Investments 4
Economic Significance
"Effects are small, eaten up by transaction costs." What can be done? From an academic perspective, economic size not always our concern. May care more about what patterns tell us about beliefs and behavior of market participants. Some effects are clearly large: Small stocks do 10x better than large stocks in January. Momentum also very large.
Interpretation
"OK, I see the patterns, but how could managers possibly predict US bond market returns? Interest rates are set by the Fed! No manager has inside info about that! What's going on?" Idea: managers use rules of thumb (e.g., "issue short-term when short-term rates are low") Actually seems to work! ... seems like CFOs are trying to take advantage of this effect
Wrong Control for Risk?
"Predictability of abnormal returns sometimes sensitive to benchmark" What can be done? Try many E[R] models. If find same pattern in abnormal returns, more likely to be genuine In some cases, risk implausible: Calendar effects, lead-lags Circumstantial evidence that some predictability reflects mispricing (e.g., see more predictability where arbitrage is more limited; see "supply response" often predicts returns)
Irrational Managers
"Rational managers, irrational markets" is a useful paradigm. But managers are human, too. Overconfidence Optimism, etc. Can an irrational managers, rational markets perspective shed further light on managerial behavior? Not a developed literature (yet) As far as indicators of ... when we're going to come out of the recession, you know, remember, I'm a P.E. [Physical Education] graduate, not an economist. So, I don't know that I can speak to that with any credibility or anything. - Bernard Ebbers (WorldCom CEO) Excerpt from a conference call with securities analysts in 2/2002 Source: CNNMoney, 1/26/05, "Ebbers: P.E. graduate, not an economist"
Violations of Weak Form EMH
...
Violations of semistrong EMH
...
Economic Planning
2. Economic planning If correct, prices give useful information about expected economic growth, discount rates, volatility Bad prices compromise business and consumer planning
Corporate Governance
3. Corporate governance If correct, securities prices can play key governance roles Incorrect prices compromise this role CEOs wrongly fired or wrongly retained... Was Enron's price correct?
Allocation of Capital
5. Allocation of capital across firms Correct prices facilitate efficient allocation of capital across and within firms and across countries wrong prices give markets misleading signals as to where economy should best allocate resources "Tobin's Q" = Market value of installed capital / replacement cost Sometimes approx with market-to-book assets ratio If market prices (numerator) are correct, Q guides capital efficiently Rule of "if Q>1 invest more, if Q<1 invest less" is similar to taking all NPV>0 investments But if prices are wrong, Q gives wrong signals
Testing Strong Form EMH
A "strong form efficient" market incorporates all information, public and private Implies that even insiders can't make abnormal profits But, many studies find that insider trades are profitable So, early evidence was that stock market is not strong form efficient In 1978, Prof. Michael Jensen wrote: "I believe there is no other proposition in economics which has more solid empirical evidence than the Efficient Markets Hypothesis." More recently, studies have turned up question marks ...
B/M Effect
B/M effect: Firms with high book equity value to market equity value have higher returns B/M = inverse of price-to-book Not attributable to beta or size Fama and French (1992), Table V, p. 446 Basic idea behind "value investing" Look for stocks with market price low relative to some measure of fundamental value (here book value) Related to long-run reversals effect Q: How do you become a high B/M stock? A: You have a long string of low returns How to interpret B/M effect? Note B/M is akin to Dividends/P, Earnings/P, CF/P And CF/P = E[R]-g in Gordon growth model So B/M effect could embody "correction of mispricing" or simply higher E[R] ... Joint-hypothesis problem especially transparent here
M&A Facts
Basic facts about U.S. mergers Like IPOs and SEOs, mergers come in waves Each wave has common elements 1960s: Conglomerates, payment in stock, high stock market 1980s: Hostile takeovers, payment in cash, low stock market 1990s: Consolidation, payment in stock, high stock market
MCI MCIC
But markets aren't always efficient... Between Nov 1996-Nov 1997 there were 24 days in which 10,000 or more shares of MCI traded 18 of those days involved significant news related to MCIC
Three Flavors of Information
Can think about three sets of information that might be used to form P* Past prices/returns All publicly available information All public AND private information Leads to three (progressively stronger) notions of market efficiency "Weak form": all info in past prices is efficiently included "Semi-strong form": all public info efficiently included "Strong form": ALL info efficiently included (incl private)
Catering
Catering: Managerial actions to boost current share price by appealing to investor sentiment Catering is more likely when: It doesn't interfere with long-run investment policy (i.e., reduce P*) Managers can profit from short-run overvaluation Managers have short horizons
.com name changes
Cooper, Dimitrov, and Rau (2001) look at firms changing name to include ".com" ".net" or "internet" during height of tech bubble Find abnormal returns of 74% in 10 days around announcement This could be rational IF name changes correspond to changes in future cash flow expectations... But this is clearly not the case here... Results also hold for firms without a real fundamental business connection to internet
Until 1960s two basic approaches to investing
Fundamental analysis Tenet: Good investments can be found through careful analysis of financial and economic data Technical analysis (a.k.a. "charting") Tenet: Can find good investments by examining past price/return patterns Until the 1960s, academia had little to add ...
