Can Behavioral Finance help to understand
them?
-
Martin Günther -
First Supervisor: Rune Lönnqvist – Universitet Uppsala
Second Supervisor: Niels Hermes - Rijksuniversiteit
Groningen
October 2008
MSc International Business & Management
Specialization: International Financial Management
Abstract
This research looks on selected theories of behavioral finance in order to answer the question whether behavioral finance can help to understand financial crises. More explicitly it is tested whether noise traders and herding behavior occur before a financial crisis starts. As financial crises, the dot.com bubble and the recent subprime mortgage crisis1 have been chosen. It is hypothesized that herding behavior and noise traders will occur before a financial crisis starts. Furthermore, it is tested whether beta as a volatility measure is relevant in financial crises. The sample consists of 20 companies being traded on the NASDAQ 100 and DAX.
The results have shown that herding behavior and noise traders come about before a financial crisis occurs on the tested stock exchanges. Furthermore, the beta coefficient has been proven to be relevant even in financial crises. Thus, the findings contribute to a clearer view about financial crises and how they might occur. Further research is however needed in order to see if the obtained results also hold for other stock exchanges affected by financial crises and for a larger sample.
Keywords: behavioral finance, financial crises, beta coefficient, stock exchange
1
Table of Contents
I Introduction ... 5 1.1 Defining a Crisis ... 6 1.1.1 Financial Crisis ... 6 1.1.2 Political Crisis ... 7 1.2 Development of a Crisis ... 7II Historical Financial Crises ... 9
2.1 before 1929 ... 9
2.2 1929 - 1997 ... 11
2.3 1998 - 2004 ... 12
2.4 After 2004 ... 13
III Theoretical Framework and Hypotheses ... 14
3.1 Efficient Market Hypothesis ... 14
3.1.1 The Classic Theory ... 14
3.1.2 Random Walk ... 14 3.1.3 Levels of Efficiency ... 15 3.1.4 Remarks ... 16 3.2 Behavioral Finance ... 16 3.2.1 Herding ... 17 3.2.2 Noise traders ... 19
3.3 Capital Asset Pricing Model ... 21
3.3.1 The Classic Theory ... 21
3.3.2 The Beta Coefficient ... 22
IV Design, methods and data ... 26
4.1 Research Question ... 26
4.2 Research Objective ... 26
4.3 Conceptual Model ... 27
4.4 Research Type ... 27
4.5 Method and Data ... 28
V Analysis and results ... 34
I Introduction
In the last 300 years, the financial world has seen many financial crises. One of the first wide-known crises was the so called Tulip Bubble which occurred in the 1630s. Followed by this, many more crises happened and currently, the financial markets are suffering from the subprime mortgage crisis which had its beginning in the United States.
Due to the many financial crises that have occurred in the past, investors have lost a lot of money and confidence in the financial markets. It would be valuable for investors if they could be able to define the starting point and possible outcome of a financial crisis in order to sell their shares before the market goes down. The latter is extremely difficult to achieve as the financial markets do not react in the same manner every time. However, behavioral finance might be able to explain some behavioral patterns which can be observed before a financial crisis occurs.
This research examines whether behavioral finance can help to understand financial crises. More explicitly, it will be tested whether herding behavior as well as noise traders can be seen before a financial crisis occurs. Further, the beta coefficient will be tested with regard to it being relevant in financial crises. This will be tested by taking the beta coefficient of selected companies under account and comparing it with the actual share price reaction during a financial crisis which occurred in the past.
The results obtained can be used for investors to see whether a stock market is close to its peak and based on this, start selling their shares to avoid high losses if a financial crisis should occur. Furthermore the results obtained, by testing the beta, will help investors to make a decision on whether to take the beta coefficient under account when creating a portfolio. In the following chapters an overview about some historical financial crises will be provided, followed by a theoretical framework and hypotheses. In the fourth chapter, the design, methods and data used will be explained and in the fifth chapter, the analysis and results will be presented. Finally, a conclusion will be drawn and some recommendations shown.
1.1 Defining a Crisis
Broadly speaking a crisis can be seen as an unstable or crucial time or state of affairs in which a decisive change is impending (Merriam-Webster, no date). There exist different types of crises which have to be taken under consideration, like e.g. a financial and political crisis. Both of these types of crises can have an affect on stock prices. In the following, these types will be explained shortly.
1.1.1 Financial Crisis
According to Mishkin (1991) a financial crisis is a disruption to financial markets in which adverse selection and moral hazard problems become much worse, so that financial markets are unable to efficiently channel funds to those who have the most productive investment opportunities. One can therefore say that a financial crisis is a disruption to the monetary system and which in the worst case can lead to a financial crash. Sometimes, a financial crisis is even followed by an economical breakdown. It can further drive the economy away from equilibrium with high output in which financial markets perform well to one in which output declines sharply (Mishkin, 1991). In the literature, various types of financial crises are mentioned: currency crises, banking crises, systematic financial crises as well as foreign indebtedness crises (Sigrist et al., 2006).
A currency crisis exists when speculative attacks are used in order to devaluate a foreign currency. As a result, the foreign government can be forced to defend its currency by either selling a large amount of foreign currency or increasing their interest rates.
A banking crisis occurs if either a huge amount of the banks in a country are insolvent or if the debtee of banks unexpectedly ask the creditors to pay their outstanding debts in large amounts of cash which leads to a financial panic.
A systemic crisis is a severe interference of financial markets with consequences for the whole economy.
Most of the time financial crises have effects on world wide stock markets. This can further be seen in section two which shortly mentions some well-known financial crises.
1.1.2 Political Crisis
Besides the above mentioned financial crises, there also exist political crises which can have an effect on stock prices. There exist different types of political crises like for example a domestically and one with regard to foreign affairs. However, in this thesis they will not be explained in detail.
In order to give a short understanding of the term “political crisis” one can say that it is triggered through critical lobbies, conflicts with reference groups or political currents (Möhrle, 2004). Unexpected events like the terror attacks from 9/11 can further disequilibrate the worldwide political system and effect international politics. As a reaction to the events, share prices on the New York Stock Exchange and around the world were falling. However, the drop in share prices can not be explained with one of the above mentioned types of a financial crisis and is rather due to destabilization of the US politic as well as the uncertain future and impact that this event caused.
As a political crisis can happen unexpectedly and even cause changes to share prices if the fundamentals of the companies are good, this thesis will be focusing on financial instead of political crises.
1.2 Development of a Financial Crisis
The development of a financial crisis can best be described according to the different stages of development which can be seen in a crisis. Kindleberger (1978) did a research on historical financial crisis in his book “Maniacs, Panics and Crashes” and based his research on a crises model which Minsky developed in 1972.
If this shock is big enough, economic losses and new profit opportunity can arise. If the profit exceeds the losses, an incentive to increase the production and investments occurs and a boom phase develops.
Extending loans in order to increase the amount of money supports the boom phase. New financial instruments are developed and granting of loans outside of banks further enhances the speculation.
