Information Production by Institutions around CEO. Turnovers and Institutional Trading as a Corporate. Governance Mechanism

Transcription

1 Information Production by Institutions around CEO Turnovers and Institutional Trading as a Corporate Governance Mechanism Thomas J. Chemmanur Gang Hu Yingzhen Li Yuyuan Zhu March 3, 2017 Professor of Finance, Fulton Hall 330, Carroll School of Management, Boston College, Chestnut Hill, MA Phone: Fax: chemmanu@bc.edu. Associate Professor of Finance, Hong Kong Polytechnic University, School of Accounting and Finance, M1038, Li Ka Shing Tower, Hung Hom, Kowloon, Hong Kong. Phone: gang.hu@polyu.edu.hk. Associate, The Brattle Group, 201 Mission Street, Suite 2800, San Francisco, CA 94105, USA. Phone: Yingzhen.Li@brattle.com. PhD Candidate, Fulton Hall 154C, Carroll School of Management, Boston College, Chestnut Hill, MA Phone: zhuyh@bc.edu. For helpful comments and discussions, we thank Agnes Cheng, Lora Dimitrova, Ronald Masulis, Richard Sias, Xuan Tian, seminar participants at Babson College, Boston College, Hong Kong Polytechnic University, Suffolk University, and University of Rhode Island, and conference participants at the European Finance Association (EFA) Meetings in Cambridge UK, the Financial Management Association (FMA) Meetings in Chicago, and the Southern Finance Association (SFA) Meetings in Puerto Rico. We alone are responsible for any remaining errors or omissions.

2 Information Production by Institutions around CEO Turnovers and Institutional Trading as a Corporate Governance Mechanism Abstract Motivated by the theoretical literature (e.g., Edmans and Manso (2011)) postulating that information production and trading by institutions serves as a corporate governance mechanism, in this paper we use proprietary institutional trading data to empirically analyze information production by institutions in the specific context of CEO turnovers. We study how institutions produce information and the settings in which information production by institutions is most effective. We find that institutional trading prior to a CEO turnover has significant predictive power for the nature of the CEO turnover (forced versus voluntary). Further, institutions produce information partly by analyzing insider trading prior to a CEO turnover, though they are able to produce additional information independently as well. Trading by institutions after a forced CEO turnover with an insider as successor CEO has significant predictive power for subsequent long-run stock returns, and realizes significant abnormal profits. Overall, our results are consistent with the notion that information production by institutions, and their trading making use of this information, is able to play an important role in corporate governance in the setting of CEO turnovers. JEL Classification: G23, G34 Keywords: Corporate Governance; CEO Turnover; Institutional Trading; Information Production

3 1 Introduction The role played by institutional investors and their trading around various corporate events has recently become the topic of considerable interest in corporate finance. One of the most important such events in the life of a firm is the CEO turnover, where the CEO either is fired (forced turnover) or leaves the firm voluntarily (voluntary turnover), since, in either case, the turnover may have a significant long-term effect on the future performance of the firm. Further, the current performance of a firm may affect the likelihood of a CEO turnover in the immediate future (at least that of a forced CEO turnover). For example, on January 28, 2015, the board of directors of fast-food chain McDonald s Corporation announced that the current CEO, Don Thompson, would be stepping down after less than three years as CEO. Further, the board also elected Steve Easterbrook, who had previously led the firm s UK and European business units, to replace Thompson as president and CEO. Finally, Thompson s departure came after the firm had recorded five straight quarters of decline in sales and its worst monthly sales in more than a decade (and a string of scandals, such as the tainted meat scandal in China). 1 The above and other examples of CEO turnovers raise many interesting questions about the governance role of institutional investors around these turnovers. Institutions may play a governance role in CEO turnovers in one of two ways: shareholder activism ( voice ) or through selling the shares of underperforming stocks ( exit ). Our interest here is in the latter: i.e., in institutional trading or exit as a governance mechanism. In particular, recent theoretical papers have suggested that, when institutions choose to exit underperforming stocks, the sale of shares may serve as a governance mechanism (Admati and Pfleiderer (2009); Edmans (2009); Edmans and Manso (2011)). Edmans and Manso (2011) argue that the widely observed fact that most firms have multiple small blockholders (rather than a single large blockholder) may indeed be optimal, even though such a shareholder ownership 1 See, e.g., the news article, McDonald s CEO Is Out as Sales Decline, Wall Street Journal, January 28,