Case Study: AOL Time Warner
January 2000: AOL proposes to buy Time Warner using its stock as currency Largest merger in history: $166 billion (mkt cap of TW) January 2001: Antitrust investigations completed, merger goes through January 2003: AOL Time Warner reports biggest yearly loss in US corporate history (-$98 billion) Shares have lost >70% since Jan. 2000 Lines up with SV model of M&A rather well
Long-Run Reversal Effect
Long-run reversal: Negative autocorrelation in abnormal returns (i.e., gamma < 0) at horizons of 3-8 years DeBondt and Thaler (1985) Compare future performance of extreme "winner" and "loser" stocks as measured over past 5 years Figure 3: Shows visually that gamma < 0 Other studies find long-run reversals in indexes Another potential violation of weak-form EMH
Characteristic Effects
Moving now to evidence relevant to semi-strong EMH ... Common question: Can (publicly known) firm "characteristics" be used to predict abnormal returns in the cross-section? Two characteristics discovered to be useful Size (market capitalization = price*shares outstanding) B/M ratio
Traditional vs Behavior Finance
Neoclassical (traditional) finance assumes investors are rational and maximize utility In other words: when making economic decisions, individuals act fully rationally Behavioral finance relaxes the assumption of full rationality
Famous Bubbles
Non-stock market bubbles Dutch tulipmania (1634-1637) Mississippi bubble (1719-1720) South Sea bubble (1720) Stock market bubbles The great crash (1929) The internet (2000)
Practical Advice
Practical advice: Kahneman and Riepe (1998) Hard to eliminate cognitive biases Trick is to develop skill of recognizing when a particular error is likely Not clear which biases are the "most" important for investor psychology Different biases matter in different settings
Overconfidence
Roll (1986) "hubris" theory of M&A An irrational mgr, rational market theory (opposite of SV (2002) approach) Main idea: Acquisitions done by bidders who most overvalue "synergies" - they are the overconfident/optimistic in valuation of combination, ignore winner's curse Many managements apparently were overexposed in impressionable childhood years to the story in which the imprisoned handsome prince is released from a toad's body by a kiss from a beautiful princess. Consequently, they are certain their managerial kiss will do wonders for the profitability of the target company... We've observed many kisses but very few miracles. Nevertheless, many managerial princesses remain serenely confident about the future potency of their kisses—even after their corporate backyards are knee-deep in unresponsive toads. -Warren Buffet (Berkshire Hathaway Inc. Annual Report, 1981)
SV Model and M&A Fact
SV (2002) model predictions ... Acquisition by stock when bidder is relatively overvalued Acquisition by cash when target is absolutely undervalued, hence likely hostile deal ... match some key features of the data Cash acquirers earn positive long-run returns, while stock acquirers earn negative long-run returns High M/B (low b/m) bidders tend to pay with stock
Model of M&A
Shleifer and Vishny (2002) offer a rational manager, irrational investor theory of M&A to explain these and other facts Key elements of model: Relative current valuations of acquirer, target Relative horizons of their managers Goal is to explain who acquires whom, how finance, whether hostile, etc.
Lottery Ticket Sales
State lottery ticket sales in Ohio increase in days after a victory by OSU football team (Arkes, Herren, and Isen (1988))
Investment and the stock market
Suppose funds have been raised from good market timing (e.g., equity issue at P>P*) Next question: How to spend the funds? Managers' horizons, the source of funds, and the use of funds all interact E.g., a rational long-term manager "stores" the market-timing gain in cash or some other "hard asset" that will retain value when share prices come down
Supply Redux
Suppose rational managers, irrational investors I.e., suppose that managers know when P=/=P* What should they do? Change supply (S)? "Market timing" In corporate finance, it basically means ... Issue securities (increase supply) when P>P* Repurchase (reduce supply) when P<P*
Where were at
The "rational managers, irrational investors" approach to corporate finance has a lot of validity Consistent with what CFOs say in surveys Consistent with actual links between financing patterns and securities prices Picks up a lot of what traditional corporate finance is missing
Availability Heuristic
The availability heuristic "Probability is estimated by the ease with which similar instances or associations can be brought to mind" Biases occur when "availability" and true frequency diverge Easily retrievable instances overestimated Implication The media can bias a rule based on recall Homicides, airline accidents, shark attacks, etc. Availability: probability estimates influenced by ease with which instances brought to mind Malmendier and Nagel (2011) find that individual experiences affect risk taking Individuals that lived through Great Depression report lower willingness to take risk and participate in stock market, and are more pessimistic about future returns
EMH Definition
The hypothesis that real capital markets actually are efficient by Fama's definition
Representativeness Heuristic
The representativeness heuristic "The probability that event X belongs to set Y is judged on the basis of how similar X is to the stereotype of Y" Biases occur when prior probabilities (base rates) are ignored Representativeness: "likelihood that event belongs in some group, estimated based on extent to which represents or is similar to stereotype for that group" In other words, base rate probability of being part of group is ignored Perhaps demand for small, volatile stocks due to representativeness Microsoft was once small, volatile stock - returned 70% in first 5 years. Similar stories for many other firms. Small, volatile stocks seem to fit this stereotype, therefore investors willing to pay high prices for them. This assessment ignores base rate probabilities! i.e., the large number of small, speculative, volatile investments that failed
Behavioral Finance
The study of less than fully rational investor behavior and its impact on asset prices and corporate finance
Other Calendar Effects
Weekend effect: Stocks do well on Friday, badly on Monday Lakonishok and Smidt (1988), Table 2 Holiday effect: Stocks do well right before holidays Lakonishok and Smidt (1988), Table 3 Turn-of-the-month effect: Stocks do best around the turn of the month
Why Average investors matter?
When arbitrage is limited, irrational investors (average investors) can affect prices ... We discussed the many limits of arbitrage P =/= P* ... Especially when they deviate systematically from rationality P =/= P* in systematic ways Systematic deviation is important. If there is no systematic deviation, then noise traders may just cancel each other out.
