Abstract

This paper provides an overview of the existing empirical and theoretical research conducted on herd behavior in financial markets. The basic aim of the paper is to look at what is specifically meant by herd behavior, to examine whether such behavior exists in the markets, to explore the success of existing studies that have investigated the concept of herd behavior, possible causes of herd behavior and lastly to study the effect of herd behavior on financial markets.

I-Introduction:

Modern Portfolio Theory developed in the 1950s through the early 1970s is a theory ofinvestmentwhich tries to maximizereturnand minimizeriskby carefully choosing differentassets. It was considered an important advance in the mathematical modeling of finance. It is a mathematical formulation of the concept ofdiversificationin investing, while selecting a collection of investment assets that have comparatively lower risk than any individual asset. MPT is based on various assumptions that are not true in reality. It assumes that investors are risk averse and prefers less risky assets. Thus, according to this assumption an investor will take on increased risk only if he is compensated by higher expected returns. Although MPT is widely used in practice in the financial industry but it has been widely challenged by fields such asbehavioral economics, and many companies that have gone bankrupt in various financial crises after using variants of MPT.

It further assumes that the investor’s risk / reward preference can be described via aquadraticutility function. The effect of this assumption is that the investor only considers expected return and thevolatility(i.e.,meanreturn andstandard deviation). Since then, much theoretical and practicalcriticismhas been pointed against it.

These traditional frameworks of studies have introduced the concept of rationality in financial markets. According to the rational models, agents always make best choices and they always update their beliefs/action as they receive new information. But the new studies have confirmed that the assumptions used in the old studies do not explain the concepts of aggregate stock market, the cross section of average returns and individual trading behavior.

Behavioral finance is a new development in the financial markets that has emerged after the difficulties faced by the use of traditional concepts and assumptions. It picks up where modern portfolio theory leaves off by describing how investors actually behave, rather than how they should behave.

The study on behavioral finance is a new emerging field which explores some psychological issues that may have an effect on the financial markets .According to this, not all agents are fully rational as opposed to classical studies. The study on Behavioral Finance has shown that Psychology and irrational behavior of participants do exist in financial markets. Many irrational reactions in financial markets can be explained by this. Without contribution of behavioral finance, certain factors and issues such as irrational reactions in the financial markets cannot be understood. An analysis has shown that how behavioral factors can help investors in order to avoid mistakes and can also be useful in finding investment strategies and improving asset performance.

According to this new concept, psychology really matters and plays an important role in financial decisions and individuals do not always perform as perfectly as they are portrayed in neoclassical literature in financial theory. Financial bubbles are prime examples of the inefficient markets.

It also makes certain aspects of financial markets more realistic by relaxing unrealistic behavioral assumptions presented in the classical models.

Purpose of the Study:

The basic objective of this paper is to study herd behavioral in financial markets that is also related to behavioral finance and psychology of agents and to explore whether such behavior causes price jumps and deviations of prices from their fundamental values and also to see if herd behavior can partially explain persistent price spikes and crashes and the resulting crazy market behavior. Herd behavior in financial markets occurs when a few early, perhaps incorrect, movements of traders induce others to ignore their private information after observing behavior of others and following the majority action, thus causing discontinuous price jumps in one direction or price deviations from the actual asset’s value.

This paper investigates the research conducted on different aspects of herd behavior by investigating the work of various authors in different years.

The paper proceeds as follows. Section II presents year wise Review of the Literature. Conclusions will be discussed in Section III.

Literature Review

In the recent years there has been increasing interest shown by the researchers on herd behavior in financial markets. After the financial crises of 1990s, many scholars have suggested that herd behavior could be the reason for excess price volatility and resulted failure of financial systems.

In this paper, literature review of the articles written on herd behavior by various authors in different years has been presented.

Year 1985-Year 1989

David S. Scharfstein & Jeremy S. Stein (1988) presented a theoretical model of herd behavior on investment. According to these authors Herd behavior is a rational behavior which can arise in a variety of context attempted by Managers in order to enhance their reputation as decision makers. Mangers mimic the investment decisions of other manager ignoring their private information about the effectiveness of different available alternatives. The study was based on assumptions by considering two types of mangers: smart who receive informational signals about the values of investment and dumb who rely purely on noisy signals. The authors have analyzed the manager’s investment decisions in the labor market by evaluating the different types of signals adopted by smart and dumb managers and by considering different objective of managers. It was found that there are multiple equilibrium after comparison with efficient investment decisions as well as reputation concerns of the managers by analyzing a series of propositions. It was also found that there can be more than one outcome associated with given economy-wide information set. The model was applied to corporate investment, the stock market and decision making within the firms. Herd behavior is found persistent in corporate investment when manager feels the investment to be excessive and is more positively disposed towards it when all his colleagues are doing the same. In the stock markets, excessive volatility may provide partial explanation for excessive stock market volatility which may results in increasing stock prices shocks. Group decision making may have a certain inherent limitations in decision making and may also offers some new insight on benefits of group decision making which may result in wide range of information.

Year 1990-Year 1994

The basic concept and theoretical research on herd behavior was introduced in these years.

Abhijit V. Banerjee(1992) developed a model to analyze the rationale behind the decision making with respect to herd behavior in which people try to do what everyone else is doing by ignoring their own information. This is rational for them because it reflects that these other decision makers may have some information they do not have.

In case of herd behavior people usually follow the decision taken by the first person. In such a case, the second person ignores his own information and joins the herd and causes a negative externality on the rest of the population. In case the second person had used his own information than this would have encouraged the rest to use their own information instead of following what others are doing.