4 structure may be suboptimal from the point of view of shareholder activism. This is because, while splitting a block (i.e., having multiple small blockholders) reduces the effectiveness of direct intervention by shareholders (shareholder activism), it increases the power of the alternative corporate governance mechanism discussed above, namely, trading by institutions (or exit). By trading on any private information that they possess, institutions and other shareholders are able to move the firm s stock price toward its fundamental value, and thus cause it to more closely reflect the effort exerted by managers (the CEO) to enhance firm value. In such a setting, if the CEO shirks or extracts private benefits, institutions and other shareholders can follow the wall street walk of voting with their feet by selling their equity holdings to retail investors. This drives down the firm s stock price, reducing the CEO s equity compensation and thus punishing him ex post. However, while a number of theoretical papers have made arguments pointing to the importance of institutional exit as a governance mechanism, there has been relatively less empirical evidence directly documenting the ability of institutional investors to produce information relevant for corporate governance and the information flows involved that allow institutional trading to play a corporate governance role. The objective of this paper is to fill this gap in the literature: we use a detailed transaction-level institutional trading database to analyze information production by institutions and the information flows between institutions, firm insiders, and the underlying firm s stock price that enables institutional trading to play a corporate governance role in the specific context of CEO turnovers. We address the following specific research questions. First, are institutions (and their affiliated analysts) able to produce private information about the current CEO s performance and his ability to manage the firm going forward: i.e., whether the CEO will be forced to depart involuntarily (possibly due to the firm s poor performance under his management) or will serve longer, and depart voluntarily. Further, if indeed institutions are able to produce valuable information about CEO turnovers, what kind of institutions, namely, investment managers (e.g., mutual fund families such as Fidelity Investments) or plan sponsors (e.g., 2

5 pension plan sponsors such as United Airlines pension plan), are better able to produce such information? Second, if indeed institutions are able to produce such information relevant for corporate governance, how do they acquire such information: Do they acquire such information only by observing the trading behavior of board members and other top corporate officers in firms performing poorly? Alternatively, are they able to obtain additional information also by independently analyzing (with the help of equity analysts employed by them) the management ability of the CEO and the firm s performance under his management? Third, after a CEO turnover, are institutions able to produce information about the ability of the incoming CEO to manage the firm in the future? Further, are institutions able to produce more precise information when the CEO turnover is forced by the board (forced turnover) versus when the CEO turnover is part of a voluntary succession plan (voluntary turnover)? Furthermore, are institutions able to produce more precise information when the incoming successor CEO is a current firm insider (who had previously served with the firm in another capacity, as was the case for the CEO turnover of McDonald s Corporation discussed above) or when he is an outsider (joining the firm for the first time)? Finally, if institutions are indeed able to produce information superior to other market participants about firms undergoing CEO turnover and their future stock return performance immediately after the CEO turnover, are they able to realize significant abnormal profits (net of trading costs) by trading on this information? The results of our empirical analysis about the settings where institutions are able to produce more precise information about the future performance of the firm clearly have important implications for institutional trading and exit as a corporate governance mechanism (since it allows us identify the situations where this mechanism will be more effective). We make use of a detailed transaction-level institutional trading database provided by Abel Noser Solutions (formerly Ancerno Ltd., or Abel/Noser Corporation) to address the above research questions. We study the pattern of institutional trading prior to a CEO turnover to analyze whether institutions are able to predict the nature of a CEO turnover 3

6 (forced versus voluntary). We also analyze the lead-lag relationship between institutional trading and insider trading, to address the question of whether institutions rely at least partially on the trading behavior of insiders for their information production. In order to analyze whether institutions are able to produce precise information about the ability of the incoming CEO to manage the firm, we analyze whether institutional trading immediately after a CEO turnover has predictive power for subsequent long-run stock returns, whether this predictive power is greater after a forced versus a voluntary turnover, and whether this predictive power is greater for turnovers where the incoming CEO is an insider versus when he is an outsider. We also analyze whether institutions are able to realize abnormal profits from their trading subsequent to CEO turnovers, whether their profits (if any) are greater after a forced versus after a voluntary turnover, and whether or not these profits are greater when the incoming CEO is an insider versus when he is an outsider. Our data covers transactional-level trading spanning twelve years from January 1999 to December 2010 originated from 1,125 different institutions, with an aggregate annualized trading principal of around $10 trillion on all U.S. domestic equity. For an average CEO turnover event, our sample institutions collectively account for about 12% of the CRSPreported trading volume within the two-year period surrounding the turnover announcement. With this dataset, we are able to track institutional trading both before and after a CEO turnover. We are also able to compute realized institutional trading profitability net of explicit trading costs (i.e., brokerage commissions) and implicit trading costs (i.e., market impact). Throughout this paper, we use a variable we call Net Buy to measure institutional trading. We define Net Buy as the number of shares purchased by institutions minus the number of shares sold by institutions, normalized by the number of shares outstanding. Our paper provides a number of new results on the information production and governance role of institutional investors around CEO turnovers. We organize our empirical tests and results into four parts, corresponding to the four sets of research questions outlined above. First, we study the pattern and informativeness of institutional trading over 4