Interpretation
"OK, I can believe that managers know when their own firm is mispriced. But are you suggesting that managers can predict the market? That's crazy!" That's not quite right ... here's the deal Mispricing is correlated across firms, because investor sentiment affects many stocks at the same time If all firms are overvalued at the same time, they'll tend to issue equity at the same time, even though each firm is responding just to its own misvalution As a result, aggregate equity issuance "predicts" the average firm's return (a.k.a. the market return), even though no individual manager has any direct info about market-level misvaluation
Statistical Significance
"Standard errors too small (t-stats too large). They don't reflect the data mining." If enough hypotheses tested, by chance some will work What can be done? To some degree, almost always a valid objection. Look at new data - e.g. see whether results hold for different time periods, different countries. Check whether novel predictions of an economic theory are borne out - e.g., January effect stronger for prior-year losers, consistent with a novel prediction of tax-loss selling. Simultaneously, gives confidence that January effect is real.
Long Term vs Short Term
Evidence above (IPOs, SEOs, repurch., stock mergers, etc.) suggests rational managerial efforts to exploit mispricing for benefit of long-term shareholders. Managers also face incentives to maximize short-term share price Compensation, job security tied to current P High P creates the opportunity for market-timing equity issuance, stock merger, etc.
Overconfidence and M&A
Malmendier and Tate (2008) examine M&A by overconfident managers Start by measuring "overconfidence" Use time at which CEOs exercise their options Idea: Rational CEOs exercise their options well before expiration date, to minimize exposure to company-specific risk. CEOs who are "longholders" are therefore bullish - i.e., overconfident or optimistic (ambiguous here)
Theoretical Underpinnings
1. If all investors are rational Rational investors value securities for their expected discounted cash flows, accurately use all info to determine P* Every rational investor demands a lot more if P<P*, a lot less (even negative demand, i.e. short-sell) if P>P* 2. If some investors are irrational, but their uncorrelated misperceptions cancel out If some "optimists" think that P<P* and an equal number of "pessimists" think P>P*, they can trade with each other without affecting P So, assuming P=P* to start with, P=P* remains 3. If arbitrage is unlimited Arbitrage: Simultaneous purchase and sale of same, or essentially similar, security in two different markets for advantageously different prices. (Sharpe and Alexander 1990) Think of arbs as big, rational investors who know P* and trade big quantities when P =/= P* So even if some investors are systematically irrational, these arbs might still enforce market efficiency
Who wins who loses
1. Who wins, who loses in investing If prices are right, average investors can't really make a mistake (as long as they stay diversified) Higher risk will be compensated with higher (expected) return But if prices aren't right, unsophisticated traders (who don't understand the game) may lose $ More generally, irrational prices can arbitrarily affect allocation of wealth in the economy
Allocation of Risk
4. Allocation of risk across investors Correct prices facilitate efficient risk sharing If prices wrong ... Harder to quantify risk, make good portfolio decisions If investors are unsophisticated/irrational... May not diversify properly
Testing SemiStrong EMH
A "semi-strong efficient" market incorporates all public information Prices to react to news quickly and accurately Example: Titanic sank April 14, 1912. Ship's owner, the International Mercantile Marine Company, spent $7.5 MM to build it Had insurance for $5 MM. IMM's share value dropped by $2.5 MM in two days (Extra .1MM = lower E[CF]s due to lost reputation?) Tests of semi-strong EMH often take form of event studies Fama, Fisher, Jensen, and Roll (FFJR 1969) is first "event study" Example: Stock splits FFJR study 940 stock split events Such events occur after steep price rises Splits are just purely cosmetic changes to the # of shares outstanding Therefore, they shouldn't in themselves affect value of company or returns going forward There are three steps to an event study Step 1. Pick a model of expected returns and calculate "abnormal returns" (AR) for each security, for each day around the event where t is the date relative to the event (e.g. -2 days, -1 day, event day, +1 day, ...) FFJR used CAPM to estimate E[R] Step 2. Take the average abnormal return across securities (N=940 for FFJR), each day, to get an average abnormal return in "event time" Step 3. Construct the cumulative abnormal return (CAR) from time k through l (k<l) Assuming the E[R] model is appropriate, then AR should be around zero (not statistically different from zero) for any day t>0 Else, a potential violation of semi-strong EMH FFJR (1969) Figures 2a, 2b On average, investors don't underreact or overreact to stock split Evidence consistent with semi-strong EMH Put differently, FFJR find that you can't use a stock split, once it becomes public info, to form a trading rule that earns abnormal returns
Testing Weak form EMH
A "weak-form efficient" market incorporates all info in past prices Rules like "buy if price falls 10% in a week, sell if rises 10%" don't earn abnormal returns Abnormal returns = returns that, on average, beat the appropriate E[R] benchmark A simple plot of stock price behavior seems to support the weak-form EMH Roberts (1959), Figures 1-4 (next slide) Can do a formal regression test to see if trends Last term is just random noise term E[R] is given by some specified model (e.g. CAPM) Weak-form EMH predicts that gamma = 0 (do a t-test) Fama (1970), Table 1, performs such tests Assumes that E[R] is simply a constant Finds that yes, gamma = 0 for return horizons between one day and two weeks Early tests generally supported weak-form EMH Authors tried very elaborate trading rules, not only trends Conclusion: Even if abnormal returns could be forecast (slightly) using past prices, trading costs would probably eat up any profits Often argued that results were economically insignificant Note: these tests face a "joint hypothesis problem" Have to maintain a hypothesis about E[R] (e.g., that it is CAPM, or a constant) in order to test the hypothesis of interest (that there are no "abnormal returns") Critique always lurking in background