A very simple model has been used in the study for identification of herd externality. The decision making in this model is sequential. The entire process of decision making is based on the history of the past decisions and their own signal if they have one. The random signals that the first decision maker have will determine the direction where rest of the crowd moves from then on. This may result in excess volatility in money asset markets and frequent and unpredictable changes in trends. Also, in this model there is no distortion in incentives as agents capture all the returns generated by their choice.

The results reveal that the equilibrium pattern of choices become inefficient because they hardly provide any useful information to the followers by making each person’s decision less responsive to their own information and hence less informative to others. This makes herd behavior less desirable from the social point of view as well.

Year 1995-Year 1999

A number of useful articles have been written on herd behavior by eminent scholars in these years.

Robert J. Shiller (1995) presented an article on herd behavior. Their article was about group thinking, how people who interact with each other regularly tend to behave and think similarly. The decision by one in a group leads the rest of the individuals to follow suit. This particular phenomenon is known as the herd behavior e.g. the trend of political beliefs on policy issues show herd behavior as well as the trend in upward moving stock. There is a psychological motivation behind people to think and behave similarly in a group. The models of Abhijeet Banergee and Sushil Bikhchandani are applied in this paper for referencing.

For understanding herd behavior it is helpful also to consider theories of information, theories that represent each group as reacting to information set common to that group. The herd behavior is important because it considers matters for which limitations of time and natural intelligence prevent each individual from individually discovering all relevant information.

There are two approaches to understand why groups at different places or time have access to different information sets. One approach is represented by models in which people acquire information by observing actions of others in their specific group, the informational cascade. In this case the first individual follows his own signal; those who follow may rationally ignore their own signals, deciding that these signals are dominated by the information revealed by their predecessor’s decision. This is known as the herd externality, which results from copying others and hiding one’s own information.

The second approach highlights the study of the mechanism of transmittal of information within groups, using the conversation analysis.

Both models show acquiring information from observing the action of others individuals in their group who precede them in a sequence.

Interpersonal communication involves group members to exchange information among themselves, they must promote a collective memory of important facts, common assumptions and conventions. Human behavior involve for an idle free flowing exchange of ideas and thoughts known as conversation. The flow of conversation serves to exchange a wide variety of information and also to reinforce memories of pieces of information to be held in common by the group.

Many of the failures of human judgment that fall under the rubric of herd behavior might be traced to the limitations imposed on human thought and memory by these patterns of communications. Opinions on matters such as how much we should diversify our portfolio and hedge risks vary through time and across groups.

The difference in mass behavior across groups may be due more differences across groups in the nature of information transmission. Different groups have in different times have different tendencies in terms of conversation patterns as well as circumstances promoting informational cascade; to transmit certain kind of information and thereby place it in collective memories.

Group may differ in complicated ways in terms of what may be called informational cascade facilitators. Conversation patterns may also vary across groups in terms of habits of revealing sources of information. A seemingly inessential group tendency to reveal or not reveal where one herds a seemingly improbable story may make an important difference in the kind of informational cascades that develop in conversation.

Thomas Lux (1995) presented their paper to formalize herd behavior and contagion of opinions in speculative markets in order to observe overall patterns of markets dynamics implied by contagion of opinion and behavior. The herd behavior among speculators exists and can be explained as an outcome of different factors. First, they can be acting rationally, second, in order to draw information from others and third, the reputational concerns may push managers to follow the crowd.

The paper formulated a basic cyclical mechanism in which overvaluation or undervaluation of assets occurs as reactions of speculators due to deviations from the equilibrium.

Firstly, the model used in this study determines the behavior of those traders who do not have access to the information of fundamentals and as a result they have to rely on the actions of others.

Secondly, the paper analyses the condition under which contagion may lead to the existence of bubbles which is a stationary state where actual price is above or below the fundamental values. At these stationary states, the market clearing price is either above or below the actual price. The price disturbance destabilizes the fundamental equilibrium and as a result the traders react to both the positive changes as well as on the resulting price increase. These bubbles grow up and eventually burst and such stationary sate does not last forever.

Jay Pil Choi (1997) highlighted two factors in their article, the herd behavior and the network externalities. Their article also states the technology adaptation process in which the effect of informational spillover interacts with network externalities.

In case of herd behavior, the history of the past decision made by predecessors, but not their basis is observable to each new decision maker. Herd behavior can naturally arise as an outcome of rational decision maker because each agent suspects that the predecessor had private information and tries to free-ride on it.

There are many economic activities that produce by products in addition to their directly intended benefits. The market behavior results in valuable information coming out, for e.g. consumers who purchase a drug of unknown quality receive information about its true quality as well as its health benefits. Thus this results in enabling agents to learn from the experiences of others. When the payoff of an agent is jointly determined by everyone else’s action the effect of informational spillover needs to recognized even at the level of individual decision making.

The authors have used The Sequential-Choice Model which is a model of technology adoption to analyze the interplay of international externalities and payoff interdependency. In order to avoid fear of being standard, each decision maker is concerned with the number of future subscribers who can act upon new information that decision can create.

The heard behavior of subsequent users influences the adoption decision of the agent who moves first. Even for risk neutral users there is bias in the first choice toward a safe technology. Such individually risk averse behavior can induce the first adopter to make suboptimal choice viewed in terms of collective welfare. As a result the first investors decision is the optimal one, and the inefficiency arises when the other follow. In contrast the model focuses on how herd behavior of subsequent users can affect the first mover’s technology choice. Thus information is created as a result of adopting a new technology. To elicit herd behavior they assume that the returns from decision are realized only after everybody has made their choice. If reward associated with each option was revealed immediately after its adoption there would be no heard behavior.