7 the one-year period before a CEO turnover. We find that institutional trading before CEO turnovers has considerable predictive power for the nature of CEO turnovers (forced versus voluntary). Greater selling by institutional investors is significantly associated with a higher probability of forced CEO turnovers. This result is robust to controlling for various variables that have been found in the prior literature to be able to predict forced CEO turnovers, including prior firm performance and insider trading. This suggests that institutional investors are able to produce valuable information about the current CEO s management ability and therefore the likelihood of a forced turnover. Further, our split-sample analysis of investment managers versus plan sponsors indicates that it is primarily investment managers who have such private information about the nature of a CEO turnover. Finally, the fact that institutions, especially investment managers, are able to produce valuable information about the management ability of the current CEO and the significant predictive power of institutional trading for the nature of CEO turnovers suggest that institutional trading is able to play a significant corporate governance role in the context of CEO turnovers. Second, we analyze, for the first time in the literature, the relationship between insider trading by corporate insiders, such as board members and top corporate officers, and institutional trading before CEO turnovers in a lead-lag framework. We find that institutional trading during the one-quarter period before a forced turnover is positively related to insider trading during the previous quarter. A 1% increase in (lead) insider trading is associated with about 0.20% increase in (lagged) institutional trading. We do not find such a result before a voluntary turnover. These results suggest that institutional investors learn from trading by corporate insiders before a forced turnover. Finally, we show that while institutional investors learn from insider trading, they are able to produce additional information independently about the CEO s ability to manage the firm in the future, since all our results on the informativeness of institutional trading hold even after we control for insider trading. Third, we study the predictive power of institutional trading immediately after CEO turnovers (six months) for the firm s subsequent long-run (one year) performance, for the 5

8 first time in the literature. We find that institutional trading immediately after a forced CEO turnover has considerable predictive power for the firm s subsequent long-run stock return performance: a 1% increase in institutional net buying after a forced turnover is associated with about 1.68% increase in the firm s abnormal stock return over the subsequent one-year period. This result is robust to controlling for various variables capturing publicly available information, as well as the extent of trading in the firm s equity by insiders. We find that there is no such predictive power of institutional trading subsequent to voluntary turnovers. Further, within forced turnovers, this predictive power of institutional trading for subsequent stock returns is confined to cases where the incoming CEOs is a firm insider. These findings suggest that institutional trading is able to play a corporate governance role after a CEO turnover most effectively in the case of a forced CEO turnover where the incoming CEO is a firm insider (rather than an outsider), given that this is the situation where institutions information advantage over retail investors is the greatest. Finally, we study the realized profitability of institutional trading around a CEO turnover, using actual transaction prices and net of brokerage commissions, for the first time in the literature. We find that institutions are able to realize positive abnormal profits around a forced turnover, even after taking commissions and other trading costs into account. For example, over the six-month and one-year horizon after a forced turnover, our sample institutions realize on average a risk-adjusted return (on maximum investment) of 1.72% and 1.59%, respectively. Further, within forced turnovers, these positive abnormal profits are realized only when the incoming CEO is a firm insider (but not when he is an outsider). In contrast, institutional investors do not earn abnormal profits around a voluntary turnover. These results suggest that the information advantage possessed by institutional investors around a forced turnover that we documented earlier translates into realized trading profits for the case where the incoming CEO is a firm insider. In summary, we are able to show that institutions not only possess an informational advantage over retail investors around forced CEO turnovers, but are also able to translate this advantage into realized trading 6

9 profits even after accounting for both explicit (brokerage commissions) and implicit (market impact) trading costs. Further, these results suggest that institutional trading is able to play a corporate governance role most effectively after forced CEO turnovers where the incoming CEO is an insider, since institutions are able to realize abnormal profits from their trading after such turnovers and therefore have the greatest incentive to engage in costly information production around such turnovers. The rest of the paper is organized as follows. Section 2 relates this paper to the existing literature and highlights its contribution relative to this literature. Section 3 describes the relevant theory and develops testable hypotheses about institutional trading around CEO turnovers. Section 4 describes the data and sample selection procedure and describes various variables used in this study. Section 5 presents our empirical tests and results. Section 6 concludes. 2 Relation to the Existing Literature and Contribution Our paper is closely related to three important literatures. The first is the literature on information production by institutions and how this allows institutions to play an important role in corporate governance through exit. The theoretical paper closest to our paper is Edmans and Manso (2011), who show that information production by competing institutions allows the firm s stock price to approach its fundamental value and serves to better incentivize firm managers (like the CEO) to exert effort to maximize its long-run value. 2 The small empirical and survey literature in this area is also related to our paper. Parrino, Sias, and Starks (2003) use quarterly holdings data to show that during the two-year period prior to a forced CEO turnover, institutions with larger fractional ownership are more likely to sell shares, which they interpret as evidence in favor of information-based trading, since larger institutions have a greater incentive to collect information about a firm s future potential, 2 Unlike Edmans and Manso (2011), Edmans (2009) and Admati and Pfleiderer (2009) analyze a setting with a single blockholder, and is therefore less directly related to the setting of our empirical analysis. 7

