The Effects of Institutional Investor Objectives on Firm Valuation and Governance

Finance and Economics Discussion Series Divisions of Research & Statistics and Monetary Affairs Federal Reserve Board, Washington, D.C. The Effects of Institutional Investor Objectives on Firm Valuation and Governance Paul Borochin and Jie Yang 2016-088 Please cite this paper as: Borochin, Paul and Jie Yang (2016). “The Effects of Institutional Investor Objectives on Firm Valuation and Governance,” Finance and Economics Discussion Series 2016-088. Washington: Board of Governors of the Federal Reserve System, https://doi.org/10.17016/FEDS.2016.088. NOTE: Staff working papers in the Finance and Economics Discussion Series (FEDS) are preliminary materials circulated to stimulate discussion and critical comment. The analysis and conclusions set forth are those of the authors and do not indicate concurrence by other members of the research staff or the Board of Governors. References in publications to the Finance and Economics Discussion Series (other than acknowledgement) should be cleared with the author(s) to protect the tentative character of these papers.

The Effects of Institutional Investor Objectives on Firm Valuation and Governance ∗ Paul Borochin Jie Yang School of Business Board of Governors of the University of Connecticut† Federal Reserve System‡ This version: November, 2016 Abstract We find that ownership by different types of institutional investor has different implications for future firm misvaluation and governance characteristics. Dedicated institutional investors decrease future firm misvaluation relative to fundamentals, as well as the magnitude of this misvaluation. In contrast, transient institutional investors have the opposite effect. Using SEC Regulation FD as an exogenous shock to information dissemination, we find evidence consistent with dedicated institutions having an information advantage. The valuation effects are primarily drivenbyinstitutionalportfolioconcentrationwhilethegovernanceeffectsaredriven by portfolio turnover. These results imply a more nuanced relationship between institutional ownership and firm value and corporate governance. Keywords: Institutional investors, investor type, dedicated, transient, misvaluation, corporate governance, blockholding, portfolio turnover, information dissemination, SEC Regulation FD JEL classification: G30, G32, G14, G38 ∗WethankBillSchwert(Editor),ananonymousreferee,ReenaAggarwal,JosephGolec,DavidRobinson,seminar participantsatGeorgetownUniversity,UniversityofBaltimore,UniversityofMassachusettsLowell,andtheOfficeof theComptrolleroftheCurrency,andparticipantsatthe2016EuropeanFinancialManagementAssociationMeeting and 2016 Financial Management Association. The ideas in this paper are solely those of the authors and do not necessarily reflect the view of the Federal Reserve System. All remaining errors are our own. †Storrs, CT 06269. Phone: (860) 486-2774. Email: paul.borochin@uconn.edu. ‡Washington, DC 20551. Phone: (202) 736-1939. Email: jie.yang@frb.gov.

The Effects of Institutional Investor Objectives on Firm Valuation and Governance This version: November, 2016 Abstract We find that ownership by different types of institutional investor has different implications for future firm misvaluation and governance characteristics. Dedicated institutional investors decrease future firm misvaluation relative to fundamentals, as well as the magnitude of this misvaluation. In contrast, transient institutional investors have the opposite effect. Using SEC Regulation FD as an exogenous shock to information dissemination, we find evidence consistent with dedicated institutions having an information advantage. The valuation effects are primarily drivenbyinstitutionalportfolioconcentrationwhilethegovernanceeffectsaredriven by portfolio turnover. These results imply a more nuanced relationship between institutional ownership and firm value and corporate governance. Keywords: Institutional investors, investor type, dedicated, transient, misvaluation, corporate governance, blockholding, portfolio turnover, information dissemination, SEC Regulation FD JEL classification: G30, G32, G14, G38

“Institutional investors are not all the same. They come in many different forms and with many different characteristics.” - Commissioner Luis Aguilar, SEC, April 2013 1 Introduction Institutional equity ownership has risen dramatically in the last 30 years, inviting a more thorough investigation into its effects on firm valuation and operations. In the 1980s, institutional investors held approximately 20%-30% of the average firm (with individual investors making up the rest). By the 2010s, over 65% of the average firm is owned by institutional investors.1 This increase in institutional holdings coincides with the growing sophistication of markets and growing importance of corporate governance. Indeed, much of the literature focuses on the informed “smart money” of the institutional investor in contrast to the less sophisticated individual “retail” investor. Barber and Odean (2008) confirm that, unlike institutions, individual investors are net buyers of attention-grabbing stocks due to limited search resources. This ability of institutional investors to gather information and impose market discipline on management should translate into improvements in market efficiency and asset valuation. For example, Sias and Starks (1997) find that institutional trading increases the speed of adjustment of information into prices and Nagel (2005) finds that short sale constraints bind for stocks with low institutional ownership, resulting in underperformance. Yet, as suggested in a speech by Commissioner Luis Aguilar of the SEC,2 not all institutional owners are alike nor do they have similar effects on firms. While many characteristics differentiate institutional investors from one another, portfolio turnover and holdings concentration are the most fundamental. For example, Bushee (1998, 2001) finds that short-term, low concentration (“transient”) investors lead to myopic investment decision-making by managers and over-weighting of near-term expected earnings to the detriment of long-term earnings. As such, myopic investment behavior leads to myopic corporate decision-making, potentially destroying long-run value for the firm. In addition, Yan and Zhang (2009) show that the pricing impact of large institutional investors is largely driven by short-term investors reacting and trading on new information. These 1Based on data from Thomson Reuters s34 Holdings Database and corroborated by Blume and Kein (2014). 2http://www.sec.gov/News/Speech/Detail/Speech/1365171515808 1

findingssuggestthat,subsetsofinstitutionalinvestorswithdifferentholdingshorizonsandportfolio concentrations may have differential impact on corporate decisions and firm valuation. In this paper, we study the impact of different institutional investor types on the valuation, corporate governance, and future performance of firms. Specifically, we focus on two fundamental investor characteristics, portfolio turnover and holdings concentration. We begin by using the Bushee (1998, 2001) classification of institutions into “transient” and “dedicated” investor types. Institutional investors are “transient” if they take small positions in the firms they hold and have high portfolio turnover. Due to a short investment horizon and lack of focus on particular firms, these investors are likely to be myopic traders looking for short-term gains. On the other hand, “dedicated” investors take highly concentrated positions in the firms they hold and have low portfolio turnover. Both of these portfolio characteristics suggest dedicated investors are more likely to invest for the long run, gathering costly firm-specific information and trading on growth potential of a firm. Bushee (1998) finds evidence consistent with this interpretation as firms with predominantlydedicated(transient)institutionalownersinvestmore(less)inR&D.Figure1shows thattransientinstitutionalinvestorscompriseanincreasinglylargeportionofthetotalinstitutional investor pool over time, making an investigation into the effects of different institutional investor types particularly relevant. We find that firms with higher percentages of transient (dedicated) institutional investors experience more (less) subsequent overvaluation and misvaluation. Specifically, using the Rhodes- Kropf, Robinson, and Viswanathan (2005) book-to-market decomposition, firms with more transient (dedicated) institutional investors experience more (less) positive firm-specific deviation from fundamental values in the following quarter, consistent with overvaluation. They also experience more (less) of an absolute deviation from fundamental value in the next quarter, consistent with misvaluation. Furthermore, firms that experience an increase in their percentage of transient (dedicated) institutional investors experience more (less) overvaluation and misvaluation in the next quarter. These effects persist after controlling for common firm characteristics that may impact firm valuation and after considering alternative explanations such as perceived growth opportunities. In the appendix, we obtain similar results using alternative definitions of overvaluation and misvaluation based on the Hirshleifer and Jiang (2010) undervalued minus 2

overvalued factor.3 We posit that the differential impact of transient versus dedicated institutional investors on subsequent firm overvaluation and misvaluation is driven by differences in information gathering. To examine the information channel, we explore the relationship between institutional investor types and firm valuation around the enactment of SEC Regulation FD. Regulation FD, enacted in 2000 by the U.S. Securities and Exchange Commission, mandates that when an issuer discloses material non-public information to specific outsiders (e.g., analysts or institutional investors), this disclosuremustbemadepublic,therebyeliminatingtheinformationaladvantagepreviouslyenjoyed by blockholders (Francis, Nanda, and Wang, 2006; Anderson, Reeb, Zhang, and Zhao, 2013). As such, this regulation addresses the problem of selective disclosure of material information and provides a natural experiment in which to test the information channel. Using a difference-in-difference framework we find that, as expected, Regulation FD has no significant impact on the relationship between dedicated investor ownership and subsequent overvaluation and misvaluation. That is, dedicated institutional investor ownership reduces overvaluation and misvaluation to a similar degree before and after the regulation, consistent with their having access to information both prior to and after Regulation FD. On the other hand, the regulation significantly reduced the subsequent overvaluation and misvaluation for firms with above-median levels of transient investors relative to the preceding time period, consistent with transient investors having lower barriers to access of information previously only enjoyed by dedicated investors. These results do not exist using a placebo event year of 1995. Although we use a model with lagged changes in institutional ownership type on overvaluation andmisvaluation, theobservedresultsmaybeduetoself-selectionbyinvestorsintocertaintypesof firmsratherthanacausaleffectofownershiptypeonfirmvaluation. Wecreatesamplesmatchedexanteonmisvaluationinadifference-in-difference-in-differenceframeworktocontrolforself-selection issues in the appendix. We obtain similar results. Having established that transient and dedicated institutional investors have differential impact 3The Hirshleifer and Jiang (2010) misvaluation measure is constructed based on incorporating managerial perceptions of firm growth and firm value as implied by equity issuance and repurchase behavior. As such, this measure approahces misvaluation from a managerial perspective while the RKRV measure approaches misvaluation from a fundamentals perspective. Given the use of distinct information sets and perspectives on misvaluation, the two misvaluation measures serve as reasonable robustness checks for each other. 3

on subsequent firm overvaluation and misvaluation, we next explore which of the associated portfolio characteristics - portfolio turnover or holdings concentration - is responsible for these findings to shed further light on the causes of differences between the two previously defined institutional types. We find that ownership by institutions in the lowest portfolio turnover tercile (i.e., long horizon investors) results in less firm-specific overvaluation as well as lower misvaluation relative to fundamentals in the subsequent quarter. Institutional ownership by investors in the highest portfolio turnover tercile also results in less overvaluation, but statistically insignificant misvaluation. Importantly, firms held by institutions that have diversified portfolios and either high or low portfolio turnover experience more future overvaluation. Finally, we explore whether institutional ownership types also have differential effects on measures of risk, corporate governance, and future firm performance. We document that firms held by transient institutional investors have more expected tail risk, higher realized returns volatility, higher average and median executive compensation, worse accruals quality, lower payout ratios, and lower leverage increases relative to those held by dedicated institutional investors. Firms held by transient investors experience positive abnormal returns relative to the 5-factor model in the subsequent quarter, whereas those held by dedicated investors experience positive abnormal performance later in the year consistent with longer investment horizons. Firms held by dedicated investors experience positive raw returns over the subsequent four quarters, whereas those held by transient investors experience negative or insignificant raw returns over the same period. These findings enable us to contribute to the ongoing debate within extant literature over whether institutional investors benefit the markets, either through improvements in market efficiency or by providing corporate governance, or whether they harm them through opportunism and pressure exerted on managers to achieve short-term results at the expense of long-term performance. While the literature generally accepts that institutional investors are more informed and allow markets to be more efficient (Sias and Starks, 1997; Nagel, 2005; Barber and Odean, 2008), there are contradicting results and arguments among both academics and practitioners over the institutional benefits to governance and costs of induced myopia. For example, Gillian (1995), Karpoff, Malatesta, and Walkling (1996), Smith (1996), and Wahal (1996) find no long-term effects fromshareholderactivismwhileNesbitt(1994)findsthatfirmstargetedbyCalPERSoutperformin 4

subsequentyears. Morerecently,Brav,Jiang,Partnoy,andThomas(2008)findapositiveabnormal return for firms targeted by hedge fund activism and Boone and White (2015) find that higher institutional ownership is associated with greater management disclosure and lower information asymmetry. Conversely, a survey of over 1,000 board members and executives around the world conducted by the Canadian Pension Plan Investment Board and McKinsey and Company (Barton and Wiseman, 2014) draws attention to the effects of short-term investor pressure on corporate decision-making and its negative effects on value maximization. In this report, 79% of respondents felt pressured to demonstrate performance over a horizon of two years or less, and 44% used a horizon of three or less years to set corporate strategy while 86% stated that a longer horizon would have improved financial performance. These seemingly dissonant effects of institutional ownership on firm performance can be explained by the differential effects we find for institutional ownership types. The conflicting findings in the growing literature above suggest that combining all institutions into one category yields different, and likely mistaken, results than those that would be obtained by more refined categorizations. Indeed, among the papers that find benefits to corporate governance, most focus on a specific type of institutional investor (e.g., CalPERS or hedge funds) rather than institutional investors as a single group. Themaincontributionofourpaperliesinidentifyingthedistincteffectsofinstitutionalinvestor types and investment styles on subsequent firm overvaluation, misvaluation, governance measures, and realized future performance. Bushee (1998, 2001) groups institutional investors based on their portfolio turnover and holdings concentration and finds that “transient” institutional investors are related to myopic corporate decision-making while “dedicated” institutional investors are not. Gompers and Metrick (2001) find that large institutional investors invest in large companies, thereby increasing the prices of large stocks. Yan and Zhang (2009) show that these trends in large institutional investors is largely driven by short-term investors reacting and trading on new information. On the other hand, Gasper, Massa, and Matos (2005) show that acquisition targets held by short-term investors experience a lower premium and Chen, Harford, and Li (2007) find that only long-term investors are related to positive post-merger performance rather than focusing on short-term gains. 5

We add to this growing literature by identifying how institutional investor types and their underlying portfolio turnover and holdings concentration affect the valuation, governance characteristics, and future performance of firms in which these institutions invest. We find that dedicated institutional investor ownership results in more accurate future firm valuation, superior characteristics related to corporate governance, and superior long-term performance of the firm. The valuation effects are primarily improved by institutional investors with higher portfolio concentration and the governance and long-term performance effects by institutions with lower portfolio turnover. Specifically, we find that long-term institutions are able to achieve long-term performance: firmswithmorelong-terminstitutionalinvestorsoutperformfirmswithfewerofthem by 2.8% on a risk-adjusted basis over the subsequent year. 2 Data 2.1 Institutional Ownership and Investor Types We start with data on institutional investor ownership obtained from the Thomson Reuters Institutional Holdings database. The Thomson Reuters holdings database covers investment companies and their security holdings as reported on their 13F forms filed with the Securities and Exchange Commission (SEC) every quarter.4 We first classify institutional investor types based on the combination of portfolio turnover and holdings concentration from Bushee (1998, 2001) and Bushee and Noe (2000). Bushee (1998, 2001) categorizes institutional investors as “transient”, “quasi-indexer”, or “dedicated” based on their investment horizons and portfolio concentration.5 Investors are classified as “transient” if they have short investment horizons reflected by high portfolio turnover and highly diversified portfolio holdings.6 Analogously, “dedicated” investors have long investment horizons reflected by low portfolio turnover and focused portfolio holdings. The third class of investors, “quasiindexers”, are long-horizon, low turnover investors that are highly diversified. We focus our analysis on dedicated (DED) and transient (TRA) institutional investors as they both have an 4All institutions conducting business in the U.S. with investments over $100 million are required to disclose their list and shares held of Section 13F securities, which include exchange-traded stocks. 5We are grateful to Brian Bushee for providing this data on his website. 6Weusethepermanentmanagerclassificationtoavoidissueswithinstitutionsthatchangeclassificationovertime. All results hold using the non-permanent (time-varying) classification. 6

active choice in their investment strategy: dedicated investors do not trade frequently, but hold specifically selected firms (and are therefore different from the passive indexers), while transient investorstradefrequently. Weexcludethequasi-indexerinstitutionalownershiptypeasthepassive ownership strategy does not imply any asset selection in either the time horizon or portfolio choice dimensions. Table I presents the top ten dedicated and transient institutional investors in our sample in decreasing order based on average portfolio size. As expected, the dedicated list of investors is comprised of investment management and insurance companies, which may be expected to hold stocks for long periods of time and some of which, like Berkshire Hathaway, are famous for it. On the other hand, many of the investment management firms on the transient list of investors are affiliated with investment banks such as UBS, Morgan Stanley, and Oppenheimer. Tofurthermotivatethedistinctionsbetweenthetwotypesofinstitutionalinvestors,weconsider empirical differences in their portfolio characteristics. Table II demonstrates that dedicated and transient institutions have different investment styles, as reflected in their significantly different portfolio characteristics. Dedicated institutional investors have average portfolio sizes almost four times larger than transient investors, though they hold fewer stocks on average. Additionally, dedicated investors hold a larger percentage of each firm on average, and at the median, than transients do. They also have a significantly higher variation in these holdings, consistent with the large positions of dedicated blockholders.7 Notably, dedicated investors hold firms with smaller average and median market capitalizations that are almost half the size of those held by transients, with a correspondingly lower variability in owned-firm capitalizations. This is consistent with dedicated investors possessing an information advantage, as smaller and younger firms are more opaque and difficult to analyze (Hadlock and Pierce, 2010; Karpoff, Lee, Masulis, 2013). As expected,dedicatedinvestorsarealsomorefocused,holdingfirmsfromfeweruniqueSIC3industries and have a much higher Hirshleifer-Herfindahl concentration in portfolio weights than do transient investors. Reassuringly, consistent with the Bushee (1998, 2001) definition, dedicated investors do indeedhavemoreconcentratedpositionsinfirmsrelativetototalsharesoutstandingandhavelower portfolio turnover than transients. All institutional investor characteristics between dedicated and 7That is, while both dedicated and transient investors hold small stakes in some firms, dedicated are more likely to be blockholders in other firms, leading to the higher variation. 7

transient institutional investors are significantly different at the 1% level. 2.2 Misvaluation Our main measure of misvaluation is the Rhodes-Kropf, Robinson, and Viswanathan (2005), hereafter RKRV, decomposition which splits the logarithm of the market-to-book ratio into three components: firm-specific error, time-series sector error, and long-run market value to book value. The RKRV decomposition is provided in equation (3) of their paper and reproduced below: m −b = m −v(θ ;a )+ v(θ ;a )−v(θ ;a )+ v(θ ;a )−b . i,t i,t i,t i,t j,t i,t j,t i,t j i,t j i,t (cid:124) (cid:123)(cid:122) (cid:125) (cid:124) (cid:123)(cid:122) (cid:125) (cid:124) (cid:123)(cid:122) (cid:125) (1) firm sector long−run This decomposition relies on a firm having a long-run, target, market-to-book ratio that equals that of its industry. This ratio is determined by a parsimonious set of valuation multiples: book value, leverage, and net income.8 The firm’s market-to-book ratio is comprised of a long-run market-to-book value determined from long-run multiples of the three variables used, with a time-varying sector-wide multiple representing sector-wide deviations and a firm-specific multiple accounting for any additional firm-specific deviations. Following Rhodes-Kropf, Robinson, and Viswanathan (2005), these firm-specific deviations capture the degree of overvaluation and misvaluation. We explore the impact of the different types of institutional investor on the firmspecificerrorcomponent. Positivefirm-specificerrorisourproxyforovervaluation,andtheabsolute value of the firm-specific error proxies for misvaluation.9 2.3 Financial Statement Data In addition to institutional investor ownership and firm valuation, we collect information on firm characteristics. We define firm characteristics based on financial statement data obtained from Standard and Poor’s Compustat North American quarterly database from 1985 to 2013. All dollar amounts are chained to 2000 dollars using CPI to adjust for inflation. We remove any firms with 8Despite the small number of multiples considered, this measure fits the cross-section of market to book ratios within industries reasonably well with an R2 of 0.80 to 0.94. 9The RKRV measure uses deviations from long-run market multiples to approach the issue of misvaluation. As such,thereisapotentialdual-hypothesisproblemregardingourchoiceofvaluationmodel. Weaddressthisissueby considering an alternative measure for overvaluation and misvaluation based on insider information from Hirshleifer and Jiang (2010) in the appendix. 8

negative book asset value, market equity, book equity, capital stock, sales, dividends, debt, and inventory. Such firms have either unreliable Compustat data or are likely to be distressed or severely unprofitable. In addition, we delete observations in which book assets or sales growth over the quarter is greater than 1 or less than -1 and remove firms worth less than $5 million in 2000 dollars in book value or market value to remove observations that have abnormally large changes due to acquisitions or small asset bases. Next, we remove outliers defined as firm-quarter observations that are in the first and 99th percentile tails for all relevant variables used in our analysis. Following standard practice in the literature, we remove all firms in the financial and insurance, utilities, and public administration industries as they tend to be heavily regulated. Merging institutional investor data to corporate financial data based on a firm’s CUSIP and year-quarter gives us a sample of 236,025 firm-quarter observations spanning 11,116 firms. Appendix A details the construction of all variables. Table III provides the summary statistics for our main sample and Table IV presents the correlation matrix. The average (median) firm in our sample has 41.5% (38.4%) institutional ownership, with 4.6% (2.9%) of the firm owned by dedicatedinstitutionalinvestorsand27.3%(27.3%)ownedbytransientinstitutionalinvestors. The correlation between dedicated and institutional ownership is -0.23.10 While there is a wide range of firm overvaluation, the average and median firm in our sample is overvalued (i.e., has positive firm-specific error) according to the RKRV measure. In Table V, we compare key characteristics of firms held by the two types of active institutional investors. We subset the data on the highest half of ownership by each investor type and find that the two active types hold firms with significantly different characteristics. This is consistent with the different investment styles and objectives evident in their different approaches to holding periodandconcentration. Notabledifferencesinthefirmseachtypechoosestoholdsignalpotential notably different effects on firm valuation, operations, and performance. Specifically, we find that dedicated institutional investors hold larger firms than transient investors do, with average total asset values of $1,763.4 and $1,434.2 million respectively. Despite this, dedicated investors hold firms with lower market values than those held by transient investors, with average market values of $1,632.5 and $1,741.9 million respectively. These two observations taken together imply a value 10While this is not high enough to warrant concerns about multicollinearity when including both dedicated and transient ownership in regression models, we test for this by including only one at a time. All results hold. 9

preference by dedicated investors relative to transient ones; indeed the log of long-run market-tobook ratio under the RKRV decomposition is significantly lower for firms held by dedicated (0.479) relative to transient institutional investors (0.578). Furthermore, firms held by dedicated investors experience almost no firm-specific overvaluation (0.004) under the RKRV measure whereas firms held by transient investors experience much more (0.140). The t-tests of means also suggest significant differences in information quality for firms held by thetwotypes: dedicatedinvestorspreferlesspopularfirmswiththeaveragenumberofinstitutional investors in firms they hold at 66.9, relative to the 83.8 in firms held by transient investors. Dedicated investors also hold firms with lower analyst following, 3.3 to 4.9 respectively, as well as a higher principal component of several opacity measures,11 consistent with an informational advantage in selecting less-followed and more opaque firms. Firms held by dedicated investors are also less likely to have a credit rating, further consistent with a preference of dedicated investors in holding firms with less publicly available information. Dedicated investors also hold firms with significantly different operating characteristics than those held by transient investors: they have higher Z-scores, higher leverage, lower cash flow dispersion, lower realized volatility, lower tai or crash risk for firm cash flowsl risk,12 lower average, maximum, and median executive pay, higher quality accruals, higher payout ratios, and higher net leverage changes. Overall, these results are strongly indicative of differing investment preferences by the two institutional investor types, motivating our study of the effects of intuitional ownership type on future firm value. The overlap between firms owned by dedicated, transient, long/short horizon, and focused/diversified institutional investors is imperfect but consistent with expectations. Firms owned by an above-median number of dedicated investors also have more long-horizon institutional investors. Firms owned by an above-median number of transient investors also have more shorthorizon institutional investors. The differences observed in Table V also hold when we subsample firms based on the dollar amount held by each institutional ownership type (i.e., value-weighted), rather than by the number of each institutional investor type (i.e., equal-weighted). 11See Karpoff, Lee, and Masulis (2013) for details on constructing principal components of opacity proxies. 12As measured by IVspread , the implied volatility spread across moneyness between OTM and ATM put mon options. See Borochin and Yang (2016) for details on constructing IVspread . mon 10

