The Efficient Market Hypothesis and its Validity in Today's Markets

Content

Table of Abbreviations

Table of Symbols

1. Introduction

2. What is the Efficient Market Hypothesis?
2.1. Definition of Market Efficiency
2.1.1. The Three Forms of Market Efficiency
2.1.2. Limits of Efficiency
2.1.3. Questions Left Unaddressed
2.2. A Short History of the Efficient Market Hypothesis

3. Explanatory Models of Market Behaviour
3.1. Expected Return Efficient Market Models
3.1.1. The Fair Game Model
3.1.2. The Submartingale Model
3.1.3. The Random Walk Model
3.1.4. The Capital Asset Pricing Model
3.1.5. The Market Model
3.2. Variance Efficient Market Models
3.3. Behavioural Models

4. Testing for Market Efficiency
4.1. Tests for Weak-Form Efficiency
4.2. Tests for Semistrong-Form Efficiency
4.2.1. Event Studies
4.2.2. Mutual Fund Studies
4.3. Tests for Strong-form Efficiency

5. Evidence of Market Efficiency
5.1. Market Anomalies
5.1.1. Serial Correlation
5.1.2. Return Seasonality
5.1.3. Stock Price Reaction to Index Inclusion
5.1.4. Neglected-Firm Effect and Liquidity Effect
5.2. Insider Information
5.2.1. Legislative Treatment of Insiders
5.2.2. Evidence from U.S. Markets
5.2.3. Evidence from International Markets
5.3. Mutual Fund Performance
5.3.1. Methodology of Mutual Fund Studies
5.3.2. Evidence of Mutual Fund Performance
5.4. Short-Term Momentum and Long-Term Reversal
5.4.1. Evidence for Trend Reversal
5.4.2. Evidence for Momentum
5.4.3. Possible Explanations for the Momentum and Reversal Effects
5.5. Evidence from the Austrian Market
5.5.1. Pichler’s 1993 Test
5.5.2. Aussenegg and Grünbichler’s 1999 Study
5.5.3. Mestel and Gurgul’s 2003 Test
5.5.4. Gurgul, Mestel, and Schleicher’s 2003 Test
5.5.5. Other Evidence
5.6. Evidence from the German Market
5.6.1. Uhlir’s 1984 Review
5.6.2. Krämer and Runde’s 1993 Study
5.6.3. Glaser and Weber’s 2003 Study
5.6.4. Other Evidence
5.7. Evidence from Bond Markets

6. Summary and Conclusion

Bibliography

Table of Abbreviations

illustration not visible in this excerpt

Table of Symbols

illustration not visible in this excerpt

1. Introduction

The efficient market hypothesis is one of the most important paradigms in modern finance and was largely accepted to hold by the early 1970s. In 1978, Michael Jensen declared his belief that “there is no other proposition in economics which has more solid empirical evidence supporting it.”^[2]

Market efficiency since then has become the basis of numerous financial models and forms the foundation of the investment strategies of many individuals and corporations. Because of the efficient market hypothesis, technical analysts have become the target of widespread criticism and passive management has seen a boom in recent years.

Despite these impressive credentials, several cracks have appeared in the efficient market edifice over recent years. More powerful statistical techniques, the advent of affordable computing and the fact that cheap data storage lead to an explosion in the amount of data available to researchers have handed the academic community more sophisticated tools for empirical studies. While modern financial research was able to discount many previous reports of market inefficiency on grounds of new statistical insights, some anomalies were confirmed and today form part of a growing body of literature at odds with the efficient market hypothesis.

It is the intention of this thesis to explain the reasoning behind and the implications of the efficient market hypothesis and give a short overview of the literature and research documenting its validity – or lack thereof – in the world’s financial markets.

Chapter 2 gives a definition of market efficiency and offers a short history of its development from the first formulation to the present day. Since models of market processes form the foundation of research into market efficiency, the most important such models are introduced in chapter 3. Chapter 4 presents different methodologies to allow the testing of the efficient market theory using empirical data. Chapter 5 gives an overview of the results of such empirical testing reported in the literature.