Aggregate Debt Market Timing
Aggregate debt issuance Can issue long-term or short-term bonds. How to decide? More survey evidence ... Graham and Harvey (2001, Table 9, p. 220) 46% "issue debt when interest rates are particularly low" 36% "issue short-term when short-term interest rates are low compared to long-term rates" 29% "issue short-term when we are waiting for long-term market interest rates to decline" Long-term share in total debt issues (dL / dL+dS) Long-term debt issues (dL) (maturity >1 yr.) Short-term debt issues (dS) Scaling isolates maturity choice decision from the level-of-debt decision Next few slides from Baker, Greenwood, Wurgler (2002)
Aggregate Equity Market Timing
Aggregate equity issuance So far, we saw that issuers and repurchasers can time their firm's return relative to a benchmark, i.e. Ri -Rm But financing patterns come in "waves." Are issuers also timing Rm? The "equity share in new issues" e/(e + d) Aggregate equity issues divided by equity plus debt issues. Annual variable. Isolates security choice decision from the level-of-investment decision Next few slides are from Baker and Wurgler (2000)
Ambiguity Aversion
Ambiguity aversion "Uncertainty is less attractive when it is difficult to quantify" But shouldn't matter whether risk is ambiguity or an objective risk Application: "Home bias" in portfolios French and Poterba (1991): US investors 94% in US stocks Huberman (2001): Hold stock in local RBOC Coval and Moskowitz (1999): Fund managers hold local stocks Finance Implications Critique from EMH-camp: Psychological biases can be combined in ad hoc way to explain almost any market anomaly Lack of one single unified theory makes beh finance susceptible to this critique Obvious difficulty in empirically attributing observed anomalies to investor psychology Observing the psychological motives behind investor actions is generally not possible
Efficient Capital Market
An efficient capital market is a market that is efficient in processing information. The prices of securities observed at any time are based on 'correct' evaluations of all information available at that time. In an efficient market, prices 'fully reflect' available information. (Fama 1970)
Summing Up
Managers appear to take advantage of market inefficiencies, particularly when deciding which securities to issue to raise funds Managers also appear to be prone to decision-making biases that affect important decisions such as investment and M&A Behavioral finance identifies both opportunities and pitfalls for corporate managers
Anchoring and conservative adjustment
Anchoring and conservative adjustment (underreaction) "People make estimates by starting from an initial value ('anchor') that is adjusted to yield the final answer. The anchor may be suggested by the formulation of the problem, or may be irrelevant." "Different starting points yield different estimates, which are biased toward the initial values." Bias occurs when adjustment is insufficient / too conservative Anchoring and conservative adjustment (underreaction) Another example: In competitive negotiations, outcome is often very close to starting point Anchoring/conservative adjustment: "Different starting points yield different estimates, which are biased toward the initial values." Conservative adjustment behind many of anomalies we've discussed...PEAD, momentum... Anchoring plays strong role in merger offers Offers made at premium to target's current price Target firm's 52-week high is a very salient number Baker, Pan, and Wurgler (2011) find that offer prices cluster at 52-week high price Likelihood of offer succeeding increases strongly for offers exceeding 52-week high. Very clear instance of anchoring
AppNet Systems
Another example of tech bubble irrationality AppNet Systems filed for IPO under symbol APPN Shares of APPN had return of 37,636% in 2 days after filing (0.007 cent/share to 20 cents/share) 1 big problem: AppNet was filing for IPO which was to take place a couple weeks in the future Current ticker symbol of APPN represented Appian Technology, an inactive circuit manufacturer trading on Nasdaq OTC Bulletin Board (penny stock) Volume increased from 200 shares traded day before filing to over 7.3 million shares traded "Net happy traders began touting APPN in chat rooms, apparently believing they were talking about AppNet. ...On Yahoo Finance's chat room...an enthusiast participant calling himself lovepennys raved: 'Just bought 50,000 shares, took 3 transactions to get it done, there r no shares out there, going to run big.'...It isn't clear why investors thought they could trade shares in a company whose IPO is weeks away." Wall Street Journal, April 1, 1999
Calendar Effects
Average returns differ within the calendar Returns are unusually high/low in certain months ... ... and certain days of month ... certain weekdays ... Discovery of calendar effects Early impetus to behavioral finance, because hard to reconcile with weak-form EMH If you plot stock returns over time, calendar effects show up as regular periodic blips January Effect:January effect: Small stocks (stocks with relatively low market capitalization) have high returns in January, on average Something to do with tax-loss selling? Emphasizes drop in demand around end of December / abatement of selling pressure in Jan Something to do with portfolio rebalancing? Emphasizes jump in demand in Jan (investing year-end bonuses etc.) September Effect:September effect: September is (by far) the worst month of the year for stocks In the US, it is the only month to have a negative average return Effect also seems to exist internationally One study of 18 developed markets finds that 15 have negative returns in September Explanation?