3 Institutional Investor Type and Firm Value Inthissection,weexaminetheimpactofdifferentinstitutionalinvestorstypesonovervaluationand misvaluation. We begin with classifying institutional investors as dedicated or transient, following Bushee (1998, 2001). Transient investors are more likely to enter and exit their positions quickly and aggressively, relying on publicly available information and strategies such as technical analysis to choose their investments (Bushee, 2001). In contrast, dedicated investors have more incentive to gather information and build relationships with their investments. Due to the information advantage that dedicated investors have relative to transient investors, we hypothesize that firms with more transient institutional investors will have more future overvaluation and misvaluation while firms with more dedicated institutional investors will have less. Furthermore, if dedicated investors possess an informational advantage an increase in dedicated institutional investors should leadtoadecreaseinovervaluationandmisvaluation,withtheoppositeeffectoccurringforincreases in transient institutional ownership. Hypothesis 1. Firms with a higher percentage of dedicated (transient) institutional investors experience less (more) overvaluation and misvaluation. Hypothesis 2. Firms that experience an increase in dedicated (transient) institutional investors experience less (more) overvaluation and misvaluation. 3.1 Baseline and Full Models Starting with a baseline model, we estimate the effects of lagged levels of institutional ownership by investor type on overvaluation and misvaluation variables of interest, Y , in the subsequent i,t quarter. We consider two measures: 1) the RKRV firm-specific valuation error, which captures the direction of valuation error with positive error indicating overvaluation, and 2) the absolute value of the RKRV firm-specific error which captures the magnitude of misvaluation regardless of direction of deviation from fundamental value. Y = α+β pDED +β pTRA +fe +fe +ε , (2) i,t 1 i,t−1 2 i,t−1 t j i,t 11

where pDED is the percentage of dedicated institutional owners relative to the total number of institutional owners, and pTRA is the percentage of transient institutional owners relative to the total number of institutional owners. We control for macroeconomic effects with year fixed effects, forseasonalitywithquarterfixedeffects,andforgeneralchangesinvaluewithindustryfixedeffects. Standard errors are double-clustered by firm and year-quarter. Next, we decompose the lagged level of ownership by type into the second lag of the level and the first lag of the change: X ≡ X +∆X . i,t−1 i,t−2 i,t−2,t−1 This allows us to study the dynamics of the effect of ownership type on overvaluation and misvaluation by examining both changes in ownership as well as levels in ownership, giving us a second version of the baseline model: Y = α+β pDED +β ∆pDED +β pTRA +β ∆pTRA +fe +fe +ε . (3) i,t 1 i,t−2 2 i,t−1 3 i,t−2 4 i,t−1 t j i,t This decomposition of the lagged level of ownership into the second lag of the level and the first lag of the change has three main advantages. First, it allows us to study the impact of levels as well as changes of ownership type on firm value. Second, it controls for any persistence in institutional ownership through the second lag level, allowing the change in ownership to act as a shock. Third, although imperfect, it serves as a first attempt to address the potential selection bias issue in controlling for the relationship between firm valuation and pre-existing ownership types. We address the potential selection bias more rigorously in Section 4.1 and in the appendix. Wefurthersupplementourbaselinemodelbyincludingcontrolsforcommonfirmcharacteristics and factors relevant to firm value. This provides us with our full model: Y = α+β pDED +β ∆pDED +β pTRA +β ∆pTRA +β lnTA i,t 1 i,t−2 2 i,t−1 3 i,t−2 4 i,t−1 5 i,t−1 +β pINST +β INSThhi +β Zscore +β Lev +β LTCR (4) 6 i,t−1 7 i,t−1 8 i,t−1 9 i,t−1 10 i,t−1 +β CFdisp +β NumEst +fe +fe +ε , 11 i,t−1 12 i,t−1 t j i,t where lnTA is log-transformed total assets of the firm to control for firm size, pINST is the percentage of institutional ownership by all types of institutional investors in the 13F database 12

to control for overall institutional investment, INSThhi is the Herfindahl-Hirschman index of institutional ownership in the firm to control for concentration of institutional ownership, Zscore is the Altman (1968) Z-score measure to control for firm financial health, Lev is the ratio of long-term debt to total assets to control for capital structure, LTCR is an indicator variable for whether the firm has a S&P long-term credit rating to control for financial constraints, CFdisp is thestandarddeviationofcashflowsovertheprior20quartersdividedbythemeanofcashflowsover the prior 20 quarters to control for cash flow volatility,13 and NumEst is the number of analysts reporting EPS forecasts for the firm in a given quarter to control for market attention. All control variables are lagged one quarter. In addition, we include year and quarter fixed effects to control for the macroeconomic environment and seasonality, respectively, and SIC3 fixed effects to control for variation across industries. Standard errors are double-clustered by firm and year-quarter. Table VI presents the results for the baseline and full models. Columns (1) and (2) report the results using the baseline model in equation (2). We see in Column (1) that firms with higher dedicated institutional ownership experience less overvaluation (-0.767) in the subsequent quarter, while those with higher transient ownership become more overvalued (0.388). That is, for a 100% increase in dedicated ownership, the logarithm of the firm’s value relative to fundamentals drops by 0.767 and for a 100% increase in transient ownership, the log firm value relative to fundamentals increases by 0.388.14 Both effects are significant at the 1% level. Using the misvaluation measure, Column (2) finds that firms with more dedicated ownership are not significantly misvalued in the subsequent quarter, while those with more transient ownership face significantly higher subsequent misvaluation (0.073). In other words, for every 100% increase in transient ownership, the logarithm of misvaluation increases by 0.073 relative to firm fundamentals, which is significant at 1%. These results are consistent with Hypothesis 1, where firms with more dedicated (transient) institutional investors face lower (higher) subsequent overvaluation and misvaluation. When we decompose lagged institutional ownership into second lagged levels and first lagged changes in equation (3), we see in Column (3) of Table VI that firms become less overvalued with both higher levels as well as larger increases in DED ownership whereas they become more 13We also consider the dispersion of earnings as well as the dispersion of sales over the same time period, with no difference in the observed results. 14Recall that Rhodes-Kropf, Robinson, and Viswanathan (2005) decompose the logarithm of the market-to-book ratio. 13

overvalued with both higher levels and larger increases in TRA ownership, significant at the 1% level. In fact, overvaluation falls by 0.871 for a 100% increase in pDED , the level of dedicated i,t−2 institutional ownership, consistent with the decrease documented in column (1), and falls further by 0.516 for a 100% increase in ∆pDED , the change in dedicated ownership. Similarly, firm i,t−1 valuerelativetofundamentalsrisesby0.453fora100%increaseinpTRA , theleveloftransient i,t−2 institutional ownership, consistent with the increase documented in column (1), and rises further by 0.410 for a 100% increase in ∆pTRA , the change in transient ownership. All effects are i,t−1 significant at the 1% level. Column (4) presents the results for misvaluation, finding a weaker but also positive misvaluation effect on pTRA , significant at the 10% level. i,t−2 Finally, Columns (5) and (6) of Table VI present the results for the full model that includes controls in equation (4). These results confirm the prior findings on the distinct effects of dedicated and transient institutional ownership on both overvaluation and misvaluation in the presence of controls. Specifically, accounting for controls, a 100% increase in the level of dedicated ownership reduces overvaluation by 0.568, and a 100% increase in the change in dedicated ownership reduces overvaluation by 0.362, both significant at the 1% level. Similarly to previous findings, a 100% increase in the level of transient ownership increases overvaluation by 0.150, and a 100% increase in the change of transient ownership increases it by 0.285 in Column (5). Column (6) shows a significant positive effect on the magnitude of misvaluation for both levels and changes of transient institutional investors. This result is consistent with, and in fact, stronger than, that in column (4). A 100% increase in the level of transient ownership increases the magnitude of misvaluation by 0.135, and a 100% increase in the change of transient ownership increases it further by 0.096. We test the robustness of our findings by repeating the analysis in Table VI using an alternative measure of overvaluation and misvaluation based on Hirshleifer and Jiang (2010). Appendix B.1 details the methodology and results. The authors hypothesize that managers who have a private signal that their firm is overvalued (undervalued) will issue (repurchase) equity to take advantage of this inside information. They propose an undervalued-minus-overvalued (UMO) factor as a portfolio that takes long positions in equity repurchasers and shorts equity issuers and finds that this factor is able to identify overvalued and undervalued firms by its explanatory power for the firm’s returns. We compute each firm’s loading on the UMO factor as an alternative measure 14

of overvaluation, and its absolute magnitude as an alternative measure of misvaluation. As this alternative measure is based on managerial behavior, the HJ measure is derived with respect to a completelydistinctsetofinformationthantheRKRVmeasure, whichisbasedonfirmfundamental multiples. Assuch,wecanexpectthetwotoserveasrobustnesschecksforoneanother. Ourresults for the two measures are mutually, and reassuringly, consistent. 3.2 Channels for Ownership Effects on Firm Value Having observed the differing effects that institutional investor types have on firm-specific overvaluation and misvaluation, we next seek to understand and test several explanations by which these differences may arise. 3.2.1 Corporate Events First, one potential reason for the observed differences is due to the occurrence of corporate events in which one type of institutional investor is involved in some capacity, resulting in both increased ownership by that investor type as well as increased (over) valuation. One such corporate event is merger and acquisition activity in which transient institutions may either engage in event-specific trading strategies or in advisory roles. To examine whether mergers and acquisitions drive the relationshipbetweeninvestortypeandmisvaluation,weobtainandincludemergersandacquisition information from the Thomson Reuters SDC M&A database. We repeat our full model analysis in equation (4) using three alternative subsamples: 1) excluding firms that have participated in M&A (as either the acquirer or target) in the current quarter, 2) excluding firms that have participated in M&A in the quarter before, during, and after the current, and 3) excluding firms that have ever participated in M&A in its entire available history. In all cases, our results are highly similar to the main findings in Table VI, suggesting that investor involvement or influence in M&A is not the main driver of investor type on misvaluation. Another such corporate event is equity offerings. It may be the case that transient investors, many of whom are large investment banks, may act in advisory roles, inflating both the valuation of the firm as well as transient ownership. To consider the effect of equity offerings, we obtain and include seasoned equity offering information from the Thomson Reuters SDC Global Issuances 15

database. Similarly to the treatment of M&A, we consider three alternative subsamples: 1) excluding firms that have issued an equity offering in the current quarter, 2) excluding firms that have issued equity in the quarter before, during, and after the current, and 3) excluding firms that have ever participated in a seasoned equity offering in its entire available history. Again, in all cases, our results are highly similar to the main findings. Altogether, these results suggest that corporate events are not the main driver of the observed effect of institutional ownership type on firm value. These results are suppressed for brevity. 3.2.2 Momentum A second potential explanation for the observed differences between different institutional investor types could be that, given the high portfolio turnover and short investment horizon characteristics of transient investors, these investors may be momentum return chasers. A momentum trend could jointly cause an increase in both current transient ownership and future firm-specific RKRV overvaluation and misvaluation.15 We create a Momentum Returns measure calculated as the cumulative return over the prior 12 to seven months following Novy-Marx (2012) to control for this potential confounding effect.16 Novy-Marx (2012) shows that intermediate-term past returns serve better in predicting future returns than recent past performance. Using intermediate-term past returns has the additional advantage of basing momentum on a period prior to our lagged institutional investor level and change measures. Columns (1) and (2) of Table VII include the Momentum Returns variable to control for momentum trends in firm value. The coefficients on both dedicated and transient institutional ownership types, their significances, and the R2’s are highly similar to those observed in our main findings in Table VI. This rules out momentum chasing as the driving explanatory mechanism for the observed overvaluation effect of transient institutional investors. As expected, Momentum Returns, itself, increases firm overvaluation significant at the 1% level as firms experiencing upward 15Momentum may increase firm-specific error due to an exacerbation of true misvaluation or due to a mechanical increase in firm-specific error caused by the slower adjustment of valuation multiples in the RKRV valuation model relative to market value. The fact that our results hold when using the alternative misvaluation measure based on Hirshleifer and Jiang (2010) that has a faster adjustment period provides additional evidence against the second possibility. 16WeadditionallyconstructaMomentumReturnsmeasurebasedonAsness,Moskowitz,andPedersen(2013)that usescumulativemonthlyreturnsoverthepastyear. Furthermore,wealsoconstructmeasuresusingabnormalreturns relative to the Fama-French 3-factor model in place of raw returns in our definitions. All results hold. 16

momentum in the past intermediate-term will on average continue to exceed their firm-specific fundamental value. 3.2.3 Firm Opacity Third, we examine the possibility that the observed differences between dedicated versus transient investors on overvaluation and misvaluation is due to different investor incentives in gathering information. As such, we would expect any information advantages that dedicated investors use in correcting valuation errors to be stronger for more opaque firms, and transient-induced mispricing to be higher. We measure Firm Opacity as the principal component of known opacity proxies following the approach of Karpoff, Lee, and Masulis (2013) to control for information asymmetry. These opacity proxies include log of firm age, log of firm size, and average bid-ask spread, standard deviation of returns, Amihud ratio, and skewness of the Fama-French 3-factor model residuals over the prior year.17 Since this principal component includes the Amihud ratio and bid-ask spread, this effectively controls for firm liquidity as well.18 Columns (3) and (4) of Table VII include the Firm Opacity measure as an additional control for availability of information. This appears to strengthen our results from Columns (5) and (6) of Table VI with higher magnitude coefficients on the levels of and changes in dedicated ownership for firm overvaluation and levels of and changes in transient ownership for firm misvaluation and the R2’s are the highest among all specifications we consider. Firm Opacity itself lowers overvaluation and increases misvaluation. This is consistent with information asymmetries playing a role in the effect of institutional investor types on firm valuation, particularly that of dedicated institutional investors with a superior information advantage. We revisit the role of information gathering on the observed results in more detail in Section 4. 17This principal component has a correlation between 70-98% with larger sets of information asymmetry proxies similar to those used by Karpoff, Lee, and Masulis (2013), including number of analysts, analyst dispersion, and analyst error. We use the more parsimonious set of proxies to maximize the number of usable observations due to the relatively low availability of analyst observations from the I/B/E/S database. 18WerepeattheanalysiswithseparatecontrolvariablesforAmihudratioandbid-askspread,andfindnodifference in results. 17

3.2.4 Perceived Growth Opportunities Finally, it may also be the case that dedicated and transient institutional investors have distinct perceptions regarding the growth and future value of firms, driving the observed differences in results. Specifically, higher perceived growth opportunities may drive both transient ownership and increased firm overvaluation and misvaluation. To address this concern, we rely on the firm’s investmentininnovationasaproxyforperceivedgrowthopportunities. WemeasureR&DIntensity, defined as annual research and development expenses as a ratio to annual sales revenue, to capture the degree to which a firm invests in innovation and growth. A firm with more R&D investment is harder to value correctly given the long-term nature of the investment and is more likely to generate diverging perceptions of growth.19 As such, R&D Intensity proxies for perceived future growth opportunities. Columns (5) and (6) of Table VII include R&D Intensity as a control for perceived growth opportunities. The results for the effect of institutional ownership type on firm-specific overvaluation and misvaluation are unchanged - dedicated (transient) institutional investors leads tosignificantlylower(higher)overvaluationinthenextquarterandtransientinstitutionalinvestors lead to significantly higher misvaluation in the next quarter. 3.3 Sector-wide Misvaluation and Long-run Value In the previous section, we study the relationship between institutional investor types and firmspecific overvaluation and misvaluation, including using a firm-specific proxy for differences in perceived growth opportunities. To further explore whether firm-specific characteristics such as perceived growth opportunities drive the observed differences between investor types and firm valuation, we examine the effect of institutional ownership type on sector-wide misvaluation and long-run market-to-book value from the remaining two components of the RKRV decomposition in equation (1). These components eliminate firm-specific cross-sectional differences, including firmspecific perceived growth opportunities, and allows us to examine institutional investor objectives at the sector level as well as in the long-run. 19Hirshleifer,Hsu,andLi(2013)findthatfirmswithhigherinnovationefficiencyhavehigherfuturepositivereturns, which are related to investor inattention and valuation uncertainty. 18