In case of informational externality the equilibrium rate of entry at any given time is owner than the socially optimal one.

Network externalities create positive payoff interdependency: the more people adopt the same technology the more valuable it is. Thus each adopter has to be concerned with how information generated by their decision has affected the adoption decision of potential users. The values of adopting each technology depend only on the intrinsic value of the technology and the final size of its network. Once the technology is used by anyone its true value is revealed and can be used by potential users. The aim is to investigate how informational externalities interact with network externalities to induce herd behavior and create biases against adoption of a new technology.

There is an issue of risky technology and the penguin effect in adoption game: each user will be reluctant to move first as long as there is a possibility that their choice may turn out to be so inferior as to make orphan their adoption. Thus they may prefer to wait for another user to choose first in order to free ride on informational externalities. The one who will make the first choice will be the most impatient user; their choice need not be the best choice for the collective welfare. The agents, who wait for others to make the decision, have the opportunity to observe the experience of others. This provides them valuable information in their own decision making.

The limitations to the study include: the assumption on information generation and its transmission is too simple to be realistic. The true value of the technologies can only be revealed through repetitive use. The adoption behavior in the presence of network externalities and informational spillover can influence the supply side of technologies also.

According to the model the less independent of network externalities to adopt earlier than those who are more concerned with network externalities, since there are less affected by the potential stranding effect. Therefore the independent users would have positive effect on other users because they could prevent inefficient risk-averse behavior in the early adoption stages.

Thirdly, additional economic factors are introduced which may influence the process of opinion formation in order to see how endogenous mechanism is responsible for the ultimate breakdown of such bubbles.

The overall nature of the market depends on the basic economic variable i.e. the actual returns including capital gains. Hence, speculators are not the blind followers of the crowd rather they observe others’ behavior in order to judge profit opportunities and also try to find out whether optimism or pessimism is prevailing in the market. Above average returns are reflected in a more optimistic attitude that promote speculators to overtake others bullish beliefs whereas below average returns are reflected in a more pessimistic attitude and discourage them to overtake and follow others’ beliefs.

Gullermo A. Calvo & Enrique G. Mendoza (1997) conducted a study on Rational Herd Behavior and the Globalization of Securities Markets.

Herd behavior is found to be adopted by the global investors whereas globalization of securities markets worsens the volatility of capital flows by strengthening incentives for herd behavior. Their study proposes a basic model of international portfolio diversification with incomplete diversification in which globalization of securities markets reduces the incentives for information gathering and generates instability in capital flows as a result of optimal herd behavior. The objective of this study is to reveal that Herd behavior is an outcome of optimal portfolio diversification that becomes more obvious as markets grow. The investors prefer to follow the market rather than taking the time and expense to make their own evaluations of each other country’s essentials. Secondly, they predict that market portfolios embodied relevant information and are afraid of the consequences of going against the market and incurring massive losses from currency collapses.

The authors have used a basic framework of mean-variance portfolio diversification to show that two properties of imperfect competition can produce equilibria in which the incentives of evaluating the variety of rumors available weakens as capital market grow. First, the expected gain made by paying the fixed cost of gathering and processing the country specific information falls as the number of countries grows. Second, if the investors face reputational costs in terms of their returns which is depending upon the performance of their portfolio as compared to the performance of the given market portfolio. In case the rumor favors another portfolio within the herding range, than the rest of the investors decide to follow the herd. Therefore, incentives for herd behavior can grow stronger as markets grow.

Simulations based on equity markets data and country credit rating propose that herd behavior can increase large capital outflows from growing markets.

Rama CONT and Jean-Philippe BOUCHUD (1998) conducted a study on Herd Behavior and aggregate fluctuations in financial markets The fluctuation in prices are not necessarily related to the variations of economic variables or arrival of new information and thus it may lead us to think that high variability in stock market returns may correspond to crowd effect or herd behavior. The basic objective behind studying the relationship between the presence of herd behavior and aggregate fluctuations in financial markets is to see: how does the presence of herding transform the distribution of returns? and also to study the implications of herding for relations between market variables such as order flow and price variability. The authors have used a simple model in which the agents group together in coalitions or clusters, once a collation is formed than all its members coordinates their individual demands so that all individual in that particular cluster have the same belief about the future movements of the asset price. In their framework of simple model, the authors have tried to examine how the existence of herd behavior among participants may lead to large fluctuations in the excess demand and give rise to a heavy tails in the distribution of stock price variations. The model predicts a relation between the fatness of the tails of asset returns and the degree of herding among market participants. The model also provides a quantitative link between the heavy tails observed in the distribution of stock market returns on one hand and the herd behavior in financial markets observed on the other hand. The study also finds a relationship between excess kurtosis of returns, the average order flow and the tendency of market participants to replicate each other.

Hans Mewis(1998) conducted a study to find out How Herd Behavior Breaks Down. Information cascades incorporate the phenomenon of herd behavior. An information cascade occurs if agents do not follow their own signals and follow what others have done earlier. Agents make their decisions after assessing the true state of the world which is based on their prior beliefs, their private signal, and the observed behavior/actions of other agents. Author has extended the standard model of information cascades as it was found to have some troubling features and few short comings. First, once the information cascade has started it runs until some exogenous variable occurs. Second, once the decision is taken by the agents than they cannot revert back. The inclusion of the option to revert leads to breakdown.

The objective of the study was to find how information is passed on within an enter-cascade and to evaluate circumstances under which breakdown occurs. The authors have extended the standard model of information cascades where agents have the option to choose between switching to a new alternative or stick to the old one. The authors have introduced two new features into the standard model of information cascade. First, the agents continue to receive signals after they have decided whether to switch. Second, firms may change their earlier decision to new and may also revert to their earlier decision.

