Securities Analyst Responses to CEO Charismatic Images:  A Symbolic Perspective

Material Information

Securities Analyst Responses to CEO Charismatic Images: A Symbolic Perspective
FANELLI, ANGELO ( Author, Primary )
Copyright Date:


Subjects / Keywords:
Analytical forecasting ( jstor )
Charisma ( jstor )
Chief executive officers ( jstor )
Investors ( jstor )
Mathematical dependent variables ( jstor )
Modeling ( jstor )
Recommendations ( jstor )
Shareholders ( jstor )
Stock markets ( jstor )
Uniformity ( jstor )

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright Angelo Fanelli. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Embargo Date:
Resource Identifier:
53207957 ( OCLC )


This item is only available as the following downloads:

Full Text




ACKNOWLEDGMENTS Despite collective belief, I believe a dissertation is a collective effort. Here, I wish to thank several people, for several reasons. More than a dissertation advisor, Henry Tosi has been a mentor, a counselor in my endless strivings with the European and American academia, an antidote to my recurrent theoretical babbling, and ultimately the major support I’ve received in 5 years – despite my continuing allegiance to anarchy. A similar fate occurred to Amir Erez and Rodney Lacey, who stoically endured my methodological and theoretical chaos. Each of them has been the right person, at the right time, with the right kind of help. Dr. Algina’s professionalism and knowledge were the means to my realization that even Italians have a chance at learning quantitative methods – eventually. Vilmos Misangyi and Remus Ilies have helped me in several important ways over the course of many years. I remain indebted to both of them for their help and, above all, for their friendship and deep humanity. Mike Loree has been my statistical Prometheus: without his appearance on the scene, I would still be cursing I/B/E/S for making their data available only in SAS format. Above all, I would still be thinking that SAS is a Scandinavian Airline. Over the last 2 years, a number of teaching assistants have given me their time in exchange for esoteric “credit hours.” Of these, Steve Cohen and Dinesh Kalera gave their time and also their dedication and creativity. Without each one of these human beings, I would not have survived the University of Florida. Acta est fabula. ii


TABLE OF CONTENTS page ACKNOWLEDGMENTS..................................................................................................ii LIST OF TABLES..............................................................................................................v LIST OF ABSTRACT......................................................................................................................vii CHAPTER 1. INTRODUCTION.......................................................................................................1 2. THEORETICAL BACKGROUND.............................................................................5 Charismatic Leadership Theory...................................................................................7 Insiders and Outsiders................................................................................................11 Social and Psychological Dynamics of the Stock Market.........................................16 CEO Charismatic Images (CCI)................................................................................20 Hypotheses of the study.............................................................................................25 Analyst Recommendations: Favorability..............................................................25 Analyst Recommendations: Uniformity...............................................................32 Analyst Forecast Accuracy...................................................................................35 3. METHODS................................................................................................................38 Sample and Data........................................................................................................39 Dependent Variables..................................................................................................41 Favorability of Analyst Recommendations..........................................................41 Uniformity of Analyst Recommendations............................................................41 Forecast Accuracy.................................................................................................42 Independent Variables...............................................................................................43 CEO Charismatic Image (CCI).............................................................................43 Charisma (factor scores).......................................................................................49 Past Performance..................................................................................................51 Outsider Status......................................................................................................52 CEO Reputation....................................................................................................53 iii


Control Variables.......................................................................................................53 CEO Controls........................................................................................................53 Firm Controls........................................................................................................54 Analyst Controls....................................................................................................54 Analytical Methods...................................................................................................55 4. RESULTS..................................................................................................................59 5. DISCUSSION AND CONCLUSIONS.....................................................................72 Theoretical Implications............................................................................................75 Implications for Practice............................................................................................79 Some Caveats for Future Research............................................................................80 APPENDIX A CONCEPTUAL NODES..............................................................................................82 B CONCEPT CATEGORIES AND SEARCH DICTIONARIES...................................84 C ITEMS OF THE CONGER-KANUNGO SCALE.......................................................88 LIST OF REFERENCES..................................................................................................90 BIOGRAPHICAL SKETCH............................................................................................99 iv


