Tài chính doanh nghiệp - Chapter 17: Capital market efficiency

Chapter 17Capital Market EfficiencyLearning ObjectivesUnderstand the concept of market efficiency.Distinguish between different categories of market efficiency.Understand the methods used to test for market efficiency. Learning Objectives (cont.)Understand the major trends in tests of market efficiency that have uncovered evidence that is ‘anomalous’ from a market efficiency viewpoint.Understand the implications of the developing field of behavioural finance for the efficient market hypothesis. Understand the implications of market efficiency for investors and financial managers.Efficient Market Hypothesis (EMH)‘EMH’: that the price of a security (such as a share) accurately reflects all available information.If the market processes new information efficiently, the reaction of market prices to new information will be instantaneous and unbiased.The concept of efficient markets with respect to information was introduced by Fama (1970).A Non-Instantaneous Price ReactionAn instantaneous price reaction would, in practice, mean that after new information becomes available it should be fully reflected in the next price established in the market.If the market often fails to react instantaneously, share traders can develop simple rules to generate excess profits. Simply purchase shares immediately after a company makes an unanticipated announcement of good news.If the reaction is not instantaneous, a positive profit will be earned.A Biased Price ReactionOverreactionA biased response of a price to information in which the initial price movement can be expected to be reversed.UnderreactionA biased response of a price to information in which the initial price movement can be expected to continue.Categories of Capital Market EfficiencyThe EMH implies that investors cannot earn abnormal returns by using information that is already available.Abnormal returns are returns over and above what would be expected as compensation for systematic risk (as suggested by the CAPM for example).The market may be efficient with respect to some sources of information, but not with respect to others.Fama (1970) provided a useful classification of market efficiency:Weak-form efficiency — the information contained in the past sequence of prices of a security is fully reflected in the current market price of that security.Semi-strong-form efficiency — all publicly available information is fully reflected in a security’s current market price.Strong-form efficiency — all information, whether public or private, is fully reflected in a security’s current market price.Categories of Capital Market Efficiency (cont.)The information content associated with each successive classification is cumulative.The implication of strong-form efficiency is that an investor cannot earn abnormal returns from having inside information. If this were true, investors would have no incentive to seek information.Paradox: the capital market can be efficient only if at least some investors believe it to be inefficient.Categories of Capital Market Efficiency (cont.)However, if the market is less than strong-form efficient, there are incentives for investors to seek information.Fama (1991) — classifies capital market efficiency by the research methodology used to evaluate efficiency.Tests of returns predictability — can future returns be predicted on the basis of past information?Event studies — analyse the behaviour of security returns around a significant event such as profit/dividend announcement or takeover bid.Tests for private information — test if investors can systematically profit from investments based on non-public information.Categories of Capital Market Efficiency (cont.)Market Efficiency and the Joint Test ProblemTesting for market efficiency involves testing for abnormal returns.We use an asset pricing model, such as the CAPM, to determine normal or expected returns.Deviations from this are classified as abnormal and could be due to market inefficiency.Market Efficiency and the Joint Test Problem (cont.)Joint test problem arises because of the possibility that the underlying asset pricing model may be incorrect.What appear to be abnormal returns due to market inefficiency may be due to a mis-specification of the asset pricing model.For example, the correct asset pricing model may be the Fama and French (1996) three-factor model.Tests of Return PredictabilityIn an efficient market, investors should not be able to earn abnormal returns by predicting future price movements.Three aspects of return predictability are:Relationship between past and future returns — in an efficient market, all information in past prices should be reflected in current prices.Seasonal effects in returns — returns should not be related to the time at which they are earned, this would lead to return predictability. Examples include, time of year, month, week and even time of day.Tests of Return Predictability (cont.)Three aspects of returns predictability are (cont.):Predicting returns on the basis of some other forecast variable — should not be able to earn abnormal returns by selecting stocks based on company size, book-to-market ratio, dividend yield or some other variable.Excess returns associated with such factors can be treated as risk premiums.Relationship between Past and Future ReturnsShort-term patternsEarly tests of short-term patterns in returns focused on random walk hypothesis.This was too strict a requirement to show market efficiency.A common way of testing for these patterns is to test for serial correlation.Relationship between Past and Future Returns (cont.)Short-term patterns (cont.)Serial correlation testsMeasure the correlation between successive price changes or, more frequently, the correlation between yields in one period and yields in a prior period.Runs tests.An alternative test for independence. Examine the sign of price changes, or yields, during a specified sample period.In general, tests for patterns suggest that successive price changes (yields) are uncorrelated.Relationship between Past and Future Returns (cont.)Short-term patterns (cont.)US evidence is mixed:Lo and MacKinlay (1988) find positive serial correlation in weekly returns.A re-examination of more recent data by Lo and MacKinlay (1999) finds the result is no longer as strong as previously found.Australian evidence is that returns are more predictable over shorter periods.Brailsford and Faff (1993) find positive serial correlation in daily returns for top 50 stocks on ASX.Brailsford (1995) examines returns over 5-minute intervals within trading day — strong evidence of positive serial correlation.Relationship between Past and Future Returns (cont.)Long-term patternsAnother group of researchers have suggested that there may be other kinds of inefficiency that will be detected only by taking a long-term view.Pricing errors that are eliminated so slowly that it could take years for the market to eliminate the error.Fama and French (1988) find evidence of this kind of behaviour in stock returns.Jegadeesh and Titman (1993) find evidence of long-run serial correlation.Several studies have found evidence of long-run share return reversals, Lee and Swaminathan (2000) and Jegadeesh and Titman (2001) for the US and Brailsford (1992) and Allen and Prince (1995) for Australia.Presence of Seasonal Effects in ReturnsMonthly share price patternsUS — the average return in January is more than 5 times larger than the average returns in the other 11 months.Australia — December and January together contribute more than half the yearly return.Possible explanation of ‘tax loss selling’:Investment strategy in which the tax rules make it attractive for an investor to sell certain shares just before the end of the tax year. Turn-of-the month effect:On average, share prices increase abnormally at the start of each month.Evidence of such effects in the US, Australia, UK, Switzerland and West Germany.Presence of Seasonal Effects in Returns (cont.)Daily share price patternsUS average returns — negative on Mondays and positive on Fridays.Recent evidence for US shows Monday provides best return and Thursday provides worst return — Rubenstein (2001).Australia — Thursdays positive, Tuesdays negative.Recent evidence that negative Tuesday return has disappeared.Large positive average returns on the days preceding public holidays.Presence of Seasonal Effects in Returns (cont.)Intraday share price patternsEvidence of systematic returns during the trading day.Hodgson (1992) finds positive returns in opening and closing trading intervals (10 minutes) and negative returns mid-morning and afternoon.Positive return on opening may be result of overreaction to news released during market’s closure.Predictions Based on Other Forecast VariablesSize effect or small-firm effectThe observation that returns on the shares of small companies exceed the returns on the shares of larger companies both before and after adjusting for beta risk.Possible explanation — shareholders in small companies require a higher expected return, to compensate for the lower liquidity of their investment.Cause of the size effect is still unclear.Predictions Based on Other Forecast Variables (cont.)The dividend yield effectDividend yield is the dividend per share divided by the share price.Beta-adjusted returns are found to be higher, the higher the dividend yield.That is, there appears to be a relationship between returns and dividend yields which cannot be explained by the capital asset pricing model.Predictions Based on Other Forecast Variables (cont.)Price–earnings effectPrice–earnings ratio — share price divided by earning per share (P/E ratio).The P/E effect — even after adjusting for beta, there is a relationship between share returns and P/E ratios.In Australia — P/E effect appears to be weaker but has been documented for larger companies.Predictions Based on Other Forecast Variables (cont.)Book-to-market effectBook-to-market ratio — book value of a company's equity divided by market value of the company’s equity.Book-to-market effect — even after adjusting for beta risk, there is a relationship between share returns and book-to-market ratios.Predictions Based on Other Forecast Variables (cont.)Book-to-market effect (cont.)Fama and French (1992) found that companies with low book-to-market ratios tended to earn low returns, while companies with high book-to-market ratios tended to earn high returns.Recent Australian evidence is mixed — Halliwell, Heaney and Sawicki (1999) finding no effect, while Faff (2001) using a later sample finds evidence of the book-to-market effect.Event StudiesEfficient market hypothesis requires security prices to adjust instantaneously to an event, such as a public announcement.Research testing whether there are any post-event abnormal returns.Such a test is generally called an event study: A research method that analyses the behaviour of a security’s price around the time of a significant event such as the public announcement of the company’s profit.Event Studies: MethodologyBasic questions to be answered for an event study:What is the (new) information?When was it announced?Were there abnormal returns associated with its announcement?Each of these questions will be considered using the announcement of annual profit as an example.Event Studies: Methodology (cont.)New information?An annual profit figure provides information only if the announced profit differs from the profit expected by investors.The effects of the expected profit will already be reflected in the share price before the announcement.Only the unexpected part of the reported profit should cause the share price to react.Event Studies: Methodology (cont.)When was it announced?Ideally, the exact moment of the event (announcement) needs to be identified.Important because the market may react in anticipation of the announcement as investors revise their expectations.The market should also react at the time of the announcement to any unanticipated information.Event Studies: Methodology (cont.)Were there abnormal returns associated with its announcement?Calculating the response of the market to the announcement.In essence, the response is the percentage change in share price in excess of (or below) the percentage change that would normally be expected.Some model of ‘normal’ security price movement is needed.Event Studies: Methodology (cont.)Tests of market efficiency are, therefore, simultaneously tests of the pricing model used to estimate what is ‘normal’.Most studies use some variant of the market model as a basis for estimating the normal rate of return on a security.Event Studies: Methodology (cont.)The standard market model: where Event Studies: Methodology (cont.)Abnormal returns before, after and at the time of the announcement are estimated.A total sample of companies and announcement dates will be gathered.At each point in event time the average abnormal return across companies is calculated.Event Studies: Methodology (cont.)Aitken, Brown, Frino and Walter (1995) studied share returns for half-hour intervals before and after profit announcements.Split sample into good and bad news sub-samples.Finding is that the only half-hour return that was statistically different from zero was the first half-hour following announcement.Unexpected news is fully incorporated into price within half an hour.Event Studies: Australian Profits and Dividends In Australia, profit and dividends are usually announced simultaneously.Brown, Finn and Hancock (1977) found:Announcements of an increase (decrease) in dividends per share results in an increase (decrease) in abnormal returns.The information content of profit and dividend announcements is increased when they are in agreement.Investors reacted quickly to new information.Event Studies: Australian Profits and Dividends (cont.)Easton and Sinclair (1989)Found that the market’s reaction to dividend announcements was weaker than the reaction to profits.Easton (1991) finds that share price response depends on interaction between dividend and profit signals.This is similar to findings in Brown, Finn and Hancock (1977).Event Studies: Australian Profits and Dividends (cont.)Other eventsCapitalisation changesAudit qualificationsTakeoversBlock tradesExtraordinary itemsMoney supply announcementsCurrent account balance announcementsTests for Private InformationStrict view of EMH requires that abnormal returns be unavailable even to investors who have private (inside) information about a company.However, it is effectively impossible to identify the date on which private information becomes available.Event study methodology cannot be applied directly to studies of private information and the EMH.Tests for Private Information (cont.)One less direct test of strong-form efficiency involves identifying those investors who might be assumed to have access to private