The speculation which has been enhanced through the boom increases the demand for products and financial investment forms like e.g. shares. Due to the enhanced demand and the supply capacity which is not very high, the prices will increase. This signals new profit opportunities which further enhances the investment activity.
The speculation continues to rise due to the expected increase in prices. This leads to more people (noise traders) entering the market as they see that investors are able to make huge profits. These speculations lead away from a rational to an irrational attitude which is affected by the “euphoria and mania” which further leads to the prices being far away from the fundamental values.
As the boom continues, the interest rates, velocity of money and the prices keep increasing until some “insiders” sell their shares to realize their profit. This causes the increase in prices to flatten. In this phase, the expectations can turn which will lead to a run on the liquid funds, by selling the speculation products. This further leads to a crash of the prices and some speculators do not have the possibility to pay back their credits. If the financial distress continues, panic will occur.
II
Historical & Current Financial Crises
Seen from a historical perspective, there have been many financial crises which have had more or less dramatic consequences for the actors in the market. One of the first and well known financial crises that occurred is the tulip mania. Following this, are crises like the South-Sea Bubble, the crash of 1929 and 1987 as well as the Internet Bubble which started in the beginning of 2000. Currently, the world wide stock markets are suffering from the subprime mortgage crisis which had its beginning in 2007 with the closure of a fund from Bear Stearns.
In the following, the various crises will be clustered according to the time frame when they occurred and described shortly.
2.1 Before 1929
In this period, three eruptions of popular speculation in financial markets can be seen: the Dutch “Tulip Mania” of the 1630s, John Law’s Mississippi scheme of the 1719s, and the South Sea Bubble of 1720 (Emmett, 2000).
Tulip Mania
John Law’s Mississippi scheme
The Mississippi Bubble of the 1719s occurred at the same time as the South Sea Bubble which will be explained later. The crisis started in 1715, when France was bankrupt after the war (Stock Market Crash, 2006). John Law established the Banque Generale in 1716 which business it would be to take deposits and issue banknotes. Equity for the bank came from the sale of stock for both cash and conversion of government debt (Erasmus School of Economics, no date). In August 1717 the company “Compagnie d’Occident”, which had been founded by Law, received the exclusive business law on the Mississippi (Aschinger, 1995). In order to finance that company, Law issued shares which could be bought from the public. Within the years, the company extended its business and the name was changed into “Compagnie des Indes”. The share price of the company increased strongly between June and November 1719. However, at the end of November, some investors started selling their shares which led to decreasing share prices. In order to support the share prices, Law started circulating more treasure money. The end of the Mississippi Bubble started in the 1720s when some large investors wanted to change their treasure money into coin money. Law knew that he was not going to be able to change all the treasure money into coin money as more treasure money has been circulated. Thus, he limited the amount of treasure money which could be converted to 100 Livres in Gold and 10 Livres in Silver per person. However, he could not limit the amount of treasure money being converted into coin money which would be transferred abroad. Due to this, the amount of coin money being circulated was reduced which limited the business operation of Law’s companies. Over the following months, the treasure money and the shares of Law’s company were devaluated. The reduction of the share price of around 50% led to a loss of confidence in Law’s system and to panic. The share price continued to decrease and in September 1721, the shares were worth 500 livres again. This is around the same price that the shares were worth before the Bubble started (Aschinger, 1995).
South Sea Bubble
a success. Holders of the short-term debt subscribed 97% of the debt into South Sea Company stock by the end of 1711 (Erasmus School of Economics, no date). The share price of the company increased from 128 Pounds to 400 Pounds within two months. However, the trading rights that the company was hyping were the monopoly of Spain. In August 1720, the share price hit the 1000 Pounds mark and insiders started to sell off their shares. Due to this, the share price collapsed to 135 Pounds. Many shareholders were ruined and the economy went into recession.
2.2 1929 – 1997
During this period occurred two well-known crises, namely the stock crashes from 1929 and 1987.
Stock crash from 1929
in investors and people having an overconfidence from the 20’s which influenced a search for easy money and made many people greedy (thinkquest, no date).
Stock Market crash from 1987
Another large stock market crash occurred in 1987. The Dow Jones lost 22.6% of its value on October 19th 1987 which can be considered as the largest one day stock market crash in history (stock market crash, 2006). As no major events or news occurred prior to the crash, the decline seems to be coming from nowhere. The efficient market hypothesis, which will be mentioned in section three, was brought into question by that event (tradingday.com, 2006). As a reason for the crash, several factors are mentioned which include illiquidity, computer trading or US trade and budget deficits (Itskevich, 2002). It is however to note that the US economy began shifting from a rapidly growing recovery to a slower growing expansion in 1986 which resulted in a soft landing as the economy slowed and inflation dropped. Further, the volatility which could be experienced on the stock market on some days in 1987 could have led to widespread nervousness leading up to the crash (tradingday.com, 2006).
2.3 1998-2004 Internet Bubble
were worth billions of dollars weeks ago, suddenly disappeared. The stock market continued to decrease until 2002 when it reached the bottom at 1114 points (Investopedia, no date).
2.4 After 2004 Subprime mortgage crisis
In 2007, the next big financial crisis, which is known as the subprime mortgage crisis, occurred. This crisis started during the middle of 2007 when the Bank “Bear Stearns” had to announce that two of its hedge funds were suffering from billion dollars of losses. Due to the subprime mortgage crisis, further banks like Merrill Lynch, UBS and Citibank had to take up cumulative value adjustments of billions of dollars. The reason for this crisis goes back until 2001 when the former chairperson of the US central bank started to lower the federal funds rate. Between 2001 and 2003, this rate went down until one percent. Due to this low interest rate, people started building houses on credit and even financially weak households were offered credits from unethical building bankrollers. When the Federal Bank started to increase the federal fund rate, the mortgage interests also increased. More and more consumers were not able to balance out their credits and foreclosure sales followed. The bubble burst and the housing prices went down. Soon the crisis broadened to the whole capital market. Banks hoarded money as they did not trust each other anymore. The risk premium went up and the US Central bank had to stand in, to ensure the liquidity of the markets with cash injections (ard.de, 2008).
III
Theoretical Framework and Hypotheses
3.1 Efficient Market Hypothesis3.1.1 The Classic Theory
The term efficient market hypothesis was established in 1970 in an article by Eugene Fama. According to the efficient market hypothesis, an efficient capital market is one in which security prices adjust rapidly to the arrival of new information and the current prices of securities should therefore reflect all information about the security (Jagric et al., 2005). This means that if new information is revealed about a company, it will be incorporated into the share price rapidly and rationally, with respect to the direction of the share price movement and the size of that movement (Arnold, 2005). It is impossible to forecast new information, and as current and past information is immediately reflected in the current share prices, abnormal profits are not able to achieve. This does however not mean that share prices are equal to true value at every point in time. It simply means that the errors that are made in pricing shares are unbiased and price deviations from true value are random (Arnold, 2005). Describing it in a different way, one can say that 50% of efficiently priced shares turn out to perform better than the market as a whole and 50% perform worse.