Bubbles
Bubble: "A fundamentally unsound commercial undertaking accompanied by a high degree of speculation" (Kindelberger) Prices go up because speculators keep buying in the belief that prices will keep going up... When bubble (inevitably) pops: Crash Every bubble is different and every bubble is the same Six bubble characteristics (identified by Kindleberger) 1. Initial displacement that grabs attention (e.g. a real boost to "fundamentals") 2. "Smart money" response 3. Channels for speculation are invented 4. Authoritative blessing 5. Crash 6. Political fallout
Categorization
Categorization: another result of limited attention, cognition Group similar (but not identical) assets together, evaluate/update as a category Basic aspect of human cognition, language Examples: Animals are categorized into "Dog" or "Cat" Shapes are categorized into "Triangle" or "Circle" etc. Why categorize? Reduces complexity Can respond to items in terms of their category, rather than as individual items Facilitates inference Don't have to be taught about novel objects if we can effectively categorize them Can use existing knowledge of items in category to infer attributes In traditional finance theory, investors do not categorize Each security viewed solely as a list of abstract statistics (mean return, variance, covariance, etc.) In reality, investors do categorize Barberis and Shleifer (2002) "Value" stocks, "Growth" stocks, "Large cap" stocks, "new economy" stocks, price-level based, etc.
Testing EMH
Direct approach: Compute P*, see if it equals P But how to forecast cash flows CF? What is the correct growth rate of CF? What is the right discount rate E[R]? Difficult or impossible to do. (estimation of value of distant cash flows, discount rate, etc will always be imprecise) Indirect approach is more common Idea is to see whether, using a given information set, one can forecast "abnormal" returns (i.e., returns over and above normal E[R]) Indirect approach avoids having to forecast CFs But still needs to take a position on the benchmark E[R]
Categorizing Securities
Does categorization affect returns? EMH: No Only a stock's own individual fundamentals matter. Stocks in the same "category" (e.g., same industry) move together because their fundamental values move together Categorization doesn't cause return comovement Does categorization affect stock returns? Behavioral finance: Yes Unsophisticated investors trade at the category level (may like "new economy" stocks today, "old economy" stocks tomorrow, etc.) Since arbitrage is limited, prices of stocks in the same category move up and down together in response to common noise trader demand pressures. Being in a category does cause return comovement with other stocks in the category, distinct from fundamentals Test of "trading-induced comovement" Barberis, Shleifer, Wurgler (2002) "S&P 500" is a distinct category S&P 500 addition does not affect fundamentals EMH: stocks added to the S&P 500 will not increase comovement with the other 499 stocks Behavioral finance: stocks added to the S&P 500 will increase comovement, because now they get traded in the same basket with those 499 other stocks Results support BF view When stock is included, it starts to comove more with S&P stocks and less with other stocks Specifically, its "beta" with respect to other S&P 500 stocks increases, and its "beta" with respect to non-S&P 500 stocks decreases To what extent do other "factors" in stock returns reflect category-level trading? Price level of shares (Green and Hwang 2009); Geographical location of firm (Pirinsky and Wang 2004); potentially many more still unknown...
Testing SemiStrong EMH Continued
Example: Mutual fund managers Rely mainly on publicly-available information Jensen (1969): Do returns on mutual funds just reflect their risk, or do some managers have a positive "alpha"? (Here, the i subscript indexes the portfolio of a given manager i, not an individual stock) Jensen (1969) 115 mutual funds over 1945-1964 uses CAPM to model E[R] Runs 115 regressions to get 115 alphas Most estimated alphas are around zero Jensen (1969) Figure 1, 2 Before expenses, the average alpha is negative After expenses, even more negative No evidence of fund manager skill Overall, early tests supported semi-strong EMH Most event studies found that mkt reacts correctly to news No systematic over- or underreaction, therefore no easy trading rules Even highly skilled investors using public info (MF mgrs) didn't make abnormal profits But here again, the "joint hypothesis problem" is big caveat to these conclusions
Limited Attention
For one individual to 100% quickly and accurately process "all public info" would require almost infinite attention/cognitive ability Human limits on attention are probably behind certain of the heuristics Surely availability Possibly representativeness and/or local representativeness Evidence of limited attention: Prices react more to more "salient" ("available") news Friday earnings announcements (DellaVigna and Pollet 2009) 15% lower immediate price response for Friday announcements Extraneous news (Hirshleifer, Lim, and Teoh 2009) Announcement date returns less sensitive to earnings news when there are more competing announcements on the same day Attention-grabbing stocks (Barber and Odean 2007) Individual investors are net buyers of attention-grabbing stocks Economically linked firms (Cohen and Frazzini 2008) Supplier firm stock prices react with lag to news about customers
Agressive Arbs?
How could one think that "arbs" are so aggressive? They will try to form "arbitrage portfolios" that include the mispriced security and opposite position in a CF substitute Idea is to exploit mispricing while hedging CF risk Example: Arbs notice that stock is underpriced What do they do? Bottom line: Three different scenarios justify EMH Still, are any of them realistic?
Interim Summary
How to mentally unify some of these findings? Calendar effects Momentum at 3-12 month horizon Reversals at long horizons B/M effect A simplistic caricature: Prices "oscillate" around E[P*] P seems to underreact in short term, overreact in long term Plus a few calendar-related blips Incorporates several important anomalies
IPOs
IPOs Insiders have high incentives to sell out when P>P*: As firm founders, they may own a lot of equity Public equity may be available at lower (risk-adjusted) cost than other types of financing IPO "underperformance" After the initial jump, average IPO has low idiosyncratic returns (underperform various benchmarks) over the 3-5 years following the issue Ritter (1991), Figure 1
Market Timing: Theory
Important points Bottom line: Rational managers who want to benefit their ongoing shareholders will take advantage of "market timing" financing opportunities. How they invest the money, however, depends on their "horizon" - their incentives to maximize short- vs. long-run value. Do managers "time" their securities issues? Survey evidence of market timing Graham and Harvey (2001, Table 8) 67% "amount by which our stock is undervalued or overvalued by the market" is an important factor in issuing stock 63% "if our stock has recently risen, the price at which we can sell is 'high'" Is market timing successful, in practice? Research has examined a wide range of security issuance (S) decisions: IPOs, SEOs, repurchases, short- and long-term debt issues ...
