University:
Boston University
Course:
CAS EC 203 | Empirical Economics 1
Academic year:

2024
Views:

161

Pages:

19
Author:

krupicom2m1n

The efficient market hypothesis and the design of investment activity 1 How should one design an investment strategy? • One option: active trading based on – – – – Information (“news”) Market trends (“momentum”) Tips from brokers Research that isolates stocks that will perform unusually well in the coming years • Another option: build or buy a diversified portfolio based on risk and return characteristics of stocks, making trades only to rebalance the portfolio’s composition and to manage tax liabilities arising from the portfolio 2 Economists and financial analysts differ widely on these strategies • Active individual investors and investment managers are working on the assumption that some or all of the four considerations above are the source of very large returns • In this lecture, we’re going to explore why some other individuals and investment managers do not agree. • Academic research is divided on this subject; many prominent financial economists also actively work with (or have formed) firms that implement their investment strategies. 3 A basic model of stock price determination and return dynamics • Stock price is the expected present discounted value of dividends (or earnings) 1 pt Et [ pt +1 + d t +1 ] = 1+ r 1 1 Et [( Et +1[ pt +2 + d t +2 ]) + d t +1 ] 1+ r 1+ r 1 2 1 1 2 ) Et pt +2 + ) Et d t + 2 = ( Et d t +1 + ( 1+ r 1+ r 1+ r ∞ 1 j = ∑ j =1 ( ) Et d t + j 1+ r 4 Components of model 1. (d,p): dividend at t and price of asset at t (stated “ex dividend”). If you own a stock at the start of period t, then you have claim to p+d. If you trade it, you sell it for p. 2. (r): the return on the stock, including compensation for its risk (discussed further below), necessary for investors to hold it. 3. (Et): the rational expectation of market investors based on a particular information set. Has the property that the “law of iterated expectations” holds in that Et (Et+1xt+k)= Etxt+k etc. 5 Implication for return pt +1 + d t +1 − pt Holding period return: ht +1 = pt pt +1 + d t +1 − pt Expected return: Et ht +1 = Et [ ] pt pt (1 + r ) − pt ] r = [= pt Unexpected return: u= ht +1 − Et ht +1 t +1 6 EMH is sometimes discussed as involving “random walks/martingales” • Random walk: pt+1=pt+et+1 • Martingale: Ept+1 |{pt, pt-1,…}= pt • Appropriate for security when it is not paying dividend (high frequency) • Usually interpreted as: “past returns should not forecast future returns” 7 EMH and the properties of expectation errors • In an efficient market, this error (unexpected return, excess return) should have conditional expectation zero given information available to all investors. • Why? If not, then one can devise a profitable trading strategy. • Suppose et+1 = b zt, where z is information that you have at t and b is the sensitivity of returns to this information. Think of z as having a zero mean, but sometimes being positive and sometimes negative. For simplicity, let b be positive so that positive z induces high returns and negative z induces low returns. 8 Trading strategy based on information • If z is positive, buy the stock, earning an extra return of bzt. Hold it for a period and then sell it. • If z is negative, short the stock, earning an extra return of bzt. Hold it for a period then sell it. • On net, you have earned r+2bzt by following this active strategy. The term 2 bzt is your profit. 9 Efficient market test • Simple and direct test Regression: h t +1 =r + ut +1 =r + bzt + et +1 Test: EMH implication that b=0 • One type of z could be past returns or any function of past returns. Others could be published earnings forecasts. Others could be events of importance to firm (earnings announcements, etc.) 10 What types of tests? • Weak-form efficiency – General: Excess returns can not be earned by using investment strategies based on historical share prices. – Specific: returns exhibit no serial dependence – Technical analysis techniques (trading rules based on patterns in stock returns) will not be able to consistently produce excess returns 11 What types of tests? • Semi-strong-form efficiency – General: Share prices adjust to publicly available new information very rapidly and in an unbiased fashion, such that no excess returns can be earned by trading on that information. – Specific: technical analysis techniques based on factors such as price, volume, and open interest will be unable to reliably produce excess returns. – Specific: adjustments to news and events must be instantaneous, not forecasting future return behavior. Methods: regressions, event studies 12 Event study • An example of an event study (cumulative average excess returns (h-r) to an event at t) • Suppose that there is an economic disaster for a firm – it owns oil tankers and one is taken over by pirates in the Gulf – then what should happen to stock returns before and after this event? • Excess returns would fluctuate around zero before the event and after the event. They would be sharply negative on the day of the event. Trading on this information the next day would be useless: it would already be in the stock price. 13 Stock splits (Fama, Fisher, Jensen, Roll, IER 1969) 14 Stock splits: cumulative average residuals 15 What types of tests? • Strong form efficiency – General: stock prices fully reflect all information and no one can earn excess returns (returns exceeding compensation for risk) – Specific: trades made by individuals associated with a firm should not be profitable (why insider trading laws?) 16 Criticisms of EMH, logically and practically • Can’t be logically true in its strongest form (prices fully reflect all available information) for reasons to be discussed in later in class lecture (Grossman and Stiglitz article) • Is subtle to test when there are time-varying expected returns required for risk reasons, as stressed by Fama (originator) and Schwert. In this case, it is hard to know whether z affects expected return or unexpected return. 17 Criticisms of EMH • Behavioral finance – Many actual market participants do not forecast rationally and use information efficiently. – They are subject to psychological biases – Hence, there are profit opportunities for more rational investors 18 A behavioral finance advocate on trading • Jay Ritter: It is very difficult to find trading strategies that reliably make money. This does not imply that financial markets are informationally efficient, however. Low frequency misvaluations may be large, without presenting any opportunity to reliably make money. As an example, individuals or institutions who shorted Japanese stocks in 1987-1988 when they were substantially overvalued, or Taiwanese stocks in early 1989 when they were substantially overvalued, or TMT stocks in the U.S., Europe, and Hong Kong in early 1999 when they were substantially overvalued, all lost enormous amounts of money as these stocks became even more overvalued. Most of these shortsellers, who were right in the long run, were wiped out before the misvaluations started to disappear. Thus, the forces of arbitrage, which work well for high frequency events, work very poorly for low frequency events. 19