the relationship between risk and return is linear. As the return increases, so does the risk. Share prices can be predicted using the CAPM model, since the relationship between the market rate and the dispersion of stock is linear. The relationship also shows that as the market return grows, so does the dispersion from the market for single stocks, while this dispersion is less when the market has a small return. On the contrary, the relationship deviates from the linear rule from 2007 onwards.

Table 7

Herding behavior test during the financial crisis

Coefficientsa

Model Beta t Sig.

1 (Constant) 440.444 0.000

-.137 -8.097 .000 .614 36.378 .000

According to the statistics, herding made its appearance during the financial crisis, which is indicated by the negative sign in the relationship between CSAD and the square of the market return. The p value for the relationship between the

18 term and the CSAD means that we fail to reject the null hypothesis that the relationship between the cross sectional average deviation and the market return is linear. Furthermore, the negative sign of the relationship between the CSAD dependent variable and the term. Herding behavior denotes the probability of Dutch and foreign investors in that specific market decided to follow the market trend, not to follow instincts, not to take risks, and to follow the crowd. This relationship proves that either the market increases in decreasing rhythm or it decreases. In the case of the Dutch market, one can say that the crisis has had an effect on the investors’ psychology, and investors seemed to follow the market

Table 8 examines the possibility of herding behavior per year in the Dutch market. As it can be seen there isn’t any specific pattern that follows the herding behavior per year.

Table 8

Herding behavior in a yearly basis

Coeff t Sig 2005 .057 .612 .542 2006 -.055 -.840 .401 2007 -.166 -2.862 .005 2008 -.083 -1.642 .102 2009 -.320 -5.905 .000 2010 .025 .545 .586 2011 -.071 -4.403 .000 2012 -.008 -.290 .772 2013 .205 5.709 .000 2014 .090 3.005 .003 2015 -.060 -1.190 .236

As it can be noted, during the economic crisis, the most negative herding behavior was noted during 2007 and 2009, with the effect making its appearance again in 2011.

19 During 2013 and 2014, a significant positive relationship appears between the square market return and the average deviation which means that stocks show greater dispersion and investors seem to diverge instead of converging (showing and anti-herding behavior). In essence, this means that during 2013 and 2014 investors had a really good opportunity to arbitrage the market by taking advantage of possible overrated stock. In essence, the Dutch market, during the years 2013 and 2014 seems to be offering great opportunities to investors who decided to take their chances in this market.

The Greek market

Having performed an analysis on the Dutch market, we proceed with the analysis of the Greek market. Figure 3 is a plot of the overall market return in the Athens stock exchange. The outcome of this result is a market that is quite volatile, due to the crisis phenomenon of the Greek economy, offering investors the opportunity for severe arbitraging but with a high amount of risk. Taking into account the descriptive statistics of the market return, it can be said that a long term investment in the market pays no return, with a significant amount of standard deviation, a fact that categorises the Athens stock exchange into the inefficient markets.

Figure 3

The plot of the average market return in the Athens stock exchange

-0.25 -0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.2 0.25 0.3

Market return

20 For the case of the Greek market, the CSSD is calculated initially and its fluctuations are presented in figure 4, throughout the years, while table 9shows the descriptive statistics.

Figure 4

CSSD over the years for the Greek market

Table 9

The descriptive statistics of the model variables

CSAD Rm CSADup CSADdown CSSD

Max 0.18 0.29 0.29 -0.22 0.29

Min 0.00 -0.22 0 -0.01 0.01

mean 0.08 -0.00 0.01 -0.02 0.04

σ 0.01 0.03 0.03 0.03 0.17

The results of the model employed by Christie and Huang (1995) for the case of the Greek market indicate that there is significant difference for the cross sectional standard deviation between the market returns of the 10%, 5% and 1% upper and lower extreme values and the rest of the market values.