The inclusion of these new features completely changes the learning process. Suppose when the managers have to decide whether they want to buy the new technology or stick to the old one especially when they do not have the complete information about the new technology than they are left with few choices that may lead to their decision making. Mangers share a common prior belief and may have private information about the quality of the new technology. The experience after switching to new technology leads managers to change and update their beliefs which may lead them to stick to the new technology or they may change their decision and return to the old technology in case they conclude switching was a mistake. At the same time in case others do not have had such bad results and do not return back to old technology than this situation may lead manager to a more optimistic attitude. In that case they may ignore the bad news and continue with their decision which results in learning from the observed behavior of others and evolve optimistic attitude among managers and such an optimistic atmosphere can also strengthen the stability of the cascade. Moreover, it was also found that an enter-cascade gains stability as it goes on and therefore, a late information release must be quite powerful to overcome the optimistic attitude on the cascade.

The results suggest that information gathering process appears to be more efficient than the standards information cascades.

Christopher Avery and Peter Zemsky (1998) conducted a study to investigate the relationship between rational herd behavior and asset pricing. The model used in the study retains the basic features of the simple model considered by BHW (p.996) with an addition of price mechanism. The study addresses whether informational cascades exist in financial markets? Can herd behavior lead to mispricing of assets? Does it lead to bubbles and crashes? Might it give an explanation for excess volatility?

When the financial market is efficient and the cost of a unit of the asset reflects all publicly available information than the asset price adjusts precisely, informational cascades are impossible and there is no herding and prices always converge to true values. In such a case, herd behavior can not be the source of excess volatility rather the volatility of prices is determined by fundamentals.

By addition of one dimension of uncertainty i.e. event uncertainty in which the market is uncertain as to whether the value of asset has changed from its initial expected value, may lead to the possibility of herd behavior. However, in this case the market does not learn about whether the asset value is high or low and such a herd behavior has little effect on asset prices. Under such circumstances one might not find connection between herd behavior and market crashes. On the other hand by introducing a combination of event uncertainty and composition uncertainty (which means that there is uncertainties to the average accuracy of traders’ information), we may identify certain states in which herd behavior can lead to bubble and crashes. The resulting uninformative herd behavior may have dramatic effects on prices. These results reveal that with the addition of multidimensional uncertainties, herd behavior has strong effect on price mechanism and this can lead to highly volatile price paths.

Xeni Dassiou (1999) conducted a research to examine the investment decisions of two mangers, in which their investment decisions have no effect on their remuneration. The study is based on different assumptions used in a model in which there are two firms A & B that are run by Mangers A & B respectively. In this study there are two types of managers, smart and dumb. The two mangers are asked to make an investment proposal in order to decide whether they want to invest in it or not.

The first central assumption in the model is that managers do know their abilities. It was assumed that agents are aware of their abilities and knowledge. This eventually reduces their chances to herd.

The second assumption is that there is a prior; i.e. one of the two states of the world is more likely to occur. This reduces the ability of the manager B to herd in case he is a dumb manger so that such a manger does not only follow the action of manager A. The higher the probability that a dumb manger B will herd on prior the greater are the chances that smart manger A deviate from the prior.

The third central assumption in the model is the perfect correlation among the signals and the fourth assumption is that the manger’s payoffs are not associated to the direct cost of these signals; this will lead the smart manager to maximize his appearance by evaluating the reported signal of his predecessor, without attempting to make the most valuable decision.

By introducing these assumption in the model, it was found that the loss occurs due to herding reduces because in this model the smart managers that play second never herd on the first manager’s action because the signal of the two smart managers are perfectly correlated. The reported strategy of the second manager will affect the reported strategy of the first manager as his signal provides the labor market with the information which is useful in evaluating the ability of the first manager.

A signaling equilibrium occurs when managers with positive ability do not herd while managers with negative ability randomize between herding and following their own signal.

Russ Wermers (1999) conducted a study is to see whether mutual funds herd in their trades and also to determine whether such herding impacts stock prices and to see whether any such impact is stabilizing or destabilizing. The trading activity of the mutual fund industry from 1975 to 1994 was used to investigate whether mutual funds herd in their trade. Portfolio holdings for all mutual funds based in United States which hold equities and which exist in these 20 years were purchased from CDA Investment Technologies.

In the average stock, a very low level of herding was found. However, significantly a higher level of herding was found when the study was focused on sub group of stocks. Also, much higher level of herding among growth oriented mutual funds was found while looking at subgroups of funds. Looking at subgroups of stocks, a much higher level of herding in small stocks was found especially on the sell-side. Further examination of subgroups of stock revealed higher levels of herding in stocks with extreme prior-quarter returns as compared to other stocks.

The study on relationship between mutual funds herding and both contemporaneous and future stock returns revealed that stock bought by herd have higher future returns during subsequent quarters than funds sell in herd. It was also found that stocks strongly bought by herd outperform those strongly sold by herd especially among small stocks.

It was found that herding by mutual funds appears to be more profitable before expenses.

Year 2000-Year 2004

After the theoretical research on herd behavior starts with the papers by Abhijit Banerjee(1992) and few other authors, a series of articles have been written in the Years 2000 to 2004.

Hubert Fromlet (2001) studied Theory and Practical Application of Behavioral Finance. The earlier study on Behavioral Finance has shown that Psychology and irrational behavior do matter on financial markets. Many reactions on financial markets that may not appear in conventional theory can be explained by behavioral finance. An analysis on behavioral finance has shown how behavioral factors can help investors in order to avoid mistakes and can also be useful in finding investment strategies and improving asset performance.

