by M Brooks · 2008 · Cited by 103 — Behavioral finance is based on the alternative notion that investors, or at least a on stocks generally exceeded the yield on long-term U.S. government.
26 pages
Missing: discourse aaa

135 KB – 26 Pages

PAGE – 1 ============
The Research Foundation of CFA Institute Literature Review ©2008, The Research Foundation of CFA Institute 1Behavioral Finance: Theories and EvidenceAlistair Byrne, CFA University of EdinburghMike BrooksBaillie Gifford & Co That behavioral finance has revolutionized the wa y we think about investments cannot be denied. But its intellectual appeal may lie in its cross-discipli nary nature, marrying the field of investments with biology and psychology. This literature review discu sses the relevant research in each component of what is known collectively as behavioral finance. This review of behavioral finance aims to focus on arti cles with direct relevance to practitioners of investment management, corporate finance, or personal financial planning. Given the size of the growing field of behavioral finance, the review is necessarily sele ctive. As Shefrin (2000, p. 3) points out, practitioners studying behavioral finance should learn to recognize their own mistakes a nd those of others, understand those mistakes, and take steps to avoid making them. The articles discussed in this review should allow the practi tioner to begin this journey. Traditional finance uses models in which the economic agents are assumed to be rational, which means they are efficient and unbiased processors of relevant informat ion and that their decisions are consistent with utility maximization. Barberis and Thaler (2003, p. 1055) note that the benefit of this framewor k is that it is fiappealingly simple.fl They also note that fiunfortunately, after years of effort, it has become clear that basic facts about the aggregate stock market, the cross-section of average re turns, and individual trading behavior are not easily understood in this framework.fl Behavioral finance is based on the alternative notion that investors, or at least a si gnificant minority of them, are subject to behavioral biases that me an their financial decisions can be less than fully rational. Evidence of these biases has typically come from cognitive psychology litera ture and has then been applied in a financial context. Examples of biases include †Overconfidence and overoptimismŠinves tors overestimate their ability and the accuracy of the information they have.†RepresentativenessŠinvestors assess situations based on superficial characteristic s rather than underlying probabilities.†ConservatismŠforecasters cling to prior beliefs in the face of new information. †Availability biasŠinvestors ov erstate the probabilities of recently obse rved or experienced events because the memory is fresh. †Frame dependence and anchoringŠthe fo rm of presentation of informatio n can affect the decision made. †Mental accountingŠindividuals allocate wealth to sepa rate mental compartments and ignore fungibility and correlation effects.†Regret aversionŠindividuals make decisions in a way that allows them to avoid fee ling emotional pain in the event of an adverse outcome. Behavioral finance also challenges the use of conventional utility functions based on the idea of risk aversion. For example, Kahneman and Tversky ( 1979) propose prospect theory as a de scriptive theory of decision making in risky situations. Outcomes are evaluated against a subjective reference point (e.g., the purchase price of a stock) and investors are loss averse, ex hibiting risk-seeking behavior in the face of losses and risk-averse behavior in the face of gains.