-0.25 -0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.2 0.25 0.3 Series1

21

Table 10

CSSD model for the Greek market

Coeff t Sig 10% lower .263 14.074 .000 upper .166 8.874 .000 5% lower .169 8.833 .000 upper .150 7.883 .000 1% lower .169 8.833 .000 upper .150 7.883 .000

In practice, this means that there are extreme positive and negative market returns that show significant difference to the central tendency of market return. In essence, herding behavior is identified for the lower market, while the upper market returns indicate significant dispersion of stock prices from the market return.

We then proceed to the application of the CSAD model. For the total market, it can be said that the model is significant at the 5% confidence level, since

F (2, 2608) = 10190, p=0.0.00>0.05

The coefficients of the relationship show a critically significant relationship between the value of the CSAD and the market rate of return, making the relationship between the two variables linear, which means that share prices can be valued using linear modeling. On the other hand, no herding behavior is observed since there is no significant relationship between the square market return and the cross sectional average deviation.

22

Table 11

Herding behavior of the Greek market

Coefficientsa

Model Beta t Sig.

1 (Constant) -54.465 0.000 .690 -2.698 .007 -.046 -139.731 .000

The results indicate a negative relationship between the market return and the cross sectional average deviation, showing that as the market rate grows/reduces, the cross sectional average deviation reduces/grows. In essence, as the market rate reduces, the greater the dispersion of the returns in the stock market.. Herding behavior is existent throughout the market since the square market return coefficient is associated significantly with the dispersion of data in a negative manner

For the up Greek market, the results are displayed in table 12. No herding effect takes place in the upper market since the beta coefficient of the term.

Table 12

Herding behavior of the Greek up market

Coefficientsa

Model Beta t Sig.

1 (Constant) -23.873 0.000 .807 40.464 .000 .155 7.749 .000

As we see, herding behavior is non-existent in the upper Greek market. The relationship between the cross sectional average deviation and the market return is not linear since there is a significant relationship between the square market return and the CSAD term. Since the upper market deals with positive returns, the market return is growing with a positive rhythm. Since the value of the market return is positive, the

23 difference between the stock returns and the market return is positive and growing with a positive pace, thus the market does not follow herding behavior and following the upper market gave the opportunity for arbitrage for the investors.

For the lower market, the results of the study indicate the existence of herding. As it was noted, during times when the market is lossy, investors were expected to follow the market and thus the distance between the market return and the share return closes down. This expectation is confirmed, since the association between the square market return and the cross sectional average dispersion is significantly negative.

Table 13

The lower market coefficients for the Greek market

Coefficientsa

Model Beta t Sig.

1 (Constant) -23.911 0.000 -.198 -7.140 .000 .706 25.437 .000

The next regression run is with respect to the Greek crisis. A dummy variable is created which takes the value 0 for all dates before 2010 and the value of 1 for the rest of the dates. For all the dates before 2010 we can see that herding behavior was also intense, with stock returns either ascending with a decelerating mode or descending.

Table 14

Herding behavior before crisis

Coefficientsa

Model Beta t Sig.

1 (Constant) _{-.135 -4.867 0.000} .713 33.734 .000 -.243 -11.503 .000

24 The results show a herd behavior during the Greek crisis since the beta coefficient is negative for the term. The negative association denotes an uncertainty among the investors leading to follow the market trend. The highly volatile market return, as it appears on the depiction of the market return over time, leads to lower dispersion of share returns.

Table 15

Herding behavior after Greek crisis

Coefficientsa

Model Beta t Sig.

1 (Constant) 0.80 2.970 .003 -.780 31.911 .000 .148 6.071 .000

Again, for the Greek market during the years of the Greek financial crisis, there is a sign of herding behavior is obvious since the coefficient of the regression between the independent variable CSAD and the dependent variable is negatively associated. This means that the dispersion gap is closing down and that the market return is either growing with a negative rhythm or declining.

The last type of analysis performed on the Greek market is going to be the per annum analysis. The effects of the economic crisis are going to be more visible through such type of test.