10 and have greater access to such information. 3 McCahery, Sautner, and Starks (2016) use survey evidence to show that institutions indeed use exit as their primary governance mechanism. Edmans, Fang, and Zur (2013) study block formation by a sample of activist hedge funds and analyze the effect of stock liquidity on governance. They show that stock liquidity increases the likelihood of block formation. However, conditional on acquiring a stake, liquidity reduces the likelihood that a blockholder governs through voice (activism) and increases the likelihood of governance through exit. Bharath, Jayaraman, and Nagar (2013) use the natural experiments of financial crises and decimalization of stocks to provide exogenous shocks to stock liquidity to empirically analyze exit as a governance mechanism, using the fact that this governance mechanism is weaker when stock liquidity is lower, and vice versa. Our paper contributes to the above literature by analyzing information production by institutions around CEO turnovers, which allows them to play a corporate governance role around such turnovers. In particular, we delve deeply into the information flow between institutions and the stock market; between institutions and firm insiders; the relative efficacy of information production by institutions around forced turnovers compared to that around voluntary turnovers; and finally, the realized profitability of information production and trading on this information by institutions, which impinges on their incentives to engage in costly information production in the first place. 4,5 The second literature our paper is related to is the accounting literature that studies how the accuracy of information produced by firm insiders or sell-side analysts affects CEO 3 Parrino, Sias, and Starks (2003) do not analyze any of the other research issues related to information production by institutional investors that we analyze here, perhaps because they use quarterly institutional holdings data for their analysis, which makes it difficult to study these issues. In addition to transaction-level institutional trading data, we also make use of insider trading data, which allows us to study the lead-lag relationship between institutional trading and insider trading around CEO turnovers for the first time in the literature. 4 A number of authors have developed theoretical analyses about the incentives of outsiders to acquire information about a firm (see, e.g., Grossman (1976), Grossman and Stiglitz (1980), Diamond and Verrecchia (1981), Verrecchia (1982), Admati and Pfleiderer (1986), and Bhushan (1989a)). Two finance papers developing theories of information production by institutions or analysts around corporate events are Brennan and Hughes (1991) and Brennan and Thakor (1990). 5 To the extent that we explore the link between institutional trading around CEO turnovers and insider trading, our paper is also related to the literature on insider trading around various corporate events: see, e.g., Li (2013). 8

11 turnover. 6 Engel, Hayes, and Wang (2003) document that accounting information appears to give greater weight in turnover decisions when accounting-based performance measures are more precise and more sensitive, relative to market-based performance measures. Farrell and Whidbee (2003) find that boards of directors use firm performance expectations (by analysts), in addition to the overall level of firm performance, to make CEO retention and firing decisions. They further show that, when the information signal from analysts is more precise, it receives more weight in boards decisions. Lee, Matsunaga, and Park (2012) argues that management forecast accuracy provides a signal regarding CEOs ability to anticipate and respond to future events, and finds a positive relationship between CEO turnover and absolute management forecast accuracy. Murphy and Zimmerman (1993) document that turnover-related changes in R&D, advertising, capital expenditures, and accounting accruals are due mostly to poor firm performance, rather than direct managerial discretion. Our paper contributes to the above literature by analyzing how the accuracy of information production by institutions is related to the nature of CEO turnovers (forced versus voluntary), the relation between information produced by institutions and the private information held by firm insiders, and the relative information advantage of institutional investors relative to retail investors around forced versus voluntary CEO turnovers. 7 The third literature our paper is related to is the broader literature on CEO turnovers. 6 There have been a number of empirical analyses regarding the extent of information production by analysts about a firm in a broader context (see, e.g., Bhushan (1989b), O Brien and Bhushan (1990), Lang and Lundholm (1996)) and the informativeness or accuracy of the information produced (see, e.g., Frankel, Kothari, and Weber (2006)). Much of the above empirical literature has focused on the cross-sectional variation in firm characteristics that lead to differences in information production by analysts and others across firms. 7 Our paper is also related to the broader empirical literature on institutional trading and information production by institutional investors around other corporate events. Gibson, Safieddine, and Sonti (2004) and Chemmanur, He, and Hu (2009) empirically analyze institutional trading around SEOs. Gallagher, Gardner, and Swan (2013) study short-term swing trades (i.e., sequence with three phases, for example, buy-sell-buy) and show that short-horizon informed trading by institutions disciplines firm management. Unlike our paper, their focus is only on short-term trading and they do not study institutional trading around CEO turnovers or any other corporate event. Further, our paper is also related to the broader literature studying the information content of institutional trading in a general setting (e.g. Yan and Zhang (2009)) and whether institutional trading increases price efficiency (e.g. Boehmer and Kelley (2009)). Ours is, however, the only paper other than Parrino, Sias, and Starks (2003) to study institutional trading around CEO turnovers, and the only one to make use of transaction-level institutional trading data. 9