PAGE – 2 ============
Behavioral Finance 2©2008, The Research Foundation of CFA Institute One aspect of the discussion about rational and irrational investors that is importan t to consider is the extent to which professional traders and money managers are subject to the same behavioral biases that are more commonly discussed in the context of individual (typically assumed uninformed ) investors. A number of articlesŠ discussed hereŠconsider this issue directly and find that professionals are far from immune to the biases. A full description of these biases and the evidence for them is beyond the sc ope of this review. Readers who would like a more detailed discussion should refer to Ba rberis and Thaler (2003) and Shefrin (2000). Although the existence of behavioral biases among some investors is an essential component of behavioral finance, a second essential strand relates to the limits to arbitrage. Traditional finance ho lds that if some (irrational) investors misprice assets, the mispricing will be corrected by the trading acti ons of rational inve stors (arbitrageurs) who spot the resulting profit opport unity, buy cheap assets, and sell expe nsive ones. Behavioral finance theory counters that mispricing may persist because arbitrage is risky and costly, which has the result of limiting the arbitrageurs™ demand for the fair-value re storing trades (Shleifer and Vishny 1997). The existing academic literature has tended to develop behavioral finance against the fifoilfl of traditional rational finance. But a number of au thors (e.g., Statman 1999a; Thaler 1999) make the case for the fiend of behavioral finance,fl arguing that because all financial th eory requires some assumptions about investor behavior, researchers should strive to make the best assumptions ab out behavior in all models ra ther than invent a subclass of models featuring empirically observ ed behavior. Despite great strides in recent years, behavioral finance does not appear to have reached the poin t of being considered in all models .Investors seeking a more comprehensive introduction to the field are directed to the review articles by Hirshleifer (2001) and Barberis and Thal er (2003) as well as to the relevant articles in the November/December1999 issue of the Financial Analysts Journal . Shefrin™s (2000) book Beyond Greed and Fear is also recommended. In the following sections, we discuss key areas in the a pplication of behavioral fina nce. We discuss the limits to arbitrage and then proceed to disc uss behavioral asset pricing theory, beha vioral corporate finance, and evidence of individual investor behavior and behavioral portfolio theory. We also discuss briefly the psychology of risk, ethics, and the emerging field of neuroe conomics. The final section of this re view provides a bibliography with a brief summary of each reference. The Limits to ArbitrageA key argument in behavioral finance is that the existe nce of behavioral biases am ong investors (noise traders) will affect asset prices and returns on a sustained basis only if limits to arbitrage also exist that prevent rational investors from exploiting short-term mispricings and, by doing so, returning prices to equilibrium values. Evidence suggests that limits to arbitrage ex ist, for example, in the failure to eliminate obvious an d straightforward mispricing situations. Mitchell, Pulvin o, and Stafford (2002) are able to document 82 cases in which the market value of a company is less than the ma rket value of the company™s stake in its subsidiary. These situations imply arbitrage opportunities leading to sw ift correction of the pricing anomal y, but the authors find a degree of persistence that indicates barriers to arbitrage. Barberis and Thaler (2 003) outline the various issues that create li mits to arbitrage. When the mispriced asset lacks a fairly priced close substitute, arbitrageurs are faced wi th fundamental risk in that th ey are unable to effectively hedge their position in the mispriced asset from adverse changes in fundamentals. Even if a close substitute is available, arbitrageurs face noise trader risk. Because trading by uninformed investors may cause the mispricing to increase before it corrects, the arbitrageur may be unable to maintain the position in the face of margin calls, especially when trading with other peop le™s capital, as in inst itutional investment management. Finally, other issues include high implementation costs for an y arbitrage trade. At the extreme, taking a short po sition in an overpriced security may be impossible if, for example, stock lendin g is prohibited or no shar es are availabl e to borrow. On the latter point, Lamont and Th aler (2003) review examples in which the market value of spun-out subsidiaries of tech companies exceeded that of the parent company that reta ined a majority stake in the spinout. In these cases, short-selling of the sp inout was difficult, expensive, or im possible, reducing or eliminating the arbitrage opportunity.

PAGE – 3 ============