25

Table 16

Herding behavior in a yearly basis

t p 2005 -0,37 -6,94 0,00 2006 -0,40 -5,17 0,00 2007 -0,25 -4,02 0,00 2008 0,38 5,36 0,00 2009 0,13 1,28 0, 2 2010 -0,23 -3,99 0,00 2011 -0,08 -2,14 0,03 2012 -0,17 3,19 0,00 2013 0,17 4,34 0,00 2014 -0,90 -32,97 0,00 2015 -0,86 -84,87 0,00

The results provide support to the previous research performed by Economou et al. (2011). Heavy herding periods occur the final 2 years of our analysis indicating the uncertainty in the country’s economics. The investors in the Greek market during 2014 and 2015 either liquefied their positions due to the volatile nature of the market or have herded the market in belief that the situation is going to turn around and they will make up for their losses.

4. Discussion

The results of the analysis show that the Dutch investors are more prone to herd behavior, compared to the investors of the Greek market. Since herding behavior can have a number of reasons to happen, no clear explanation can be given of what caused this behavior. The absence of herding behavior in the Greek market can be interpreted as a belief by the investors that the local stock market cannot act as a

26 safety net during times of recession, while on the other hand the growing dispersion between share returns and market return gives the opportunity for arbitraging.

The difference in mentality can be seen in the period after the crisis for the Dutch investors who have utilized the market return as a safety net instead of investing and looking for arbitrage opportunities. On the other hand, the investors in the Greek market showed no trust towards the total market return, by not following neither the normal market nor the upper market. This practically means that the Greek market may be more vulnerable to arbitraging compared to the Dutch economy.

The results also prove the point made by Economou et al. (2011) which shows that during the down market people tend to herd the market rate of return, maybe because of belief that the situation is going to turn around and the market is going to return into becoming profitable.

Finally, the per annum analysis of the Greek market has indicated significant herding during the whole period, with some exceptions during the years 2008, 2009 and 2013. The heaviest herding appears during the years 2015, 2012 and 2010, corresponding to years where significant events have occurred in the country, such as elections, entering the bailout programs etc. On the other hand, the Dutch market shows good dispersion with respect to the market return, and the positive square market return association signifies that the total dispersion is growing with accelerating rhythm, offering opportunities for increased profits.

In general, this kind of analysis is not very enlightened about the reasons behind the herding behavior.

Further research

The proposition for further research would be to include the market and the stock volatilities in the equation that measures the share return dispersion. That way, researchers are going to have a better viewpoint on the reasons that have caused this herding phenomenon.

27

References

Bae, K.H., Karolyi, G.A., Stulz, R.M., 2003. A new approach to measuring financial market contagion. Review of Financial Studies 16, 717–764.

Baur, D., 2006. Multivariate market association and its extremes. Journal of International Financial Markets, Institutions and Money 16, 355–369.

Bikhchandani, S., Hirshleifer, D., Welch, I., 1992. A theory of fads, fashion, custom and cultural change as informational cascades. Journal of Political Economy 100, 992–1026.

Bohl, M.T., Brzeszczynski, J., 2006. Do institutional investors destabilize stock prices? Evidence from an emerging market. Journal of International Financial Markets, Institutions and Money 16, 370–383.

Brunnermeier, M., 2001. Asset Pricing Under Asymmetric Information. Oxford University Press.

Brunnermeier, M., 2009. Deciphering the liquidity and credit crunch 2007–2008. Journal of Economic Perspectives 23, 77–100.

Caparrelli, F., D’Arcangelis, A.M., Cassuto, A., 2004. Herding in the Italian stock market: a case of behavioral finance. The Journal of Behavioral Finance 5, 222–230. Chang, E.C., Cheng, J.W., Khorana, A., 2000. An examination of herd behavior in equity markets: an international perspective. Journal of Banking and Finance 24, 1651–1679.