12 In a recent paper, Pan, Wang, and Weisbach (2015) document that the stock return volatility of a firm goes up around a CEO turnover, which they interpret as reflecting incrased uncertainty about the firm s new leadership. They further show that post-ceo turnover, stock return volatility declines with CEO tenure, which they attribute to the market learning about the CEO s ability over time. Unlike our paper, they do not study information production or trading by institutions around CEO turnovers. Parrino (1997) provides evidence on the factors that affect the likelihood of a forced versus voluntary turnover and shows that the likelihood of a forced turnover increases with industry homogeneity. Borokhovich, Parrino, and Trapani (1996) study the announcement effect of CEO turnovers, and show that forced turnovers in which an outsider is appointed to replace a fired CEO are associated with a larger announcement return. Huson, Malatesta, and Parrino (2004) show that operating performance relative to other firms deteriorates prior to CEO turnovers, and improve after these turnovers: see Denis and Denis (1995) for somewhat similar results. Huson, Malatesta, and Parrino (2004) also show that the degree of improvement in a firm s operating performance after a CEO turnover is positively related to the level of institutional shareholdings, the presence of an outsider-dominated board, and the appointment of an outsider (rather than an insider) as CEO. Jenter and Kanaan (2015) show that, contrary to the prediction made by standard economic theories, CEOs are fired after bad firm performance caused by factors beyond their control. Ertugrul and Krishnan (2011) find significant variation in the prior stock returns of firms that dismiss their CEOs and study reasons why boards may dismiss CEOs early (i.e., in the absence of significant poor prior stock performance). 8 Our paper contributes to this literature by delving deeper into the consequences of information production and trading by institutions around CEO turnovers and the corporate governance 8 To the extent that activist investors are often involved in CEO turnovers, our paper is also indirectly related to the literature on shareholder activism. See Gillan and Starks (1998) and Karpoff (2001) for surveys of the shareholder activism literature. It is important to point out that focus of this paper is on institutional trading ( exit ) as a governance mechanism and not the role of shareholder activism by institutions and how such activism is related to CEO turnover: see, e.g., Helwege, Intintoli, and Zhang (2012) for an example of empirical research on shareholder activism and its effect on CEO turnover. Given the nature of our data, however, it is unlikely that our results are driven by the shareholder activism effects of institutional trading. 10

13 role played by institutional trading in this setting. In summary, ours is the first paper to analyze the information production role of institutional investors around CEO turnovers and its implications for institutional trading and exit as a governance mechanism. In contrast to our paper, much of the existing literature has focused on the activism role of institutional investors around CEO turnovers. While Parrino, Sias, and Starks (2003) also compare institutional selling of shares around forced CEO turnovers versus that around voluntary turnovers, all our other results are novel to the literature. In particular, ours is the first paper to analyze the predictive power of institutional trading after a CEO turnover for subsequent stock returns, and also the first paper to analyze the realized profitability of institutional trading around CEO turnovers. It is also the first paper to analyze the lead-lag relationship between institutional trading and insider trading. The results of our empirical analysis of information production and trading by institutions around CEO turnovers shed light on the corporate governance role of institutional trading in three important ways. First, when institutions are able to produce more precise information about the future fundamental value of the firm under a given CEO, the board is able to make more accurate decisions regarding CEO turnover, since the information produced by institutions gets reflected in the firm s stock price, thus enabling the board to make the right CEO turnover decisions. For example, the theoretical analysis of Chemmanur and Fedaseyeu (2016) shows that, when board members have access to a public signal (such as the stock price) and when this signal is more informative, the board makes better CEO turnover decisions from the point of view of shareholder value maximization. Our results showing significant selling of a firm s equity by institutions in the market immediately prior to a forced CEO turnover, and significant predictive power of pre-turnover institutional trading for the nature of the CEO turnover (forced versus voluntary) suggest that institutions are able to play the above information production role to a significant degree. Second, after a CEO turnover, if institutions are able to produce more precise information after a certain type of 11

14 CEO turnover (e.g., forced versus voluntary; insider versus outsider successor CEO), and this information about the firm s long-run value under that CEO gets reflected in the firm s stock price, then the CEO is provided with stronger high-powered incentives to enhance the firm s long-run value in some situations than in others. The results of our analysis on the predictive power of institutional trading after a CEO turnover for the firm s subsequent stock return performance suggest that institutions are able to perform a governance role through trading most effectively after forced CEO turnovers and within forced turnovers, primarily when the incoming CEO is a firm insider rather than an outsider to the firm. Third, our results showing that institutions are able to realize significant abnormal profits from trading on their private information around forced CEO turnovers suggest that institutions have strong incentives to engage in costly information production about the CEO s ability to manage the firm, which, in turn, enables institutional trading to act as an effective corporate governance mechanism. 3 Theory and Hypotheses Development In this section we will discuss the underlying theoretical motivation and develop the hypotheses we test in our empirical analysis. By analyzing how information production by institutions affect the firm s stock price and the interaction between the information produced by institutions and the private information held by firm insiders, we hope to be able to identify the channels through which institutional trading is able to serve as a governance mechanism in the specific context of CEO turnovers. While we are not aware of any formal theoretical models of information production by institutional investors specifically around CEO turnovers, models of information production by institutions around other corporate events can be adapted to the context of CEO turnovers. One such model is that of Chemmanur and Jiao (2011), who develop a theoretical analysis of information production and trading by institutions around seasoned equity offerings 12