The objective of the paper was to emphasize that psychology really matters and plays an important role in financial decisions. The study of behavioral finance is a new emerging field and it explores some psychological issues in financial markets that are not present in the conventional financial analysis. Without contribution of behavioral finance, certain factors and issues of financial markets cannot be understood.

In the classical and neoclassical economic literature, human being- referred as homo oeconomicus are considered as rational that are making right choices under the perfect conditions. This thinking in financial theory explains situation where market participant get together and set price that works best for every player. On the other hand, Behavioral finance makes certain aspects of financial markets more realistic by relaxing unrealistic behavioral assumptions used in the homo oeconomicus tradition. According to this theory, individuals do not always function as perfectly as they are portrayed in neoclassical literature in financial theory. Financial bubbles are prime examples of the inefficient markets. Below are some of the typical phenomenons for behavioral finance:

Heuristic dealing with information: There is wide use of heuristics which can be defined as the use of experience and practice efforts to solve problems and to improve performance

Varying availability of information: Information may not be available to every participant in the financial market. Different participants may act differently depending on the information they have.

Preference of certain news: Analyst and fund managers have forecasts and they want their own analysis to be true in reality. In order to show their forecast as predicted correctly they may not give importance to new information that may not be in line with their original position.

Differences in interpretation: Analysts & Fund Managers may have different interpretations of the same situations and as a result they may react differently to such differences.

Anchoring: Expectations are usually the anchor for quick explanation by active players and their deviations may have impact on market prices and can turn into a shock.

Representativeness: Human beings have the tendency to give more importance to certain developments, reports, or statements than they really deserve.

Overconfidence and control illusion: It is the tendency of people to believe that they can control a situation which in reality they cannot. Overconfidence is also linked with certain risks.

Disposition effect: It is linked with the tendency of human being in realizing good deals early and aversion to recognizing bad deals and cutting losses may take too long. This is due to strong associated personal commitment.

Home bias: Investors have the tendency to prefer their domestic markets over possible better returns associated with investment abroad.

Following the herd: This is the most recognized observation in financial markets. New investors follow the decisions taken by the previous successful investors by ignoring their private information and in this way follow the crowd.

Nicholas Barbis & Richard Thaler (2002) conducted a survey of behavioral finance based on the analysis of the earlier research conducted by different authors on this topic.

The traditional framework of studies introduced the concept of rationality in financial markets. According to the rational models, agents update their belief when they receive new information and they always make best choices that are normatively acceptable. New studies have confirmed that the assumptions used in the classical models do not explain the concepts of aggregate stock market, the cross section of average returns and individual trading behavior

Behavioral finance is a new development in the financial markets that has emerged in response to the difficulties faced by the use of traditional concepts and assumptions. According to Behavioral finance, not all agents are fully rational.

This review paper evaluates the recent work done in this new and rapidly growing field of behavioral finance. The series of theoretical paper show that in an economy where rational and irrational trader act together, irrationality can have strong and long term impact on prices. This literature on limits to arbitrage has become one of the important part in behavioral finance. Two years ago, it was thought that concept of Efficient Markets Hypotheses had to be true because of the forces of arbitrage but now due to this new development of concept we have come to know that the limits of arbitrage can results in substantial mispricing. The second building block is the concept of psychology of the agents which force them to act differently from one another.

V.V Chari and Patrick J. Kehoe (2002) wrote an article on Herd behavior and how it is presented in many markets. Herd behavior accounts for an extensive fraction of the volatility in capital flows to up-and-coming markets. The model has been subjected to two apparently devastating critiques.

In case of the first critique the continuous investment is that in the basic model, herds disappear if simple zero one investment decision are continuous. Thus if investment decisions are continuous than herds disappear. The investment critique seems to be quite strong because the scale of investment can often be changed easily.

On the other hand the price critique is that herds disappear if, as seems natural, other investors can observe asset market prices. Once agents are allowed to trade in financial markets, prices reveal information and herds disappear .The price critique is strong because it suggests that herd behavior will not be observed in financial markets.

In this study, the authors have replaced the exogenous timing where investors can move in a pre specified order by the endogenous timings in which investors can move freely. This has resulted in both critiques to overturn. Under this timing there is a tradeoff between investing and waiting: waiting is potentially beneficial because investors can gain information but it is costly because of discounting. In case of continuous investment there can be herds of investment and that even with prices and asset trade there can be herds of investment. In both the cases the reason is that at some point the gains from waiting for more information are outweighed by the cost from waiting due to discounting and investors choose to invest with relatively little information.

The information sharing critique implies that there are other ways of sharing information besides having investors infer information from investment decisions. In this case if investors are allowed to communicate with each other there will have no incentive to misrepresent their information and there is no essential reason for information to get trapped. In case of exogenous timings there is a equilibrium in which all investors truthfully reveal their signals and there are no herds. This can be over turned by allowing endogenous timings and assuming that there is a small early mover advantage in that the rate of return earned on the risky project is higher when there are fewer investors. Thus with information sharing there can be herds.

In the base line model of herd Behavior, with endogenous timings, waiting and receiving information is beneficial because investors have the option of not investing if the signals are sufficiently low. This can be beneficial to benefit the non investment option value. The cost of waiting comes from a kind of discounting in that investors forgo the flow of return from investing. One critique of baseline model is that the result depends critically on the action space being coarse relative to the signal space. If the action space is continuous then there can be no herd of investment. The critique is handled by dropping the exogenous timing assumption and as a result this critique is overturned.

The paper has taken a step towards the developing models of herd behavior that can be used in applied situations. The study has achieved this by answering the two main critiques of the early models.