Behavioral Finance ©2008, The Research Foundation of CFA Institute 3Behavioral Asset PricingWhereas academics talk about asset pricing and about explaining the cr oss-section of stock returns, for practitioners, the same issues fall under the simpler heading of fistock pick ing.fl If behavioral biases among investors cause mispricing of stocks in a predictable fashion, then active managers may have the scope to beat the market by using strategies based on these sources of mispricing. Investor Sentiment. One important issue is whethe r investor sentiment has the potential to affect stock returns, which is considered self-evide nt by most practitioners. But traditio nal finance theory has little role for sentiment in asset pricing. Recent behavioral literature (Baker and Wurgler 2006; Kumar and Lee 2006; Tetlock 2007) suggests evidence of investor sentiment affecting stock re turns. The effect is most pronounced for stocks that are difficult to value and/or hard to arbitrage. This category includes sma ll stocks, young stocks, unprofitable stocks, and extreme- growth stocks. When investor sentiment is high, subsequent returns for these types of stocks tend to be relatively low, and vice versa. Causes of swings in investor sentimen t vary and, in some cases, can be quite trivial. Hirs hleifer and Shumway (2003) present evidence that daily retu rns across the world™s markets are affected by the weather in the city of the country™s leading stock exchange. Unfortunately, a strategy to exploit this predictability in returns involves quite frequent trading, and trading costs ma y well eliminate any available gains fo r most investors. Kamstra, Kramer, and Levi (2003) provide similar eviden ce, showing that returns in various countries through the year are related to hours of daylightŠa result possibly driven by the occurrence of seasonal affective disorder. The effect of sentiment is evident in various aren as. For example, Gemmill an d Thomas (2002) show that noise trader sentiment, as proxied by re tail investor fund flows, leads to fluc tuations in the disc ount of closed-end funds. Of note, one measure of sentiment that does not predict returns is the current sentimentŠbullish or bearishŠof investment newslett er writers. Rather, recent past returns pr edict the sentiment of the writers, which, in turn, has no correlation with future returns (Clarke and Statman 1998). Under- and Overreaction. Another key area of behavioral resear ch relates to the extent to which investors under- or overreact to information in pricin g securities. The available em pirical evidence appears to suggest short-term (up to 12 months) return continuati ons, or momentum (e.g., Jegadeesh and Titman 1993), but longer term (three- to five-year) reversals (e.g., De Bondt and Thaler 1985; Lakonishok, Shleifer, and Vishny 1994). This evidence poses something of a challenge for behavioral researchers to come up with a theory that explains initial underreaction but long er term overreaction and rebuts Fama™s (1998) contention that a market that overreacts about as much as it underr eacts can be regarded as broadly efficient. Various behavioral models have been developed to expl ain the empirical findings. In Barberis, Shleifer, and Vishny (1998), investors suffer conser vatism bias and use the representative ness heuristic. Conservatism means that individuals are slow to change th eir beliefs in the face of new eviden ce and can explain why investors would fail to take full account of the implications of an earn ings surprise. The representativeness heuristic means that individuals assess the probability of an event or situation based on superficial characteristics and similar experiences they have had rather than on the unde rlying probabilities. This approach can mean that investors, seeing patterns in random data, could extrapolate a company™s recent po sitive earnings announcements further into the future than is warranted, creating overreaction. Daniel, Hirshleifer, and Su brahmanyam (1998) present a related model based on overconfidence and biased self-attribution. Overconfidence leads investors to overwe ight their private information in assessing the value of securities, causing the stock price to ove rreact. When public information arrives, mispricing is only partially corrected, giving rise to underreaction. Furthermore, biased self-attribution means that when public information confirms the initial private signal, investor confiden ce in the private signal rises, lead ing to the potential for overreaction.

PAGE – 4 ============
Behavioral Finance 4©2008, The Research Foundation of CFA Institute Finally, Hong and Stein (1999) present a model popula ted by finews watchers,fl th ose who base their trades on private information but not past pr ices, and fimomentum trader s,fl those who base their trades on past price trends. News spreads slowly among the news watchers, causing initial underreactio n, but it is followed by momentum buying that can crea te an eventual overreaction. Related empirical work includes Drem an and Berry™s (1995) study that finds an asymmetry of response to earnings surprise between low and high P/E stocks. Low P/E (i.e., value) stocks respond most favorably to a positive earnings surprise, suggesting the low P/E status may be the result of prior overreaction to negative news. Lee and Swaminathan (2000) show that turnover levels provide a link between valu e and momentum effects. Winners with high past volume experience reversals at five-year hori zons, consistent with initia l underreaction and eventual overreaction. They argue also that as stocks decline in popularity, trading volume drops off and the stocks become neglected value stocks. Taffler, Lu, an d Kausar (2004) document market underreaction to the bad news contained in going-concern-modified audit report s. The underreaction may be the result of the limits to arbitrage in the sample companies, predominantly small loser stocks, but th e authors cannot rule out the behavioral explanation of investors (professional and individual) being in denial of the implications of the going-concern opinion. Other articles attempt to explain short-term momentum in returns, arguably the most difficult empirical finding to reconcile with traditional rational finance theory. Grinblatt and Han (2005) argue that prospect theory, and the resulting tendency of investor s to hold losing positions and sell winners, explains the momentum effect. This trading behavior of investors me ans prices underreact to news and mo mentum occurs as the mispricing slowly corrects. For example, when good news emerges about a stock, selling by inve stors who, subject to