3.1.2 Random Walk
As mentioned before, share prices are impossible to forecast as they are based on the announcement of new information. Thus, share prices follow a random walk. This occurs, as the share price at any one time reflects all available information and it only changes if new information will be announced. As the new piece of information is independent of the last piece of information, prices follow a random walk (Arnold, 2005).
for dependence but their results were always that reliable and profitable forecasts being made from past movements is not possible (Arnold, 2005).
3.1.3 Levels of efficiency
Having shortly described the efficient market hypothesis as well as the random walk theory, the three levels of efficiency in addition to their tests are also worth noticing. They were produced in 1970 by Eugene Fama, in order to define the extent to which markets are efficient. The three levels of efficiency as mentioned by Fama are: weak-form, semi-strong form and strong-form. These three levels are based on different types of investment approaches which were designed to produce abnormal returns (Arnold, 2005).
Weak Form
A market is efficient in the weak form if historical price or return information is incorporated in the current share price. This means that no investor will be able to make abnormal returns by developing trading rules which are solely based on historical price information (jurikres, no date)
The weak form efficiency has been tested for example by the filter approach or the Dow Theory and it is to say that the market seems to be efficient in the weak form.
Semi-Strong Form
If a market is efficient in the semi-strong form, investors should not be able to earn excess risk-adjusted returns if their decisions are based on information that has already been made public. Neither technical analysis nor fundamental analysis would provide a predictable edge (financial education, 2008)
Strong Form
In this level, all relevant information, including the privately held information is reflected in the share price. Even insiders are not able to make abnormal profits if the market is strong-from efficient.
Concerning the strong form it is to say that the market is not fully efficient. There exists insider dealings as seen e.g. at the Enron scandal and insiders are able to make abnormal profits. Nevertheless, as insider trading is forbidden, most of the time the insider trading will be uncovered and the involved persons are prosecuted.
3.1.4 Remarks
Various studies have been undertaken in order to test the efficient market hypothesis. These studies looked for example at the reaction of the stock market to the announcement of various events such as earnings, stock splits or takeovers (Russel and Torbey 2002). The results were that security prices seemed to adjust to new information within a day of the event of announcement which is consistent with the efficient market hypothesis. On the other side, there have also been authors arguing that there is little, if any, correlation between the greatest aggregate market movement and public release of important information. The efficient market hypothesis also seems to be inconsistent with many events in stock market history. As mentioned in chapter two, the theory was brought into question, when the stock markets dropped drastically in 1987 without any new information being published.
Behavioral finance might therefore be able in helping to understand why financial crises happen and explain them better as the efficient market hypothesis does.
3.2 Behavioral Finance
As mentioned before, dominant models like the efficient market hypothesis fail to explain some of the past financial crises and occurrences on the stock markets. Behavioral finance tries to explain the events on the financial markets by taking human behavior under account. In this regard, methods and cognitions from psychology and sociology are also being considered. According to the behavioral finance theory, do market participants not always act rational. This is because of their psychological and mental restrictions, which only let them act with limited rationality (Roßbach, 2001). Malkiel (2003) states in his article “The Efficient Market Hypothesis and its Critics” that individuals see a stock price rising and are drawn into the market in a kind of “bandwagon effect”. The rise in the US stock market during the late 1990s is described to be the result of psychological contagion leading to irrational exuberance. Thus, behavioral finance seems to be a reason why financial crises can happen in an efficient market.
In order to test whether behavioral finance can help to understand financial crises it will be looked at in this thesis, if herding behavior and noise traders can be seen before a financial crises occurs. In the following, these two theories will be explained in more detail.
3.2.1 Herding
The word “herding” describes the observance that investors seem to be following a herd (other investors) when making investment decisions which leads to investors investing or disinvesting in the same assets.
As an example for demonstrating herding behavior we assume that there exist two pubs which are lying nearby each other. One evening, we decide to go to pub A as this one has been recommended by our guidebook. As we arrive there, nobody is sitting in pub A, we observe though that pub B which is nearby is very busy. Based on our observance, we change our minds and go to pub B instead of pub A.
But why does herding behavior exist? A reason can be due to the fact that people are afraid of making mistakes. It is therefore easier to go with the flow and thereby making the same mistakes then following an own decision and later being the only one who made a mistake (Cassuto et al., 2004).
There exist two different views of herding which have to be differentiated: non-rational and rational view.
The non-rational view centers on investors psychology and assumes that investors behave like lemmings which follow each other blindly without making any rational analysis. On the other side, the rational view, centers on externalities and that an optimal decision making is being distorted by information difficulties or incentive issues (Welch and Devenow, 1995). In between these two views, there exists the intermediate view which holds that decision makers are near-rational, economizing on information acquisition costs by using heuristics and that rational activities by third-parties cannot eliminate this influence (Welch and Devenow, 1995).
In the literature many articles can be found about herding behavior with different results. For example do Drehmann, Oechssler and Roider (2003) test the theory of informational cascades in financial markets. Their result was that the presence of a flexible market price prevents herding behavior. Walter and Weber (2006) investigated in their study the trading activity of German mutual funds to see whether German fund managers are engaged in herding behavior. They found the highest level of buy-side herding during the boom phase, whereas sell-side herding is most pronounced during crash periods. A study from Demirer and Kutan (2005) examines the presence of herd formation in Chinese stock markets. Their findings support rational asset pricing models and market efficiency. Uchida and Nakagawa (2007) found in their study about the herding behavior in the Japanese loan market that irrational herding was only observed during the bubble period.
whether the amounts traded is higher before a financial crisis occurs. This would be consistent with the findings of Uchida and Nakagawa as well as Walter and Weber who both observed herding behavior during crash/ bubble periods. The following hypothesis is thus formulated as:
H1: Amounts of shares traded is excessively higher before a financial crisis occurs.
When looking on past financial crises in chapter two it was visible that the share price always increased before the bubble burst. This could possibly be explained by herding behavior which will be tested in hypothesis number one, as mentioned above. In this context it is interesting to test whether there is a correlation between the share price and trading volume. This will especially be of interest, if the first hypothesis is proven right. As a correlation method, the Pearson method will be used.
3.2.2 Noise traders
Another explanation for stock prices not reflecting the fundamental value of the company can be due to noise traders. These are uninformed investors buying and selling financial securities at irrational prices, thus creating noise in the price of securities (London South East, no date). There are two classes of traders that have to be distinguished according to this theory which are the informed and uninformed traders. The first trade shares in order to bring them towards their fundamental value whereas the latter can behave irrationally and create noise in share prices and thereby generate bias in the pricing of shares (Arnold, 2005). This is in some way contradictory to what the efficient market hypothesis believes as stated above. In an efficient market we expect no investor types to perform persistently better or worse than other investor types as the necessary information is available to everyone (Bae et al., 2006).
their study that these traders lose relatively little when prices are close to true values, but they suffer massive losses when new information takes on extreme values. This leads to noise traders occasionally making small profits but ultimately to suffer even greater losses.