15 (SEOs). 9 We adapt the model of Chemmanur and Jiao (2011) to analyze institutional trading around CEO turnovers. 10 The economic setting of this adapted model can be described as follows. Consider a situation in which a firm is managed by an incumbent CEO, and who is being evaluated in terms of his ability to manage the firm by both firm insiders (e.g. board members) and outsider investors (such as institutions). Firm insiders may come into possession of private information about the CEO s actions and ability to manage the firm, and may trade on this information in the stock market. 11 Outsiders in the equity market consist of two types of investors. The first type of investors are institutional investors, who have the ability to produce noisy information about the current CEO s ability to manage the firm going forward including the possibility of the firm undergoing a CEO turnover (either due to the CEO being forcibly replaced or due to his retiring voluntarily). In the case of a CEO turnover, institutions may also be able to produce information about the incoming (new) CEO, and his or her ability to manage the firm subsequent to the turnover. The precision of information produced by institutions may be lower than that of the private information held by firm insiders, so that, while information production is likely to help institutions reduce their information disadvantage with respect to firm insiders, it may not eliminate it. The second type of investors are retail investors, who do not have any ability to produce 9 In Chemmanur and Jiao (2011), institutions engage in costly information production about the firm s future performance around a SEO. Since the equity market consists of institutional investors and liquidity traders (retail investors), and market makers cannot distinguish between the orders coming from the two kinds of investors, the price of the firm s equity in the days prior to the SEO will only imperfectly impound the information produced by institutional investors. Firm insiders, who set the firm s SEO offering price, therefore are able to infer (but only partially) some of the information produced by outsiders (whether it is favorable or unfavorable) by observing the firm s stock price prior to the SEO and use it to infer the potential demand for the firm s equity in the SEO itself. 10 While it is relatively straightforward to adapt the Chemmanur and Jiao (2011) model to the specific context of CEO turnovers, we do not present such a formal model here due to space limitations. 11 If the CEO is of poorer quality, he is more likely to be fired by the board (i.e., the CEO turnover is more likely to be a forced rather than to be a voluntary turnover), since, in this case, the board may decide that the firm s performance is unlikely to improve under the current CEO s management. Board members may trade on their private information about the CEO s ability. While trading by such insiders is constrained by the relevant insider-trading regulations, there is nevertheless a significant amount of such trading. 13

16 information about the intrinsic value of the firm, and are therefore at a disadvantage with respect to both institutions and insiders. Retail investors are essentially liquidity traders in the equity market in the economic setting we study here, similar to their role in market microstructure models such as Kyle (1985). We further assume that institutions are able to observe insiders trading (to the extent that they have to report their trades as prescribed by the insider trading regulations); conversely, insiders are able to observe the change in institutions stock holdings in the firm. The price of the firm s stock in the equity market is set by a market maker who is uninformed to begin with, but who sets the stock price to break even (after observing the aggregate order flow of trades in the firm s equity), again similar to the price-setting rule in market microstructure models. While the market maker cannot fully separate informed trading (originating from either firm insiders or institutional investors) from uninformed trading (originating from retail investors), the price of the firm s equity will reflect, to some degree, the information held by institutional investors. The information flow between the firm, its insiders, and institutional investors and the resulting determination of the firm s stock price in the stock market (by the market maker) is depicted in Figure 1. Since the institutions producing information about a firm and its CEO will trade on this information and are more likely to sell equity in the firm if their information is unfavorable (and more likely to buy equity if their information is favorable), this means that the net buy (number of shares bought minus number of shares sold, normalized by the number of shares outstanding) prior to a turnover will have predictive power for the nature of CEO turnovers (forced versus voluntary). In other words, institutional net buy prior to the turnover will be algebraically lower in the case of forced rather than voluntary turnovers. In fact, this net buy may even turn negative, in which case institutions may sell more shares prior to a forced turnover rather than a voluntary turnover. This is the first hypothesis we will test here (H1). Further, if indeed institutions are able to produce valuable information about the firm and its CEO, those institutions such as investment managers that employ equity 14