TheRKRVsectorovervaluationisdefinedbyRhodes-Kropf,Robinson,andViswanathan(2005) as the difference between the log of firm value derived from current sector multiples and the log of firm value derived from long-run sector multiples. The RKRV long-run market-to-book value is the difference between the log of fundamental value derived from long-run multiples and the log of book value. Similar to the firm-specific misvaluation measure, we create a sector-wide misvaluation measure as the absolute value of the sector-wide overvaluation measure. Table VIII presents the results. Columns (1) through (3) present the baseline results following equation (3) for the sector overvaluation and sector misvaluation measures, as well as the firm’s long-run market-to-book value. Column (1) shows that the lagged level of transient ownership increases subsequent sector-wide overvaluation by 0.038 while an increase in transient ownership increases it by 0.053 both significant at the 5% level without controls. Column (2) finds that the lagged level of transient ownership reduces sector-wide misvaluation by 0.032 significant at the 1% level. Theresultssuggestthatwhiletransientownershipincreasesfuturesector-wideovervaluation, the magnitude of the overvaluation (i.e., misvaluation) falls in the future. Column (3) finds that both higher levels of dedicated and transient ownership lead to higher long-run market-to-book value, significant at the 1% level, with a similar effect for an increase in transient ownership. The inclusion of our controls from equation (4) in Columns (4) through (6) of Table VIII absorbs some of the effects. In Column (4), the level of transient ownership increases sector-wide overvaluation by 0.038 and the change in transient ownership increases it by 0.060, both significant at the 5% level. These results are consistent with those observed in Column (1). However, in Column (5) with the inclusion of control variables, transient investors no longer significantly reduce sector misvaluation. This suggests that the previous result in Column (2), where the lagged level of transient investors reduces sector misvaluation, is due to publicly known firm characteristics (as measured by the controls), rather than any specific influence by the transient investors. Finally, when we include controls in Column (6), we find that the level of dedicated ownership decreases long-runmarkettobookby0.107,andthechangeindedicatedownershipdecreaseslong-runmarket to book by0.095. Meanwhile, thelevel of transientownership increases long-runmarket tobookby 0.083 and the change increases it by 0.109. In other words, after controlling for firm characteristics, dedicated institutional investors appear to have a value preference while transient institutional 19

investors do not.20 4 The Role of Information in Institutional Valuation In the preceding section we have identified the differential effects of institutional ownership types on firm value and have tested their robustness against alternative explanations. We found that Firm Opacity strengthens our findings, and now seek to further test the information channel. InOctober2000theSECimplementedaselectivedisclosureandinsidertradingrule, informally known as Regulation FD, which prohibited selective disclosure of information to institutions and required all material disclosures to be public. As a result, this regulation effectively serves as an exogenous shock to the information advantage previously enjoyed by some well-connected institutions, improving market efficiency (Heflin, Subramanyam, Zhang, 2003), reducing flow of private information to analysts (Francis, Nanda, and Wang, 2006) and reducing informed trading in firms with blockholders and analyst access (Anderson, Reeb, Zhang, and Zhao, 2013). We take advantage of this exogenous shock to examine the importance of dedicated and transient ownership for firm overvaluation and misvaluation. Specifically, we hypothesize that by reducing the information advantage to well-connected dedicated institutions, this regulation will result in lower misvaluation for transient institutional owners that do not possess these connections to the same degree that dedicated intuitional owners do.21 4.1 Exogenous Shock to Institutions’ Information Asymmetry The implementation of Regulation FD (RegFD) allows for a difference-in-difference framework. We identify three potential treatment and control groups designated by three indicator variables, isDED, isTRA, and TvD. The first set is comprised of firms with above-median dedicated institutional ownership at the end of 1999, for which the respective control group is all firms with 20Recallthatthelong-runvaluecomponentoftheRKRVmarket-to-bookdecompositionreliesonlong-runindustrywide multiples. While firm-specific and, to a lesser extent, time-series sector misvaluation can be driven by trends in institutional ownership, allowing more for a causal interpretation of the results, it is unlikely that institutional ownership influences long-run industry-wide multiples, making the results for long-run value more consistent with strategicportfolioselection. Wefurthertestthecausalinterpretationforinstitutionalownershiptypeonfirm-specific misvaluation in Section 4. 21Since the Bushee (1998, 2001) definition of dedicated institutions includes concentrated ownership of firms, this makes them likely to be blockholders and therefore more likely to receive private information through firms prior to Regulation FD consistent with Anderson, Reeb, Zhang, and Zhao (2013). 20

below-median dedicated ownership. The second, of those with above-median transient institutional ownership with a corresponding below-median transient ownership control. The third set is drawn from a subsample of firms with only above-median dedicated, or transient, institutional ownership, but not both. The treatment group in this subsample are those with above-median transient ownership and the control group are those with above-median dedicated ownership. We define a post-event indicator, isFD, that takes the value of one from 2001 through 2005 and 0 from 1995 through 1999. We remove 2000 from the sample to avoid any anticipation or adjustments taken by the firms. We also restrict the sample to observations between 1995 and 2005 to get a balanced sample prior to and following the event. We follow a standard difference-in-differences model by including an interaction term between the treatment indicator and the post-event indicator as the difference-in-difference variable in studying our measures of misvaluation, Y : i,t Y = α+β isFD +β isType +β isFD ×isType +β pDED i,t 1 t 2 i 3 t i 4 i,t−2 +β ∆pDED +β pTRA +β ∆pTRA +β lnTA 5 i,t−2,t−1 6 i,t−2 7 i,t−2,t−1 8 i,t−1 (5) +β pINST +β HHI INST +β Zscore +β Lev 9 i,t−1 10 i,t−1 11 i,t−1 12 i,t−1 +β LTCR +β CFdisp +β NumEst +fe +fe +ε 13 i,t−1 14 i,t−1 15 i,t−1 t j i,t where isFD is the indicator variable for the implementation of Regulation FD after 2000Q4 and captures effects that would occur in firms regardless of treatment effect, isType is the indicator variable for whether the firm has an above-median level of institutional ownership by type (i.e., isDED, isTRA, and TvD) as of 1999Q4 and captures possible differences between treatment and control groups in the absence of a policy change, and isFD×isType is the difference-in-difference variable of interest that captures the effect of the policy on the treatment group net of control. One potential complication is changes in treatment/control classification over the estimation period of 1995 through 2005. We control for this by benchmarking the treatment and control classifications to 1999Q4. In other words, a firm is classified based on their institutional ownership at 1999Q4, regardless of how their ownership changes. Additionally, we include the firm’s level and change in dedicated and transient institutional ownership to account for any actual changes in institutional ownership that may occur. The coefficients β through β absorb effects due to a 4 7 21

firm’s potential change in in ownership composition after the 1999 classification. We also control for firm characteristics as discussed in Section 3.1. Finally, we control for macroeconomic effects and seasonality with year and quarter fixed effects respectively and for industry trends with SIC3 fixed effects. Standard errors are double clustered by firm and year-quarter. Panel A of Table IX presents the results of our difference-in-difference model in equation (5) for the RKRV overvaluation and misvaluation measures. Columns (1) and (2) report findings for the effect of Regulation FD on the valuation of firms held by dedicated institutional investors. As we expect, there is no significant effect on the interaction term between the implementation of RegFD and above-median pre-implementation ownership by the dedicated institutional type both for overvaluation and for misvaluation. This implies that the introduction of RegFD did not change the information available to the better-informed dedicated investors relative to the rest of the market. However, when we consider the effects of Regulation FD on the valuation of firms with abovemedianpre-implementationownershipbytransientinstitutionalinvestors,wefindsignificantresults (at the 5% level) for the expected direction of RKRV overvaluation in Column (3). That is, by mandating public disclosure of material information - thereby reducing the information asymmetry between the less-informed transient and more-informed dedicated institutional investors - RegFD increasesthetransientinvestors’informationsetandreducestheirovervaluationaftertheregulation is implemented. There is no significant effect of transient institutional investor ownership on the magnitude of misvaluation reported in Column (4). Columns (5) and (6) of Table IX compare the valuation effects of above-median transient institutional ownership to that of above-median dedicated ownership in 1999Q4, TvD, around the RegFD policy change. The treatment group is comprised of only the isTRA firms and the control group is comprised of only the isDED firms. This smaller subsample reduces our number of observations to 39,899 firm-quarters. However, this approach allows us to more cleanly address the potential information advantage enjoyed by dedicated institutional investors by directly comparing them to transient ones. Column (5) shows the effect of the policy change on overvaluation, and findsthatovervaluationdecreases forTvD, whichis tosayit fallsfortransient-owned firmsrelative to dedicated-owned ones. This effect is significant at the 10% level. Furthermore, Column (6) 22

shows that the magnitude of misvaluation falls much more strongly for transient-owned relative to dedicated-owned firms, significant at the 1% level. These results suggest that following Regulation FD and the easier availability of information, the overvaluation and misvaluation observed previously with transient investors are mitigated. In Panel B of Table IX, we check the robustness of the above difference-in-difference results around the RegFD event with a placebo test. Specifically, we pick a counterfactual year for the exogenous shock event and repeat the difference-in-differences analysis to test whether the effect we observed around 2000, the implementation of RegFD, is spurious. We select 1995 as the counterfactual year and conduct the analysis over three-year pre- and post-event windows to minimize overlap with the timeframe considered in the main analysis in Panel A. The counterfactual analysis in Panel B of Table IX shows only a weakly significant negative effect on the direction of RKRV misvaluation for firms with above-median transient ownership in Column (3). More notably, there is no effect in Columns (5) and (6) on the change in direction and magnitude of RKRV misvaluation on firms held by transient versus dedicated institutions. This absence of an effect in a counterfactual year supports the validity of our analysis in Panel A and suggests that information availability plays a large role in observed misvaluation effects by the different institutional investor types. There are two potential explanations for our results thus far: either institutional investor types causedistinctvalueeffectsonthefirmstheyowninthesubsequentquarter,ortheysimplychooseto ownfirmsthataredistinctlydifferenttobeginwith(i.e.,selectionbias)andthesedifferencespersist. While our analysis is predictive and uses both lagged levels and changes of explanatory variables, we seek to further distinguish between these two potential explanations. To do this, we use the Mahalanobis (1936) distance measure to match each firm prior to the enactment of Regulation FD in 2000 that has above-median dedicated or transient ownership with a contemporaneous control firm that does not, but is similarly misvalued both at the firm and sector levels and has a similar size and book-to-market ratio. We create an alternative set of matches using propensity score matching. WediscussthematchingprocessqualityandresultsinAppendixB.2. Thematched-firm tests aroundRegulation FDin theappendixshowthat theseovervaluation andmisvaluation effects are not merely due to institutional self-selection into firms. The overvaluation and misvaluation 23

of firms owned by transient institutions falls relative to that of matched control firms around the implementation of the regulation in the difference-in-difference analysis, and this decrease is significantly greater than that for firms owned by dedicated institutions in the difference-indifference-in-difference. Overall the findings around the Regulation FD exogenous shock to the informational advantage of dedicated institutional investors point to both future overvaluation and future misvaluation falling for firms owned by transient institutions. This is to be expected as the SEC regulation puts transientinvestorsonamoreevenfootingwiththeirbetter-informeddedicatedcounterparts. These results also support the interpretation of asymmetric information as one channel for the observed differences between firms owned by dedicated and transient investors. Furthermore, matched-firm analysis rules out self-selection by institutions into misvalued firms as the explanation. 4.2 Individual Characteristics of Institutional Investor Types Having established that dedicated and transient institutional investors have different impacts on firm overvaluation and misvaluation, we examine the individual investor characteristics that form the dedicated and transient classifications, specifically investor portfolio turnover and holdings concentration. By considering the individual portfolio characteristics separately, we test whether the information advantage enjoyed by dedicated institutions documented in Section 4 is primarily due to their concentration or duration of ownership. To do this, we classify institutions based on the individual characteristics of portfolio turnover and concentration of ownership. FollowingYanandZhang(2009),wemeasuretheportfolioturnoverofeachinstitutionalinvestor based on the prior four quarters. In each quarter, we calculate the aggregate purchase and sale of each institutional investor based on the changes in portfolio value of purchases and sales between the current and previous quarter: (cid:88) Ag buys = |S P −S P −S ∆P |, whereS > S k,t k,i,t i,t k,i,t−1 i,t−1 k,i,t−1 i,t i,t i,t−1 i∈N k (cid:88) Ag sells = |S P −S P −S ∆P |, whereS ≤ S k,t k,i,t i,t k,i,t−1 i,t−1 k,i,t−1 i,t i,t i,t−1 i∈N k whereS isthenumberofsharesheldbyinvestork infirmiinquartertandP istheshareprice k,i,t i,t 24

of firm i in quarter t. We then take the minimum between the aggregate purchase and aggregate sale in each quarter and divide by the average portfolio value between the current and previous quarter, providing us with a churn rate for each institution in each quarter: min(Ag buys ,Ag sells ) k,t k,t CR = k,t (cid:80) S k,i,t Pi,t+S k,i,t−1 Pi,t−1 i∈N k 2 We take the average churn rate over the past four quarters as our measure for portfolio turnover. Similarly,followingBushee(1998,2001),wemeasureholdingsconcentrationofeachinstitutional investor. We start with the institutional holding of each firm in the portfolio, that is the number of shares held in a firm that institutional investor as a ratio over the shares outstanding for the firm: S k,i,t IOHold = k,i,t S i,t whereS isthenumberofsharesheldbyinvestork infirmiinquartertandS isthenumberof k,i,t i,t shares outstanding for firm i in quarter t. To calculate holdings concentration, we take the average of the squared institutional holding over the institutional investor’s portfolio of firms:22 1 (cid:88) IOConc = IOHold2 . k,t N k,i,t k i∈N k This measure has the interpretation that if the investor tends to be a blockholder in the firms of their portfolio, then on average they take concentrated positions in their holdings. Hypothesis 3. Firms with a higher percentage of long-horizon or focused (short-horizon or diversified) institutional investors experience less (more) overvaluation and misvaluation. Hypothesis 4. Firms that experience an increase in long-horizon or focused (short-horizon or diversified) institutional investors experience less (more) overvaluation and misvaluation. We form terciles based on portfolio turnover in each quarter, and select the top turnover tercile as the “short-horizon” institutional type, and the bottom tercile as the “long-horizon”. Similarly, we form terciles based on holdings concentration in each quarter and take the top tercile as the 22For robustness, we also take the sum of the squared institutional holdings and all results hold. 25

“focused” institutional type, and the bottom as the “diversified”. For each firm in our sample, we then compute the percentage held by each of these institutional investor types. We also identify the joint percentage held by institutions in specific terciles of both portfolio turnover and concentration. That is, we estimate the percentage of the firm owned by long-horizon and focused institutional investors as that held by institutions that are both in the lowest portfolio turnover tercile and the highest concentration tercile. We create analogous ownership percentages for the three other possible combinations of institutional investor types based on portfolio turnover and holdings concentration. Finally, as before, we decompose the lagged level of each single-dimensional institutional ownership type (long/short horizon, focused/diversified portfolio) into a second lag of the level and first lag of the change in the ownership by a given institutional investor type. While correlation between these fundamental types is approximately 40% (see Table IV), the results are not driven by multicollinearity as they are robust to the inclusion of only one type at a time. These results are suppressed for brevity. Table X presents the effects of these fundamental institutional ownership types on RKRV overvaluation and misvaluation in the full sample, as well as the pre- and post- RegFD subsamples. Panel A presents the full-sample results for next quarter’s firm-specific RKRV overvaluation and the absolute magnitude of misvaluation. Column (1) of Panel A shows that overall the lagged level and change in long-horizon institutional ownership surprisingly results in higher subsequent overvaluation significant at the 1% and 10% levels respectively, but notably so does lagged change in short-horizon institutional ownership, also significant at the 1% level. There is no ownership effect on the absolute magnitude of misvaluation in Column (2) for long-horizon ownership but both the lagged level and change in short-horizon institutional ownership increases the magnitude of misvaluation significant at the 5% level in both cases. Columns (3) and (4) of Panel A present analogous full-sample findings for the focused and diversified institutional types. Consistent with expectations, we find that lagged levels and changes in focused institutional ownership decrease overvaluation, significant at the 10% and 1% levels respectively. Also consistent with prior results, lagged levels and changes in diversified ownership increase overvaluation with 1% level of significance for both. However, in Column (4) we find that 26

it is only the diversified institutional types that reduce the magnitude of firm misvaluation, with both the level and change effects significant at the 5% and 10% levels respectively. Columns (5) and (6) consider the joint effects of institutional portfolio turnover and concentration. Notice that long-horizon and focused investors are more likely to be dedicated whereas short-horizon and diversified investors are more likely to be transient investors. We find that the effect of focused institutional ownership reduces subsequent overvaluation in Column (5) regardless of its joint pairing with long- or short-horizon ownership. Similarly, diversified ownership increases subsequent overvaluation regardless of pairing with horizon types. Consistent with Column (4), we find in Column (6) that diversified ownership reduces the magnitude of misvaluation regardless of any joint effects with institutional investment horizon. This result is surprising given that we previously found that TRA investors, those most likely to be short and diversified, increase the magnitude of misvaluation. Given the information advantage for TRA investors with the enactment of Regulation FD, we split our full sample into pre- and post- RegFD to test whether the disclosure requirement had an effect on the fundamental institutional ownership types. Panel B presents results for the pre-RegFD (pre-2000) subsample and Panel C presents results for the post-RegFD (post-2000) subsample. The results along the horizon dimension are similar to that of the full sample (see Columns (1) and (2)). However, when we turn our attention to focused and diversified investors, we see that prior to the enactment of Regulation FD, ownership by diversified investors in Columns (3) and (4) as well as that by short-horizon diversified investors in Columns (5) and (6) increases the magnitude of misvaluation. It is only in the post-RegFD period that this reverses in Panel C. These results suggest that the concentration of institutional ownership is primarily responsible for the information advantage enjoyed by DED investors and reinforce the idea that Regulation FD provided TRA institutions with the information that previously flowed only to DED investors. This importance of blockholder status in obtaining private information prior to the implementation of RegFD is consistent with the findings of Anderson, Reeb, Zhang, and Zhao (2013). Overall these findings present mixed evidence for Hypotheses 3 and 4, due to the regime change in the distribution of information after RegFD. To a large degree, the effects of long-horizon and focused ownership appear to occur prior to the implementation of the SEC regulation. The effects 27

of diversified institutional ownership in reducing misvaluation only appear after its passage. 5 Institutional Investor Types, Governance, and Performance In the preceding sections we consider a puzzle in the literature regarding the effect of institutional investors on firm value. Several papers document the ability of institutional investors to acquire information, improve governance, and make markets more efficient (Lakonishok, Shleifer, Vishny, 1992; Nesbitt, 1994; Sias and Starks, 1997; Nagel, 2005; Barber and Odean, 2008) while other research finds results suggestive of opportunism without long-term benefit, and with potentially destabilizing value effects (Gillian, 1995; Karpoff, Malatesta, and Walkling, 1996; Smith, 1996; Wermers, 1999; Barton and Wiseman, 2014). The differential effect on firm value by institutional investor type that we document provides a view towards reconciling this tension in the literature. Institutional investors are indeed not all the same: some reduce overvaluation and misvaluation, while others stoke it. Examining different institutional investor types can also help refine prior findings regarding the role of institutions and blockholders on corporate governance and long-term firm performance. Some studies have shown that institutional investors improve governance (Gillan and Starks, 2000; Hartzell and Starks, 2003), while others find no effect (Agarwal and Knoeber, 1996; Karpoff, Malatesta, and Walkling, 1996). Recent work in asset pricing research also demonstrates that holding period matters for performance studies (Kamara, Korajcyk, Lou, Sadka, 2016; Chaudhuri and Lo, 2016). In this section, we examine some common firm governance and performance measures to gain additional insight on the roles that different institutional investor types play. Table XI Panel A reports findings about the effect of institutional investor type on key characteristics related to firm performance and risk as well as corporate governance. We find that firms held by dedicated institutions have lower realized volatility and lower average, median, and maximum executive compensation. Firms held by transient institutions have higher tail risk, as proxied by the IVspread implied volatility spread between out-of-the-money and in-themon money puts, higher realized volatility, higher executive pay, lower payout ratios, and lower leverage increases than those in firms held by dedicated institutions.23 23Borochin and Yang (2016) find that change in the implied volatility spread across option moneyness predicts 28

While the information advantage enjoyed by dedicated institutional investors prior to the implementationofRegFDclearlyplaysaroleintheireffectsonfirmovervaluationandmisvaluation, it is not obvious that the firm characteristics considered in Table XI should be impacted due to having private information. We test this by subsampling into pre- and post- RegFD subsamples in Panels B and C respectively, and find that the effects of dedicated and transient ownership on the firm characteristics related to good governance are largely consistent through time. The only notable exception to this is the payout ratio whose negative coefficients become similar in magnitude post-2000. Overall, these results suggest that some, but not all, institutional investors provide governance benefits, consistent with Chen, Harford, and Li (2007). In Table XII, we again deompose investors along investor horizon and portfolio concentration and consider the joint pairings between the two portfolio characteristics. We find that long-horizon ownership levels and changes are associated with lower historical volatility, higher payout ratios, and net leverage increases, largely invariant to the dimension of the concentration of ownership. At the same time, short-horizon ownership is associated with higher tail risk, higher executive compensation, and lower payout ratios, also largely invariant to portfolio concentration. Some, but not all, institutional investors do provide governance benefits consistent with Chen, Harford and Li (2007). Therefore, while the effects of dedicated and transient ownership on firm overvaluation and misvaluation are shown to be largely due to the holdings concentration dimension in Table X, the corporate governance effects of these institutional investor types appear largely due to the portfolio turnover dimension as shown in Table XII. The changes in over- and misvaluation effects found for dedicated and transient ownership around the implementation of RegFD, coupled with their strong relation to the holdings concentration dimension, suggest that the concentration dimension in particular was sensitive to the information effect from RegFD. The largely time-invariant firm characteristics in Table XI, coupled with their strong relation to the portfolio turnover dimension, suggest that institutional variation on investment horizon was not as big a driver of informational asymmetries addressed by RegFD. Next, we test the economic significance of institutional ownership types by considering the raw leverage increase and is a significant signal of firm quality and is related to other quality firm characteristics. 29

and risk-adjusted performance of firms held by each institutional type over the following year. This analysiscapturesthelong-termeffectsofinstitutionalownershiptypeonthefirmandisparticularly relevant to other market participants investing alongside these institutions. In calculating riskadjusted returns, we benchmark to the P´astor and Stambaugh (2003) 5-factor model. Table XIII reports the raw performance over each of the next four quarters in Columns (1) through (4) and the 5-factor abnormal performance in Columns (5) through (8). We find that firms held by transient institutions have insignificant or negative raw returns over the next four quarters. Conversely, firms held by dedicated institutions have positive raw returns over the same period. The results become more compelling when we consider abnormal, rather than raw, performance. Transient institutional ownership results in highly significant and positive abnormal returns in the first quarter, consistent with the previously documented myopic objectives of transient institutions (Bushee, 1998, 2001). However, these firms have no abnormal performance in any subsequent period considered. On the other hand, firms held by dedicated institutions have significant abnormal performance in the second and third quarters, consistent with their long-term value objectives. Market participants may therefore be able to observe the type of institutional ownership and use it as a guide for the horizon of expected future performance. Finally, we follow the monthly raw and risk-adjusted returns of firms over the next year in Figures2and3. Figure2tracksthemonthlyreturnsandabnormalreturnsforfirmswithdedicated and transient institutional investors in Panels A and B, respectively. Consistent with the analysis in Table XIII, firms with more dedicated institutional investors realize raw returns of 17.7% over the following year, whereas firms with more transient institutional investors realize raw returns of 14.7%. That is, firms with more dedicated investors achieve 3.0% higher returns than those with more transient investors. However, when we turn to risk-adjusted or abnormal returns, firms with more dedicated (transient) institutional investors earn returns of 4.1% (3.8%), resulting in a difference of 0.3%. Figure 3 repeats the previous figure by separating institutional investors based on portfolio turnover and holdings concentration. Panels A and B of Figure 3 present the raw and riskadjusted returns for firms with more long-horizon or short-horizon institutional investors. Panels C and D present the raw and risk-adjusted returns for firms with more focused or diversified 30