2. What is the Efficient Market Hypothesis?

In 1953 Maurice Kendall published a study^{[3] [4]} in which he found that stock price movements followed no discernible pattern, that is, they exhibited no serial correlation. Prices were as likely to go up as they were to go down on any given day, irrespective of their movements in the past. These results lead to the question of what, exactly, influenced stock prices. Past performance obviously did not. In fact, had this been the case, investors could have made money easily. Simply building a model to calculate the probable next price movement would have enabled market participants to gain large profits without (or with reduced) risk. On the other hand, if everybody could have done so, stocks that were about to rise would have risen instantly, because large numbers of investors would have wanted to buy them, while those holding the stock would not have wanted to sell. This mechanism suggests that the market “prices in” the performance data that is already available about a stock.

2.1. Definition of Market Efficiency

The concept of efficiency adopted for this thesis is one regarding the incorporation of information into security prices.

Generalizing from the results of the above paragraph leads to the proposition that any available information which could influence a company’s stock performance should already be reflected in said company’s stock price. In an efficient market, therefore, security prices should equal the security’s investment value, where investment value is the discounted value of the security’s future cash flows as estimated by knowledgeable and capable analysts.^[5]

Under this definition, the one thing that can still influence stock prices is new information. When new information about a company becomes available, the above process makes stock prices move immediately to reflect the new situation.^[6] Naturally, this new information needs to be unpredictable; otherwise the prediction about the new information (which is itself a piece of information) would already have caused share prices to change.

These considerations suffice to formulate the efficient market hypothesis. In its original postulation, it stated that “an efficient market is one in which stock prices fully reflect available information.”^[7] Later texts have weakened this definition to allow for prices to be sufficiently different from full-information prices for investors to become informed.^[8]

A good description of market efficiency and the underlying mechanics is the one by Cootner (1964):

“If any substantial group of buyers thought prices were too low, their buying would force up the prices. The reverse would be true for sellers. Except for appreciation due to earnings retention, the conditional expectation of tomorrow’s price, given today’s price, is today’s price.

In such a world, the only price changes that would occur are those that result from new information. Since there is no reason to expect that information to be non-random in appearance, the period-to-period price changes of a stock should be random movements, statistically independent of one another.”

In a perfect market, these criteria are obviously fulfilled. In such a market transactions and information are costless, implying that market participants have full information and can react to news without incurring costs. Nevertheless, while perfect markets are a sufficient assumption for market efficiency, they are not a necessary condition.^[9]

2.1.1. The Three Forms of Market Efficiency

In economic and financial theory a distinction is made between three forms of market efficiency. The basis of this separation is what is meant by the term “all available information”. Each stronger form of efficiency incorporates all weaker forms of efficiency.

In weak-form efficient markets stock prices reflect market trading data and information derived from it. Examples of market trading data are past prices, volume or short interest.^{[10] [11]} This data is generally easily available and, according to this theory, should therefore be reflected in current prices. If weak-form efficiency holds, stock prices should be composed only of three components – the last period’s price, the expected return on the stock and a random error term which has an expected value of zero. This random error is due to new, unexpected information released in the period observed. Their relationship can be expressed as follows:^[12]

illustration not visible in this excerpt

Weak-form efficiency is a prerequisite for many asset pricing models (CAPM, APT) and for option valuation models following the Black/Scholes and Cox/Ross/Rubinstein methods.^[13]

Semistrong-form efficient markets reflect all publicly available information influencing the company’s value in stock prices immediately. Above and beyond the requirements of weak-form efficiency, in this model data about the firm’s products, operations, balance sheet, patents, etc. is priced in.