17 Information contained in reported insider trading Information about changes in institutional holdings Insiders and Board Private information CEO and Firm Public information (e.g. Firm performance) Institutional Investors (Information Producers) Insider trading Public information about firm performance and other firm variables Institutional trading Stock Price (Set by market makers after observing total order flow from informed and uninformed trades) Figure 1: Information flow between the firm, insiders, institutional investors, and the stock market analysts (buy-side analysts) are more likely to be able to produce such information compared to plan sponsors or smaller institutions, which are less likely to employ such analysts (H2). If institutions are indeed able to generate valuable information about the management ability of a firm s CEO so that institutional trading has predictive power for the nature of a CEO turnover, an interesting question arises: What is the mechanism through which they acquire this information and the order of this information acquisition relative to that of the 15

18 relevant firm insiders such as corporate board members? Chemmanur and Fedaseyeu (2016) develop a model of corporate boards and CEO turnover where individual board members have private information about the CEO s ability, and the decision of the board to fire the CEO or retain him reflects the aggregation of board members private signals. 12 The empirical analysis of Li (2013) provides support for the above notion that board members indeed have private information about CEO ability by showing that insider trading by corporate board members also has predictive power for the nature of CEO turnovers. Given the above, there are two possibilities. One possibility is that institutions acquire their information only by observing and analyzing the trading behavior of corporate insiders such as board members. 13 Alternatively, institutions may be able to extract some information about CEO ability by analyzing trading by firm insiders and may produce additional information independently with the help of equity analysts employed by them. We test whether institutions are indeed able to obtain any valuable information about CEO ability by analyzing the pattern of trading by firm insiders (H3). In particular, we analyze the lead-lag relationship between insider trading and institutional trading. If indeed institutions are learning about the current CEO s ability by observing insider trading in the firm s equity, one interesting question that arises is whether they learn more from insider trading around forced CEO turnovers (H4A) or around voluntary turnovers (H4B). Given that insiders themselves can be expected to have less of an information advantage over other equity market participants around voluntary turnovers (since the succession plan involved in a voluntary turnover is largely observable by all market participants), one would expect such learning to occur to a greater extent around forced turnovers. We now turn to the question of whether institutions are able to produce superior information about the incoming CEO subsequent to a turnover. If institutions have more 12 Chemmanur and Fedaseyeu (2016) show, however, that the aggregation of board members private signals is often inefficient if the CEO can impose a cost on individual board members who spoke or voted against him in the board meeting. 13 According to the disclosure requirement of the current SEC insider trading regulation, a Form 4 must be filed within two business days following a change in ownership of securities or derivative securities (including the exercise or grant of stock options) for corporate insiders. 16

19 favorable information about the quality of an incoming (new) CEO, they are likely to buy more shares in the equity of the firm after the CEO turnover in order to profit from this private information. Thus, a greater net buy of the firm s equity by institutions after a CEO turnover will be positively related to subsequent long-run stock returns if institutions indeed are able to produce valuable information about the incoming CEO s ability. Thus, a larger net buy will reflect a more favorable assessment by institutions about the new CEO s ability. In conducting the above analysis, we control for the extent of insider trading (as well as variables related to prior firm performance), thus allowing us to assess whether institutions are able to generate valuable information about incoming CEO ability over and above that they may be able to obtain by analyzing insider trading by board members and other top corporate officers of the firm. An interesting question that arises in this context is whether, if institutions indeed have an information advantage over retail investors in assessing the incoming CEO s quality, it is greater around a forced CEO turnover or around a voluntary CEO turnover. Typically voluntary turnovers are part of a planned CEO succession so that even if institutions have an advantage over retail investors in assessing how the new CEO will perform after a voluntary turnover, this information may be staggered over time. Consequently, it is likely that a significant amount of information becomes publicly available about the incoming CEO around a voluntary CEO turnover so that it is easier for even retail investors to assess his ability. This, in turn, implies that institutions are less likely to be able to generate a significant information advantage over retail investors about the incoming CEO in the case of a voluntary turnover. On the other hand, in the case of a forced turnover, since the turnover is likely to be somewhat unexpected, there is likely to be a smaller amount of publicly available information about the ability of the incoming CEO to manage the firm after the turnover. This means that, in a forced CEO turnover, institutions are more likely to be able to generate a significant information advantage over retail investors about the future stock market performance of the firm under the new CEO during the period immediately following the 17