institutional investors. We see that firms with more long-horizon institutional investors realize raw (abnormal) returns of 19.3% (5.7%) over the following year, whereas firms with more short-horizon institutional investors realize raw (abnormal) returns of 13.3% (2.9%) over the next year. This leaves firms with more long-horizon institutional investors with raw (abnormal) returns that are 6.0% (2.8%) higher than firms with more short-horizon investors. In contrast, firms with more focused institutional investors realize raw (abnormal) returns of 15.2% (2.7%) while firms with more diversified institutional investors realize raw (abnormal) returns of 16.1% (4.9%) over the next year. This results in firms with more focused institutional investors earning raw (abnormal) returns0.9%(2.1%)lowerthanfirmswithmorediversifiedinstitutionalinvestors. Thenetnegative effect of focused investors is offset by the net positive effect of long-horizon investors, resulting in the 0.3% abnormal returns observed for dedicated investors in Figure 2. These figures support the earlier finding that the higher firm performance observed with dedicated institutional investors, which are defined by Bushee (1998) as focused, long-term investors, is driven by lower portfolio turnover rather than by higher holdings concentration. Furthermore, this superior long-horizon performance more broadly suggests that long-horizon investors are able to accomplish their investment objectives. In addition to providing insight about the causes of dedicated investor performance, these results have wider implications regarding the existence of portfolio manager ability. While we are able to demonstrate the ability of long-horizon managers to achieve long-horizon returns, our data may not be captured at a sufficiently high enough frequency to address the ability of short-horizon managers to achieve short-horizon returns. At any rate, we do not find any superior short-term performance by firms they hold in our sample. 6 Conclusion Institutional ownership can have starkly different effects on firm valuation, depending on the type and objectives of the institutional investor. Dedicated institutional investor ownership reduces future misvaluation, while transient institutional investor ownership increases it. Using the Bushee (1998, 2001) classification of institutional investor types, we show these effects on the direction and magnitude of the firm-specific valuation error relative to fundamentals from Rhodes-Kropf, Robinson, and Viswanathan (2005). Specifically, dedicated institutional ownership is correlated 31

with lower future firm-specific misvaluation and long-run valuation multiples, whereas transient institutional ownership has the opposite relationship. We find similar effects for the absolute magnitude of misvaluation, with dedicated institutional ownership reducing the magnitude of firmspecific misvaluation and transient institutional ownership increasing it. Furthermore, decomposing the level of institution ownership by type into the prior period’s level and the change with respect to the current period, we find that changes in ownership type have the same effect as the level. This suggests that the effects of institutional ownership types become stronger as the degree of institutional ownership of that particular type becomes more pronounced. These results persist and become stronger in the presence of measures of information quality, suggesting a connection between ownership type and the ability to access firm-specific information. These results are robust to the inclusion of control variables, the use of alternative misvaluation measure, and the examination of alternative explanations. We examine this relationship between firm misvaluation and institutional investor type further usingtheexogenousshockoftheSEC’sintroductionofRegulationFairDisclosure(RegFD)in2000, which requires firms to disclose material information publicly. Prior to this regulation, firms could reveal information to preferred institutional owners, contributing to the informational advantage enjoyedbythem. Wefindevidenceconsistentwithaninformationaladvantageenjoyedbydedicated institutional owners prior to the introduction of RegFD in 2000. The policy change reduces both future overvaluation and misvaluation experienced by firms with transient institutional owners. Wealsoconsidermorefundamentalinstitutionalinvestortypesalongthedimensionsofportfolio turnover and portfolio concentration. We find that ownership by focused institutions is largely related to lower overvaluation and misvaluation, with the opposite being true for ownership by diversified institutions. That is, of the two portfolio characteristics underlying the transient and dedicated investor types, portfolio concentration drives more of the observed relationship between type and future overvaluation and misvaluation. Finally, we demonstrate that institutional ownership types have effects on key firm characteristics that are related to corporate governance. Dedicated (transient) institutions hold firms with less (more) tail risk, historical volatility, and mean, median, and maximum executive compensation. Additionally, dedicated institutions hold firms with less earnings management, 32

higher payouts, and higher leverage increases. We find that these characteristics are stable over time and not sensitive to the information effects from RegFD. Furthermore, we find that they are largely due to the portfolio turnover dimension than the portfolio concentration dimension. These differential institutional ownership effects also have real performance implications for investors. Over the subsequent year, firms held by transient institutions experience abnormal positive performance only in the first quarter, whereas those held by dedicated institutions have positive abnormal and raw returns in later quarters. This difference in performance appears to be also largely due to the portfolio turnover dimension, as firms held by long-horizon institutions outperform those held by short-horizon ones over subsequent year. This is consistent with the objectives of transient (short-term) and dedicated (long-term) institutions, but has profound implications on the expected effects of institutional ownership for other market participants. This also demonstrates that long-horizon institutional investors are able to achieve superior long-term performance, adding to our understanding of investment manager ability. These results provide a new perspective on the multifaceted nature of institutional investors. We find that there are distinct roles that different types of institutional investors play in setting firm valuations. Institutional types also have distinct effects on corporate governance and future firm performance. Separating institutional ownership by type helps towards resolving standing arguments in the literature about the role of institutional investors in the financial markets. 33

References Agrawal, A., and C. Knoeber, 1996, Firm Performance and Mechanisms to Control Agency Problems between Managers and Shareholders. Journal of Financial and Quantitative Analysis, 31, 377-397. Almeida, H., M. Campello, B. Laranjeira, and S. Weisbenner, 2011, Corporate Debt Maturity and the Real Effects of the 2007 Credit Crisis, Critical Finance Review, 1, 3-58. Altman, E., 1968, Financial Ratios, Discriminant Analysis, and the Prediction of Corporate Bankruptcy, Journal of Finance, 23, 589-609. Anderson, R., D. Reeb, Y. Zhang, and W. Zhao, 2013, The Efficacy of Regulatory Intervention: EvidencefromtheDistributionofInformedOptionTrading,Journal of Banking and Finance, 37, 4337-4352. Asness,C.S.,T.J.Moskowitz,andL.H.Pedersen,2013,ValueandMomentumEverywhere,Journal of Finance, 68, 929-985. Barber, B., andT.Odean, 2008, AllthatGlitters: TheEffectofAttentionandNewsontheBuying Behavior of Individual and Institutional Investors, Review of Financial Studies, 21, 785-818. Barton, D., and M. Wiseman, 2014, Short-termism: Insights from business leaders, Harvard Business Review, 92, 44-77. Blume,M.E.andD.B.Keim,2014,TheChangingNatureofInstitutionalInvestors,CriticalFinance Review, 6, forthcoming. Boone, A.L. and J.T. White, 2015, The Effect of Institutional Ownership on Firm Transparency and Information Production, Journal of Financial Economics, 117, 508-533. Borochin, P., and J. Yang, 2016, Options, Equity Risks, and the Value of Capital Structure Adjustments, Journal of Corporate Finance, forthcoming. Brav, A., W. Jiang, F. Partnoy, and R. Thomas, 2008, Hedge Fund Activism, Corporate Governance, and Firm Performance, Journal of Finance, 63, 1729-1775. Bushee, B., 1998, The Influence of Institutional Investors on Myopic R&D Investment Behavior, The Accounting Review, 73, 305-333. Bushee, B., 2001, Do Institutional Investors Prefer Near-Term Earnings over Long-run Value?, Contemporary Accounting Research, 18, 207-246. Bushee, B. and C.F. Noe, 2000, Corporate Disclosure Practices, Institutional Investors, and Stock Return Volatility, Journal of Accounting Research, 38, 171-202. Carhart, M.M., 1997, On Persistence in Mutual Fund Performance, Journal of Finance, 52, 57-82. 34

Chaudhuri, S., and A. Lo, 2016, Spectral Portfolio Theory, working paper. Chen, X., J. Harford, and K. Li, 2007, Monitoring: Which Institutions Matter?, Journal of Financial Economics, 86, 279-305. Francis, J., D. Nanda, and X. Wang, 2006, Re-examining the Effects of Regulation Fair Disclosure using Foreign Listed Firms to Control for Concurrent Shocks, Journal of Accounting and Economics, 41, 271-292. Gaspar, J., M. Massa, and P. Matos, 2005, Shareholder Investment Horizons and the Market for Corporate Control, Journal of Financial Economics, 76, 135-165. Gillian, S.L., 1995, Shareholder Activism Through the Proxy Mechanism: An Empirical Investigation, working paper. Gillian, S.L. and L.T. Starks, 2000, Corporate Governance Proposals and Shareholder Activism: The Role of Institutional Investors, Journal of Financial Economics, 57, 275-305. Gompers, Paul A., and A. Metrick, 2001, Institutional Investors and Equity Prices, Quarterly Journal of Economics, 116, 229-259. Hadlock, C.J., and J.R. Pierce, 2010, New Evidence on Measuring Financial Constraints: Moving Beyond the KZ Index, Review of Financial Studies, 23, 1909-1940. Hartzell, J., and L. Starks, 2003, Institutional Investors and Executive Compensation. Journal of Finance, 58, 2351-2374. Heflin, F., K.R. Subramanyam, and Y. Zhang, 2003, Regulation FD and the financial information environment: Early evidence, The Accounting Review, 78:1, 1-37. Hirshleifer, D., P. Hsu, and D. Li, 2013, Innovative Efficiency and Stock Returns, Journal of Financial Economics, 107, 632-654. Hirshleifer, D. and D. Jiang, 2010, A Financing-Based Misvaluation Factor and the Cross-Section of Expected Returns, Review of Financial Studies, 23, 3401-3436. Kamara, A., R. Korajczyk, X. Lou, and R. Sadka, 2016, Horizon Pricing, Journal of Financial and Quantitative Analysis, forthcoming. Karpoff, J., G. Lee, and R. Masulis, 2013, Contracting under asymmetric information: Evidence from lockup agreements in seasoned equity offerings. Journal of Financial Economics, 110, 607-626. Karpoff, J., P. Malatesta, and R. Walkling, 1996, Corporate Governance and Shareholder Initiatives: Empirical Evidence, Journal of Financial Economics, 42, 365-395. 35

Lakonishok, J., A. Shleifer, and R. Vishny, 1992, The Impact of Institutional Trading on Stock Prices. Journal of Financial Economics, 32, 23-43. Mahalanobis,P.,1936,Onthegeneraliseddistanceinstatistics,ProceedingsoftheNationalInstitute of Sciences of India 2:1, 49-55. Nagel,S.,2005,ShortSales,InstitutionalInvestorsandtheCross-SectionofStockReturns,Journal of Financial Economics, 78, 277-309. Nesbitt, S., 1994, LongTermRewardsfromShareholderActivism: AStudyoftheCalPERSEffect, Journal of Applied Corporate Finance, 6, 75-80. Novy-Marx, R., 2012, Is Momentum Really Momentum? Journal of Financial Economics, 103, 429-453. P´astor, L. and R.F. Stambaugh, 2003, Liquidity Risk and Expected Stock Returns, Journal of Political Economy, 111, 642-685. Rhodes-Kropf, M., D.T. Robinson, and S. Viswanathan, 2005, Valuation Waves and Merger Activity: The Empirical Evidence, Journal of Financial Economics, 77, 561-603. Sias, R.W., and L.T. Starks, 1997, Return Autocorrelation and Institutional Investors, Journal of Financial Economics, 46, 103-131. Smith, D.G., 1996, Corporate Governance and Managerial Incompetence: Lessons from Kmart, North Carolina Law Review, 74, 1038-1139. Wahal, S., 1996, Pension Fund Activism and Firm Performance, Journal of Financial and Quantitative Analysis, 31, 1-23. Wermers, R., 1999, Mutual Fund Herding and the Impact on Stock Prices. Journal of Finance, 54, 581-622. Williamson, R. and J. Yang, 2016, Financing Constraints and Acquisitions, working paper. Yan, X. and Z. Zhang, 2009, Institutional Investors and Equity Returns: Are Short-term Institutions Better Informed, Review of Financial Studies, 22, 893-924. Compustat, Compustat Capital IQ/North America/Fundamentals Quarterly, Wharton Research Data Services (WRDS), wrds-web.wharton.upenn.edu/wrds/about/databaselist.cfm CRSP, CRSP/CRSP Monthly Stock, Wharton Research Data Services (WRDS), wrdsweb.wharton.upenn.edu/wrds/about/databaselist.cfm Thomson Reuters, Thomson Reuters Institutional (13f) Holdings/s34 Master File, Wharton Research Data Services (WRDS), wrds-web.wharton.upenn.edu/wrds/about/databaselist.cfm 36

ThomsonReuters,IBES/Unadjusted Summary History,WhartonResearchDataServices(WRDS), wrds-web.wharton.upenn.edu/wrds/about/databaselist.cfm 37

Appendix A This appendix details the construction of our control variables below. Summary statistics of these variables are reported in Table III. Calculations based on: Compustat Capital IQ/North America/Fundamentals Quarterly, Thomson Reuters Institutional (13f) Holdings/s34 Master File, and IBES/Unadjusted Summary History, Wharton Research Data Services (WRDS), wrdsweb.wharton.upenn.edu/wrds/about/databaselist.cfm Total Assets Assets - Total (ATQ) * Adjustment to 2000 Dollars Ln Total Assets (LnTA) ln{Assets - Total (ATQ) * Adjustment to 2000 Dollars} Total Market Capitalization Price-Close-Quarter (PRCCQ) * Common Shares Outstanding (CSHOQ) * Adjustment to 2000 Dollars Market Value Total Market Capitalization + Total Book Debt % Owned by Institutional Investors Total number of shares held by institutional investors / Shares outstanding % Owned by Dedicated Institutional Investors Number of shares held by institutional investors classified as “Dedicated” based on Bushee (2001) / Total number of shares held by institutional investors % Owned by Transient Institutional Investors Number of shares held by institutional investors classified as “Transient” based on Bushee (2001) / Total number of shares held by institutional investors % Owned by Long (Short) Horizon Institutional Investors Total # shares held by long (short) horizon institutional investors / Shares outstanding where long (short) horizon is calculated based portfolio turnover, as defined in Yan and Zhang (2009) % Owned by Focused (Diversified) Institutional Investors Total # shares held by focused (diversified) institutional investors / Shares outstanding where focused (diversified) is based on the HHI of portfolio weights for each institutional investor HHI-Index of Institutional Investors (cid:16) (cid:17)2 (cid:80) Numberofsharesheldbyinvestori i Totalnumberofsharesheldbyinstitutionalinvestors 38

RKRV (Firm-specific) Overvaluation Firm-specific error from the Rhodes-Kropf Robinson, and Viswanathan (2005) decomposition of the log of the market-to-book ratio into three components: firm-specific error, time-series sector error, and long-run market-to-book. A positive (negative) firm-specific error indicates an over- (under-) valued firm. RKRV (Firm-specific) Misvaluation Absolute value of RKRV (Firm-specific) Misvaluation. A larger absolute firm-specific error indicates a more misvalued firm. RKRV Sector Overvaluation Sector-wideerrorfromtheRhodes-KropfRobinson,andViswanathan(2005)decompositionofthe log of the market-to-book ratio into three components: firm-specific error, time-series sector error, andlong-runmarket-to-book. Apositive(negative)sectorerrorindicatesanover-(under-)valued sector. RKRV Sector Misvaluation Absolute value of RKRV Sector Misvaluation. A larger absolute sector error indicates a more misvalued sector. RKRV Long-run MtB Long-run market-to-book from the Rhodes-Kropf Robinson, and Viswanathan (2005) decomposition of the log of the market-to-book ratio into three components: firm-specific error, time-series sector error, and long-run market-to-book. Altman’s Zscore 3.3*PretaxIncome(PIQ)+1.0*NetSales(SALEQ)+1.4*RetainedEarnings(REQ)+1.2*WorkingCapital TotalAssets(ATQ) where Working Capital = Current Assets-Total (ACTQ) - Current Liabilities-Total (LCTQ) Long-term Debt / TA Long-Term Debt-Total (DLTTQ) / Total Assets (ATQ) Has Long-term Debt Credit Rating 1, if firm has a S&P Long-term credit rating, and 0, otherwise Cash Flow Dispersion Standard Deviation (Cash flow (OIBDPQ) over the past 20 quarters) / Average(Cash flow (OIBDPQ) over the past 20 quarters) Number of Analyst Estimates Number of analyst EPS estimates for the relevant quarter based on IBES data Firm Opacity First principal component of log firm age, log firm size, bid-ask spread, standard deviation of returns, Amihud ratio, and skewness of 3-factor residuals following Karpoff, Lee, and Masulis (2013) to create a single proxy variable increasing in firm opacity Momentum Return Cumulative monthly returns between R[t-7] to R[t-12], as based on Novy-Marx (2012) 39

R&D Intensity Annual R&D Expense (XRD) / Annual Sales (SALE) Firm Concentration (cid:16) (cid:17)2 Numberofsharesinvestedbyinstitutionalinvestor Sharesoutstanding Following Bushee (1998), used as a measure for investor concentration within a firm. Portfolio Turnover (cid:16) (cid:17) min (cid:80) ∆ ∆ po p r o tf r o t l f i o o lio va v lu al e u f e ro fr m om bu b y u in yi g n i g , (cid:80) ∆ ∆ po p r o tf r o t l f i o o lio va v lu al e u f e ro fr m om se s ll e i l n li g n i g i i i i divided by average portfolio value between t−1 and t Following Yan and Zhang (2009), used as a measure for portfolio turnover. IVspread mon Spreadbetweenquarterlyaverageimpliedvolatilityfromshort-termout-of-the-moneyputoptions and in-the-money put options. Following Borochin and Yang (2016), used as a measure of cash flow tail risk. 40

Appendix B This appendix contains additional robustness tests of our main findings. In particular, Section B.1 presents findings using an alternative measure of misvaluation based managerial issuance and repurchase behavior following Hirshleifer and Jiang (2010). Section B.2 presents matched-firm results by misvaluation to rule out self-selection of institutional investor types into misvalued firms. Appendix B.1 Alternative Measure for Misvaluation We test the robustness of our findings by repeating the analysis in Table VI using an alternative measureofovervaluationandmisvaluation. HirshleiferandJiang(2010)hypothesizethatmanagers who have insider information that their firm is overvalued will issue equity to take advantage of this. Conversely, managers will repurchase equity if they have private information about firm undervaluation. Theauthorsproposeanundervalued-minus-overvalued(UMO)factorasaportfolio that takes long positions in equity repurchasers and shorts equity issuers and finds that this factor is able to identify overvalued and undervalued firms by its explanatory power for the firm’s returns. We take the negative of the beta on the UMO factor - that is the negative of the coefficient of the firm’s exposure to the UMO factor - as our measure of HJ misvaluation to be consistent with the direction of the RKRV misvaluation measure (i.e., in the direction of overvaluation). This HJ measure is based on managerial repurchase and issuance behavior, rather than on fundamental multiplesaswiththeRKRVmeasure. SincetheHJmeasureisderivedwithrespecttoacompletely different set of information than the RKRV measure, we can expect the two to serve as robustness checks for one another. To estimate the HJ measure for misvaluation, we estimate the firm’s exposure to Hirshleifer and Jiang’s (2010) UMO factor using rolling 60-month regressions: r = α +β MKTRF +β SMB +β HML +β UMD +β UMO +ε . i,t i 1,i,t t 2,i,t t 3,i,t t 4,i,t t 5,i,t t i,t A positive coefficient on UMO proxies for undervaluation. However, for consistency with the direction of the RKRV measure, we define the HJ measure of misvaluation as −βˆ such that it 5 remains positive (negative) in the direction of overvaluation (undervaluation). As with the RKRV 41

measure, we take the absolute value of the HJ measure to proxy for the magnitude of misvaluation. Table B.I displays the results of estimating equations (2), (3), and (4) for the HJ measure of overvaluation and its absolute value proxying for misvaluation. Column (1) finds results consistent with Column (1) of Table VI, with the lagged level of transient institutional ownership resulting in a positive and significant HJ coefficient, consistent with more overvaluation. Column (2) reports similar results for the absolute value of the factor loading as a proxy for the magnitude of misvaluation. A 100% increase in the level of transient ownership increases the loading on the overvaluation factor by 0.228 and the magnitude of misvaluation by 0.460. This is consistent with prior findings for the RKRV measure for transient institutional ownership. There are no statistically significant effects for dedicated institutional ownership, implying that dedicated institutional investor ownership does not significantly increase overvaluation or misvaluation in the subsequent quarter. Columns (3) and (4) of Table B.I repeats the decomposition of the lagged level of ownership into the second lag of the level and the first lag of the change in the level of ownership to estimate the dynamic effect of institutional ownership on overvaluation and misvaluation. The positive coefficient on the overvaluation factor loading is consistent with prior findings in Column (3) of Table VI, though only the level of transient ownership has a significant effect. However, both the level and change in transient ownership have the expected positive relationship with the magnitude of misvaluation in Column (4), consistent with the results in Column (4) of Table VI. Finally, columns (5) and (6) of Table B.I use the HJ valuation measures in the estimation of the full model with controls, as in equation (4). Column (5) of Table B.I finds the level of transient institutional ownership leading to a subsequent increase in overvaluation significant at the 5% level, consistent with prior results. Column (6) finds stronger results that both the level and change of transient institutional ownership increases the magnitude of misvaluation while the level of dedicated ownership reduces it in the subsequent quarter, all significant at the 1% level. We therefore establish a robust relationship between institutional investor ownership by type, both in levels and changes of ownership, and both overvaluation and misvaluation at the firm level. 42