Strong-form efficiency postulates that stock prices reflect all relevant information, including information which is only known to company insiders. This definition implies that these insiders, who are privy to information before it becomes known to the rest of the market, also cannot earn any excess profits. In theory, if these individuals tried to trade on their information, the market would recognize the attempt and prices would adjust before the trade could go through. In practice, the insider trading rules of the SEC and similar regulatory bodies aim at preventing insiders from profiting from their superior knowledge by prohibiting insider trading for said individuals, their relatives and anybody who is supplied with their information. In the academic community there are also proponents of the view that, if insider information is not available to investors, strong-form efficiency can be considered to hold regardless of whether the availability of insider information would lead to excess returns.^[14]

Strong-form efficiency, naturally, is the most contested of the three models. For it to hold, weak and semistrong-form efficiency have to hold first.

2.1.2. Limits of Efficiency

As was already stated in section 2.1 of this text, strict interpretation of market efficiency implies that an efficient market is one in which stock prices fully reflect all available information.^[15] As was also indicated above, weaker interpretations have been proposed since this first formulation.

A somewhat more practical interpretation and in fact the one adopted by a fair part of the academic community is that in an efficient market stock prices reflect all available information to a degree that prevents investors from earning excess returns net of transaction costs. The exact composition of the transaction costs considered varies from author to author. Most members of the academic community agree that the minimum trading commissions that floor traders on major exchanges must pay should define a minimal tolerance band around strict-sense efficient prices. Prices that come to lie within that bandwidth, according to this interpretation, would still be considered efficient even if different from strictly correct prices.

Grossman and Stiglitz^[16] go even further and argue that with costly information, for a market to come close to efficiency some arbitrageurs must be able to earn excess profits. They show that only if for some investors arbitrage is profitable, they will, through their actions, keep prices close to their theoretically ideal value for the rest of the market. Should on the other hand no investor be able to earn abnormal profits which pay for the costly procurement and analysis of information, no investment in information will take place. This implies that the assumption of costless information, as present in much of the theoretical literature, is a necessary condition for prices to fully reflect all available information. Since costless information is unrealistic in practice, so are fully efficient capital markets. The question, therefore, should not be whether a given market is efficient but rather to what degree efficiency holds.

2.1.3. Questions Left Unaddressed

In real-world markets the term “all available information” from the above definition must include insider information, which obviously is available to some individuals. Current evidence indicates that the theoretical ideal for such a market, strong-form market efficiency, is too strict a concept to correctly describe reality.

With that in mind, Schwartz^[17] asks whether markets are more efficient when a particular bit of information is known to no market participant or when it is known to some. In the first case, he argues, there is an equal distribution of information and gaining excess profits due to that bit of information is impossible for all investors. In the second case, markets have broader, more complete information, but offer excess profits for investors privy to it.

This consideration is similar to what might be asked regarding rules which forbid insider trades. Such regulation is in place in most western capital markets. One might wonder whether markets should be considered more efficient if investors who hold information which was not yet made public are allowed to trade – thus bringing prices closer to their “real” value – or if they are not.

2.2. A Short History of the Efficient Market Hypothesis

Bachelier (1900) laid the theoretical groundwork for the efficient market hypothesis, which was postulated half a century later by Maurice Kendall (1953). Kendall found that stock prices evolved randomly and that his data offered no way to predict price movements. While the explanation for this phenomenon, the efficient market hypothesis, initially seemed counterintuitive to the academic community, scholars quickly embraced the theory and began to document its validity in real-world markets by studying empirical data.^[18]

These early studies got a strong methodological boost from the formulation of the capital asset pricing model of Sharpe (1964), Lintner (1965) and Mossin (1966). The CAPM allowed researchers to price securities in efficient markets and spawned a large number of studies based on this and subsequent asset pricing models. Most such studies found evidence consistent with the hypothesis and by the early 1970s, markets were largely considered to be efficient in the semistrong form, as defined by Fama (1970).