20 turnover. We therefore expect the predictive power of institutional trading to be stronger in the case of a forced CEO turnover compared to that in the case of a voluntary turnover (H5). We now analyze whether, within forced turnovers, the information advantage of institutional investors is greater when the incoming CEO is a firm insider (who is currently working within the firm in some capacity other than CEO) or an outsider to the firm. On the one hand, we would expect the total uncertainty regarding how the new CEO would manage the firm going forward to be greater for an outsider successor CEO compared to that for an insider successor CEO. This is because there is likely to be more information available to investors regarding the track record of an insider currently working as a top manager in the firm (albeit in a position different from that of a CEO), compared to that for an outsider (where there is no such information available at all about his ability to work within the firm as its future CEO). However, the total uncertainty faced by investors about a CEO s future performance in managing a firm can be split into two components. We refer to the first component as information-related uncertainty, by which we mean uncertainty that can be reduced significantly due to information production by institutions. We refer to the second component as residual (pure) uncertainty, by which we mean the portion of total uncertainty that cannot be reduced by information production on the part of institutions. 14 Whether the information advantage of institutions about CEO turnovers (and therefore the predictive power of institutional trading) is greater when the new CEO is an insider (H6A) or when the new CEO is an outsider (H6B) will depend not upon the total 14 It is useful to illustrate these two components of uncertainty with an example. Consider first the example Satya Nadella taking over CEO of Microsoft. Prior to Satya Nadella taking over as CEO, he was head of the cloud computing division of the firm. Given that he took over as CEO while he was an insider, it is likely that institutions could produce significant information enabling them to predict his future performance as CEO of Microsoft. Consider next a second example, namely, that of Meg Whitman taking over as CEO of Hewlett-Packard (HP). Given that she was joining as CEO from outside, and that the firm that she was previously CEO of (namely, Ebay) is in a somewhat different industry than HP, much of the total uncertainty facing investors about the future performance of HP is likely to have been residual (pure) uncertainty since it would have been hard for even institutional investors to predict how HP would perform under Meg Whitman. In other words, it is likely that the information-related uncertainty was greater in the case of CEO turnover at Microsoft while the total (and also residual) uncertainty was likely greater in the case of CEO turnover at HP. 18

21 uncertainty faced by institutions about a new CEO s future performance, but on whether the information-related component of total uncertainty is greater or less when the new CEO is an insider rather than in the case when he is an outsider. Next, we analyze whether institutions are indeed able to capitalize on any information advantage they may have over retail investors and realize abnormal profits around a CEO turnover by trading on this information advantage after the turnover. First, we analyze the realized profitability of institutional trading after a forced versus a voluntary turnover. If, as we hypothesized under H5, institutions are able to generate an information advantage over retail investors only in the case of a forced turnover, then we expect institutions to realize abnormal profits only by trading after a forced CEO turnover (H7). Further, if we compare realized institutional trading profitability after forced turnovers where the incoming CEO is an insider versus profitability where the incoming CEO is an outsider, the relative magnitude of these realized profits will depend upon whether the information-related component of uncertainty is greater when the incoming CEO is an insider (H8A) or it is greater when he is an outsider (H8B). It is worth pointing out here that, if the realized profitability of institutional trading is driven by information production around a CEO turnover, then we expect these empirical results on institutional trading profitability to be consistent with the findings of our empirical analysis on the predictive power of institutional trading (in other words, we expect H8A to hold if H6A holds; conversely, we expect H8B to hold if it is H6B that holds). Note that, whether institutions are able to make abnormal trading profits or not and the situations under which they are able to make greater profits gives us insight into their incentives to produce information and therefore into the situations in which institutional trading (or exit) is most effective in acting as a corporate governance mechanism. 19

22 4 Data and Summary Statistics 4.1 CEO Turnover Sample The CEO turnover data in this study is largely based on the ExecuComp database from Compustat. ExecuComp reports on an annual basis, among other information, the names of the CEOs of S&P 1500 companies (firms removed from the S&P 1500 Index but still trading continue to be covered by ExecuComp). Based on this information, we identify CEO turnovers during the period. We next search for the exact announcement date and reasons for each CEO turnover in Factiva. We remove CEO turnovers due to deaths and changes in control such as M&A and spin-offs. We follow Parrino (1997) and Huson, Malatesta, and Parrino (2004) to classify each CEO turnover to be either voluntary or forced. The classification is based on the following process. (1) If the Wall Street Journal or other business and trade press reports that the CEO was fired, terminated, forced out from the position, or departed due to policy differences, the turnover is classified as forced. (2) For the remaining cases, if the departing CEO is under the age of 60 and the news articles do not report the reason for the departure as death, poor health, or the acceptance of another position, the turnover is classified as forced. (3) If the news articles report that the CEO is retiring, but the firm does not announce the retirement at least six months prior to the succession, the turnover is classified as forced. (4) For those that do not fit into (1), (2), or (3), the turnover is classified as voluntary. Finally, for cases (2) and (3) we further investigate the turnover by reading news articles around the announcement date, so that potential misclassification errors can be minimized. After merging with CRSP, we have 426 forced CEO turnovers and 1,160 voluntary CEO turnovers from 2000 to 2009 (that is, 1,586 turnovers in total). Table 1 reports summary statistics of the CEO turnover data as well as other firm characteristics. There is no significant difference in firm size, defined as the logarithm of total assets, between firms in the forced turnover subsample and firms in the voluntary turnover 20