Torben Lutje, University of Hannover, Germany (May 2004) identified different characteristics of herding versus non-herding asset managers. The study based on a questionnaire survey on German asset managers’ attitude towards herd behavior and distinguished between herding asset managers who try to be good and non herding managers who try to be better than their peers/competitors. It was realized that herding exists in practice and is not only theoretical problems and is perceived by a great majority of asset managers. Herding asset managers who follow the trend are more concerned about their reputation and career concerns as compared to non herding asset manager since they believe in beneficial effects of such investment behavior. As they do not try to be better than their competitors so they show less working effort as compared to non-herding asset managers who wish to do better and show more working efforts and typically claim to be more ambitious. Herding asset managers follow the trend and decisions of others so they ignore substantive private information they possess and base their investment decisions on fundamental information-especially on the investment decisions of other market players. Herding managers focus on short term investment horizon as compared to non-herding behavior. Herding asset manager regard themselves as more risk averse and are expected to have a higher loss aversion than non-herding ones and their investment decision is more biased by a disposition effect. It was also found that although herding managers are more risk averse, but in a tournament they are willing to take more risk.

Richard W. Sias (2004) conducted a study on Institutional Herding. Institutional herding is defined as a group of investors following each other into and out of the same securities over some period of time. The primary goal of this study was to find if herding exists in institutions investors. In this study institutional herding was tested by directly examining the cross-sectional correlation between institutional demands over adjacent quarters. The data for this study was gathered from returns, shares outstanding, and firm capitalization taken from the Center for Research in Security Prices (CRSP) monthly tapes for all NYSE, AMEX and NASDAQ stocks. The study reveals that Institutional demand for a security is positively correlated with the demand for security in the adjacent quarter. The fraction of institutions buying securities this quarter positively convary (across securities) with fraction of securities buying last quarter if financial investor follow each other(herd) or themselves into and out of the same securities. The results of the study show that the cross sectional correlation between fraction of institutions buying over adjacent quarter can be decomposed into portions that come from following their own trade and following other institutional investors’ trade. The decomposition reveals that both factors play an important role in explaining the relationship between institutional demands between adjacent quarters.

Analysis of changes in portfolios weights show that institutional investor’s tendency to follow their own lag trades is not necessarily an outcome of time series correlation in the net flows and investing these net flows in their existing portfolios. The results of the analysis also show that herding does not result only from time series and cross sectional correlation in the flow of investors. The study suggests that institutional demand is strongly related to lag demand as compared to lag returns. It was also found that herding reduces over time and may differ among different types of investors.

The theoretical foundation for institutional herding can be dived into five categories- informational cascades, security characteristics, fads, investigative herding and reputational herding and it can effect security selection decision and asset prices. Informational cascade result from institutional investors following other because they gather information from each other, reputational herding is the result of reputational cost faced by the institutional investors acting different from the herd, characteristic herding occurs when institutional investors are attracted to securities with specific characteristics, fads is the consequence of the tendency of intuitional investors to herd as a result of fads and investigative herding occurs when institutional investors information is positively & cross-sectionaly correlated because of the same signals. After evaluation across capitalizations, it was found that institutional herding primarily results from institutions’ gathering information from each other’s trades.

Bogachan Celen and Shachar Kariv (2004) wrote a paper which distinguishes between Informational Cascades and Herd Behavior. An informational cascades occurs when an infinite sequence of individuals completely ignore their private information and find it optimal to follow the decision of their predecessors as their belief is so strong that no signal can outweigh it. Whereas in Herd Behavior, an infinite sequence of individuals not necessarily ignore their private information while making a decision i.e. they may have acted differently if they had realized a different private signal. Another important distinction between herd and cascades is that a cascade ceases social learning as the decisions are completely identical and uniform. In contrast, herd behavior may provide some information from the action of individuals.

Informational cascades are harder to identify as they are defined in term of unobservable beliefs, whereas herd behaviors are defined in terms of observable beliefs.

This paper reports an experimental test of how individuals learn from each other by distinguishing informational cascades and herd behavior by using techniques that are only available in the laboratory. The experiment was run at the Experimental Economics Laboratory of the Centre for Experimental Social Sciences by recruiting 40 subjects from undergraduate economics classes at the New York University.

The subjects were asked to choose between the cutoffs such that Action A will be chosen if the signal received is greater than the cutoff and single B otherwise. Signal A is the profitable action when the sum of all subjects’ signal is positive and B if it is not. The cutoff data helped in determining which subject exhibit cascade behavior irrespective of their signal. Hence, cascades behavior is identifiable by the choice of cutoff while herd behavior is identifiable from the action taken by the subjects. In the laboratory experiment study, herd behavior was found to occur more frequently as do cascades. Since theoretical result shows that an informational cascades is impossible, but they are a reality. A generalized Baysian model to take account for human error was used by allowing the possibility that subjects can make errors and they incorporate the possibility that other are making errors into their beliefs which provides an explanation of cascade behavior. It was found that in a laboratory subjects give more weighted to their private information as compared to the observed behavior of others, but over the time they have a tendency to move towards Baysian updating.

Year 2005-Year 2009

A number of articles have been published on herd behavior in these years based on recent studies by using different methods providing new perspectives on the topic.