Whats an average investor
In behavioral finance, an "average investor" Doesn't know P* May form irrational expectations of future CFs, leading to incorrect estimates of fundamental value May also have odd preferences (e.g. may not make expected-utility-maximizing decisions, given her beliefs) Matters of terminology In economics, an investor is deemed "irrational" if she doesn't form rational expectations, and then maximize her expected-utility, given those expectations Bottom line: investors sometimes make irrational decisions
M&M Theorem
In efficient markets, market timing is not possible. This is closely related to the Modigliani & Miller theorem: The market value of a company is independent of its capital structure. Interpretations of M&M: Firm value is determined by real assets and growth opportunities, not by security choice The total amount of pizza is unaffected by how it is sliced Some key assumptions: No taxes Investment policy (firm's cash flows) fixed Bankruptcy is not costly Efficient capital markets Symmetric information ... Argument can be extended to the full range of capital structure choices, not just D/E ratio Debt maturity structure (long-term vs. short-term) irrelevant Secured/unsecured debt choice irrelevant Convertible/nonconvertible debt choice irrelevant Preferred/common stock choice irrelevant Dividend policy choice irrelevant (related argument)
M&M not reality
In inefficient markets, the M&M theorem doesn't hold In inefficient markets, firms can reduce their overall cost of capital by "market timing" - i.e., switching toward whatever source of capital is currently available at lowest cost
EMH Math
In math, EMH is assertion that P = E[P*] P* is the fundamental value - the present value of all cash flows that investors rationally expect to receive E[R] is the rational discount rate (based on riskiness of firm) Note: EMH does not say anything about which E[R] or E[CF] to use Just that what the market uses is "right" I.e., EMH asserts that P equals the best possible estimate of P* that can be made using a given "information set"
Investor Sentiment
Investor sentiment: General investor mood High "sentiment" associated with bubble Low "sentiment" associated with crash Hard to pin down. (What is today's sentiment?) How do swings in "investor sentiment" affect stocks? Baker and Wurgler (2006) create an "investor sentiment index" Create index from variables thought to reflect sentiment of investors: Closed-end fund discount Number of IPOs Avg 1st day IPO return Equity share of new issues Turnover Dividend premium Prediction is that sentiment should most affect stocks whose valuations are highly subjective, and stocks that are difficult to arbitrage Evidence is consistent with this hypothesis Small stocks, young stocks, high volatility stocks, non-dividend-paying stocks, distressed stocks, and unprofitable stocks are most sensitive to sentiment
Real World Investor Behavior
Lack of diversification In Odean sample, median household owns 2.61 stocks Among households with direct stock ownership, only 31% hold any international stock Individuals sell winners and hold losers (disposition effect) Using Finnish data, Grinblatt et al. (2011) find that high IQ investors: Hold greater number of different stocks (diversify) More likely to hold mutual funds (i.e., diversify) Tend to hold more small, value, and low beta stocks
Lead-Lag Effect
Lead-lag effect: Large stocks "lead" small stocks over a horizon of a few weeks In statistical lingo, there's "positive cross-autocorrelation" between abnormal returns on large stocks this week and small stocks next week Lo and MacKinlay (1990), Table 4, p. 196 Another "pattern" apparent in past returns Potential violation of weak-form EMH The lead-lag effect could contribute to an industry-level momentum Large stocks "drag around" small ones ... maybe small stocks are slow to react to industry news? Then high returns on industry's large stocks today, high returns on industry's small stocks next week Hou (2007) Evidence of industry-level momentum Moskowitz and Grinblatt (1999), Table 2, p. 1260
Dividends
Like firm names, dividend policy is irrelevant to firm value in perfect and efficient markets Modigliani and Miller (1961) Standard rationales for dividends are based on Asymmetric information, agency costs, transaction costs, institutional investment constraints Is catering also a rationale? Baker and Wurgler (2004): Managers "cater" to investor sentiment in dividend policy They initiate/continue dividends when investors prefer (put higher valuations on) dividend payers They omit dividends when investors prefer nonpayers Analogous to name changes (same finding for stock splits and price levels) Main finding: Nonpayers initiate dividends when existing payers are trading at a premium
Local Representativeness
Local representativeness "The essential characteristics of the process will be represented, not only globally in the entire sequence (50% B, 50% G in the population), but locally in each of its parts." Results from reliance on "law of small numbers" Small samples perceived to represent population to same extent as large samples "exaggerate how likely it is that a small sample resembles the parent population from which it is drawn" Local representativeness Another example: Gambler's fallacy "After a long run of red, black is now due" "After a long bear market, a bull market is due" Another example: Belief in "hot hand" in basketball "He's hot - give him the ball" Evidence says doesn't exist Gilovich, Vallone, and Tversky (1985) Local representativeness and overreaction Clear implications in finance Money pours into mutual funds that have recently beaten the average Predicts investors believe in non-existent variation in quality of fund managers Fit fictitious stereotype of "good investment" Money moves toward stocks that have performed well and away from those that have done badly Potential explanation for momentum And long-run mean reversion as prices get too high or too low, arbitrage kicks in: "after long bear market, bull market is due"
M&A and stock market
M&A is a type of investment that allows us to make finer tests of these predictions Can observe the financing (cash or stock) for each deal (can't do this for capex investment) Can separate and study the valuations of the acquirer and the target (at least, until merger consummated) Can ponder horizons of acquirer and target mgmt. Does the "rational managers, irrational investors" perspective shed light on M&A patterns?