refielhsriHehtnodesabdenfiederanoitaulavsimdnanoitaulavrevO .selbairavlortnochtiwsrotsevnilanoitutitsnifosepytnonoitaulavsimfonoitamitsE:I.BelbaT A .rotcaf OMU )0102( gnaiJ dna refielhsriH eht htiw gnola rotcaf-4 )7991( trahraC a no desserger era mrfi hcae rof snruteR .rotcaf OMU )0102( gnaiJ dna hcusatebOMUeht foevitagen ehtsekat noitaulavrevo cfiiceps-mrfiJHehT .mrfi deulav)-revo(-rednunasetacidni rotcafOMUehtno ateb)evitagen(evitisop .noitaulavsim eht fo edutingam eht serutpac noitaulavrevo eht fo eulav etulosba ehT .mrfi deulav )-rednu( -revo na setacidni erusaem )evitagen( evitisop a taht detartnecnocgnivahsadeziretcarahcera,)1002,8991(eehsuBnidenfiedsa,srotsevnilanoitutitsnidetacideD .1.2noitceSnidenfiederasepytrotsevnilanoitutitsnI smrfiehtnisgnidlohdefiisrevidgnivahsadeziretcarahcerasrotsevnilanoitutitsnitneisnart,tsartnocnI .revonrutwolhtiwsoiloftropriehtnismrfiehtnisgnidloh nihtiw srotsevni lanoitutitsni fo rebmun latot eht ot evitaler si srotsevni lanoitutitsni tneisnart ro detacided fo egatnecreP .revonrut hgih htiw soiloftrop rieht ni eht ta ecnacfiingiS .retrauq-raey dna mrfi htob yb deretsulc dna sesehtnerap eht ni detroper era srorre dradnatS .A xidneppA ni denfied era slortnoc llA .mrfi a .*** yb level %1 dna ,** yb level %5 ,* yb detacidni si level %01 JH JH JH JH JH JH .lavsiM .lavrevO .lavsiM .lavrevO .lavsiM .lavrevO )6( )5( )4( )3( )2( )1( 3901.0 4460.0 1−t,isrotsevnI .tsnIdetacideDybdenwO% )8390.0( )4251.0( ***9733.0- 8381.0 9701.0 5370.0 2−t,isrotsevnI .tsnIdetacideDybdenwO% )9801.0( )2881.0( )6501.0( )2271.0( 8441.0- 6090.0 6821.0 7990.0 1−t,isrotsevnI .tsnIdetacideDybdenwO%∆ )3390.0( )0621.0( )7511.0( )2621.0( *** 5954.0 *** 8722.0 1−t,isrotsevnI .tsnItneisnarTybdenwO% )2550.0( )6280.0( *** 4366.0 ** 8172.0 *** 7835.0 *** 9703.0 2−t,isrotsevnI .tsnItneisnarTybdenwO% )8950.0( )6501.0( )3560.0( )7590.0( *** 3133.0 3670.0 *** 2472.0 3990.0 1−t,isrotsevnI .tsnItneisnarTybdenwO%∆ )9640.0( )2380.0( )7250.0( )7570.0( ***6780.0- 0400.0 1−t,istessAlatoTgoL )4600.0( )1010.0( ***2902.0- 1860.0- 1−t,isrotsevnIlanoitutitsnIybdenwO% )5230.0( )1060.0( * 7480.0 **1371.0- 1−t,isrotsevnIlanoitutitsnIfoxednI-IHH )3050.0( )8670.0( ***5610.0- ***5700.0- 1−t,ierocs-ZnamtlA )7100.0( )6200.0( *** 7532.0 **8802.0- 1−t,iAT/tbeDTL )1740.0( )8180.0( 4400.0- 4200.0 1−t,ignitaRtiderCTLevaH )7810.0( )7030.0( *** 3710.0 *6110.0- 1−t,inoisrepsiDwolFhsaC )9300.0( )7600.0( *** 2620.0 *** 6740.0 1−t,isetamitsEtsylanAfo .oN )2010.0( )8610.0( *** 1122.1 4743.0 *** 0337.0 4981.0 *** 6067.0 0032.0 tnatsnoC )8201.0( )7032.0( )1901.0( )3312.0( )0001.0( )8391.0( Y Y Y Y Y Y ?stceffEdexiFretrauQ Y Y Y Y Y Y ?stceffEdexiFraeY Y Y Y Y Y Y ?stceffEdexiFyrtsudnI 070631 070631 991261 991261 623761 623761 .sbO .oN 1371.0 3750.0 6501.0 4840.0 5401.0 6740.0 2RdetsujdA 43

Appendix B.2 Matched-firm Analysis Inthissection, weseektodisentangletwopotentialexplanationsforourresults: eitherinstitutional investor types cause distinct value effects on the firms they own in the subsequent quarter, or they simply choose to own firms that are distinctly different to begin with (i.e., selection bias) and these differences persist. While our analysis is predictive and uses both lagged levels and changes of explanatory variables, we seek to further distinguish between these two potential explanations. To do this, we use a matching algorithm to find a control firm with a similar prior overvaluation at both the firm and sector levels, but different institutional ownership type for each (treatment) firm held by a particular institutional ownership type around the Regulation FD shock. This is similar to the approach taken by Almeida, Campello, Laranjeira and Wesibenner (2011) and Williamson and Yang (2016). Specifically, we use the Mahalanobis (1936) distance measure to match each firm prior to the enactment of Regulation FD in 2000 that has above-median dedicated ortransientownershipwithacontemporaneouscontrolfirmthatdoesnot,butissimilarlymisvalued both at the firm and sector levels and has similar size and book-to-market ratio. The advantage of using the Mahalanobis distance matching algorithm is that this algorithm ensures a match on each of the matching characteristics, rather than relying on a propensity score.24 As summarized in Table B.II, the match quality is good: there are no significant differences between the sample and control firms for above-median dedicated (isDED), above-median transient (isTRA), or above-median transient and below-median dedicated institutional ownership (isTvD) across the characteristics being matched. We examine differences in next-period misvaluation by ownership type (isDED, isTRA, TvD) before and after the implementation of Regulation FD. Similarly, we also test this difference for Mahalanobis-matched control firms that are similarly overvalued in the current period, but do not haveabove-medianownershipbythesameinstitutionalownertype. Next, wetestthedifference-indifference between the treatment and control firms. This difference-in-difference provides us with a cleanermeasureofwhetherthereisasignificantmisvaluationchangearoundRegFDimplementation due to each ownership type. Furthermore, our framework allows us to test the difference in 24Forrobustness,wealsocreatepropensityscoresforeachoftheseownershiptypesandidentifycontrolfirmsusing propensityscorematching. This,however,doesnotnecessarilyenforcetherequirementthatex-antefirmmisvaluation be similar across the sample and control firms, only that the propensity scores are similar. Nevertheless, the results are similar to the Mahalanobis approach and are therefore suppressed for brevity. 44

the difference-in-difference for isDED against the difference-in-difference for isTRA, essentially providing us with a difference-in-difference-in-difference. This is the difference between firms with above-median dedicated ownership versus above-median transient ownership, taking into account both the time effect (i.e., pre- and post- RegFD) as well as the selection bias (i.e., relative to their matched control firms). This framework is thus robust against the explanation of institutional self-selection for our findings of the effect of ownership type on next quarter’s overvaluation and misvaluation. Table B.III summarizes the results of this difference analysis of next quarter’s firm-specific overvaluation and misvaluation relative to control firms chosen using Mahalanobis distance matching based on contemporaneous RKRV firm- and sector-specific overvaluation, size, and book to market ratio. Panel A presents the results for next quarter’s RKRV firm-specific overvaluation measure. We observe that the firms with high dedicated ownership (isDED=1), in Row (a), have no significant overvaluation one year prior to the implementation of RegFD in Column (1), and substantially higher overvaluation one year post in Column (2). The difference between the two, reported at 0.208 in Column (3), is also statistically significant indicative of an increase in overvaluation after the implementation of RegFD. This result may be due either to an increase in market efficiency as was the intention of RegFD,25 or to the overall increase in overvaluation in the market. The matched control firms do demonstrate a similar increase in overvaluation between pre- and post- RegFD periods of 0.180 in Column (6). Notably, there are no significant differencein-differences in the changes in overvaluation of firms held by dedicated investors relative to the controls around RegFD in Column (7), suggesting the absence of an informational effect due to Regulation FD on dedicated institutional investors. We next consider the future overvaluation of firms with high transient ownership (isTRA = 1), in Row (b), which are significantly overvalued both pre- and post- RegFD in Columns (1) and (2) respectively. The difference of 0.083 is significant at the 5% level in Column (3). However, we see a much greater increase in overvaluation for the matched control firms in Columns (4) through (6) with a difference in control firm overvaluation around the implementation of RegFD of 0.306 significant at the 1% level. This larger change in the control firms implies that there is an 25By mandating the uniform disclosure of material information, RegFD removes the informational advantage in finding undervalued firms previously enjoyed by dedicated institutions. 45

overall reduction in the overvaluation of firms held by transients around RegFD in the differencein-difference test in Column (7), with a magnitude of -0.223 significant at the 1% level. That is, when controlling for the self-selection of institutional types into firms using firm characteristics and contemporaneous firm- and sector-specific overvaluation, we find that the implementation of RegFD results in transient institutional ownership resulting in significantly less future firm-specific overvaluation, consistent with Table IX. Wefurthercomparethedifference-in-differenceresultsinColumn(7)fortransientanddedicated institutions to obtain a difference-in-difference-in-difference between the two institutional investor types relative to matched controls around the implementation of RegFD in Column (8). We find a coefficient of -0.251 significant at the 1% level, implying that the decrease in future overvaluation of firms held by transients relative to matched controls, when compared to that of firms held by dedicated institutions relative to controls, itself decreased markedly around the implementation of the RegFD disclosure requirement. In other words, transient institutional ownership results in less overvaluation after RegFD than dedicated ownership, controlling for contemporaneous overvaluation. Finally, we refine the sample to exclude firms that have both above-median dedicated and transient ownership, creating the indicator variable TvD which takes the value of 1 only when the firm has above-median transient ownership and below-median dedicated ownership and 0 only when the firm has above-median dedicated ownership and below-median transient ownership. This additional restriction removes potential confounding effects of observing the effects of both institutional investor types on the same firm. The results for TvD, in Row (c), in Columns (1) through (7) closely match those of isTRA firms, with overvaluation falling relative to matched controlfirmsaftertheimplementationofRegFD.Thesefindingsforthetransient-heldfirmssuggest that transient institutional investors experienced an information effect on future overvaluation around the implementation of RegFD. Panel B of Table B.III considers a similar matched-firm analysis for the next quarter’s firmspecificRKRVmisvaluation. Wefindthatfirmswithabove-mediandedicatedownershipinRow(a) do not experience a significant change in the absolute magnitude of misvaluation relative to control firms in the difference-in-difference analysis in Column (7) of Panel B. Firms with above-median 46

transientownershipinRow(b),ontheotherhand,experienceareductionof0.175inthemagnitude of misvaluation relative to control firms significant at the 1% level in Column (7). Furthermore, the difference-in-difference-in-difference test in Column (8) also finds a reduction of 0.121 in the absolute magnitude of future firm-specific misvaluation between the transient-held firms and their controls relative to dedicated-held firms and their controls around RegFD. When we exclude firms withbothabove-mediantransientanddedicatedownershipintheTvDanalysisinRow(c)inPanel B, we also observe a 0.164 reduction in the magnitude of misvaluation with a 5% significance level. The matched-firm tests show that these overvaluation and misvaluation effects are not merely due to institutional self-selection into firms based on current valuation since we select control firms usingcontemporaneousfirm-andsector-specificovervaluation. Theovervaluationandmisvaluation of firms owned by transient institutions falls relative to that of matched control firms around the implementation of the regulation in the difference-in-difference analysis, and this decrease is significantly greater than that for firms owned by dedicated institutions in the difference-indifference-in-difference. Table B.II: Mahalanobis distance matching algorithm. Using Mahalanobis distance matching, the single nearest neighborisidentifiedasacontrolfirmforeachtreatmentfirm. Columns(1),(4),and(7)reportthemeansofmatching characteristics for the actual DED, TRA, or TvD firms, respectively. Columns (2), (5), and (8) report the means of matching characteristics for the matched nearest neighbor to the actual DED, TRA, or TvD firms, respectively. Columns(3),(6),and(9)reportsthep-valuesofthet-testofdifferenceinmeansofmatchingcharacteristicsforDED, TRA,orTvDmatching,respectively. isDEDisequalto1ifthepercentageofdedicatedinstitutionalinvestorswithin a firm falls into the upper tercile and 0 otherwise. isTRA is equal to 1 if the percentage of transient institutional investorswithinafirmfallsintotheuppertercileand0otherwise. isTvDcontrastsisTRA=1firmsagainstisDED=1 firmsandisequalto1ifthepercentageoftransientinstitutionalinvestorswithinafirmfallsintotheuppertercileand thepercentageofdedicatedinstitutionalinvestorswithinafirmfallsintothebottomtercileand0ifthepercentageof transientinstitutionalinvestorswithinafirmfallsintothebottomtercileandthepercentageofdedicatedinstitutional investors within a firm falls into the upper tercile. isDED=1 isTRA=1 isTvD=1 Actual Matched p-val Actual Matched p-val Actual Matched p-val (1) (2) (3) (4) (5) (6) (7) (8) (9) RKRV(Firm)Overvali,t−1 0.062 0.066 0.609 0.155 0.151 0.574 0.177 0.167 0.399 RKRVSectorOvervali,t−1 0.137 0.138 0.927 0.152 0.150 0.464 0.150 0.147 0.528 LogTotalAssetsi,t−1 5.236 5.217 0.254 5.242 5.230 0.532 5.183 5.184 0.986 Book-to-MarketRatioi,t−1 0.556 0.551 0.258 0.502 0.504 0.681 0.497 0.499 0.698 47

gnihctamecnatsidsibonalahaMgnisU .0002nidetcaneDFnoitalugeRgnisugnihctamecnatsidsibonalahaMrednunoitamitseecnereffid-ni-ecnereffiD:III.BelbaT ,1=ARTsi,1=DEDsinodesabsmrfitnemtaertrofdnuoferaslortnoC .mrfitnemtaerthcaerofmrfilortnocasadefiitnedisirobhgientseraenelgniseht,mhtirogla ta ecnacfiingiS .)9002( nesreteP ni sa retrauq-raey dna mrfi htob yb deretsulc dna sesehtnerap eht ni detroper era srorre dradnatS .ylevitcepser ,1=DvTsi dna .*** yb level %1 dna ,** yb level %5 ,* yb detacidni si level %01 eht noitaulavrevOVRKR:AlenaP DED-ART lortnoC-tnemtaerT lortnoC tnemtaerT ffid-ni-ffid-ni-ffiD ffid-ni-ffiD erP-tsoPry-1 DFgeR-tsoPry-1 DFgeR-erPry-1 erP-tsoPry-1 DFgeR-tsoPry-1 DFgeR-erPry-1 )8( )7( )6( )5( )4( )3( )2( )1( 820.0 *** 081.0 *** 862.0 ** 880.0 *** 802.0 *** 262.0 450.0 1=DEDsi :tnemtaerT)a( )740.0( )430.0( )630.0( )730.0( )330.0( )330.0( )630.0( ***152.0- ***322.0- *** 603.0 *** 474.0 *** 961.0 ** 380.0 *** 213.0 *** 922.0 1=ARTsi :tnemtaerT)b( )570.0( )940.0( )130.0( )730.0( )730.0( )930.0( )430.0( )140.0( ***992.0- *** 772.0 *** 444.0 *** 761.0 220.0- *** 682.0 *** 803.0 1=DvTsi :tnemtaerT)c( )860.0( )240.0( )350.0( )150.0( )450.0( )650.0( )260.0( noitaulavsiMVRKR:BlenaP DED-ART lortnoC-tnemtaerT lortnoC tnemtaerT ffid-ni-ffid-ni-ffiD ffid-ni-ffiD erP-tsoPry-1 DFgeR-tsoPry-1 DFgeR-erPry-1 erP-tsoPry-1 DFgeR-tsoPry-1 DFgeR-erPry-1 )8( )7( )6( )5( )4( )3( )2( )1( 450.0- ** 850.0 *** 576.0 *** 616.0 400.0 *** 926.0 *** 526.0 1=DEDsi :tnemtaerT)a( )730.0( )620.0( )420.0( )120.0( )620.0( )020.0( )220.0( **121.0- ***571.0- *** 121.0 *** 067.0 *** 936.0 *550.0- *** 995.0 *** 356.0 1=ARTsi :tnemtaerT)b( )950.0( )040.0( )720.0( )420.0( )220.0( )920.0( )320.0( )620.0( ***461.0- *** 901.0 *** 837.0 *** 926.0 550.0- *** 146.0 *** 796.0 1=DvTsi :tnemtaerT)c( )550.0( )630.0( )530.0( )030.0( )140.0( )730.0( )240.0( 48

Table I: List of top ten dedicated and transient institutional investors by average portfolio size. Panel A lists the top ten dedicated institutional investors and panel B lists the top ten transient institutional investors. Institutional investor types are defined in Section 2.1. Dedicated institutional investors, as defined in Bushee (1998, 2001), are characterized as having concentrated holdings in the firms in their portfolios with low turnover. In contrast, transient institutional investors are characterized as having diversified holdings in the firms in their portfolios with high turnover. Investors are sorted based on their average portfolio size. Calculations based on: Thomson Reuters, Thomson Reuters Institutional (13f) Holdings s34 Master File, Wharton Research Data Services (WRDS), wrdsweb.wharton.upenn.edu/wrds/about/databaselist.cfm PanelA:Top10Dedicated(DED)InstitutionalInvestorsByAveragePortfolioSize Average Name PortfolioSize($) FidelityManagement&Research 366B CapitalResearch&Management 130B WellingtonManagement 121B JennisonAssociates 31B BerkshireHathaway 27B StateFarmMusualAutomobileInsurance 27B HarrisAssociates 16B SanfordBernstein&Co 14B BankofNewYorkAssetManagement 12B SoutheasternAssetManagement 12B PanelB:Top10Transient(TRA)InstitutionalInvestorsByAveragePortfolioSize Average Name PortfolioSize($) Blackrock 190B MorganStanley 84B JanusCapital 60B PutnamManagement 58B PacificInvestmentManagementCo. 42B Oppenheimerfunds 39B UBSWarburg 37B InvestorsResearchCorp 36B MarsicoCapitalManagement 34B AIMManagement 31B 49

Table II: Comparison of means for portfolio characteristics of dedicated institutional investors versus transient institutional investors. Institutional investor types are defined in Section 2.1. Dedicated institutional investors, as defined in Bushee (1998, 2001), are characterized as having concentrated holdings in the firms in their portfolios withlowturnover. Incontrast,transientinstitutionalinvestorsarecharacterizedashavingdiversifiedholdingsinthe firms in their portfolios with high turnover. Significance at the 10% level is indicated by *, 5% level by **, and 1% level by ***. Calculations based on: Thomson Reuters, Thomson Reuters Institutional (13f) Holdings s34 Master File, Wharton Research Data Services (WRDS), wrds-web.wharton.upenn.edu/wrds/about/databaselist.cfm DEDInvestors TRAInvestors Sig TotalPortfolioValue($M) 9431.6 2581.5 *** NumberofStocksinPortfolio 182.0 247.2 *** AverageFirmHoldings(%) 5.250 2.460 *** MedianFirmHoldings(%) 3.114 1.756 *** Std. Dev. FirmHoldings(%) 6.847 2.413 *** AverageFirmSize($M) 8784.8 17376.6 *** MedianFirmSize($M) 3023.5 5631.8 *** Std. Dev. FirmSize($M) 15899.0 32817.8 *** NumberofSIC3IndustriesinPortfolio 34.5 47.1 *** HHIofPortfolio 0.142 0.035 *** AverageFirmHoldingsConcentration 0.052 0.007 *** AveragePortfolioTurnover 0.041 0.122 *** 50

Table III: Sample statistics of institutional investor and firm characteristics. Institutional investor types are defined in Section 2.1. Dedicated institutional investors, as defined in Bushee (1998, 2001), are characterized as having concentrated holdings in the firms in their portfolios with low turnover. In contrast, transient institutional investors arecharacterizedashavingdiversifiedholdingsinthefirmsintheirportfolioswithhighturnover. Firmcharacteristics are defined in Appendix A. No. Obs Mean StdDev 1% 25% 50% 75% 99% TotalAssets($M) 236025 2582.6 12839.5 11.6 63.4 214.6 963.0 41117.0 MarketValue($M) 209482 3023.1 15107.5 9.4 73.2 260.7 1129.6 53706.2 No. ofInstitutionalInvestors 234969 92.3 142.5 1.0 15.0 43.0 112.0 727.0 %OwnedbyInstitutionalInvestors 230250 0.415 0.278 0.002 0.169 0.384 0.646 0.966 %OwnedbyDedicatedInst. Investors 236025 0.046 0.066 0.000 0.008 0.029 0.061 0.286 %OwnedbyTransientInst. Investors 236025 0.273 0.146 0.000 0.182 0.273 0.361 0.667 %OwnedbyLongHorizonInst. Investors 234969 0.273 0.144 0.000 0.191 0.250 0.325 0.857 %OwnedbyShortHorizonInst. Investors 234969 0.388 0.149 0.000 0.313 0.400 0.478 0.750 %OwnedbyFocusedInst. Investors 234969 0.436 0.217 0.147 0.261 0.379 0.571 1.000 %OwnedbyDiversifiedInst. Investors 234969 0.162 0.121 0.000 0.053 0.162 0.265 0.386 HHI-IndexofInstitutionalInvestors 232496 0.126 0.147 0.015 0.041 0.072 0.146 0.809 RKRVOvervaluation 236025 0.087 0.789 -2.151 -0.346 0.117 0.570 1.855 RKRVSectorOvervaluation 236025 0.018 0.325 -0.876 -0.168 0.041 0.222 0.791 RKRVLong-RunMtB 236025 0.531 0.434 -0.661 0.288 0.535 0.807 1.570 Altman’sZ-score 218482 1.107 5.665 -25.238 0.468 2.454 4.052 7.427 Long-termDebt/TA 232735 0.159 0.167 0.000 0.002 0.114 0.266 0.640 HasLTDebtCreditRating 236025 0.235 0.424 0.000 0.000 0.000 0.000 1.000 CashFlowDispersion 220816 0.450 1.411 -4.791 0.208 0.390 0.695 5.625 No. ofAnalystEstimates 236025 4.432 5.561 0.000 0.000 2.000 6.000 25.000 OpacityMeasure 198016 -0.156 0.441 -0.562 -0.431 -0.292 -0.052 1.699 MomentumReturn 222731 0.081 0.495 -0.693 -0.160 0.030 0.234 1.664 R&DIntensity 230381 0.097 0.373 0.000 0.000 0.004 0.065 1.927 IVspreadmon 59600 0.052 0.145 -0.418 -0.002 0.063 0.121 0.381 RealizedVolatility 81411 0.526 0.257 0.172 0.347 0.468 0.647 1.377 AverageExecutiveCompensation($M) 75079 29.4 262.1 0.0 3.1 7.6 19.6 212.3 MaximumExecutiveCompensation($M) 75079 114.1 1180.3 0.0 8.7 22.9 64.0 872.0 MedianExecutiveCompensation($M) 75079 9.0 38.4 0.0 1.5 3.6 8.6 79.5 AccrualsQuality 223249 -0.009 0.276 -0.926 -0.085 -0.008 0.062 0.973 PayoutRatio 228561 -0.003 0.049 -0.225 -0.009 -0.001 0.005 0.140 NetLeverageIncrease 205873 0.117 0.255 0.000 0.000 0.000 0.121 1.324 51