Arnold (2005) also mentions that in order to reinforce the power of the uninformed investor to push the market up and up, the informed investor seeing a bubble developing often tries to get in on the rise but they also try to get out before the uninformed investors and before the crash finally occurs. Caginalp and Ilieva (2008) on the other side state in their articles “The Dynamics of Trader Motivations in Asset Bubbles” that the cash supply of bidders diminishes and the cash supply of the offerers increases as the bubble forms which leads to their suggestion that the bubble is fueled by the cash of the momentum players. This is further supported by Komároni (2004) who mentions that when a stock market bubble occurs, the intensity of noise trading occurs too.
The second hypothesis will determine whether informed investors sell out before a financial crisis occurs. In order to test this, informed investors will be described as institutional investors, as they are believed to have more expertise and knowledge about the company and the market overall and should therefore be able to foresee that the share prices are too far away from the fundamental values. In order to test this hypothesis, 5 minutes trading data from three stock exchanges (DAX, NASDAQ 100 and FTSE 100) will be gathered from available databases. Informed investors should be identifiable by increased volatility of the stock exchanges in the period before the crash occurred, due to the fact that trading high amounts of shares have a bigger impact on the index then trading small amounts. According to Campbell (2007) it is often assumed that large trades are carried out by institutions while small trades more likely reflect individual buying or selling. More information about the data gathering as well as method which is going to be used can be found in chapter 4.5 Method and Data. The following hypothesis is formulated as:
H2: Informed investors step out earlier then uninformed investors, thus selling their shares before a financial crisis occurs.
3.3 C 3.3.1 T The cap in the ea based o by Harr relation Accordi riskless not be th The CA unique classes: There unsystem away w market rates as Capital As The Classic pital asset p arly 1960s a n the idea th ry Markowi nship betwee ing to the C investment he same or APM theory risk. The are factor matically a when an inv or undiver s these have So Fig sset Pricin c Theory ricing mode and it mark hat not all r itz (Perold, en the return CAPM theo t plus a risk above the n y divides th following f rs which ffect them vestor adds sifiable risk e an affect urce: Investo gure 1: Risk C ng Model el (CAPM) ks the birth o risks should 2004). In g n on a secur ory the expe k premium. I necessary re he portfolio figure will systematica (Hamberg, more secu k. Example on all comp opdia.com Classes was develo of asset pric d affect asse general it is rity and the ected return Investors w eturn. risk into t help to un ally affect 2004). The urities to his es for system panies. The oped by Wi cing theory et prices and to say that risk of that n of investo will not inves
two differen nderstand a t securities e part of th s portfolio matic risks e other risk illiam Sharp (Fama and d builds on the CAPM t security. ors is the sa st if the exp nt risk class and differen s and oth e volatility can be mar are inflatio k which disa pe and John French, 20 the portfoli theory exp n Lintner 04). It is io theory lains the
securities are added which leads to a diversified portfolio is called unique, diversifiable or unsystematic risk.
Thus, the total risk is defined as: systematic risk + unsystematic risk.
According to the CAPM theory, an investor can balance out the risk by adding more securities to his portfolio. This is due to the fact that the unsystematic risk can be diversified. The net risk position of an investor is thus only the systematic risk of his portfolio. The CAPM formula is the following:
Expected return= riskless return + beta x (market return – riskless return)
or
E (Ri) = Rf + ßi [E(Rm) – Rf]
whereas,
E (Ri) = expected return Rf = riskless return
ßi = beta
Rm = market return
The beta (ß) coefficient will be explained in the following part.
3.3.2 Beta coefficient
risky. The shares will neither enjoy the same level of upswing nor suffer the same downward movement than the market.
Companies which use to have a beta less than 1.0 are normally producing goods which people can not abstain from, regardless of their financial situation, like e.g. groceries. Luxury goods producing companies or car companies tend to have a beta that is greater than 1.0 as people do not buy those goods if the market is in a downturn, however they buy more of these goods, when the markets experiences an upward stream.
There exist a couple of studies which tested the CAPM theory. Xu and Yang (2006) tested in their study if the CAPM holds true in the Shanghai Stock Exchange. They found that the expected returns and betas are linear related with each other during their tested period and thus found support for the CAPM theory. Feinberg and Tokic (2002) evaluated in their study if beta is a valid measure of systematic risk. This was tested by looking at two extreme single-day drops in stock prices which were due to systematic rise shocks to the market. They found that stocks with higher betas decreased relatively more than stocks with lower betas.
In this study it will be tested whether the beta coefficient as a volatility measure is relevant in financial crises. Instead of looking on single-day drops, it will be evaluated whether the price fall is less or higher than what the beta value indicates in a financial crisis incident. It will therefore be looked on the share price development during the whole crisis period (1999 – 2003). The beta will be calculated by using historical stock price data. The actual price fall of the share will then be compared with what the beta coefficient indicated. This leads us to the following hypotheses:
H3: The beta coefficient is relevant in financial crises as the price fall is less or the same than what the beta value indicates.
H4: The beta coefficient is not relevant in financial crises as the price fall is higher than what the beta value indicates.
On the other side, if the price fall is higher than the beta coefficient indicates, an investor will perform worth than he expected and thus the beta coefficient can not be seen to be relevant in financial crises.
This leaves the following set of hypotheses which will be tested: Table 1: Hypotheses
Hypotheses
H1: Amounts of shares traded is excessively higher before a financial crisis occurs.
H2: Informed investors step out earlier then uninformed investors, thus selling their shares before a financial crisis occurs.
H3: The beta coefficient is relevant in financial crises as the price fall is less or the same than what the beta value indicates.
H4: The beta coefficient is not relevant in financial crises as the price fall is higher than what the beta value indicates.
Since the introduction of the CAPM, it has been a hot debated topic. There are people who believe that it is a useful theory while others think that the CAPM can not help in creating a good portfolio. This chapter will be completed by giving a short overview about some advantages and disadvantages of this theory.
3.3.3 Advantages and disadvantages of the CAPM Theory
The advantage of the CAPM theory is that the calculated expected return can give an advice of what an investor can expect and the beta can help him to decide whether the company is risky or not and if he wants to invest money in that company. Furthermore, the CAPM can help to give an advice of how a company will perform compared to the market and an investor is able to reduce the systematic risk.
method based upon historic data and using historic data for estimating future relationship is doubtful (Arnold, 2005).
IV
Design, methods and data
4.1 Research QuestionThe main research question which is going to be answered in this thesis is:
Financial Crises: Can behavioral finance help to understand them?
As there have been various financial crises in the world of which some are not explainable by the efficient market hypothesis it will be interesting to find out whether behavioral finance can help to understand and explain financial crises. It will therefore be tested whether there is a link between specific behavioral finance theories and financial crises. Further, it will be tested whether the beta coefficient as a volatility measure is relevant in financial crises by taking the actual change of share prices of companies under account and comparing them with what the beta coefficient proposes.
In the previous section, hypotheses have been shaped with the use of appropriate literature on relevant topics. Similar studies within the behavioral finance theory testing the kind of hypotheses like number one and two were not found in the available literature. With regard to hypotheses number three and four, similar studies were found though only comparing the beta coefficient of companies with a single day drop.