the disposition effect, are inclined to sell winners wi ll slow the pace at which the good news can be reflected in a higher stock price. The authors find that a proxy for unrealized gains, which will de termine investors™ disposition to sell or hold, can explain the leve l of momentum profits. Representativeness Bias and fiGood Companies.fl The representativeness heuristic involves individuals assessing situations based on superficial char acteristics rather than unde rlying probabilities. One possible manifestation of this inclination is the assumpti on that the shares of a figood companyfl will be a good investment. Shefrin and Statman (1995) show that survey respondents believ e that the shares of companies that do well in the annual Fortune magazine survey of corporate reputation will prove to be good investments. Their findings indicate that these companies tend to be larg e companies (past winners) with low book-to-market ratios, which are characteristics linked empirically to poor subsequent returns. More recent work is somewhat mixed. Anderson and Smith (2006) fi nd that the shares of the Fortune survey™s most admired companies outperform the S&P 500 Index in the periods following publication of the survey results, whereas Statman, Fisher, and Anginer (2008) use a longer sample and find results consistent with Shefrin and Statman (1995). Cooper, Dimitrov, and Rau (2001) show that investors can be influenced also by the name a company adopts, again consistent with the representativeness heuristic. Their analysis of 95 companies that changed to dot-com (.com) names during 1998 and 1999 finds that these compan ies earned statistically sign ificant and sizably positive abnormal returns that did not appear to reverse in the following 120 trading da ys. They note that adoption of the dot-com name appears to lead to fiinvestor mania.fl Not all of the companies that changed names had substantial involvement with the internet, but th e extent to which they did was not related to the share price response. The Equity Risk Premium. The relatively high level of the equ ity risk premiumŠthat is, the excess return of equities over bonds or T-bill sŠis another empirical finding regarded by some authors in the traditional finance literature as a puzzle. Behavior al theories may offer some solution to the puzzle. Benartzi and Thaler (1995) argue that loss-averse investors who evaluate their portfolio on a regu larŠat least annualŠbasis will require a high risk premium to be induced to invest in equities. For these investor s, losses are weighed more heavily than comparable-sized gains, and given the distribution of gains and losses at sh ort horizons, investors who regularly evaluate their portfolios will often be confronted with painful losses.

PAGE – 5 ============
Behavioral Finance ©2008, The Research Foundation of CFA Institute 5Asness (2000) presents an ex planation for time variation in the equity risk premium based on the idea that the relative yield on stocks versus bonds wi ll reflect the experience of each generati on of investors with each asset class, particularly in terms of volatility. The risk premium at any point in time is ar gued to be determined by the relative volatility of stocks and bonds over the pa st 20 years (i.e., the personal experience of the majority of current investors). The results are shown to be robust when changing the horizon to between 10 and 30 years. The results can explain why stocks previously yielded more than bonds but in the more recent past have had the opposite relationship. (See also Zhiyi Song™s literature review on the Equity Risk Premium: Behavioral Corporate FinanceBehavioral finance also has applications in analysis of corporate finance decisions. As Baker, Ruback, and Wurgler (2007) note, the extension of behavioral ideas to corpor ate finance has taken two distinct paths. The first path, which takes the view that investors are less than fully ra tional, analyzes the corporat e financing decisions made by management in response to the behavior of investorsŠth at is, the rational managers make decisions in response to the mispricing of securities by behaviorally biased investors. The second path holds that corporate managers can be subject to behavioral biases and that some of the corporate financ e transactions they undertake are the result of those biases. For example, managers may make certain decisions because they are overconfident about their abilities or the prospects for their firm or because they are loss averse. Bake r et al. (2007) note that the second, fiirrational managers,fl path is somewhat less developed th an the first path, which focuses on managerial responses to market mispricing. Rational Managers and Irrational Investors. The rational managers/irrational investors school of thought has its main implications in terms of corporate financial structure and the timing of securities issues. For example, Baker and Wurgler (2000) fi nd that the share of equity issues relative to total equity and debt issues is high before periods of low equity market returns, su ggesting that companies time their equity issues to take advantage of positive investor sentiment and market misp ricing. These results suggest also that corporate capital structure often reflects the cu mulative outcome of past attempts to time the equity market rather than some target capital structure (Baker and Wurgler 2002). Baker and Wurgler (2004) argue that dividend policy may be influenced by managers ficateringfl to the demands of investors. According to the authors, managers rationally cater to investor demand by paying dividends when investors put higher prices on payers and not pa ying when investors prefer nonpayers. The authors show that the lagged dividend premiumŠthe difference between the average market-to-book ratio for dividend payers