Actual Probability Assessments
Many systematic biases from Bayesian ideal have been detected People use "heuristics" or "rules of thumb" when making probabilistic judgments May be useful on average, but lead to many systematic biases ... Heuristic: A mental shortcut or "rule of thumb" to simplify decision making Why do investors behave irrationally? Human beings have limited attention, cognition, and time Financial world is challenging decision environment filled with countless pieces of information, calculations, decisions, etc.
Momentum
Momentum: Abnormal returns on individual stocks are positively autocorrelated (i.e., gamma > 0) at a 3 to 12-month horizon Jegadeesh and Titman (1993) Table 1 "Big" effect. Often greater than 15% average annual return difference between highest and lowest momentum portfolios (!) Just reflects risk? I.e. just bad control for E[R]? Are high momentum stocks just somehow "riskier"? No pattern between momentum and size or beta Jegadeesh and Titman (1993), Table 2 Another apparent violation of weak-form EMH
CAPM to get E{R}
Most popular model of expected return Also, embodies much intuition Assumptions: Investors are rational, and like higher portfolio mean return, dislike portfolio variance. Called "mean-variance" investors. Rational investors will diversify, do don't particularly care about performance of any one asset. Only care about overall portfolio return. Investors have homogenous expectations regarding future in terms of means, variances, covariances (and correlations) To repeat: No E[R] reward for idiosyncratic risk Hedging intuition: low beta stocks don't get punished when times are bad (market drops), making them desirable hedges. This pushes up today's price P, pushes down E[R] Prefer high returns when times are bad, therefore, securities that pay off in good times (high beta stocks) must give higher returns to induce investors to hold them CAPM just a one-period model. Can think of it as a prediction for any horizon - day, week, year ... CAPM and other E[R] models can in principle be applied to any investment asset (stocks, bonds, REITs, racehorses, parking lots, etc.)
Biases must be systematic
Noise trader biases only affect prices if arbitrage is limited AND biases in valuing assets are systematic Barber, Odean, Zhu (2009) "Systematic noise" find that individual traders (noise traders) often agree with each other whether to buy or sell stock In a given month, random subgroups of individual investors display similar buying and selling patterns Agreement is higher than would be expected by chance
Name Changes
One example of catering: Name changes Recall ".com" example from 1st day of class Cooper, Dimitrov, Rau (2001) Examine ".com" name changes in 1998, 1999 (Table I) Find large value increase (around +74%), even for firms w/o a real, fundamental business connection to Internet (Figure 1) Deleting ".com" After Internet bubble burst (starting in March 2000), firms started to change their names back (Table II, Figure 2) Again, large value increase (+20% ballpark) A few "double-dippers" changed twice!
Optimism
Optimism "Beliefs tend to be biased in the optimistic direction" Most severe among the young Experience is a harsh teacher? Optimists prone to an illusion of control Exaggerated sense of how much they control fate Underestimate the role of chance Svenson (1981) examines opinion of own driving ability In US sample 93% rated themselves as more skillful than the median driver (69% in Swedish sample) Optimism and overconfidence Implication for finance: Rational people don't trade much Realize that they are probably trading at an informational disadvantage Rational theory predicts only "better-than-avg" should trade, except for liquidity, rebalancing, or tax reasons Behavioral view is that optimism, overconfidence in one's skill explains why people trade so much Optimism and overconfidence Barber and Odean (2000) analyze 1991-1996 data on 78,000 households at a discount broker Before trading costs, average investor beats market by 0.8% per year After fees, trails market by -1.5% Most active (most overconfident?) traders trail market by -6.5% due to fees Barber and Odean (2001) gender differences Men are more overconfident, trade more often (45% more), earn lower returns after fees (0.93% annually)
Overconfidence
Overconfidence "People set their subjective confidence intervals too tightly. They are 'surprised' too often." Some professionals are reasonably calibrated Shiller: In interviews with investors, "I find that overconfidence is apparent when I interview investors; they seem to express overly strong opinions and summary judgments."