Table IV: Pairwise correlation matrix of institutional investor and firm characteristics. Institutional investor types are defined in Section 2.1. Institutional investor types are defined in Section 2.1. Dedicated institutional investors, as defined in Bushee (1998, 2001), are characterized as having concentrated holdings in the firms in their portfolios withlowturnover. Incontrast,transientinstitutionalinvestorsarecharacterizedashavingdiversifiedholdingsinthe firms in their portfolios with high turnover. Firm characteristics are defined in Appendix A. (1) (2) (3) (4) (5) (6) (7) (1)TotalAssets($M) (2)MarketValue($M) 0.8525 (3)No. ofInstitutionalInvestors 0.5612 0.7070 (4)%OwnedbyInstitutionalInvestors 0.0665 0.0879 0.4728 (5)%OwnedbyDedicatedInst. Investors -0.0685 -0.0715 -0.1651 -0.1305 (6)%OwnedbyTransientInst. Investors -0.0157 -0.0199 0.0359 0.1958 -0.2288 (7)%OwnedbyLongHorizonInst. Investors -0.0053 0.0032 -0.0831 -0.3022 0.0498 -0.4702 (8)%OwnedbyShortHorizonInst. Investors -0.0394 -0.0465 0.0132 0.2457 -0.0530 0.6165 -0.6475 (9)%OwnedbyFocusedInst. Investors -0.1351 -0.1388 -0.4282 -0.6035 0.2811 -0.3261 0.4049 (10)%OwnedbyDiversifiedInst. Investors 0.2235 0.2470 0.6227 0.6798 -0.2283 0.1933 -0.2502 (11)HHI-IndexofInstitutionalInvestors -0.0869 -0.1029 -0.3217 -0.4685 0.1021 -0.1229 0.2206 (12)RKRVOvervaluation 0.0162 0.0850 0.3028 0.3125 -0.0687 0.0636 -0.1199 (13)RKRVSectorOvervaluation -0.0464 -0.0274 0.0212 0.0711 0.0009 0.0143 -0.0570 (14)RKRVLong-RunMtB -0.1951 -0.1113 -0.1275 -0.1565 0.0298 0.0434 0.0443 (15)Altman’sZ-score 0.0549 0.0698 0.1434 0.1507 0.0291 -0.1390 0.0378 (16)Long-termDebt/TA 0.0600 0.0311 0.0560 0.0616 0.0035 -0.0528 0.0198 (17)HasLTDebtCreditRating 0.2850 0.2788 0.5016 0.3452 -0.0953 0.0167 -0.0819 (18)CashFlowDispersion -0.0111 -0.0128 -0.0247 -0.0067 0.0130 -0.0307 0.0392 (19)No. ofAnalystEstimates 0.2623 0.3401 0.7202 0.5678 -0.1700 0.1472 -0.1917 (20)OpacityMeasure -0.1333 -0.1401 -0.3579 -0.5021 0.1751 -0.2727 0.3500 (21)MomentumReturn -0.0075 0.0039 0.0195 0.0408 -0.0202 0.1198 -0.0756 (22)R&DIntensity -0.0333 -0.0268 -0.0428 -0.0319 0.0109 0.0655 -0.0338 (8) (9) (10) (11) (12) (13) (14) (9)%OwnedbyFocusedInst. Investors -0.3213 (10)%OwnedbyDiversifiedInst. Investors 0.2056 -0.7601 (11)HHI-IndexofInstitutionalInvestors -0.2125 0.4284 -0.4776 (12)RKRVOvervaluation 0.1123 -0.2767 0.3457 -0.2617 (13)RKRVSectorOvervaluation 0.0408 -0.0188 0.0061 -0.0502 -0.0779 (14)RKRVLong-RunMtB 0.0136 0.1601 -0.1929 0.1153 -0.1094 -0.0225 (15)Altman’sZ-score -0.0595 -0.0214 0.1746 -0.1433 0.0354 0.0512 -0.2858 (16)Long-termDebt/TA -0.0228 -0.0305 0.0823 -0.0278 -0.0194 0.0226 -0.1149 (17)HasLTDebtCreditRating 0.0341 -0.3317 0.4600 -0.2270 0.1267 0.0338 -0.2219 (18)CashFlowDispersion -0.0298 0.0521 -0.0181 0.0083 -0.0427 -0.0025 -0.0856 (19)No. ofAnalystEstimates 0.1464 -0.5045 0.6442 -0.3675 0.3359 0.0176 -0.1137 (20)OpacityMeasure -0.3026 0.5842 -0.5517 0.4409 -0.2699 -0.1117 0.1709 (21)MomentumReturn 0.0989 -0.0429 0.0434 -0.0523 0.1208 0.0851 0.0417 (22)R&DIntensity 0.0510 -0.0127 -0.0432 0.0088 0.0090 -0.0265 0.2292 (15) (16) (17) (18) (19) (20) (21) (16)Long-termDebt/TA 0.0557 (17)HasLTDebtCreditRating 0.1190 0.3987 (18)CashFlowDispersion 0.1931 0.0326 0.0002 (19)No. ofAnalystEstimates 0.1075 0.0587 0.4300 -0.0174 (20)OpacityMeasure -0.1227 -0.0123 -0.2662 0.0395 -0.4013 (21)MomentumReturn 0.0388 -0.0203 -0.0002 0.0059 0.0049 -0.0655 (22)R&DIntensity -0.3792 -0.1439 -0.1067 -0.1132 -0.0115 0.0073 -0.0157 52

TableV:Comparisonofmeansforfirmswithdedicatedinstitutionalinvestorsversusfirmswithtransientinstitutional investors. InstitutionalinvestortypesaredefinedinSection2.1. Dedicatedinstitutionalinvestors,asdefinedinBushee (1998, 2001), are characterized as having concentrated holdings in the firms in their portfolios with low turnover. In contrast, transient institutional investors are characterized as having diversified holdings in the firms in their portfolios with high turnover. Sorting by firm based on its percentage of dedicated institutional investors, isDED is equal to 1 if the percentage of dedicated institutional investors within a firm falls into the upper tercile and 0 otherwise. Similarly, sorting by firm based on its percentage of transient institutional investors, isTRA is equal to 1 if the percentage of transient institutional investors within a firm falls into the upper tercile and 0 otherwise. Firm characteristics are defined in Appendix A. Significance at the 10% level is indicated by *, 5% level by **, and 1% level by ***. isDED=1 isTRA=1 Sig TotalAssets($M) 1763.4 1434.2 *** MarketValue($M) 1632.5 1741.9 ** No. ofInstitutionalInvestors 66.9 83.8 *** %OwnedbyInstitutionalInvestors 0.418 0.412 *** %OwnedbyDedicatedInst. Investors 0.088 0.015 *** %OwnedbyTransientInst. Investors 0.187 0.378 *** %OwnedbyLongHorizonInst. Investors 0.296 0.230 *** %OwnedbyShortHorizonInst. Investors 0.333 0.466 *** %OwnedbyFocusedInst. Investors 0.458 0.418 *** %OwnedbyDiversifiedInst. Investors 0.147 0.170 *** HHI-IndexofInstitutionalInvestors 0.130 0.124 *** RKRVOvervaluation 0.004 0.140 *** RKRVSectorOvervaluation 0.016 0.018 RKRVLong-RunMtB 0.479 0.578 *** Altman’sZ-score 1.376 0.987 *** Long-termDebt/TA 0.162 0.154 *** HasLTDebtCreditRating 0.201 0.228 *** CashFlowDispersion 0.460 0.480 ** No. ofAnalystEstimates 3.3 4.9 *** OpacityMeasure -0.142 -0.168 *** MomentumReturn 0.018 0.150 *** R&DIntensity 0.098 0.091 *** IVspreadmon 0.039 0.061 *** RealizedVolatility 0.497 0.577 *** AverageExecutiveCompensation($M) 13.8 25.7 *** MaximumExecutiveCompensation($M) 51.1 100.1 *** MedianExecutiveCompensation($M) 5.2 8.7 *** AccrualsQuality -0.012 -0.002 *** PayoutRatio 0.131 0.090 *** NetLeverageIncrease -0.001 -0.005 *** 53

)5002( nahtanawsiV dna ,nosniboR ,fporK-sedohR eht no desab denfied si noitaulavsiM .srotsevni lanoitutitsni fo sepyt no noitaulavsim fo noitamitsE :IV elbaT VRKR .koob-ot-tekram nur-gnol dna ,rorre rotces seires-emit ,rorre cfiiceps-mrfi :stnenopmoc fo sepyt eerht otni oitar koob-ot-tekram eht fo noitisopmoced tnadneped eht sa noitaulavrevo cfiiceps-mrfi eht esu )5( dna ,)3( ,)1( snmuloC .noitisopmoced VRKR eht morf rorre cfiiceps-mrfi eht ot srefer noitaulavrevO cfiiceps-mrfi eht fo eulav etulosba eht esu )6( dna ,)4( ,)2( snmuloC .mrfi deulav )-revo( -rednu a setacidni rorre cfiiceps-mrfi )evitisop( evitagen A .elbairav detacideD .1.2 noitceS ni denfied era sepyt rotsevni lanoitutitsnI .noitaulavsim eht fo edutingam eht serutpac sihT .elbairav tnadneped eht sa noitaulavrevo .revonrut wol htiw soiloftrop rieht ni smrfi eht ni sgnidloh detartnecnoc gnivah sa deziretcarahc era ,)1002 ,8991( eehsuB ni denfied sa ,srotsevni lanoitutitsni fo egatnecreP .revonrut hgih htiw soiloftrop rieht ni smrfi eht ni sgnidloh defiisrevid gnivah sa deziretcarahc era srotsevni lanoitutitsni tneisnart ,tsartnoc nI .A xidneppA ni denfied era slortnoc llA .mrfi a nihtiw srotsevni lanoitutitsni fo rebmun latot eht ot evitaler si srotsevni lanoitutitsni tneisnart ro detacided dna ,** yb level %5 ,* yb detacidni si level %01 eht ta ecnacfiingiS .retrauq-raey dna mrfi htob yb deretsulc dna sesehtnerap eht ni detroper era srorre dradnatS .*** yb level %1 VRKR VRKR VRKR VRKR VRKR VRKR .lavsiM .lavrevO .lavsiM .lavrevO .lavsiM .lavrevO )6( )5( )4( )3( )2( )1( 5420.0- ***8667.0- 1−t,isrotsevnI .tsnIdetacideDybdenwO% )5050.0( )7701.0( 9400.0- ***6765.0- 3430.0- ***8078.0- 2−t,isrotsevnI .tsnIdetacideDybdenwO% )8650.0( )9690.0( )1150.0( )6221.0( 0301.0- ***0263.0- 2201.0- ***2615.0- 1−t,isrotsevnI .tsnIdetacideDybdenwO%∆ )7470.0( )1790.0( )4680.0( )6521.0( *** 6270.0 *** 7783.0 1−t,isrotsevnI .tsnItneisnarTybdenwO% )8720.0( )1960.0( *** 8431.0 ** 5941.0 * 7450.0 *** 5254.0 2−t,isrotsevnI .tsnItneisnarTybdenwO% )4230.0( )8260.0( )1030.0( )3770.0( *** 6590.0 *** 6482.0 3030.0 *** 5904.0 1−t,isrotsevnI .tsnItneisnarTybdenwO%∆ )5920.0( )0050.0( )5130.0( )4770.0( *** 4460.0 ***1470.0stessAlatoTgoL 1−t,i )0900.0( )5110.0( ***7943.0- *** 1154.0 1−t,isrotsevnIlanoitutitsnIybdenwO% )2430.0( )2350.0( *** 8512.0 ***9554.0- 1−t,isrotsevnIlanoitutitsnIfoxednI-IHH )7330.0( )5440.0( ***2400.0- 5100.0- 1−t,ierocs-ZnamtlA )7000.0( )3100.0( ***0401.0- ***0712.0- 1−t,iAT/tbeDTL )0720.0( )0640.0( 6300.0- *** 8480.0 1−t,ignitaRtiderCTLevaH )7610.0( )1020.0( *** 4600.0 ***8610.0- 1−t,inoisrepsiDwolFhsaC )6100.0( )4200.0( 1900.0- *** 1772.0 1−t,isetamitsEtsylanAfo .oN )7800.0( )1310.0( 7662.0 ** 2527.0 *** 3665.0 8743.0 *** 4575.0 8643.0 tnatsnoC )5602.0( )3992.0( )3281.0( )5762.0( )4371.0( )6252.0( Y Y Y Y Y Y ?stceffEdexiFretrauQ Y Y Y Y Y Y ?stceffEdexiFraeY Y Y Y Y Y Y ?stceffEdexiFyrtsudnI 763581 763581 600322 600322 520632 520632 .sbO .oN 3490.0 8532.0 8850.0 1660.0 6850.0 5160.0 2RdetsujdA 54

,fporK-sedohR eht no desab denfied si noitaulavsiM .selbairav lortnoc evitanretla htiw srotsevni lanoitutitsni fo sepyt no noitaulavsim fo noitamitsE :IIV elbaT dna ,rorre rotces seires-emit ,rorre cfiiceps-mrfi :stnenopmoc fo sepyt eerht otni oitar koob-ot-tekram eht fo noitisopmoced )5002( nahtanawsiV dna ,nosniboR cfiiceps-mrfi eht esu )3( dna )1( snmuloC .noitisopmoced VRKR eht morf rorre cfiiceps-mrfi eht ot srefer noitaulavrevO VRKR .koob-ot-tekram nur-gnol etulosba eht esu )4( dna )2( snmuloC .mrfi deulav )-revo( -rednu a setacidni rorre cfiiceps-mrfi )evitisop( evitagen A .elbairav tnadneped eht sa noitaulavrevo ni denfied era sepyt rotsevni lanoitutitsnI .noitaulavsim eht fo edutingam eht serutpac sihT .elbairav tnadneped eht sa noitaulavrevo cfiiceps-mrfi eht fo eulav soiloftropriehtnismrfiehtnisgnidlohdetartnecnocgnivahsadeziretcarahcera,)1002,8991(eehsuBnidenfiedsa,srotsevnilanoitutitsnidetacideD .1.2noitceS .revonruthgihhtiwsoiloftropriehtnismrfiehtnisgnidlohdefiisrevidgnivahsadeziretcarahcerasrotsevnilanoitutitsnitneisnart,tsartnocnI .revonrutwolhtiw ni denfied era slortnoc llA .mrfi a nihtiw srotsevni lanoitutitsni fo rebmun latot eht ot evitaler si srotsevni lanoitutitsni tneisnart ro detacided fo egatnecreP %5 ,* yb detacidni si level %01 eht ta ecnacfiingiS .retrauq-raey dna mrfi htob yb deretsulc dna sesehtnerap eht ni detroper era srorre dradnatS .A xidneppA .*** yb level %1 dna ,** yb level VRKR VRKR VRKR VRKR VRKR VRKR .lavsiM .lavrevO .lavsiM .lavrevO .lavsiM .lavrevO )6( )5( )4( )3( )2( )1( 1400.0- ***3185.0- 2910.0 ***9137.0- 5600.0- ***3806.0- 2−t,isrotsevnI .tsnIdetacideDybdenwO% )7750.0( )3890.0( )2660.0( )6801.0( )3160.0( )9201.0( 0601.0- ***8063.0- 7090.0- ***6514.0- 6611.0- ***1433.0- 1−t,isrotsevnI .tsnIdetacideDybdenwO%∆ )2670.0( )8790.0( )4880.0( )0111.0( )2280.0( )2701.0( *** 3131.0 ** 6051.0 *** 0191.0 0420.0- *** 6221.0 6330.0- 2−t,isrotsevnI .tsnItneisnarTybdenwO% )0330.0( )0460.0( )8630.0( )7760.0( )1630.0( )1860.0( *** 8980.0 *** 2192.0 *** 8921.0 *** 4742.0 *** 9580.0 *** 8242.0 1−t,isrotsevnI .tsnItneisnarTybdenwO%∆ )4030.0( )4050.0( )8230.0( )8150.0( )8130.0( )1940.0( *** 6460.0 ***1470.0- *** 5380.0 ***7990.0- *** 0660.0 ***9170.0stessAlatoTgoL 1−t,i )1900.0( )6110.0( )3900.0( )5210.0( )1900.0( )7110.0( ***9453.0- *** 7154.0 ***8633.0- *** 7454.0 ***6653.0- *** 8664.0 1−t,isrotsevnIlanoitutitsnIybdenwO% )6430.0( )2350.0( )4630.0( )5550.0( )9430.0( )3450.0( *** 4322.0 ***3954.0- *** 4171.0 ***5982.0- *** 9922.0 ***3844.0- 1−t,isrotsevnIlanoitutitsnIfoxednI-IHH )6430.0( )2540.0( )7830.0( )4940.0( )0530.0( )2640.0( ***4400.0- *4200.0- ***2400.0- **5300.0- ***1400.0- **6200.0- 1−t,ierocs-ZnamtlA )8000.0( )4100.0( )7000.0( )4100.0( )7000.0( )3100.0( ***1601.0- ***5012.0- ***6341.0- ***4302.0- ***8490.0- ***8512.0- 1−t,iAT/tbeDTL )3720.0( )2640.0( )2820.0( )1940.0( )6720.0( )3740.0( 8200.0- *** 4180.0 5720.0- *** 4621.0 6400.0- *** 9180.0 1−t,ignitaRtiderCTLevaH )7610.0( )1020.0( )5810.0( )1220.0( )9610.0( )5020.0( *** 7600.0 ***0610.0- *** 8400.0 ***6310.0- *** 9600.0 ***7610.0- 1−t,inoisrepsiDwolFhsaC )6100.0( )4200.0( )7100.0( )4200.0( )7100.0( )5200.0( 3700.0- *** 4182.0 9600.0- *** 9962.0 5900.0- *** 3082.0 1−t,isetamitsEtsylanAfo .oN )8800.0( )3310.0( )3900.0( )8310.0( )0900.0( )3310.0( 6200.0 *** 9551.0 1−t,isnruteRmutnemoM )2400.0( )2720.0( *** 6011.0 ***5782.0- 1−t,iyticapOmriF )1110.0( )0810.0( 9010.0- 0020.0- 1−t,iytisnetnID&R )8800.0( )8410.0( 5202.0 *** 3136.0 4852.0 *** 3139.0 0272.0 ** 2337.0 tnatsnoC )9661.0( )1342.0( )4712.0( )6782.0( )2912.0( )7513.0( Y Y Y Y Y Y ?stceffEdexiFretrauQ Y Y Y Y Y Y ?stceffEdexiFraeY Y Y Y Y Y Y ?stceffEdexiFyrtsudnI 992181 992181 327651 327651 070571 070571 .sbO .oN 0690.0 2242.0 2001.0 9062.0 4790.0 3652.0 2RdetsujdA 55