relative to the average for nonpayersŠ is positively related to dividend in itiations. The authors argue also that investors™ time-varying demand for dividends is relate d to sentiment. When the dividend premium is high, investors are seeking companies that exhibit characteristic s of safety, and when it is low, investors are seeking maximum capital growth. Shleifer and Vishny (2003) present a model that s eeks to explain merger and acquisition (M&A) deals in behavioral terms. In the model, stocks are mispriced and management perceives and responds to the mispricing. The authors argue that M&A decisions and decisions about methods of financing deals are driven by misvaluations of the participating companies; for e xample, acquisitions will in volve payment in stock wh en valuations are high. The model suggests that acquisitions for stock are made by overvalued co mpanies and target companies tend to be less overvalued. The model is able to explain many of the observed characte ristics of the M&A market. Behavioral finance also has implications for the mark et for IPOs. These offerings are widely documented as showing high first-day returns, usually taken to imply that the issues are un derpriced at the offering price. One puzzle is why issuers and pre- IPO shareholders are prepared to tolerate th is fimoney left on the tablefl phenomenon. Loughran and Ritter (2002) propose a model based on prospe ct theory in which issuers are likely to net the amount of money left on the table by an underpriced offering to gether with the figainfl in their wealth that comes from the rise in the price of the shares th at they retain in the company. The ne t amount will often be a positive sum

PAGE – 6 ============
Behavioral Finance 6©2008, The Research Foundation of CFA Institute with the increase in value of the re tained holdings exceeding the difference between the offer price and the market price for the shares sold in the offering. Furthermore, the most underpri ced offerings tend to be those in which the offer price has been revised up in the face of strong demand from the pr ice set out in the prospectus. Therefore, the original pre-IPO shareholders can of fset the loss of the underpricing with the good news that their total wealth is higher than was previously expected. Ljungqvist and Wilhelm (2005) provide some support for this hypothesis in that issuers of underpriced offerings often use th e IPO underwriter for subsequ ent equity issues, suggestingthey are not unhappy with the service received. Irrational Managers. Despite the suggestion by Baker and Wurgle r (2004) that the irrational managers school of behavioral corporate finance is currently underdeveloped, the theory can be regarded as having a relatively long history. For example, Roll™s (1986 ) fihubris hypothesisfl of takeovers is based on the idea of overconfidence among managers, which leads them to overestimate the gain s to be made from corporate activity. More recently, Doukas and Petmezas (2007) calculate a measure of management overconfidence and find overconfident managers™ companies earn lower merger announcement returns and have poorer long-term share price performance. Self-attribution bias also appears to be at pl ay, in that returns are lower for serial acquirers (five or more deals in three years) than for first-time deals. Another example of the managerial overconfidence idea relates to project appraisal and internal investment decisions. Malmendier and Tate (2 005) argue that overco nfident management overestimates the returns on investment projects and views external funds as too cost ly. They tend to overinvest when internal funds are abundant but refrain from investing when external funds are required. Th e authors use management™s personal financial exposure to company-specific risk as a proxy for overconfidence and find that investment by overconfident CEOs is closely related to cash flow. Investor Behavior and Behavioral Portfolio TheoryIn this section, we look at the trad ing and portfolio construction behavior of investors without regard to whether or not that behavior has a lasting impact on market pric es. We consider the evidence for professional investors, fiduciaries (such as pension fund trus tees), and individual investors. Professional Investors. Categorizing market participants as info rmed and uninformed traders, or noise traders and arbitrageurs, encourages the perspective of professional investors as the rational, informed arbitrageurs. But plenty of evidence is available of behavioral biases being displayed by professional investor s, even in fireal moneyfl situations. Hong, Kubik, and Stein (2005) find that mutual fund ma nagers herd in terms of the stocks that they buy or sell during a particular quarter. Co val and Shumway (2005) show that tr aders at the Chicago Board of Trade (CBOT) are loss averse and inclined to take more risk in the afternoon if they have had losses in the morning. Their action has at least a short-term effect on prices. In an experimental setting, Haigh and List (2005) show that CBOT traders display myopic loss aversion to a gr eater degree than do students . Garvey and Murphy (2004) find evidence of the disposition effe ctŠthe tendency to sell winners and ho ld losersŠamong a group of profitable proprietary traders. The tendency to sell winners and hold losers lowers the returns the traders earn. Fiduciaries. A large proportion of institutional funds are co ntrolled by trustee boards or other fiduciary committees that are potentially subject to behavioral biases. Importantly in this context, individual behavioral biases operate alongside biases that may occur as a re sult of the dynamic of gr oup (or committee) decision making. For example, Wood (2006) provides an interestin g discussion of the behavior al biases that may affect investment committees.