Post Earnings Announcement Drift
Post-earnings announcement drift: In few months after the earnings announcement, stocks earn abnormal returns + CAR if was good announcement - CAR if was bad Bernard (1992) Figure 1 Looks like "underreaction" to information in an earnings announcement Drift "spikes" again at subsequent earnings announcement dates In other words, prices do not fully reflect the extent to which current changes in quarterly earnings predict future changes in quarterly earnings. Bernard (1992) Figure 4 Hardly the "quick and accurate" semi-strong EMH response PEAD related to momentum effect Perhaps momentum is PEAD Research finds some, but not total overlap Controlling for momentum, still some PEAD, and vice-versa PEAD related to B/M effect + abnormal returns on high B/M stocks concentrated around earnings announcements Earnings surprises systematically positive for high B/M firms LaPorta et al. (1997) Table I, p. 864 Supports "correction of mispricing" view of B/M effect Convincing evidence against EMH? Concerns 1. Wrong control for risk (E[R] model wrong; jt. hyp. prob.) 2. Statistically insignificant (not robust; data mining) 3. Economically insignificant (small effects)
Prediction 1
Prediction (1): Overconfident managers are more likely to conduct mergers that have high prob. of failure (and negative expected return) MT find support for this "Longholder" CEOs more likely to complete diversifying mergers (widely regarded as dubious value-creation strategy) Table 5
Prediction 2
Prediction (2): Overconfident managers more likely to do acquisitions when have extra cash Table 4
Prediction 3
Prediction (3): Market reaction to merger bid announcement is lower for overconfident managers The "rational markets" side of the story Table 6
Optimism
Q: How good a driver are you? Above or below average? ... People are more optimistic about outcomes ... that they believe they can control ... to which they are highly committed ... (corporate management!) Unlike many other telecommunications kingpins of the 1990's, who cashed out hundreds of millions of dollars in inflated stock ahead of unsuspecting investors, Mr. Ebbers apparently believed that he could keep the company afloat one way or another. He kept buying WorldCom shares even as the company's house of cards began to topple. - New York Times 3/3/2004
Rational Probability Assessments
Rational investors have "unbiased" expectations at any given time Unbiased: Expectations are correct on average No systematic mistakes World is ever-changing ... Rational investors use "Bayes'(s) rule" to adjust to new info Use laws of probability to revise expected future cash flows in response to all relevant information Bayes' rule in general: P(A|B) = [P(B|A)P(A)]/P(B) Rule mathematically expresses how rational individuals should update beliefs in response to news
Repurchases
Repurchases When P<P*, rational managers can benefit ongoing shareholders by reducing supply Repurchaser performance Over 3-5 years following repurchase, repurchasing firms do outperform; returns are abnormally high Ikenberry et al. (1995), Figure 1 Again, investors seem to recognize repurchases as "good news," just not to full extent (cf. SEO reaction) Looks a lot like repurchases are successfully timed to exploit undervaluation
SEOs
SEOs In seasoned equity offering at P>P*, ongoing shareholders gain by selling overvalued shares SEO "underperformance" Almost exactly the same magnitude as IPO underperformance! Loughran and Ritter (1995), Tables 1 and 2 Investors do recognize potential for market timing - SEOs are viewed as "bad news," signals of overpricing - but they don't react negatively enough
Size Effects
Size effect: Low market cap. firms have higher returns Not attributable to higher beta Fama and French (1992), Table I, p. 434 Size effect closely related to January effect Outside January, no size effect Outside small stocks, no January effect
Sporting outcomes and stock returns
Sporting event outcomes and stock returns? Outcome of sporting matches should not affect stock returns in an efficient market Psychological evidence finds a strong link between soccer outcomes and mood Good mood optimism about future outcomes Turns out international soccer match outcomes have implications for national stock market returns Loss in World Cup elimination stage leads to next-day abnormal stock return of -0.49% (Edmans et al. 2007) Weaker effect found after losses in intern'l cricket, rugby, & basketball Effect seems to come entirely from mood - controlling for pre-game expectations ensures that results not attributable to economic factors such as reduced productivity
Sunshine and stock returns
Sunshine and stock returns Efficient markets view says sunshine should have no effect on stock returns Psychological evidence predicts that sunny weather is associated with upbeat moods People in a good mood tend to evaluate future prospects more optimistically than people in a bad mood Turns out that there are in fact higher returns on sunny days Hirshleifer and Shumway (2003) document this relationship using a sample of 26 different countries and examining daily cloudiness relative to expected cloudiness for that day of year Magnitude is quite large: difference in return for hypothetical perfectly cloudy year vs perfectly sunny year is 8.7% vs 24.8% Evidence can't be explained by any potential efficient markets explanation
Market Timing: Who Gains?
Suppose current price of E (per share) is $100, but long-run (fundamental value) is only $50 N shares outstanding. What can CFO do? Suppose issues n new shares at $100 and keeps proceeds in cash. Long-run price per share (ie P*) is now (50N + 100n)/(N+n) which is > 50. Ongoing shareholders gain in the long run. For them, price was going to go down to $50 anyway, but now it goes down not quite as far. New shareholders lose in the long run. They paid $100 for shares destined to go down, but they don't fully recapture the cash they added to firm. They share it with N others. Important points The example keeps the $100 in cash. This "stores" all long-term market timing gain for ongoing investors. Other extreme: Assuming company has perfect access to financing, then will have already taken all pos npv projects. Investment in assets (potentially overpriced assets) of firm likely to be negative npv. Eliminates all (or at least some) of long-run gain, but may help keep share price high in short-run. In-between case: Invest new equity in an asset that is less overvalued than firm average. Preserves some gain for long-term investors.
Reaction to new and non-news
We said before that under semi-strong EMH: P reacts quickly and accurately to E[P*] news P doesn't react to anything else (no "excess volatility") "What moves stock prices?" Cutler, Poterba, Summers (1989) look at both implications Does "news" affect P? Sometimes (Table 3) Does P ever move w/o news? Yes (Table 4) A clear example of "excess volatility" 5/3/98 Sunday NYT reports on cancer "breakthrough" Front page article of new cancer-curing drug from EntreMed Stock price closed at 12.063 on Friday, closed near 52 on Monday But same news had been reported >5 mos. earlier in scientific journals and NYT itself (page A28 11/28/1997) Initial report led to price jump of 28% Enthusiasm spills over to other biotech stocks Nasdaq Biotech Index jumps 7.5% (contagious/correlated optimism) 11/12/1998 WSJ reports other labs fail to replicate results Price drops, but still above 11/28/1997 prices Stock realized permanent gain in price due to increased publicity - even though no new info was presented