What is important when forming hypotheses is the fact that the data needed in order to answer them is available. More information about the dataset can be found in the section 4.5: Method & Data.
4.2 Research Objective
The research objective of this study is to find out whether behavioral finance can help to understand financial crises. Further, the objective is to observe if the beta coefficient as a volatility measure is relevant in financial crises.
exchanges. Furthermore, investors will be able to know if the beta coefficient is relevant in financial crises, thus supporting their decision whether to base their trading strategy on the beta coefficient.
4.3 Conceptual Model
The conceptual model is a graphical representation of the core components of the research (noise traders, herding, beta coefficient) and the assumed relations between them. The corresponding model for this study is the following:
Herding Financial crisis + /‐ Outcome Beta of Company Noise traders +/‐
Figure 2: Conceptual model
The definitions related to the above mentioned concepts can be found in the corresponding hypotheses (chapter III).
4.4 Research type
in this study are going to be achieved through an analytic survey which attempts to test a theory by taking the logic of the experiment out of the laboratory and into the field (Gill and Johnson, 2006). Overall, nomothetic methods are used. These methodologies have an emphasis on the importance of basing research upon systematic protocol and technique.
4.5 Method and Data
As mentioned in the third chapter, four hypotheses will be tested in this thesis. Due to data gathering problems, two stock exchanges (DAX and NASDAQ 100) will be considered for the first, third and fourth hypotheses whereas the FTSE 100 will be included as well in the second hypothesis. The latter stock exchange is not used for the other three hypotheses as historical stock price information is difficult or impossible to collect for this stock exchange. Most financial websites which offer historical prices for companies being traded on the FTSE 100 only do this back to the year 2003. This period is too short in order to define a valid beta which would be the requirement for testing hypotheses number three and four. Further, for testing hypothesis number one a financial crisis which already is over is needed, as the average trading volume of selected companies before the crisis, in between the crisis and after the crisis are going to be compared. Thus, testing this hypothesis on the current subprime mortgage crisis is not possible due to the fact that this crisis is not over yet. Considering this, only stock exchanges where historical prices that reach back a long period (beginning of 1990s) can be gathered, are used which gave the choice for the DAX and NASDAQ 100. 20 companies from the DAX and NASDAQ 100 have been chosen, in order to test the first, third and fourth hypothesis. These companies operate in different sectors: chemical, retail, medical, car manufacturing, insurance, software, construction, electronics, pharmaceutical, banking as well as airline.
Different sectors have been chosen because of the fact that the beta coefficient varies within those sectors. It will be interesting to test whether there will be different results according to different sectors when testing hypotheses number three and four.
Table 2: Overview of companies and sectors chosen
Company Sector Description
Lufthansa
Airline Lufthansa is one of the biggest and worldwide operating companies in the aviation sector. Deutsche Bank Banking Deutsche Bank is the biggest financial institution
in Germany. Mercedes Benz also referred to as Daimler Car manufacturing Mercedes Benz is a German car manufacturing company producing cars under the following brands: Mercedes, Smart and Maybach. BMW Car manufacturing BMW is a German car manufacturing company producing cars under the following brands: BMW, Mini and Rolls‐Royce.
Paccar Car manufacturing Paccar is the biggest truck manufacturing company in the US.
Henkel Chemical Henkel is a worldwide operating company situated in Germany with three main business segments: detergent, cosmetics and adhesives. BASF Chemical BASF is a German Chemical company producing
among others chemicals and adhesives. Foster Wheeler Construction Foster Wheeler is an international operating
plant construction company.
Apple Electronics Apple is a US company producing electronics. Allianz Insurance Allianz is a large financial service provider with
its core business being insurances.
Münchener Rückversicherung Insurance Münchener Rückversicherung is the world’s largest reinsurance company.
Patterson Companies Medical Patterson is a distributor serving the dental, veterinary and rehabilitation supply markets. Bayer Pharmaceutical Bayer is a German chemical and pharmaceutical
company.
Genzyme Pharmaceutical Genzyme is a biotechnology company being specialized in orphan drugs. These are drugs which treats diseases that are very rare. Staples Retail Staples is a US retailer for office supplies. Whole Foods Market Retail Whole Foods market is a food retailer of natural
and organic products.
Costco Retail Costco is a US department store chain.
SAP Software SAP is producing software for businesses and is the biggest software company in Europe. Oracle Software Oracle is the third largest software company in
the world being well‐known for its database program, Oracle Database.
Microsoft Software Microsoft is a worldwide software company
being well‐known for its operating system Windows.
The data necessary for testing the four hypotheses are obtained from websites containing financial market data. On handelsblatt.com, historical share price information of German and US companies on a daily basis can be obtained. Further, the end of day courses of the DAX and the NASDAQ 100 will be gathered from this website. From the Russian Full Service Investment Company website, the data necessary for testing the second hypothesis will be obtained. On this website, it is possible to find historical and recent trading information of the DAX, NASDAQ 100 and FTSE 100, on a minute, hourly, daily or weekly basis. The only drawback of this website is that it offers the information for the period 2002-2008.
As a financial crisis, the dot.com bubble as well as the recent subprime mortgage crisis will be used. These two crises have been chosen because of the difficulty of getting historical share price information for all stock exchanges, as mentioned above. Historical five minutes trading data, necessary for testing hypothesis number two is hard to find. Most companies and websites which offer this information charge high prices for it. Only some websites like the ones mentioned above offer this service for free.
The data and information used in this thesis reaches up to the 5th September 2008. The subprime mortgages crises is not over yet and it can be expected that further information will be released which will lead to a downturn of the worldwide stock markets. This should be kept in mind when considering the subsequent results and analysis.
It is vis this per NASDA on the Between The bot the NAS crash, a the aver more pe During are afra conside expecte Hypothe a 5 min and NA FTSE 1 reached 2007. T the NAS Figu So
ible that the riod, the do AQ 100 and 7th March n the 8th of ttom is reac SDAQ 100. after the cras
rage trading eople see th the whole p aid in trad rable amou d to increas esis number nute basis fo ASDAQ 100 00 reached d its peak on The bottom l SDAQ 100 ure 3: DAX a ource: hande e DAX and ot.com bubb d the beginn 2000 for th March and ched on the . Thus, for t sh and the p g volume w he opportun period of th ding shares unts of mon se again, as r two will b or the period 0 have reach d its peak on n the 31st O line for the
The per the ave instituti shares w investor It is exp their sha The bet by using share p (DAX a using a noisy da two var calculat When th price m less/ the Figur Sou rcentage ch erage chang onal invest which affec rs. pected that t ares earlier ta value wh g historical rice inform and NASDA shorter per ata is highe riables are u ting the beta he betas of ovement of e same or m re 4: FTSE 10 urce: Hande ange of the ge defined. tors selling ct the stock the volatilit than uninfo ich is neede share price mation betw AQ 100) wi riod (weekl er. Covarian used to calc a coefficien the compan f the compan more as com 00, NASDAQ lsblatt.com e stock exch If the ave their share k index to ty increases ormed inves ed for solvi e informatio ween 1988 a ill be used. ly or daily) nce and var culate beta ( nt can be fou nies have be nies during mpared to th Q 100 and DA hanges with erage chang es. Those in a higher ex before a cr stors.