Chiang, T.C., Zheng, D., 2008. An empirical investigation of herd behavior: evidence from international perspective, Working paper, Drexel University.

Chiang, T.C., Zheng, D., 2010. An empirical analysis of herd behavior in global stock markets. Journal of Banking and Finance 34, 1911–1921.

Chiang, T.C., Li, J., Tan, L., 2010. Empirical investigation of herding behavior in Chinese stock markets: evidence from quantile regression analysis. Global Finance Journal 21, 111–124.

28 Christie, W.G., Huang, R.D., 1995. Following the pied piper: do individual returns herd around the market? Financial Analysts Journal 51, 31–37.

De Long, B.J., Shleifer, A., Summers, L.H., Waldmann, R.J., 1990. Positive feedback investment strategies and destabilizing rational speculation. Journal of Finance 45, 379–395.

Demirer, R., Kutan, A.M., 2006. Does herding behavior exist in Chinese stock markets? Journal of International Financial Markets, Institutions and Money 16, 123– 142.

Demirer, R., Kutan, A.M., Chen, C., 2010. Do investors herd in emerging stock markets? Evidence from the Taiwanese market. Journal of Economic Behavior and Organization 76, 283–295.

Devenow, A., Welch, I., 1996. Rational herding in financial economics. European Economic Review 40, 603–615.

Gleason, K.C., Mathur, I., Peterson, M.A., 2004. Analysis of intraday herding behavior among the sector ETFs. Journal of Empirical Finance 11, 681–694.

Hirshleifer, D., Teoh, S.H., 2003. Herd behaviour and cascading in capital markets: a review and synthesis. European Financial Management 9, 25–66.

Hodrick, R.J., Prescott, E., 1997. Post-war U.S. Business cycles: an empirical investigation. Journal of Money, Credit, and Banking 29, 1–16.

Hwang, S., Salmon, M., 2004. Market stress and herding. Journal of Empirical Finance 11, 585–616.

Karolyi, G.A., Stulz, R., 1996. Why do markets move together? An investigation of US–Japanese stock return comovements. Journal of Finance 51, 951–986.

Lea S.E.G, R.M Tarpy, Webley P., “The individual in the economy: A textbook of economic posychology, New York, NY,US, Cambridge University Press

Morelli, D., 2010. European capital market integration: an empirical study based on a European asset pricing model. Journal of International Financial Markets, Institutions and Money 20, 363–375.

29 Morris, S., Shin, H.S., 2003. Global games: theory and applications. In: Dewatripont, M., Hansen, L., Turnovsky, S. (Eds.), Advances in Economics and Econometrics. Cambridge University Press.

Nofsinger, J.R., Sias, R.W., 1999. Herding and feedback trading by institutional and individual investors. Journal of Finance 54, 2263–2295.

Pedersen, L., 2009. When everyone runs for the exit. International Journal of Central Banking 5, 177–199. 460

F. Economou et al. / Int. Fin. Markets, Inst. and Money 21 (2011) 443–460 Shefrin, H., 2000. Beyond Greed and Fear: Understanding Behavioral Finance and the Psychology of Investing. Oxford University Press.

Shleifer, A., 2007, Inefficient Markets. Oxford University Press.

Sewell M. (2007) “ An Introduction to Behavioral Finance” http://www.behaviouralfinance.net/behavioural-finance.pdf

Tan, L., Chiang, T.C., Mason, J.R., Nelling, E., 2008. Herding behavior in Chinese stock markets: an examination of A and B shares. Pacific-Basin Finance Journal 16, 61–77.

Welch, I., 2000. Herding among security analysts. Journal of Financial Economics 58, 369–396. Wermers, R., 1999. Mutual fund herding and the impact on stock prices. Journal of Finance 54, 581–622.

Zhou, R.T., Lai, R.N., 2009. Herding and information based trading. Journal of Empirical Finance 16, 388–393.