Marco Cipriani and Antonio Guarino (2005) studied Herd Behavior in a Laboratory Financial Markets. Herd behavior increases the importance of early decisions and is focused on the previous history of trades, whereas contrarian behavior goes against the previous history of trades and reduces its importance. Agents always find it favorable to trade on the difference between their own information and the available information. The experimental study on herd behavior in a laboratory financial market where a sequence of subjects trades an asset whose value is unknown is consistent and in line with the results of theoretical research conducted on herd behavior which predicts that agents should never herd. In frictionless laboratory market in which traders trade only for informational reasons, herd behaviors rarely occurs. The experimental analysis based on the model by Lawrence Glosten and Paul Milgrom (1985) in which they recruited 216 students as subjects from undergraduates’ courses in all disciplines at New York University and University College London to run 16 sessions and used three different treatments in these sessions. In the laboratory market study, subjects decide whether they want to buy or sell depends upon the private information on the value of a security and upon the history of past trades. The herding in such cases can be detected by observing the way in which the subjects use their private information and react to the decisions of the previous traders. Three different treatments were used and their effect on trading was observed. In the first treatment, Asset price was fixed and was not updated on the basis of the order flow, it was found that informational cascade arise i.e. subjects should buy despite a negative signal or sell despite a positive signal which occurred after a trade imbalance higher than or equal to 2, or equal than or lower to -2.

In the second treatment the price was flexible in a Bayesian way on the basis of order flow and was updated after each trade decision in which rational subjects should always follow their signal i.e. they should buy after seeing a positive signal and should sell after seeing a negative signal. An information cascade cannot occur in such case. Therefore, in this treatment, when a subject decided to buy, the price was updated that he had seen a positive signal and when he decided to sell, the price was updated assuming he had seen a negative signal. It was found that when price is flexible subjects ignore their private information and herd much less frequently than when the price is held constant. In the experimental laboratory study when the price is flexible, subjects sometimes choose not to trade or they sometime choose not to follow their private information, a proportion of which is contrarian behavior in which the subjects neglect their private information and go against the market.

The third treatment used was the control treatment where subjects could not observe the history of trades or decisions of those who traded before them.

The result of the study are consistent with theoretical predictions and suggests that herd behaviors rarely occurs in laboratory markets as subjects decided not to follow their private information and a proportion of these trades against the private information and in some cases they even abstain from trading. These results suggest that to in order to understand herd behavior in financial markets, we must also look for other explanations such as reputation concerns etc.

Andreas Park & Hamid Sabourian(2006) provided a complete characterization of trading behavior and a new perspective on herding in financial markets with efficient prices.

When markets crash without any significant shifts in the economics conditions than it shows that the investors behaved like a herd. Rational herding occurs when traders/agents follow the crowd and act against their private information.

In this paper the authors have followed the microstructure literature and employed a stylized specialist sequential security trading model and argued that people may subject to herding only if the outside information leads them to believe that extreme conditions are more likely than the moderate ones. The authors have characterized conditions under which this behavior arises and have also tried to show the moving pattern of prices with and without herding. The authors define herding as any history-induced switch of opinion/behavior in the direction of crowd.

The authors have analyzed the pattern of Price movements during herding and show that prices become more volatile when herding occurs relative to a situation in which ignore the public information and herding does not occur. The authors have also provided numerical simulations of price-paths and found that some traders may change their trading modes during herding, prices become history dependent. Herding results in price paths that are very sensitive to changes in some key parameters and it slows down convergence to the true value if the herd does not move in the right direction.

The authors have analyzed different kind of behaviors of traders by conditional signal distributions. With a U-shaped conditional signal distribution, the recipient’s expectations become more volatile and the traders start following majority and acts like a herd. A hill shaped distribution achieves opposite results by making recipients expectation more stable. Lastly, the signal recipient remains on one side of the market irrespective of public information and behave like ‘contrarian. The model used by the authors is restricted by the likelihood of middle-signal types and imply that agents with extreme signals always take the same action and do not herd or behave as contrarians. Their model does not imply that all traders act alike. They have showed that herding can occur only if the signal distribution is U- Shaped and there is sufficient amount of noise.

In this paper the authors have also tried to investigate the contrarian behavior conditions for which a trader after observing a trading history, changes his actions and acts against the crowd and remains on one side of the market irrespective of the public information.

Overall, the authors have concluded that herding occurs only if the outside information leads them to believe that extreme conditions are more likely than the moderate ones. Herding has a significant effect on prices and it can also affect the process of learning.

Robert J. Weiner (2006) conducted a study that focuses on measuring and assessing the tendency of speculators to herd (trade in the same direction as a group and flock (trade in the same direction by subgroups of speculators) in the international oil market which eventually causes volatility in oil prices. The Commitment of Traders (COT) data was used in the study which is consist of the open positions of large players in options and futures markets and is collected by CTFC database, covering the period 1993-1997 of the petroleum market which is a simple and volatile international commodity market.

The past few years have witnessed the worst turmoil in the international financial systems. One reason for these crises could be due to the weaknesses in social, political, and economic systems. An alternative view is that these crises are generated by the financial systems itself, arising from herding, or flocking. The other perspective behind these crises is the role of speculators which may aggravate or cause financial stability and it should be controlled by taking policy measures.

Speculative behavior falls under two categories. First speculators are engage in noise trading or liquidity trading and their buying and selling activities are based on factors other than changes in market fundamentals. Second, speculators herd or flock by watching and observing each other than market fundamentals.

Herding is a widespread social phenomenon, often observed when people only follow what others are doing. The finance literature has developed two hypotheses regarding herding behavior among traders, or flocking (trade in the same direction by subgroups of speculators). According to asymmetric information hypothesis, poorly informed traders make their decision by following the acts of well informed traders to take similar position. In contrast, according to the monitoring/incentive hypothesis investors make their decisions knowing that their incentives are based on their performance.