By the early 1980s, however, doubts appeared about this earlier stance. Several effects which are described later in this text, like serial correlation and the turn-of-the-year effect, were observed in practice. In his 1978 article, Jensen already hinted at these rising doubts regarding the efficiency of capital markets when he said:

“…we seem to be entering a stage where widely scattered and as yet incohesive evidence is arising which seems to be inconsistent with the theory.”^[19]

Over the course of the 1990s and the first years of the new millennium, several anomalies saw intensive scrutiny which solidified their existence and persistence until they could no longer be explained away by even the staunchest of supporters of the efficient market theory. Among these anomalies are the turn-of-the-year effect (and the related time of the month effect, day-of-the-week effect and time-of-the-day effect) and the trend of stock prices to exhibit short-term momentum and long-term reversal. The last decade brought many new insights regarding especially the latter phenomenon and to the present day the reason for its existence has not been satisfactorily established.

3. Explanatory Models of Market Behaviour

Tests of market efficiency rely on one or more asset pricing models for the derivation of their conclusions. Each empirical study analyzing an aspect of market efficiency, therefore, is a joint test of market efficiency and the asset pricing model used.^[20] This chapter gives an overview of the models most commonly employed in the academic literature covering the topic.

3.1. Expected Return Efficient Market Models

Expected return efficient market models are by far the most common frameworks used in market efficiency studies. The following section gives an overview of the most important such models in the literature but does not claim completeness.

3.1.1. The Fair Game Model

Fama (1970) lists three models describing the characteristics inherent to an efficient market – the fair game, random walk and martingale/submartingale models. What they have in common is their reliance on expected returns for the description of market efficiency. Putting this common characteristic into mathematical notation yields the following equation:^[21]

illustration not visible in this excerpt

This above definition leads to viewing expected return efficient market models as “fair game” models. The rationale of fair game efficient market models is that no trading system based solely on the information set

illustration not visible in this excerpt

3.1.2. The Submartingale Model

The submartingale model is a special case of the fair game efficient market model and the martingale model. Adding to (2) the assumption that^[22]

illustration not visible in this excerpt

are always positive in this model, no trading strategy involving only one security and a cash holding can outperform a simple buy-and-hold strategy.^[24] The same cannot be said of the fair game efficient market model, where securities may well have negative expected returns. This is especially true of securities which offer a negative correlation with the market, thus compensating their low or negative returns by providing a reduction of total risk when added to a well-diversified portfolio.^[25]

Due to this difference, the submartingale model provides venues for testing other than those supported by the more general fair game model described above.

3.1.3. The Random Walk Model

Another special case of the “fair game” efficient market model, the random walk model requires that successive price changes be independent and identically distributed. Formally, these conditions can be expressed as follows:

illustration not visible in this excerpt^[26]

The random walk model does not imply that past prices contain no information about future return distributions. On the contrary, if the random walk hypothesis is confirmed, past prices constitute the best information for forecasts. What the model does imply is that the past price sequence cannot be used to obtain information about future price sequences.^[27]

The random walk model thus adds conditions to the more general fair game model which naturally makes it less likely to pass tests for market efficiency.

3.1.4. The Capital Asset Pricing Model

The capital asset pricing model (CAPM) was formulated by William Sharpe (1964), John Lintner (1965) and Jan Mossin (1966) and has had a turbulent history since its inception. Although it does not fully withstand empirical tests, it is still in widespread use in the investment community because of the insights it offers and because its accuracy suffices for many applications.^[28]

The CAPM is based on the assumption that investors hold diversified portfolios. If so, the variance of a security is not an appropriate measure of its riskiness, as investors are more concerned with the security’s contribution to their portfolios’ variances. As a measure of this contribution, the model uses the security’s covariance with a value-weighted portfolio of all securities in the market, standardized over this market portfolio’s variance.