ing the third on on a mon and 1997 f Monthly sto might lead riance need (Beta= cova und in the ap een defined the dot.com he beta valu AX Chart 2005 hin each 5 m ge decrease nvestors usu xtent than risis occurs d and fourth nthly basis. for each co ock returns d to changes to be defin ariance/ var ppendix. d it will be c m bubble. E ue. The form
5 - 2008 NASDA minutes wil es, this is ually trade the volume as informe h hypothesi . In order to ompany as are used, a s in beta an ned for each riance). The compared w Either the sh mer would DA AQ 100 F ll be calcula seen as a higher am e traded by ated and sign for ounts of y private
d investors will sell
s will be ca o calculate t well as the as returns ca nd the risk h company e actual me
volatility measure is relevant in financial crises whereas the latter is an evidence for beta not being relevant in financial crises..
An overview about the dataset is provided in the table below: Table 3: Characteristics of dataset
Variable Frequency Source
Trading Volume Daily Handelsblatt.com /
boerse-online.de
Share prices 5 min Data, Daily & Monthly Russian Full Service Investment Company & Handelsblatt.com
The analyses are made through use of Microsoft Excel 2007. The dataset was constructed in Excel, and calculations were made there.
V
Analysis and results
This study focuses on behavioral finance and if the related theory can help to understand financial crises. More explicitly, it was tested in this study whether herding behavior and noise traders occur before a financial crisis takes place and thus can be seen as a cause of them. Further, the beta coefficient was tested on its relevance during financial crises. In this chapter, the results of the analysis of the four set up hypotheses are presented.
The first hypothesis, which deals about the amount of shares being excessively higher before a financial crisis occurs, was tested by taking the average trading volume in the period 1999 – 2004 under account. More specifically, the chosen period is 7th March 2000 – 12th March 2004 for the DAX and 27th March 1999 – 7th October 2003 for the NASDAQ 100. These dates have been determined according to the peak and bottom reached of the stock exchanges. The average trading volume in this study refers to the average number of shares being dealt with on the Frankfurter Stock Exchange and NASDAQ 100. In order to compare the trading volume before, in between and after the event, the average trading volume beginning on 7th March 1999 (a year before) and ending on 12th March 2004 (a year after) will be taken under consideration. In the same way, the dates for the NASDAQ 100 have been defined.
As a statistical tool, the mean of the trading volume before, between and after the internet bubble was defined and compared. The results for the companies being traded on the DAX and NASDAQ 100 are shown in figures 5 - 13.
Figure 5: DAX Companies – Average volume before burst of internet bubble I
1 year 9 months 6 months 3 months 2 months 1 months 1 week
Allianz
Münchener Rückversicherung Henkel
Figure 6: DAX Companies – Average volume before burst of internet bubble II 0,00 2.000.000,00 4.000.000,00 6.000.000,00 8.000.000,00 10.000.000,00 12.000.000,00
1 year 9 months6 months3 months2 months1 months 1 week
Daimler BMW Bayer Deutsche Bank BASF Lufthansa
Looking at the average trading volume of the ten chosen companies (Figure 5 and 6) it is visible that it did increase in the observed period. The volume reached its maximum for most of the companies, three months before the event. Only the trading volume of Daimler, Deutsche Bank and Lufthansa increased further a week before the event.
Figure 7: DAX Companies – Average trading volume before, in between and after the event
0,00 1.000.000,00 2.000.000,00 3.000.000,00 4.000.000,00 5.000.000,00 6.000.000,00 1 year before crisis 3 months before crisis In between Crisis 1 year after Crisis Allianz Münchener Rückversicherung Daimler BMW Bayer Deutsche Bank Henkel SAP BASF Lufthansa
Figure 8: NASDAQ 100 Companies – Average volume before burst of internet bubble I 0 2.000.000 4.000.000 6.000.000 8.000.000
1 year 9 months 6 months 3 months 2 months 1 months 1 week
Apple Costco Genzyme Staples
Figure 9: NASDAQ 100 Companies – Average volume before burst of internet bubble II
0 10.000.000 20.000.000 30.000.000 40.000.000 50.000.000 60.000.000
1 year 9 months 6 months 3 months 2 months 1 months 1 week
Microsoft Oracle
Figure 10: NASDAQ 100 Companies – Average volume before burst of internet bubble III
0 100.000 200.000 300.000 400.000 500.000 600.000 1 year 9 months 6 months 3 months 2 months 1 months 1 week Foster Wheeler Paccar Patterson Companies Whole Foods
by investors seeing that the share price is far away from the fundamental value and they therefore start selling their shares.
Figures 11/12: NASDAQ 100 Companies – Average trading volume before, in between and after the event I
0 1.000.000 2.000.000 3.000.000 4.000.000 5.000.000 6.000.000 7.000.000 Apple Costco Genzyme Staples 0 10.000.000 20.000.000 30.000.000 40.000.000 50.000.000 60.000.000 Microsoft Oracle
Figures 13: NASDAQ 100 Companies – Average trading volume before, in between and after the event II
0 500.000 1.000.000 1.500.000
1 year 3 months in between 1 year
Foster Wheeler Paccar
Patterson Companies Whole Foods
volume of six companies increases after the event compared to in between, while four companies show a lower average trading volume. As these four companies (Apple, Costco, Staples and Foster Wheeler) operate in different sectors, the made observance can not be sector specific.
Overall, taken the above shown results from the DAX and NASDAQ 100 companies under account, hypothesis number one seems to be supported. The average trading volume increases for every tested company in the period before the event. Nevertheless, it is difficult to define the time when the trading volume increases, as it is different for each company. Some experience an increase months before the event occurs whereas other first experience an increase days before the bubble burst. The observation made that the trading volume declined near the peak, can be explained by the fact that big investors leave the share in an early stage, which has been tested in hypothesis number two. As these investors hold larger amounts of shares compared to individual investors, there selling should affect the overall trading volume.
As the share prices usually always increase before a crisis, the reason for this has been explained in chapter one, it is further interesting to see whether there is a correlation between the share price and the trading volume. According to the results shown above, there does seem to be a slightly correlation as the average trading volume seems to increase in the same time period as the share price. For testing the correlation, the Pearson correlation method, which is the best known one, has been selected. It is the common measure of the correlation between two variables. As variables, the share price and trading volume on a daily basis for the period 1999 – 2004 has been chosen. The results are shown in table 3.