)5002(nahtanawsiVdna,nosniboR,fporK-sedohRehtnodesabdenfiedsinoitaulavsiM .srotsevnilanoitutitsnifosepytnonoitaulavsimfonoitamitsE:IIIVelbaT )1(snmuloC .koob-ot-tekramnur-gnoldna,rorrerotcesseires-emit,rorrecfiiceps-mrfi :stnenopmocfosepyteerhtotnioitarkoob-ot-tekramehtfonoitisopmoced snmuloC .yrtsudnideulav)-revo(-rednunasetacidnirorrerotces)evitisop(evitagenA .elbairavtnadnepedehtsanoitaulavrevorotcesseires-emitehtesu)4(dna )3( snmuloC .noitaulavsim eht fo edutingam eht serutpac sihT .elbairav tnadneped eht sa noitaulavrevo cfiiceps-rotces eht fo eulav etulosba eht esu )5( dna )2( ,srotsevnilanoitutitsnidetacideD .1.2noitceSnidenfiederasepytrotsevnilanoitutitsnI .elbairavtnadnepedehtsaeulavkoob-ot-tekramnur-gnolehtesu)6(dna tneisnart ,tsartnoc nI .revonrut wol htiw soiloftrop rieht ni smrfi eht ni sgnidloh detartnecnoc gnivah sa deziretcarahc era ,)1002 ,8991( eehsuB ni denfied sa tneisnart ro detacided fo egatnecreP .revonrut hgih htiw soiloftrop rieht ni smrfi eht ni sgnidloh defiisrevid gnivah sa deziretcarahc era srotsevni lanoitutitsni detropererasrorredradnatS.AxidneppAnidenfiederaslortnocllA .mrfianihtiwsrotsevnilanoitutitsniforebmunlatotehtotevitalersisrotsevnilanoitutitsni .*** yb level %1 dna ,** yb level %5 ,* yb detacidni si level %01 eht ta ecnacfiingiS .retrauq-raey dna mrfi htob yb deretsulc dna sesehtnerap eht ni VRKR VRKR VRKR VRKR VRKR VRKR nuR-gnoL rotceS rotceS nuR-gnoL rotceS rotceS eulaV .lavsiM .lavrevO eulaV .lavsiM .lavrevO )6( )5( )4( )3( )2( )1( ***6601.0- 8800.0 5710.0 *** 8401.0 3600.0 6310.0 2−t,isrotsevnI .tsnIdetacideDybdenwO% )8430.0( )9510.0( )2420.0( )5130.0( )2610.0( )1520.0( ***5490.0- 9500.0- 0840.0- 1620.0 7010.0- 5530.0- 1−t,isrotsevnI .tsnIdetacideDybdenwO%∆ )2230.0( )9220.0( )1150.0( )3930.0( )3120.0( )8240.0( *** 9280.0 7310.0- ** 8730.0 *** 1460.0 ***7130.0- ** 4830.0 2−t,isrotsevnI .tsnItneisnarTybdenwO% )1020.0( )4110.0( )4810.0( )4810.0( )0110.0( )7610.0( *** 4901.0 7700.0- ** 6950.0 *** 7170.0 5710.0- ** 9250.0 1−t,isrotsevnI .tsnItneisnarTybdenwO%∆ )4510.0( )6610.0( )4420.0( )9810.0( )5410.0( )1120.0( ***8770.0- 9000.0 ***3310.0stessAlatoTgoL 1−t,i )6300.0( )5200.0( )2400.0( 4410.0- ***1550.0- *** 4920.0 1−t,isrotsevnIlanoitutitsnIybdenwO% )8410.0( )5600.0( )7900.0( *** 2260.0 *** 0830.0 ***6730.0- 1−t,isrotsevnIlanoitutitsnIfoxednI-IHH )8510.0( )1900.0( )4210.0( ***2500.0- ***6200.0- ** 5100.0 1−t,ierocs-ZnamtlA )6000.0( )5000.0( )7000.0( *** 9603.0 ** 2740.0 ** 4360.0 1−t,iAT/tbeDTL )4120.0( )7810.0( )5130.0( *** 1630.0 *** 9510.0 *** 5710.0 1−t,ignitaRtiderCTLevaH )2800.0( )6300.0( )7500.0( ***3800.0- 6000.0- 4100.0- 1−t,inoisrepsiDwolFhsaC )1100.0( )6000.0( )9000.0( *** 5130.0 9100.0 * 0600.0 1−t,isetamitsEtsylanAfo .oN )3400.0( )0200.0( )3300.0( *** 8016.0 *** 7641.0 3120.0 *** 1152.0 *** 8171.0 4820.0tnatsnoC )7360.0( )0520.0( )1850.0( )0660.0( )1320.0( )5450.0( Y Y Y Y Y Y ?stceffEdexiFretrauQ Y Y Y Y Y Y ?stceffEdexiFraeY Y Y Y Y Y Y ?stceffEdexiFyrtsudnI 763581 763581 763581 600322 600322 600322 .sbO .oN 5324.0 1361.0 2523.0 7933.0 0351.0 6023.0 2RdetsujdA 56

noitalugerCESsihT.DFnoitalugeRgnisustluserniamehtstneserpAlenaP .0002nidetcaneDFnoitalugeRgnisunoitamitseecnereffid-ni-ecnereffiD:XIelbaT noitaulavsim dna noitaulavrevO .cilbup edam eb serusolcsid mrfi lla taht gnitadnam yb snoitutitsni tnatropmi yb deyojne egatnavda lanoitamrofni eht decuder rof 0 sa denfied si DFgeRsi .oitar koob-ot-tekram eht fo noitisopmoced )5002( nahtanawsiV dna ,nosniboR ,fporK-sedohR eht morf rorre cfiiceps-mrfi eht esu .DFnoitalugeRfonoitacilppaehtnisyaledronoitapicitnaynaroftnuoccaotsisylanaehtmorf0002evomereW .5002ot1002sraeyrof1dna9991ot5991sraey detartnecnocgnivahsadeziretcarahcera,)1002,8991(eehsuBnidenfiedsa,srotsevnilanoitutitsnidetacideD .1.2noitceSnidenfiederasepytrotsevnilanoitutitsnI eht ni sgnidloh defiisrevid gnivah sa deziretcarahc era srotsevni lanoitutitsni tneisnart ,tsartnoc nI .revonrut wol htiw soiloftrop rieht ni smrfi eht ni sgnidloh .mrfi a nihtiw srotsevni lanoitutitsni fo rebmun latot eht ot evitaler si srotsevni lanoitutitsni detacided fo egatnecreP .revonrut hgih htiw soiloftrop rieht ni smrfi )1( snmuloC .pihsrenwo detacided naidem-evoba htiw esoht rof 1 ot lauqe si DEDsi ,srotsevni lanoitutitsni detacided fo egatnecrep sti no desab mrfi yb gnitroS dna)3(snmuloC .)1=DEDsi(srotsevnilanoitutitsnidetacidedfoegatnecrepegralybdenwosmrfignomanoitaulavsimnoDFgeRfotcapmiehtenimaxe)2(dna )5( snmuloC .)1=ARTsi( srotsevni lanoitutitsni tneisnart fo erahs naidem-evoba na yb denwo smrfi gnoma noitaulavsim no DFgeR fo tcapmi eht enimaxe )4( nahtrehtar,5991dnuorasretnectahtstluserobecalpehtstneserpBlenaP .smrfi1=DEDsitsniagasmrfi1=ARTsignomaDFgeRfotcapmiehttsartnoc)6(dna denfied era slortnoc llA .8991 ot 6991 sraey rof 1 dna 4991 ot 2991 sraey rof 0 sa denfied si DFgeRsi ereH .kcohs suonegoxe ruo rof kcehc ssentsubor a sa ,0002 %5 ,* yb detacidni si level %01 eht ta ecnacfiingiS .retrauq-raey dna mrfi htob yb deretsulc dna sesehtnerap eht ni detroper era srorre dradnatS .A xidneppA ni .*** yb level %1 dna ,** yb level stluseRDFnoitalugeRniaM:AlenaP VRKR VRKR VRKR VRKR VRKR VRKR .lavsiM .lavrevO .lavsiM .lavrevO .lavsiM .lavrevO )6( )5( )4( )3( )2( )1( 4440.0 ***0321.0- *** 4541.0 *** 5921.0 *** 4451.0 *** 0990.0 DFgeRsi )0640.0( )3030.0( )2510.0( )9420.0( )1510.0( )2520.0( ***9060.0- ***6560.0- DEDsi )9010.0( )5310.0( 9810.0- 5610.0 DEDsixDFgeRsi )5610.0( )0910.0( ** 8320.0 *** 5280.0 ARTsi )9010.0( )9410.0( 7410.0- **0440.0- ARTsixDFgeRsi )9410.0( )2120.0( *** 7020.0 *** 5351.0 DvTsi )3700.0( )0120.0( ***8030.0- *4050.0- DvTsixDFgeRsi )2010.0( )9920.0( 8230.0 **7183.0- 7250.0 ***7526.0- *** 5723.0 ***6383.0- 2−t,i.vnI .tsnIdetacideD% )2620.0( )3851.0( )8280.0( )5821.0( )2680.0( )4131.0( 0920.0- 6320.0- 3850.0- **3053.0- 6761.0 2061.0t,1−t,i.vnI .tsnIdetacideD%∆ )8140.0( )9561.0( )9131.0( )7831.0( )0621.0( )0341.0( *8130.0- *** 3771.0 ** 2570.0 ** 5331.0 *** 5211.0 *** 3272.0 2−t,i.vnI .tsnItneisnarT% )1910.0( )4860.0( )1830.0( )1060.0( )8730.0( )8950.0( 2010.0 *** 9751.0 5050.0 *** 8942.0 * 2970.0 *** 4753.0 t,1−t,i.vnI .tsnItneisnarT%∆ )5120.0( )2510.0( )9540.0( )2850.0( )6240.0( )7060.0( *** 1191.0 4167.0 7323.0 9092.0 3953.0 8133.0 tnatsnoC )2360.0( )1684.0( )1972.0( )9693.0( )3082.0( )7593.0( Y Y Y Y Y Y ?slortnoCrehtO Y Y Y Y Y Y ?stceffEdexiFretrauQ Y Y Y Y Y Y ?stceffEdexiFraeY Y Y Y Y Y Y ?stceffEdexiFyrtsudnI 99893 99893 21567 21567 21567 21567 .sbO .oN 4791.0 6042.0 4390.0 4452.0 8690.0 3452.0 2RdetsujdA 57

.deunitnocXIelbaT stluseRDFnoitalugeRobecalP:BlenaP VRKR VRKR VRKR VRKR VRKR VRKR .lavsiM .lavrevO .lavsiM .lavrevO .lavsiM .lavrevO )6( )5( )4( )3( )2( )1( 0220.0 1210.0- **9720.0- 0100.0- 4110.0- 6900.0- DFgeRsi )5120.0( )0520.0( )9310.0( )7910.0( )0410.0( )1020.0( ***5950.0- **1130.0- DEDsi )3010.0( )2510.0( 1800.0- 7910.0- DEDsixDFgeRsi )2510.0( )1910.0( 0400.0- *** 0690.0 ARTsi )2010.0( )9810.0( 9220.0 *2140.0- ARTsixDFgeRsi )6410.0( )2120.0( *** 9750.0 *** 9731.0 DvTsi )0510.0( )8520.0( 6030.0 4710.0- DvTsixDFgeRsi )5020.0( )5920.0( *** 7172.0 **9803.0- 0770.0- ***4644.0- ** 4751.0 **0582.0- 2−t,i.vnI .tsnIdetacideD% )3960.0( )1551.0( )0760.0( )7701.0( )3260.0( )4411.0( *** 3322.0 ***7103.0- 1860.0- ***0493.0- *** 7731.0 ***2852.0t,1−t,i.vnI .tsnIdetacideD%∆ )4570.0( )8580.0( )4940.0( )1860.0( )4740.0( )7360.0( 1000.0 *** 6212.0 4400.0- *** 4982.0 8810.0 *** 5654.0 2−t,i.vnI .tsnItneisnarT% )2040.0( )7960.0( )4230.0( )9660.0( )3330.0( )5460.0( 9640.0 *** 5302.0 9610.0 *** 4962.0 5130.0 *** 2304.0 t,1−t,i.vnI .tsnItneisnarT%∆ )7940.0( )7010.0( )6920.0( )0550.0( )8920.0( )8750.0( *** 6526.0 1534.0 *** 4116.0 0042.0 *** 2236.0 6062.0 tnatsnoC )2012.0( )5023.0( )2361.0( )2192.0( )2161.0( )3782.0( Y Y Y Y Y Y ?slortnoCrehtO Y Y Y Y Y Y ?stceffEdexiFretrauQ Y Y Y Y Y Y ?stceffEdexiFraeY Y Y Y Y Y Y ?stceffEdexiFyrtsudnI 02602 02602 13004 13004 13004 13004 .sbO .oN 4870.0 3532.0 8760.0 1742.0 4170.0 9542.0 2RdetsujdA 58

Table X: Estimation of misvaluation on types of institutional investors with control variables. Overvaluation and misvaluation use the firm-specific error from the Rhodes-Kropf, Robinson, and Viswanathan (2005) decomposition of the market-to-book ratio. Long- (short-) horizon institutional investors, as defined in Yan and Zhang (2009), are characterized by having low (high) portfolio turnover. Focused (diversified) institutional investors, as defined in Bushee (1998, 2001), are characterized by having high (low) average holdings in invested firms. Percentage of institutional investor type is relative to the total number of institutional investors within a firm. All controls are defined in Appendix A. Panel A reports the results using observations across all years. Panels B and C report the resulting using observations prior to and post 2000, respectively, to sub-sample into pre and post Regulation FD periods. Standard errors are reported in the parentheses and clustered by both firm and year-quarter. Significance at the 10% level is indicated by *, 5% level by **, and 1% level by ***. PanelA:AllYears RKRV RKRV RKRV RKRV RKRV RKRV Overval. Misval. Overval. Misval. Overval. Misval. (1) (2) (3) (4) (5) (6) %Long-termInst. Inv.i,t−2 0.3403 *** -0.0199 (0.0635) (0.0415) ∆%Long-termInst. Inv.i,t−1 0.0910 * -0.0244 (0.0544) (0.0319) %Short-termInst. Inv.i,t−2 0.0273 0.0871 ** (0.0682) (0.0339) ∆%Short-termInst. Inv.i,t−1 0.1423 *** 0.0623 ** (0.0504) (0.0262) %FocusedInst. Inv.i,t−2 -0.1013* 0.0458 (0.0536) (0.0363) ∆%FocusedInst. Inv.i,t−1 -0.2325*** 0.0383 (0.0462) (0.0279) %DiversifiedInst. Inv.i,t−2 1.9342 *** -0.1865** (0.0972) (0.0791) ∆%DiversifiedInst. Inv.i,t−1 1.5232 *** -0.1007* (0.0813) (0.0519) %LongandFoc. Inst. Inv.i,t−2 -0.0022 0.0157 (0.0583) (0.0425) ∆%LongandFoc. Inst. Inv.i,t−1 -0.2364*** 0.0076 (0.0580) (0.0356) %LongandDiv. Inst. Inv.i,t−2 3.4606 *** -0.2753* (0.2091) (0.1459) ∆%LongandDiv. Inst. Inv.i,t−1 2.1724 *** -0.1407 (0.1643) (0.0917) %ShortandFoc. Inst. Inv.i,t−2 -0.1479** 0.1186 *** (0.0685) (0.0442) ∆%ShortandFoc. Inst. Inv.i,t−2 -0.1067* 0.0768 ** (0.0548) (0.0369) %ShortandDiv. Inst. Inv.i,t−2 1.2900 *** -0.2857*** (0.1467) (0.0969) ∆%ShortandDiv. Inst. Inv.i,t−1 1.2652 *** -0.1277* (0.1132) (0.0684) Constant 0.7178 ** 0.2673 0.9773 *** 0.2606 0.9333 *** 0.2538 (0.2864) (0.2172) (0.2715) (0.2188) (0.2678) (0.2192) OtherControls? Y Y Y Y Y Y QuarterFixedEffects? Y Y Y Y Y Y YearFixedEffects? Y Y Y Y Y Y IndustryFixedEffects? Y Y Y Y Y Y No. Obs. 156723 156723 156723 156723 156723 156723 AdjustedR2 0.2603 0.0992 0.2831 0.0994 0.2807 0.0999 59

TableXcontinued. PanelB:Priorto2000 RKRV RKRV RKRV RKRV RKRV RKRV Overval. Misval. Overval. Misval. Overval. Misval. (1) (2) (3) (4) (5) (6) %Long-termInst. Inv.i,t−2 0.1103 * 0.0736 * (0.0602) (0.0382) ∆%Long-termInst. Inv.i,t−1 -0.0557 0.0322 (0.0542) (0.0270) %Short-termInst. Inv.i,t−2 0.2030 *** 0.0416 (0.0547) (0.0334) ∆%Short-termInst. Inv.i,t−1 0.2561 *** 0.0258 (0.0511) (0.0241) %FocusedInst. Inv.i,t−2 -0.3774*** -0.0393 (0.0546) (0.0324) ∆%FocusedInst. Inv.i,t−1 -0.3776*** -0.0322 (0.0410) (0.0271) %DiversifiedInst. Inv.i,t−2 1.3256 *** 0.1626 ** (0.1080) (0.0725) ∆%DiversifiedInst. Inv.i,t−1 1.1059 *** 0.1307 ** (0.0856) (0.0510) %LongandFoc. Inst. Inv.i,t−2 -0.2531*** -0.0002 (0.0609) (0.0412) ∆%LongandFoc. Inst. Inv.i,t−1 -0.3639*** 0.0005 (0.0541) (0.0323) %LongandDiv. Inst. Inv.i,t−2 2.4602 *** 0.5233 *** (0.2173) (0.1379) ∆%LongandDiv. Inst. Inv.i,t−1 1.5762 *** 0.3054 *** (0.1545) (0.0876) %ShortandFoc. Inst. Inv.i,t−2 -0.0768 -0.0634* (0.0689) (0.0345) ∆%ShortandFoc. Inst. Inv.i,t−2 -0.0263 -0.0677** (0.0525) (0.0307) %ShortandDiv. Inst. Inv.i,t−2 1.3002 *** 0.0848 (0.1563) (0.0989) ∆%ShortandDiv. Inst. Inv.i,t−1 1.1361 *** 0.1000 (0.1216) (0.0778) Constant 0.6985 *** 0.5650 *** 1.2141 *** 0.6447 *** 0.9957 *** 0.6442 *** (0.1723) (0.1443) (0.1645) (0.1446) (0.1661) (0.1424) OtherControls? Y Y Y Y Y Y QuarterFixedEffects? Y Y Y Y Y Y YearFixedEffects? Y Y Y Y Y Y IndustryFixedEffects? Y Y Y Y Y Y No. Obs. 58929 58929 58929 58929 58929 58929 AdjustedR2 0.2522 0.0743 0.2731 0.0748 0.2690 0.0756 60

TableXcontinued. PanelC:Post2000 RKRV RKRV RKRV RKRV RKRV RKRV Overval. Misval. Overval. Misval. Overval. Misval. (1) (2) (3) (4) (5) (6) %Long-termInst. Inv.i,t−2 0.3911 *** 0.0126 (0.1029) (0.0730) ∆%Long-termInst. Inv.i,t−1 0.1166 -0.0183 (0.0926) (0.0606) %Short-termInst. Inv.i,t−2 -0.1507 0.1200 ** (0.1037) (0.0569) ∆%Short-termInst. Inv.i,t−1 0.0264 0.0900 ** (0.0744) (0.0458) %FocusedInst. Inv.i,t−2 -0.0439 0.0808 (0.0799) (0.0583) ∆%FocusedInst. Inv.i,t−1 -0.2296*** 0.0921 * (0.0735) (0.0478) %DiversifiedInst. Inv.i,t−2 2.0162 *** -0.2928*** (0.1358) (0.1098) ∆%DiversifiedInst. Inv.i,t−1 1.6246 *** -0.2200*** (0.1207) (0.0710) %LongandFoc. Inst. Inv.i,t−2 -0.0181 0.0376 (0.1054) (0.0766) ∆%LongandFoc. Inst. Inv.i,t−1 -0.2869** 0.0109 (0.1154) (0.0714) %LongandDiv. Inst. Inv.i,t−2 3.9262 *** -0.4268** (0.2798) (0.2058) ∆%LongandDiv. Inst. Inv.i,t−1 2.4976 *** -0.2585* (0.2520) (0.1390) %ShortandFoc. Inst. Inv.i,t−2 -0.3483*** 0.2963 *** (0.1237) (0.0890) ∆%ShortandFoc. Inst. Inv.i,t−2 -0.2853*** 0.2451 *** (0.0980) (0.0673) %ShortandDiv. Inst. Inv.i,t−2 0.8518 *** -0.4357*** (0.2126) (0.1431) ∆%ShortandDiv. Inst. Inv.i,t−1 1.1159 *** -0.2689*** (0.1765) (0.1018) Constant 0.1674 0.3036 0.3249 0.3064 0.3806 0.2918 (0.3863) (0.2891) (0.3693) (0.2887) (0.3550) (0.2902) OtherControls? Y Y Y Y Y Y QuarterFixedEffects? Y Y Y Y Y Y YearFixedEffects? Y Y Y Y Y Y IndustryFixedEffects? Y Y Y Y Y Y No. Obs. 91213 91213 91213 91213 91213 91213 AdjustedR2 0.3026 0.1361 0.3218 0.1373 0.3219 0.1386 61