PAGE – 8 ============
Behavioral Finance 8©2008, The Research Foundation of CFA Institute Investors with weak preferences or limited knowledge may use simple rule s of thumb to deal with investment choice. Benartzi and Thaler (2001) document experiments of inve stors using a naïve 1/ n diversification strategy in which they allocate contributions equally among the fu nds offered in their 401(k) pension plans. This type of strategy means the fund range has a strong influence on the investors™ chosen asse t allocation. Huberman and Jiang (2006), using a larger and more appropriate dataset, find evidence instead for a conditional 1/ n approach in which investors choose three or four fu nds from the range offered and then allocate equally among them. In this case, fund range has less influence on asset allocation. One particularly puzzling e xample of investor behavior observed in 401(k) plans is the enthusiasm of many participants for investing their contribu tions in the stock of their employer. Although some of the account balances invested in employer stock are explained by the fact that employer matching contributions are often made in the form of stock, with restrictions on sa le, studies (e.g., Benartzi 2001) find pa rticipants voluntarily allocating their own contributions to employer stock. Obvi ously, not only does this strategy re present an underdiv ersified portfolio, but the investment also has a strong correlation with em ployees™ labor income. Particip ants should, ideally, look to hedge the labor income risk they face rather than fid ouble upfl with portfolio invest ments. Explanations for this strange investment approach include na ïve extrapolation of the strong past performance of the company™s shares (Benartzi 2001) and employees underestimating the risk of the employ er™s share, possibly because of their familiarity with the company (Benartzi 2001; Huberman 2001). Employees may also perceive implicit advice (or an endorsement) in the fact that employer matching cont ributions are made in stock. In addition to evidence of 401(k) investors following dubious investment strategies , substantial evidence shows inertia that leads participants to stic k with default options in terms of savin gs rates and investment funds. In many cases, the default funds will be cash or money market funds, which are arguably t oo conservative for long-term saving (e.g., Madrian and Shea 2001). One of the most influential programs of research on the trading behavi or of individual investors has been conducted by Barber and Odean, who managed to obtain the trading records of 35,0 00 investors with accounts at a discount brokerage. The authors find evidence of excessive trading re ducing returns (Barber and Odean 1999, 2000; Odean 1999) and attribute the result to overconfid ence. Psychology research typically finds men are more overconfident than women, and consistent with this, Barber and Odean (2001) find that men trade more than women and earn lower returns. Barber and Odean (1999; Od ean 1998) also find evidence of the disposition effect (as do Shefrin and Statman 1985), in which investors are reluctant to realize losses and tend to sell winners and hold losers. In the discount brokerage data, stocks sold tended to do better subsequ ently than those retained or used to replace those sold. Many individual investors use mutual funds rather than investing in indi vidual stocks. Selecting mutual fund managers is, however, not necessarily an easier task than picking stocks. Rabin (2002) argues that many investors believe in the filaw of small numbersfl and are prone to overestimate the extent to which a short sequence of observations is likely to be characteri stic of the underlying da ta-generating process. In the context of mutual fund performance, this view is likely to le ad these investors to reac t to short-term performance records in hiring or firing funds, even those in which perf ormance histories are uninformative as to future performance. This finding is similar to that discussed above in th e context of fiduciar y decision making. Mutual fund investors are fooled by more than uninformative past perf ormance figures. Cooper, Gulen, and Rau (2005) show that mutual fund na me changes are often designed to latch on to the current fihotfl investment styles. They find that in the year after a fund changed its name, money flows from investors increased substantially. The funds experienced no significant im provement in performance, and in ma ny cases, the holdings of the fund did not match the style implied by the new name. Further empirical support for the notion that fund select ion decisions are affected by behavioral biases comes from Frazzini and Lamont (forthcoming 2008), who argue that mutual fund investor cash flow s represent fidumbmoney.fl They find that mutual fund investors tend to re allocate their cash to fund s that own stocks with low future returns.

135 KB – 26 Pages