Table 4: Pearson Correlation – DAX and NASDAQ 100 companies
Allianz Münchener Rückversicherung Daimler BMW Bayer
Correlation
Pearson ‐0,287 ‐0,175 0,274 ‐0,157 0,234
Apple Foster Wheeler Costco Genzyme Microsoft
Correlation
Pearson ‐0,148 0,038 0,038 ‐0,176 ‐0,485
Henkel Deutsche Bank SAP BASF Lufthansa
Correlation
Pearson 0,005 0,231 ‐0,352 0,273 0,088
Oracle Paccar Patterson Staples Whole Foods
Correlation
The correlation coefficient can be interpreted in the following way: Table 5: Correlation coefficients
-1.0 - 0.7 strong negative association. -0.7 - 0.3 weak negative association. -0.3 + 0.3 little or no association. +0.3 + 0.7 weak positive association. +0.7 +1.0 strong positive association.
Thus, for most of the companies there exist little or no association between share price and trading volume. The results obtained by testing the average trading volume correlates with the results obtained by the Pearson correlation test. The average trading volume was increasing in the same time period as the share price but it was rather unstable. While the share prices were increasing steadily until the bubble burst, the average trading volume of most companies was volatile.
Hypothesis number one is supported as the trading volume does increase before a financial crisis occurs, although it was expected that the trading volume would increase steadily until the bubble burst. That this was not observed can be due to the fact that the market is after all efficient and investors do not solely follow others without being rational. Another explanation could be that institutional investors step out earlier as they expect the bubble to burst which leads to a lower trading volume. This was tested in the second hypothesis yet for another event.
The second hypothesis which deals about noise traders was tested by taking 5 minutes trading data for the period 2007 – 2008 under account. More explicitly, the chosen period is 12th September 2008 – 18th April 2008 for the DAX, 12 July 2007 – 18th April 2008 for the FTSE 100 and 31st July 2007 – 10th April 2008 for the NASDAQ 100.
This period was chosen, as the selected stock exchanges reached its peak and preliminary bottom during this time. The DAX reached its peak on the 12th December 2007 and bottom on the 18th March 2008, the FTSE 100 reached its peak on the 12th October 2007 and bottom on the same date as the DAX and the NASDAQ 100 reached its peak on the 31st October 2007 and preliminary bottom on the 10th March 2008.
compare the average development of the 5 minutes trading data. The chosen timeframe is shorter compared to the first hypothesis as it is expected that institutional investors rather step out just before the event occurs due to the fact that they do not want to miss on the profit possibility. These are the highest during the upstream on a market which usually is observed just before a bubble burst.
During the period when the subprime mortgage crisis affected the specific stock exchanges, the 5 minutes trading data was collected and the percental change within every 5 minute data was calculated. This trading data contains the actual pit of the selected stock exchanges in the chosen period on a 5 minutes base. It was expected that there would be a decrease in average percental change before the event as institutional investors will step out earlier and sell their shares which is due to their expert knowledge. As these investors usually trade huge amount of shares at once, this should be reflected in the stock exchanges going down in the moment of sales which further leads to a negative change within the 5 minutes trading data. The results can be seen in figures 14-16.
Figure 14: NASDAQ 100 – 5 minutes trading data results
‐0,0100% 0,0000% 0,0100% 0,0200% 0,0300% 0,0400% 0,0500% 0,0600%
NASDAQ 100
NASDAQ 100Figure 15: FTSE 100 – 5 minutes trading data results ‐0,0050% 0,0000% 0,0050% 0,0100% 0,0150% 0,0200% 0,0250% 0,0300% 0,0350%
FTSE 100
FTSE 100The average percental change in the 5 minute trading data does also increase at the FTSE 100. The closer the period until the day, when the stock exchange goes on a downturn, the higher the percental change. However the difference compared to the NASDAQ 100 is that on the day, when the stock exchange reaches its highest level, the volatility measured is negative. This could be an indicator for institutional investors selling their shares, as they assume that the share prices will fall in the future.
Figure 16: DAX – 5 minutes trading data results
‐0,01000% 0,00000% 0,01000% 0,02000% 0,03000% 0,04000%
DAX
DAXFigure 17: DAX, FTSE 100 and NASDAQ 100 end of day quotes 0,00 2.000,00 4.000,00 6.000,00 8.000,00 10.000,00 FTSE 100 DAX NASDAQ 100
The above figures show the end of day quotes of the chosen stock exchanges for the selected period. Comparing this figure with the previous shown ones it is obvious that the three indexes continued to rise until the event started compared to the 5 minutes trading data which showed to be volatile in the tested period. Thus, this volatility can be a sign for institutional investors selling off their shares earlier compared to individual investors, as their selling order only have a short affect on the stock index. Once the price has fallen, due to their selling order, other individual investors will step in and buy those shares as they believe that they are able to make some profit. This will bring the overall share price back in line.
It is to add that in all of the three cases, noise traders and institutional investors respectively can be detected. Nevertheless, on the DAX and the NASDAQ 100 the results were not as obvious as could be seen on the FTSE 100.
Table 6: DAX & NASDAQ 100 - Percental Change within period
DAX Price Index 07/03/00: 8064,97 NASDAQ Price Index 27/3/00: 4704,73
DAX Price Index 12/3/03: 2202,96 NASDAQ Price Index 07/10/02: 804,65
Percentage Change: ‐72,68% Percentage Change: ‐82,90%
Table 7: Overview Beta
Allianz Münchener Rückversicherung BASF SAP Henkel
BETA 1,34 1,02 0,84 0,39 0,79
Expected Change ‐97,27% ‐74,11% ‐61,25% ‐28,00% ‐57,18%
Realised Change ‐86,23% ‐75,78% ‐35,67% ‐92,05% 3,83%
Difference 11,04% ‐1,66% 25,58% ‐64,05% 61,01%
Lufthansa Deutsche Bank Bayer Daimler BMW
Beta 0,98 0,98 0,51 1,18 1,13
Expected Change ‐71,28% ‐71,07% ‐37,38% ‐86,07% ‐82,21%
Realised Change ‐68,30% ‐64,58% ‐74,07% ‐62,33% ‐16,67%
Difference 2,98% 6,49% ‐36,69% 23,74% 65,54%
Staples Whole Foods Patterson Companies Paccar Oracle
Beta 0,93 0,1 0,52 0,68 1,26
Expected Change ‐77,46% ‐8,48% ‐43,47% ‐56,41% ‐104,28%
Realised Change ‐37,68% 2,17% 27,11% ‐30,57% ‐91,29%
Difference 39,78% 10,66% 70,58% 25,84% 12,98%
Apple Foster Wheeler Costco Genzyme Microsoft
Beta 1,03 0,67 0,28 1,26 1,45
Expected Change ‐85,23% ‐55,20% ‐23,03% ‐104,14% ‐120,56%
Realised Change ‐90,13% ‐75,00% ‐39,70% ‐57,54% ‐57,67%
Difference ‐4,90% ‐19,80% ‐16,67% 46,60% 62,90%
Investors that based their portfolio on beta would in most of the cases have lost less then expected during the observed period. Thus, hypothesis number three is supported as in most of the cases, the price fall was less or the same according to what the beta coefficient indicated.
Besides testing whether beta as a volatility measure is relevant in financial crises it is further interesting to see, if there is a difference in realized change among the different companies. As mentioned in chapter three, companies which have a lower beta are expected to be less effected by financial crises. For the results please see table 8.