information, and then determining whether they earn abnormal returns. In 1995, amendment to Corporations Act requiring company directors to disclose trades in own company stock.Inefficiency would see directors profiting from trades, selling before drops and buying before rises in share price. Tests for Private Information (cont.)Evidence suggests directors’ sales were profitable, but purchases were not.Evidence also shows that directors profit from takeovers, executive directors first and later non-executive directors, as they may have poorer access to information.Tests for Private Information (cont.)Evidence relating to both Australia and the US supports the conclusion that neither market is strictly (strong-form) efficient.This is not surprising, because inside information will be costly (and often impossible) for an outsider to obtain, and, because of legal implications, may prove costly for an insider to use.Behavioural Finance and the EMH Behavioural finance — study of human fallibility in competitive markets.Market sentiment plays a role in market inefficiency.Why do we observe herd mentality behaviour, which may be irrational? Behavioural finance suggests that arbitrage is risky when market participants are behaving irrationally.Behavioural Finance and the EMH (cont.)This irrational behaviour can arise from a belief that a share price will continue to rise (fall) and if it is not purchased (sold) immediately a profit (loss) will be forgone (ensured).This type of behaviour can lead to the formation of what is often referred to as asset price bubbles.Behavioural Finance and the EMH (cont.)Price ‘bubbles’Share prices might display bubbles:Prices show a strong tendency to rise for a period, possibly followed by a decrease, which may be quite sudden.As a result, the price departs from the true, fundamental value of the asset.While the first feature can be observed, it does not necessarily imply the presence of the second feature.Whether price bubbles occur is a controversial issue.Behavioural Finance and the EMH (cont.)Bubbles can be rational if there is a belief of high and increasing future profits.These beliefs justify rising share prices.If there is a change in expectations, such as profits not materialising or growing, the share price will fall dramatically, like a bursting bubble. Behavioural Finance and the EMH (cont.)Surveys of behavioural finance framework can be found in Shleifer (2000), Hirshleifer (2001) and Shiller (2002).Fama (1998) defends EMH against behavioural finance criticisms:Price over-reactions and under-reactions are as common as each other — consistent with EMH.Many results indicating market inefficiency are methodology-driven and as such are not true reflections of inefficiency.Substitute for EMH requires a theory of price formation that can be tested.Implications of evidence on the EMHSuppose that, notwithstanding the evidence on anomalies, the stock market is semi-strong-form efficient.Then there are clear implications for investors in securities and for financial managers.Implications of the EMH for Investors in SecuritiesCharting, or technical analysisPlotting a share’s historical price record on a chart and then using this as the basis for predictions as to the likely future short-term course of prices.However, the alleged benefits of this type of analysis are dubious.Batten and Ellis (1996) find that for Australian shares, technical trading rules did not deliver abnormal returns after transaction costs.Lo, Mamaysky and Wang (2000) find small positive returns in US data.Implications of the EMH for Investors in Securities (cont.)Fundamental analysisBased on the belief that the market either ignores some publicly available information, or systematically misinterprets that information.Therefore, careful analysis of available information may reveal mis-priced securities, and therefore excess returns can be made by the skilled fundamental analyst.Implications of the EMH for Investors in Securities (cont.)Random selection of securitiesWhile the EMH asserts that all securities are ‘correctly’ priced, given the information available, this does not imply that investors should select their investments randomly.A randomly-chosen portfolio is likely to have a similar risk to the market portfolio, however, this may not suit the risk preferences of the investor.Investors should also consider their tax position when selecting investments, which is unlikely to be suitable if investments are selected randomly.Implications of the EMH for Investors in Securities (cont.)Buy-and-hold policiesA strategy in which shares are bought and then retained in the investor’s portfolio for a long period.Barber and Odean (2000, 2001) find support for buy and hold — their study finds 20% of households that trade most regularly earn 8% lower return than the average for whole sample.An inflexible buy-and-hold policy is not optimal.Portfolio will need to be rebalanced at times because, as share prices change ov