Employing the parametric (correlation) and nonparametric (count) methods, findings of the research finds little evidence of herding in heating oil futures markets among noncommercial markets. The results of the study also indicate that evidence of such behavior among participants depends upon the approach adopted. Overall, the data provides support for the monitoring/incentive approach of flocking behavior in which investors make their decisions knowing that their incentives are based on their performance and provides weak support for the asymmetric information theory which is based on the observation of well informed traders.

Conclusion:

For a long time, the entire functioning of the financial markets was dominated by the concept of Efficient Market Hypothesis but the recent empirical investigations and factual observations have declined the trust in the classical theories which denies the existence of deviations from the perfectly rational scheme of behavior The literature on herding reveals that asset markets are driven by animal spirit, where investors behave irrationally and copy the actions of other participants. Both Financial economists and market participants reportedly believe that imitative behavior is common in financial markets which has led some researchers to emphasize that market participant are engaged in nonrational herd behavior. An examination of empirical study on herd behavior and explosion of research papers in the last several years argue that herd like behavior exists in financial markets in which a slight change in public information is sufficient to induce all the agent to follow the lead can hinder the flow of useful information to the followers. Such sequential actions have disappropriate effect in the long run as the entire decision process is based on the earliest decision.

A number of articles written on herd behavior in different years have explored many aspects of herd behavior. Studies reveal that herd behavior may cause fluctuations in prices and returns. The significant fluctuation in prices and returns are not necessarily an outcome of variations in fundamental variables or arrival of new information, this lead us to think that such variations may correspond to crowd effect or herd behavior.

There are quite a few disadvantages associated with herd behavior. Studies have shown that herd behavior may result from private information not publicly shared. Research has shown that individuals acting sequentially on the basis of their private information and public knowledge based on observing actions of others, may end up choosing the option that is not as profitable as perceived and socially undesirable.

Studies have also shown that herding behavior may lead to bubbles and crashes in the market. Many observers that large stock market trends often begin with frenzied buying (bubbles) or selling that lead to crashes. Many observers believe that theses extreme shifts in the markets are due to existence of herd behavior of agents.

Herd behavior revealed that herd behavior should not be adopted by the investors, because an investor is generally better off if he does not follow the herd and focus on his own information.

References

• Avery, Christopher. & Zemesky, Peter. (1998). Multidimensional Uncertainty and Herd Behavior in Financial Markets. The American Economic Review, Vol. 88, 724-748.
• Barberies, Nicholas. & Thaler, Richard. (2002). A survey of Behavioral Finance. National Bureau of Economics Research, Working Paper 9222.
• Benerjee, Abhijit V. (1992). A Simple Model of Herd Behavior. The Quarterly Journal of Economics, Vol. 107, 797-817.
• Calvo, Guillermo A. & Mendoza, Enrique G. (1997). Rational Herd Behavior and the Globalization of Securities Markets. Discussion Paper 120, Institute for Empirical Macroeconomics.
• Celen, Bogachan. & Kariv, Shachar. (2004). Distinguishing Informational Cascades from Herd Behavior in the Laboratory. The American Economic Review, Vol. 94, 484-498.
• Chari, V.V. & Kehoe, Patrick J. (2002). On the Robustness of Herds, Federal Reserve Bank of Minneapolis, Research Department, Working Paper 622.
• Choi, Jay Pil. (1997). Herd behavior, the penguin effect, and the suppression of informational diffusion: an analysis of informational externalities and payoff interdependency. RAND Journal of Economics, Vol. 28, 407-425.
• Cipriani, Marco. & Antonio, Guarino. (2005). Herd Behavior in a Laboratory Financial Market. American Economic Review, Vol. 95(5), 1427-1443.
• CONT, Rama. & BOUCHAUD, Jean-Philippe. (1997). Herd Behavior and aggregate flucatuations in financial markets. Journal of Macroeconomics, 4(2), 170(2000).
• Dassiou, Xeni. (1999). The Impact of Signal Dependence and Own Ability Awareness on Herding Behavior: A Tale of Two Managers. Managerial and Decision Economics, Vol. 20, 379-395.
• Fromlet, Hubert. (2001). Behavioral Finance-Theory and Practical Application. Business Economics, 36, 63.
• Lutje, Torben. (2004). To BE Good or To Be Better Asset Managers’ Attitudes Towards Herding. Department of Economics, University of Hannover, Germany. Working Paper No. 297.
• Lux, Thomas. (1995). Herd Behavior, Bubbles and Crashes. The Economics Journal, Vol. 105, 881-896.
• Mewis, Hans.(1998). The Stability of Informational Cascades: How Herd Behavior Breaks Down. Discussion Papers.
• Park, Andreas. & Sabourian, Hamid. (2006). Herd Behavior in Efficient Financial Markets. University of Toronto, Department of Economics, Working Paper Series-tecipa-249.
• Scharfstein, David. & Stein, Jeremy C. (1988). Herd Behavior and Investment. Massachusetts Institute of Technology (MIT), Sloan School of Management in its series Working papers. WP 2062-88.
• Scharfstein, David. & Stein, Jeremy C. (1988). Herd Behavior and Investment. Massachusetts Institute of Technology (MIT), Sloan School of Management in its series Working papers. WP 2062-88.
• Shiller, Robert. (1995). Conversation, Information, and Herd Behavior. Cowel Foundation Paper 909.
• Sias, Richard W. (2004). Institutional Herding. The Review of Financial Studies, Vol. 17, 165-206.
• Weiner, Robert J. (2006). Do Birds of a Feather Flock Together? Speculative Herding in the world of Oil Markets. Discussion Paper.
• Wermers, Russ. (1999). Mutual Fund Herding and the Impact on Stock Prices. The Journal of Finance, Vol. LIV, No.2.