.elbairav tnadneped eht sa daerpsVI sesu )1( nmuloC .selbairav lortnoc htiw srotsevni lanoitutitsni fo sepyt no scitsiretcarahc mrfi fo noitamitsE :IX elbaT nom eht serutpac erusaem sihT .elbairav tnedneped eht sa raey tsap eht revo ytilitalov dezilaer sesu )2( nmuloC .stneve hsarc dna ksir liat rof seixorp erusaem sihT tnednepedehtsanoitasnepmoclaireganamnaidemdna,mumixam,egarevaehtfogollarutanehtesu)5(hguorht)3(snmuloC .mrfiehtfoksirlacirotsihllarevo tuoyapehtsesu)7(nmuloC .ytilauqslaurccanignisaercedsierusaemsihT .elbairavtnadnepedehtsaytilauqslaurccaehtsesu)6(nmuloC .ylevitcepser,elbairav tnedneped eht sa stessa latot ot sesaercni egarevel ten sesu )8( nmuloC .elbairav tnedneped eht sa ,sesahcruper erahs dna sdnedivid htob sedulcni hcihw ,oitar ,8991( eehsuB ni denfied sa ,srotsevni lanoitutitsni detacideD .egarevel gnisaercni fo noitcerid eht ni scimanyd erutcurts latipac serutpac erusaem sihT .elbairav erasrotsevnilanoitutitsnitneisnart,tsartnocnI .revonrutwolylemertxednasoiloftropriehtnismrfinitnemtsevniegarevaegralgnivahsadeziretcarahcera,)1002 lanoitutitsni tneisnart ro detacided fo egatnecreP .sgnidloh oiloftrop defiisrevid ylhgih dna revonrut oiloftrop hgih gnivah sa deziretcarahc srotsevni lanoitutitsni C dna B slenaP .sraey lla morf snoitavresbo gnisu stluser eht stroper A lenaP .mrfi a nihtiw srotsevni lanoitutitsni fo rebmun latot eht ot evitaler si srotsevni xidneppAnidenfiederaslortnocllA .sdoirepDFnoitalugeR-tsopdna-erpotnielpmas-busot,ylevitcepser,0002tsopotroirpsnoitavresbognisusluserehttroper ,** yb level %5 ,* yb detacidni si level %01 eht ta ecnacfiingiS .retrauq-raey dna mrfi htob yb deretsulc dna sesehtnerap eht ni detroper era srorre dradnatS .A .*** yb level %1 dna sraeYllA:AlenaP teN naideM mumixaM egarevA egareveL tuoyaP slaurccA evitucexE evitucexE evitucexE dezilaeR esaercnI oitaR ytilauQ noitasnepmoC noitasnepmoC noitasnepmoC ytilitaloV nomdaerpsVI )8( )7( )6( )5( )4( )3( )2( )1( 4200.0- ***6201.0- 5910.0- **1237.1- 9523.1- 1381.1- ***0182.0- 9060.0 2−t,i.vnI .tsnIdetacideD% )8200.0( )2220.0( )2120.0( )5357.0( )6758.0( )0177.0( )9180.0( )4950.0( 8100.0 ***7360.0- 0510.0- ***3640.2- *3361.1- **1842.1- **7402.0- 4700.0- 1−t,i.vnI .tsnIdetacideD%∆ )8400.0( )0020.0( )6020.0( )8366.0( )9466.0( )0126.0( )6780.0( )4360.0( ***4110.0- ***9122.0- * 7520.0 *** 6866.1 *** 6612.2 *** 4400.2 *** 2373.0 *** 9250.0 2−t,i.vnI .tsnItneisnarT% )6100.0( )0510.0( )9410.0( )7812.0( )6252.0( )9822.0( )9220.0( )5310.0( ***1120.0- ***0341.0- 1610.0 *** 9934.1 *** 5296.1 *** 8395.1 *** 4351.0 *** 2511.0 1−t,i.vnI .tsnItneisnarT%∆ )2200.0( )6010.0( )4210.0( )6871.0( )1212.0( )0691.0( )2520.0( )2610.0( ***9510.0- 4800.0- 7600.0 ***8918.2- 4380.0- 8385.1- *** 8697.0 ***3970.0tnatsnoC )6400.0( )3720.0( )6410.0( )1353.0( )2494.0( )0025.0( )8720.0( )5510.0( Y Y Y Y Y Y Y Y ?slortnoCrehtO Y Y Y Y Y Y Y Y ?stceffEdexiFretrauQ Y Y Y Y Y Y Y Y ?stceffEdexiFraeY Y Y Y Y Y Y Y Y ?stceffEdexiFyrtsudnI 043351 553341 836251 09475 79875 79875 36426 68654 .sbO .oN 2230.0 5921.0 1420.0 4654.0 3224.0 1154.0 3077.0 3780.0 2RdetsujdA 62

.deunitnocIXelbaT 0002otroirP:BlenaP teN naideM mumixaM egarevA egareveL tuoyaP slaurccA evitucexE evitucexE evitucexE dezilaeR esaercnI oitaR ytilauQ noitasnepmoC noitasnepmoC noitasnepmoC ytilitaloV nomdaerpsVI )8( )7( )6( )5( )4( )3( )2( )1( 2200.0- ***1060.0- 6500.0 *5676.1- 2474.1- 1553.1- **7732.0- 0010.0 2−t,i.vnI .tsnIdetacideD% )7300.0( )8220.0( )2020.0( )2339.0( )8980.1( )6679.0( )9201.0( )1280.0( 3400.0- *3530.0- 7120.0- **9019.1- 5968.0- 0610.1- **3591.0- 4630.0- 1−t,i.vnI .tsnIdetacideD%∆ )8500.0( )8020.0( )7810.0( )6828.0( )7408.0( )9477.0( )4090.0( )4490.0( ***8210.0- ***7261.0- *** 7940.0 *** 2801.2 *** 0809.2 *** 0856.2 *** 4863.0 3720.0 2−t,i.vnI .tsnItneisnarT% )4200.0( )2510.0( )3210.0( )7433.0( )8693.0( )7253.0( )0630.0( )9220.0( ***7520.0- ***8201.0- ** 4130.0 *** 0797.1 *** 4182.2 *** 2651.2 ** 1901.0 *** 5921.0 1−t,i.vnI .tsnItneisnarT%∆ )3300.0( )4110.0( )2410.0( )7162.0( )4382.0( )2252.0( )1940.0( )5230.0( 5900.0- 4000.0 ** 0040.0 ***0204.1- 5055.0- **2164.1- *** 6476.0 ***3533.0tnatsnoC )4800.0( )3810.0( )8910.0( )6370.0( )0647.0( )4837.0( )1230.0( )5810.0( Y Y Y Y Y Y Y Y ?slortnoCrehtO Y Y Y Y Y Y Y Y ?stceffEdexiFretrauQ Y Y Y Y Y Y Y Y ?stceffEdexiFraeY Y Y Y Y Y Y Y Y ?stceffEdexiFyrtsudnI 49675 07135 75875 25071 86271 86271 43711 1717 .sbO .oN 7120.0 2231.0 9940.0 0244.0 2614.0 5434.0 6457.0 0050.0 2RdetsujdA 63

.deunitnocIXelbaT 0002otroirP:ClenaP teN naideM mumixaM egarevA egareveL tuoyaP slaurccA evitucexE evitucexE evitucexE dezilaeR esaercnI oitaR ytilauQ noitasnepmoC noitasnepmoC noitasnepmoC ytilitaloV nomdaerpsVI )8( )7( )6( )5( )4( )3( )2( )1( 0500.0- ***6302.0- *3390.0- **1857.2- 6840.2- 3567.1- 3730.0- 7521.0 2−t,i.vnI .tsnIdetacideD% )6400.0( )7240.0( )0150.0( )5942.1( )8562.1( )0521.1( )2690.0( )0380.0( 7700.0 ***4321.0- 7030.0- **9169.2- *8811.2- *8860.2- 7350.0 3840.0 1−t,i.vnI .tsnIdetacideD%∆ )1800.0( )7930.0( )6250.0( )6532.1( )3261.1( )5960.1( )7601.0( )9880.0( ***3010.0- ***7552.0- 0400.0 *** 4740.1 *** 8894.1 *** 2613.1 *** 8482.0 *** 6760.0 2−t,i.vnI .tsnItneisnarT% )2200.0( )5910.0( )4420.0( )5642.0( )2992.0( )2662.0( )4120.0( )9510.0( ***3710.0- ***7071.0- 7000.0- *** 0379.0 *** 8770.1 *** 8610.1 *** 2611.0 *** 8801.0 1−t,i.vnI .tsnItneisnarT%∆ )2300.0( )3510.0( )8810.0( )1602.0( )0762.0( )2042.0( )2420.0( )1910.0( ***1710.0- 2300.0- ***7860.0- ***4848.2- 7562.0- **1086.1- *** 7140.1 5810.0 tnatsnoC )1400.0( )8640.0( )4020.0( )8754.0( )6896.0( )7207.0( )8020.0( )1810.0( Y Y Y Y Y Y Y Y ?slortnoCrehtO Y Y Y Y Y Y Y Y ?stceffEdexiFretrauQ Y Y Y Y Y Y Y Y ?stceffEdexiFraeY Y Y Y Y Y Y Y Y ?stceffEdexiFyrtsudnI 98198 03248 17288 97973 36183 36183 74184 35463 .sbO .oN 9240.0 6241.0 3720.0 8684.0 7654.0 1094.0 0248.0 6540.0 2RdetsujdA 64

.elbairav tnadneped eht sa daerpsVI sesu )1( nmuloC .selbairav lortnoc htiw srotsevni lanoitutitsni fo sepyt no scitsiretcarahc mrfi fo noitamitsE :IIX elbaT nom eht serutpac erusaem sihT .elbairav tnedneped eht sa raey tsap eht revo ytilitalov dezilaer sesu )2( nmuloC .stneve hsarc dna ksir liat rof seixorp erusaem sihT tnednepedehtsanoitasnepmoclaireganamnaidemdna,mumixam,egarevaehtfogollarutanehtesu)5(hguorht)3(snmuloC .mrfiehtfoksirlacirotsihllarevo tuoyapehtsesu)7(nmuloC .ytilauqslaurccanignisaercedsierusaemsihT .elbairavtnadnepedehtsaytilauqslaurccaehtsesu)6(nmuloC .ylevitcepser,elbairav tnedneped eht sa stessa latot ot sesaercni egarevel ten sesu )8( nmuloC .elbairav tnedneped eht sa ,sesahcruper erahs dna sdnedivid htob sedulcni hcihw ,oitar ni denfied sa ,srotsevni lanoitutitsni noziroh )-trohs( -gnoL .egarevel gnisaercni fo noitcerid eht ni scimanyd erutcurts latipac serutpac erusaem sihT .elbairav ,)1002 ,8991(eehsuBnidenfied sa ,srotsevnilanoitutitsni )defiisrevid(desucoF .revonrutoiloftrop)hgih(wol gnivahybdeziretcarahcera ,)9002(gnahZ dna naY lanoitutitsniforebmunlatotehtotevitalersiepytrotsevnilanoitutitsnifoegatnecreP .smrfidetsevninisgnidlohegareva)wol(hgihgnivahybdeziretcarahcera .retrauq-raey dna mrfi htob yb deretsulc dna sesehtnerap eht ni detroper era srorre dradnatS .A xidneppA ni denfied era slortnoc llA .mrfi a nihtiw srotsevni .*** yb level %1 dna ,** yb level %5 ,* yb detacidni si level %01 eht ta ecnacfiingiS teN naideM mumixaM egarevA egareveL tuoyaP slaurccA evitucexE evitucexE evitucexE dezilaeR esaercnI oitaR ytilauQ noitasnepmoC noitasnepmoC noitasnepmoC ytilitaloV nomdaerpsVI )8( )7( )6( )5( )4( )3( )2( )1( *** 1500.0 *** 7980.0 1800.0- ** 2402.1 4258.0- 2701.0- ***8595.0- 8550.0 2−t,i.vnI .tsnIdesucoFdnamret-gnoL% )9100.0( )8710.0( )3410.0( )1674.0( )1045.0( )8884.0( )1570.0( )1140.0( *** 9010.0 *** 1260.0 9700.0- 6330.0 ***7012.1- *1327.0- ***6273.0- 9310.0- 1−t,i.vnI .tsnIdesucoFdnamret-gnoL%∆ )4200.0( )1410.0( )4110.0( )5273.0( )8224.0( )4093.0( )1170.0( )8040.0( *** 4710.0 *** 3727.0 0750.0- 1467.0- 2171.0- 9713.0- ***8233.0- 5540.0 2−t,i.vnI .tsnIdefiisreviDdnamret-gnoL% )9400.0( )2660.0( )2940.0( )7065.0( )6696.0( )2516.0( )3860.0( )2830.0( 6000.0 *** 2133.0 4720.0- 0461.0- 7000.0 4030.0- 2120.0- 7410.0- 1−t,i.vnI .tsnIdefiisreviDdnamret-gnoL%∆ )0600.0( )2840.0( )2330.0( )8614.0( )9054.0( )0314.0( )8270.0( )0740.0( 2300.0- ***6770.0- 8710.0 *** 4100.2 *** 3349.1 *** 9678.1 9920.0 *** 1251.0 2−t,i.vnI .tsnIdesucoFdnamret-trohS% )2200.0( )5310.0( )9410.0( )8033.0( )6593.0( )3453.0( )4040.0( )2820.0( 3500.0- ***7640.0- 7610.0 *** 4701.1 *** 0690.1 *** 1640.1 3600.0 *** 4070.0 2−t,i.vnI .tsnIdesucoFdnamret-trohS%∆ )0400.0( )7110.0( )5210.0( )3542.0( )2982.0( )5762.0( )0830.0( )6520.0( 2200.0 ***0490.0- 2720.0 *** 7573.2 *** 2593.3 *** 5731.3 9010.0 ** 6180.0 2−t,i.vnI .tsnIdefiisreviDdnamret-trohS% )0400.0( )1630.0( )8230.0( )9993.0( )7754.0( )3424.0( )9740.0( )2330.0( ***1610.0- **2160.0- ** 8550.0 *** 5702.2 *** 9444.2 *** 7383.2 **9231.0- *** 1131.0 1−t,i.vnI .tsnIdefiisreviDdnamret-trohS%∆ )8400.0( )9920.0( )7420.0( )1262.0( )1692.0( )6972.0( )5350.0( )0620.0( ***3810.0- 6940.0- 7300.0 ***3152.3- *** 5830.0 ***5616.1- *** 8079.0 ***5690.0tnatsnoC )9400.0( )6130.0( )6710.0( )2643.0( )7700.0( )3323.0( )3720.0( )8810.0( Y Y Y Y Y Y Y Y ?slortnoCrehtO Y Y Y Y Y Y Y Y ?stceffEdexiFretrauQ Y Y Y Y Y Y Y Y ?stceffEdexiFraeY Y Y Y Y Y Y Y Y ?stceffEdexiFyrtsudnI 043351 553341 836251 09475 79875 79875 36426 68654 .sbO .oN 6130.0 7031.0 1420.0 4454.0 1914.0 8744.0 2467.0 4780.0 2RdetsujdA 65

dloh-dna-yub eht esu )4( hguorht )1( snmuloC .selbairav lortnoc htiw srotsevni lanoitutitsni fo sepyt no ecnamrofrep mrfi erutuf fo noitamitsE :IIIX elbaT .retrauq tnaveler eht revo snruter ylhtnom gnidnuopmoc yb detaluclac era snruter dezilaeR .ylevitcepser ,tsop rtq-4 dna ,rtq-3 ,rtq-2 ,rtq-1 rof snruter dezilaer evitaler detaluclac era snruter lamronbA .ylevitcepser ,tsop rtq-4 dna ,rtq-3 ,rtq-2 ,rtq-1 rof snruter lamronba dloh-dna-yub eht esu )8( hguorht )5( snmuloC egareva egral gnivah sa deziretcarahc era ,)1002 ,8991( eehsuB ni denfied sa ,srotsevni lanoitutitsni detacideD .ledom rotcaf-5 hguabmatS dna rotsaP eht ot gnivahsadeziretcarahcsrotsevnilanoitutitsnierasrotsevnilanoitutitsnitneisnart,tsartnocnI .revonrutwolylemertxednasoiloftropriehtnismrfinitnemtsevni fo rebmun latot eht ot evitaler si srotsevni lanoitutitsni tneisnart ro detacided fo egatnecreP .sgnidloh oiloftrop defiisrevid ylhgih dna revonrut oiloftrop hgih dna mrfi htob yb deretsulc dna sesehtnerap eht ni detroper era srorre dradnatS .A xidneppA ni denfied era slortnoc llA .mrfi a nihtiw srotsevni lanoitutitsni .*** yb level %1 dna ,** yb level %5 ,* yb detacidni si level %01 eht ta ecnacfiingiS .retrauq-raey drawroF drawroF drawroF drawroF drawroF drawroF drawroF drawroF rtQ-4 rtQ-3 rtQ-2 rtQ-1 rtQ-4 rtQ-3 rtQ-2 rtQ-1 snruteR .bA snruteR .bA snruteR .bA snruteR .bA snruteR snruteR snruteR snruteR )8( )7( )6( )5( )4( )3( )2( )1( 7310.0 1120.0 *** 3830.0 5910.0 * 8620.0 2020.0 * 0220.0 * 0120.0 2−t,i.vnI .tsnIdetacideD% )9310.0( )6410.0( )9310.0( )9110.0( )3410.0( )6510.0( )9210.0( )4110.0( 3410.0 ** 5750.0 * 3140.0 6000.0 2630.0 9010.0 *** 4121.0 6920.0- 1−t,i.vnI .tsnIdetacideD%∆ )1320.0( )4320.0( )2320.0( )1810.0( )4940.0( )5630.0( )8630.0( )0030.0( 7610.0- 7700.0- 1210.0- 8000.0 3010.0- **1030.0- *4910.0- 3310.0- 2−t,i.vnI .tsnItneisnarT% )2010.0( )1010.0( )2900.0( )1700.0( )5310.0( )9110.0( )9010.0( )2900.0( 3410.0- 9400.0 4010.0- *** 7520.0 4610.0- **2350.0- 2400.0 0100.0 1−t,i.vnI .tsnItneisnarT%∆ )3310.0( )3110.0( )5900.0( )6700.0( )5320.0( )2120.0( )4910.0( )1510.0( 6110.0- ***8930.0- **4420.0- ***1820.0- 7130.0- 1400.0 0200.0- 2510.0 tnatsnoC )3210.0( )1210.0( )5900.0( )4700.0( )0040.0( )1150.0( )9430.0( )3920.0( Y Y Y Y Y Y Y Y ?slortnoCrehtO Y Y Y Y Y Y Y Y ?stceffEdexiFretrauQ Y Y Y Y Y Y Y Y ?stceffEdexiFraeY Y Y Y Y Y Y Y Y ?stceffEdexiFyrtsudnI 428211 558311 454311 424801 124351 486051 350051 668241 .sbO .oN 9600.0 7500.0 6500.0 3500.0 4300.0 4630.0 4720.0 9250.0 2RdetsujdA 66

45% 40% 35% 30% 25% 20% 15% 10% 5% 0% srotsevnI lanoitutitsnI latoT ot egatnecreP 1Q 5891 1Q 6891 1Q 7891 1Q 8891 1Q 9891 1Q 0991 1Q 1991 1Q 2991 1Q 3991 1Q 4991 1Q 5991 1Q 6991 1Q 7991 1Q 8991 1Q 9991 1Q 0002 1Q 1002 1Q 2002 1Q 3002 1Q 4002 1Q 5002 1Q 6002 1Q 7002 1Q 8002 1Q 9002 1Q 0102 1Q 1102 1Q 2102 1Q 3102 % of Dedicated Institutional Investors % of Transient Institutional Investors Figure 1: Percentage of transient or dedicated institutional ownership to total institutional ownership over time. Dedicatedinstitutionalinvestors,asdefinedinBushee(1998,2001),arecharacterizedashavingconcentratedholdings in the firms in their portfolios with low turnover. In contrast, transient institutional investors are characterized as having diversified holdings in the firms in their portfolios with high turnover. Calculations based on: Thomson Reuters,Thomson Reuters Institutional (13f) Holdings/s34 Master File,WhartonResearchDataServices(WRDS), wrds-web.wharton.upenn.edu/wrds/about/databaselist.cfm 67

Panel A 20.0% 16.0% 12.0% 8.0% 4.0% 0.0% 0 1 2 3 4 5 6 7 8 9 10 11 12 snruteR dloh‐dna‐yuB Firms w/ More DED Investors Firms w/ More TRA Investors Firms w/ More DED Investors ‐ Firms w/ More TRA Investors Months Panel B 5.0% 4.0% 3.0% 2.0% 1.0% 0.0% ‐1.0% 0 1 2 3 4 5 6 7 8 9 10 11 12 snruteR lamronbA dloh‐dna‐yuB Firms w/ More DED Investors Firms w/ More TRA Investors Firms w/ More DED Investors ‐ Firms w/ More TRA Investors Months Figure 2: Monthly returns of firms with more transient or dedicated institutional investors. Panel A shows the buyand-hold monthly returns for firms with more transient institutional investors than the median firm and firms with morededicatedinstitutionalinvestorsthanthemedianfirm. PanelBshowsthebuy-and-holdabnormalreturnsbased on the Pastor and Stambaugh (2005) 5-factor model. Dedicated institutional investors, as defined in Bushee (1998, 2001), are characterized as having large average investment in firms in their portfolios and extremely low turnover. Incontrast,transientinstitutionalinvestorsareinstitutionalinvestorscharacterizedashavinghighportfolioturnover and highly diversified portfolio holdings. Calculations based on: CRSP, CRSP/CRSP Monthly Stock, Wharton Research Data Services (WRDS), wrds-web.wharton.upenn.edu/wrds/about/databaselist.cfm 68

Panel A Panel B 20.0% 16.0% 12.0% 8.0% 4.0% 0.0% 0 1 2 3 4 5 6 7 8 9 10 11 12 snruteR dloh‐dna‐yuB 6.0% Firms w/ More Long Horizon Investors Firms w/ More Short Horizon Investors Firms w/ More Long Horizon Investors ‐ Firms w/ More Short Horizon Investors 5.0% 4.0% 3.0% 2.0% 1.0% 0.0% 0 1 2 3 4 5 6 7 8 9 10 11 12 Months snruteR lamronbA dloh‐dna‐yuB Firms w/ More Long Horizon Investors Firms w/ More Short Horizon Investors Firms w/ More Long Horizon Investors ‐ Firms w/ More Short Horizon Investors Months Panel C Panel D 20.0% 16.0% 12.0% 8.0% 4.0% 0.0% ‐4.0% 0 1 2 3 4 5 6 7 8 9 10 11 12 snruteR dloh‐dna‐yuB 5.0% Firms w/ More Focused Investors Firms w/ More Diversified Investors Firms w/ More Focused Investors ‐ Firms w/ More Diversified Investors 4.0% 3.0% 2.0% 1.0% 0.0% ‐1.0% ‐2.0% ‐3.0% 0 1 2 3 4 5 6 7 8 9 10 11 12 Months snruteR lamronbA dloh‐dna‐yuB Firms w/ More Focused Investors Firms w/ More Diversified Investors Firms w/ More Focused Investors ‐ Firms w/ More Diversified Investors Months Figure 3: Monthly returns of firms across different institutional investor characteristics. Panels A and B show the buy-and-hold monthly returns and abnormal returns, respectively, for firms with more long-horizon institutional investors than the median firm and firms with more short-horizon institutional investors than the median firm. Panels C and D show the buy-and-hold monthly returns and abnormal returns, respectively, for firms with more focused institutional investors than the median firm and firms with more diversified institutional investors than the median firm. Abnormal returns are calculated relative to the Pastor and Stambaugh (2005) 5-factor model. Calculations based on: CRSP, CRSP/CRSP Monthly Stock, Wharton Research Data Services (WRDS), wrdsweb.wharton.upenn.edu/wrds/about/databaselist.cfm 69