Apart from an investment in securities, the CAPM also offers the possibility of a long or short position in a risk-free asset. These assumptions lead to the following equation:^[29]

illustration not visible in this excerpt

Put into words, the expected return on a security can be split up into the return on the risk-free asset and the excess return over the risk-free rate of the market portfolio times the security’s standardized covariance with that market portfolio. Thus, beta is a factor expressing the sensitivity of the security’s returns to movements in the excess returns over the risk-free rate of the market portfolio.

[...]

---

^[1] To facilitate the study of basic literature quoted in this thesis, symbols were generally not changed from the notation used in the papers quoted. Nonetheless, where ambiguities would have resulted from this policy, clarifying alterations were made so each symbol has only one, unambiguous meaning.

^[2] Jensen (1978), p. 95.

^[3] Cp. Bodie/Kane/Marcus (2002), pp. 340-43.

^[4] Cp. Kendall (1953).

^[5] Cp. Sharpe/Alexander (1990), pp. 79.

^[6] Since trading is a non-continuous process which involves discrete instances of price setting, the exact meaning of “immediately” can be clearly specified. That is, the first trade in a security after new information about that security has been made public should already take place at the new, informationally correct price. While this interpretation of market efficiency is the one most generally used in the literature, it is not the only one available. May (1991), for example, requires of efficient markets that they anticipate information. He states that markets can be considered efficient if the price correction process induced by a piece of new information is completed before that information becomes known. He does not, however, elaborate on how this adjustment process is to come about. (Cp. May (1991), p. 314. [In German])

^[7] Cp. Ross/Westerfield/Jaffe (2002), p. 341.

^[8] Cp. Grossman (1976) and Grossman/Stiglitz (1980).

^[9] Cp. Fischer (2003), p. 48 [in German].

^[10] Adapted from Bodie/Kane/Marcus (2002), pp. 342-343 and Ross/Westerfield/Jaffe (2002), pp 343-347. See also Fama (1970).

^[11] Many authors limit the information set under weak-form efficiency to past prices. While the choice between these two definitions is important for some applications, it is inconsequential for this thesis.

^[12] in this expression is the absolute return on the security. If interpreted as the percentage return, the formula reads .

^[13] Cp. Pichler (1993), p. 118. [In German]

^[14] Cp. Fama (1970), pp. 409-10. See also section 2.1.3 below.

^[15] Cp. Ross/Westerfield/Jaffe (2002), p. 341.

^[16] Cp. Grossman/Stiglitz (1980), p 393.

^[17] Cp. Schwartz (1970), p. 423.

^[18] Cp. Ziemba (1994), pp. 200-201.

^[19] Cp. Jensen (1978), p. 95.

^[20] Cp. Fama (1991), p. 1575-1576.

^[21] Cp. Fama (1970), p. 384-387. For a discussion of Fama’s models, see LeRoy (1976) and Fama (1976).

^[22] Adapted from Fama (1970), p. 386.

^[23] If (5) holds as an equation, the stochastic process follows a martingale.

^[24] In Jensen (1987), p. 97, Jensen defines this model as follows: , where is the required return on the asset for period and is the expected end of period price conditional on information set . This notation shows clearly that the amount a security’s price in one period is expected to exceed the prior period’s price is exactly the return required on that asset.

^[25] Information taken from a lecture in Corporate Finance by o.Univ.-Prof. Dr. Peter Steiner, Institute for Banking and Finance, Karl-Franzens Universität Graz, held on November 11, 2003, in Graz.

^[26] Adapted from Fama (1970), pp. 386-387.

^[27] Cp. Uhlir (1979), p. 38.

^[28] Cp. Bodie/Kane/Marcus (2002), p. 259 and pp. 263-274.

^[29] The assumptions listed here are naturally simplified. In addition to those mentioned, the CAPM also relies on the assumptions of rational investors, perfect markets, identical holding periods, zero taxes and transaction costs and homogeneous expectations.

^[30] For most applications beta is assumed to be stable over time despite empirical findings to the contrary. The same is true of the risk-free rate.