Table 8: Comparing Beta
Company Sector Beta Realised Change
Microsoft Software 1,45 ‐57,67% Allianz Insurance 1,34 ‐86,23% Oracle Software 1,26 ‐91,29% Genzyme Pharmaceutical 1,26 ‐57,54% Daimler Car manufacturing 1,18 ‐62,33% BMW Car manufacturing 1,13 ‐16,67% Apple Electronics 1,03 ‐90,13% Münchener Rückversicherung Insurance 1,02 ‐75,78% Lufthansa Airline 0,98 ‐68,30% Deutsche Bank Banking 0,98 ‐64,58% Staples Retail 0,93 ‐37,68% BASF Chemical 0,84 ‐35,67% Henkel Chemical 0,79 3,83% Paccar Car manufacturing 0,68 ‐30,57% Foster Wheeler Construction 0,67 ‐75,00% Patterson Companies Medical 0,52 27,11% Bayer Pharmaceutical 0,51 ‐74,07% SAP Software 0,39 ‐92,05% Costco Retail 0,28 ‐39,70% Whole Foods Retail 0,10 2,17%
It was expected that share prices of companies with a higher beta would be affected to a higher extent. Thus, according to this study there does not seem to be a difference in how high versus low beta companies reach in financial crises. Beta can however still be used as a volatility measure as the expected change in share prices was less or almost the same as what the beta coefficient proposed.
Overall, the results obtained did not all give strong support for the tested hypotheses. It could be proven that herding behavior exists before a financial crisis occurs. Concerning institutional investors, the results are not very clear. As the 5 minutes trading data from the observed stock exchanges turned out to be very volatile, this points in the direction of institutional investors selling off their shares earlier. This connection has been drawn as the stock exchanges increased during the same period. However, it was expected to receive clearer results. The test of the beta coefficient turned out to be clear as the expected share price change of the chosen companies were less or came close to what the beta coefficient proposed. However, in this study there did not seem to be a difference in how low versus high beta companies reached in financial crises.
An overview of the formulated hypotheses and their results is provided in table 9. Table 9: Overview results
Hypotheses Results
H1: Amounts of shares traded is excessively higher before a financial crisis occurs
Supported
H2: Informed investors step out earlier then uninformed investors, thus selling of their shares before a financial crisis occurs
Supported
H3: The beta coefficient is relevant in financial crises as the price fall is less or the same than what the beta value indicates.
Supported
H4: The beta coefficient is not relevant in financial crises as the price fall is higher than what the beta value indicates.
The results of the tested hypotheses are presented in the following figure: Financial crisis Herding + Noise traders Outcome + Beta of Company Is relevant in financial crises
Figure 18: Results of Hypotheses
VI Conclusion
The aim of this paper was to answer the set up research question, whether behavioral finance can help to understand financial crises. As there have been various financial crises in the past of which some are not explainable by the commonly used theory of an efficient market, it was interesting to test if behavioral finance can help to understand these occurrences. In order to answer the research question four hypotheses have been formulated. The results obtained are consistent with the related literature and turned out as expected.
Hypothesis number one which dealt about the amounts traded being excessively higher before a financial crisis occurs was supported. The average trading volume of the chosen companies increased before the bubble burst however the period turned out to be different. For some companies, the trading volume first increased some days before the burst of the bubble whereas for other companies this observance could be made weeks before. Overall, the results are in common with the study of Walter and Weber and Uchida and Nakagawa who both found that herding behavior exists during a bubble period.
Hypothesis number two was also supported as negative changes within the 5 minutes trading data on the three chosen stock exchanges was observed. Even if the results were expected to be stronger, the supply of the offerers increased just before the bubble burst which is a sign for noise traders. The results are in common with Caginalp and Ilieva who found that bubbles are fueled by the cash of momentum players and Komároni who stated that the intensity of noise trading occurs at the same time as stock bubble markets occurs.
Hypothesis number three who tested the validity of beta as a volatility measure in financial crises was supported as well whereas hypothesis number four was not supported. From the chosen companies, the beta coefficient was right in 70% of the cases. Thus, beta can be of help for investors when creating their portfolio. The results are in common with existing literature about Beta as Xu and Yang or Feinberg and Tokic also found support for the Beta coefficient.
market circumstances like a financial crisis and can be considered to be useful for investors when creating their portfolio.
VII Recommendations
Bibliography
Books / Publications
• Arnold G (2005), Corporate Financial Management, 3rd
Edition, London: Pearson Education Limited.
• Aschinger G (1995), Börsenkrach und Spekulation - Eine ökonomische Analyse, Verlag Franz Vahlem GmbH.
• Bae H K, Yamada T, Ito K (2006), How do Individual, Institutional, and Foreign Investors in and Lose in Equity Trades? Evidence from Japan, International Review of Finance, 6:3–4, 2006: pp. 129–155.
• Bloomfield R, O’Hara M, Saar G (2005), The Limits of Noise Trading: An Experimental Analysis, Johnson Graduate School of Management, Cornell University & Stern School of Business, New York University.
• Caginal G & Ilieva V (2008), The dynamics of trader motivations in asset bubbles, Journal of Economic Behavior and Organization, Vol. 66, Issue ¾, p641-p656.
• Cassuto A, D’Arcangelis M A, Caparrelli F (2004), Herding in the Italian Stock Market - A Case of Behavioral Finance, The Journal of Behavioral Finance 2004, Vol. 5, No. 4, 222–230.
• Demirer R, Kutan M A (2005), Does herding behavior exist in Chinese stock markets?, Journal of International Financial Markets, Institutions and Money, Volume 16, Issue 2, April 2006, Pages 123-142.
• Dupernex S (2007), Why Might Share Prices Follow A Random Walk?, Student Economic Review, Vol. 21.
• Emmett R B (2000), Great Bubbles: reactions to the South Sea Bubble, the Mississippi scheme and the tulip mania affair, The Economic History Review, Vol. 54, No. 4.
• Fama F E, French R K (2004), The Capital Asset Pricing Model: Theory and Evidence, Journal of Economic Perspectives, Volume 18, Number 3, Pages 25-46.
• Feinberg M, Tokic D (2002), Beta and Return: One-day effect, University of Texas-Pan American.
• Gill J & Johnson P (2006), Research Methods for Managers, 3rd
edition, London: Sage Publications.
• Hamberg M (2006), Strategic Financial Decisions, Malmö: Daleke Gafiska AB.
• Hirschey M (1998), How much is a tulip worth?, Financial Analysts Journal, vol. 54, no. 4.
• Jagric T, Podobnik B, Kolanovic M (2005), Does the Efficient Market Hypothesis Hold? – Evidence from Six Transition Economies, Eastern European Economics, vol. 43, no. 4, pp. 79–103.
• Kindleberger C P (1978), Maniacs, Panics and Crahses: A history of financial crises, 1st edition, John Wiley & Sons Inc.
• Komáromi G (2004), Anatomy of Stock Market Bubbles, University of Veszprém.
• Malkiel B G (1973), A Random Walk Down Wall Street, 1st