LIST OF TABLES Table page 2-1. Internal members vs. stock market actors..............................................................12 2-2. Construction of CCI scores.....................................................................................45 3-1. Correlation Matrix, CCI and Conger-Kanungo subscales......................................51 4-1. Parameter estimates and variance components for the null model.........................59 4-2. Descriptives and correlations, level2......................................................................60 4-3. Parameter estimates for Favorability of Recommendations (unstandardized coefficients in parentheses)......................................................................................62 4-4. Parameter estimates for Uniformity of Recommendations (unstandardized coefficients in parentheses)......................................................................................66 4-5. Parameter estimates for Forecast Accuracy (unstandardized coefficients in parentheses)..............................................................................................................68 4-6. Charisma factor, parameter estimates for all dependent variables (unstandardized coefficients in parentheses)......................................................................................70 C-1. Items from the C-K scale used in the study...........................................................88 C-2. Items from the C-K scale used in the study (continued)........................................89 v


LIST OF FIGURES Figure page 2-1 Favorability of analyst recommendations................................................................26 2-2 Uniformity of analyst recommendations..................................................................32 2-3 Analyst forecast accuracy.........................................................................................35 vi


Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy SECURITIES ANALYST RESPONSES TO CEO CHARISMATIC IMAGES: A SYMBOLIC PERSPECTIVE By Angelo Fanelli August 2003 Chair: Henry L. Tosi, Jr. Major Department: Management This dissertation focuses on the effects of CEO charisma on the stock market. Although the business press often insists on the importance of CEO charisma for firm performance, the academic literature presents few studies on this subject, and none on the relationship between CEO charisma and the stock market. The study focuses on how CEO charisma affects an important antecedent of stock market performance, the perceptions and judgments of securities analysts. From a theoretical viewpoint, the dissertation extends and integrates Charismatic Leadership Theory (CLT) by focusing on external stakeholders (i.e., securities analysts) rather than internal organizational members. Moreover, the dissertation develops a conceptualization of CEO charisma, CEO Charismatic Image (CCI), based on a text analysis of letters to the shareholders. The hypotheses of the study focus on three dependent variables of interest: the favorability of analyst recommendations, the vii


uniformity (standard deviation) of analyst recommendations, and the forecast accuracy of individual analysts. The sample of the study comprises 367 US firms undergoing a CEO succession in the period 1990-1999. For each of these firms, data on favorability, uniformity and forecast accuracy were collected. Letters to the shareholders were content-analyzed to obtain CEO Charismatic Images scores. Data on control variables at multiple levels (analyst, CEO, firm) were also collected. The main result of the study is that CEO charisma significantly affects the perceptions of securities analysts in ways that lead them to individually recommend the firm’s stock to investors in a more favorable way; and, collectively, to present to investors more uniform judgments. These effects, however, are not accompanied by increased accuracy in forecasting the future performance of the firm on the part of the analysts. CEO charisma represents therefore a problematic phenomenon in the stock market, possibly leading to investor overconfidence. viii


CHAPTER 1 INTRODUCTION With Dunlap staying on, he gets to do what he does best: turn around underperforming companies and make them additive to earnings. Moreover, the oft-questioned longevity of this story is put to rest; he will look to build shareholder value right here. [] We are reiterating our strong buy rating Analyst report, March 2 nd , 1998 In our opinion, Dunlap’s removal is a long term positive for Sunbeam. The credibility lost under Dunlap may never have been regained. [] In our view, this company has a wealth of opportunity [] and the removal of Dunlap is a step toward realizing that opportunity [rating: hold]. Same analyst’s report, June 15 th , 1998 This study focuses on the effects of CEO charisma on the stock market. In the last two years, CEOs and their charisma have been at the center of a heated debate in the US, especially within and around the stock market. As The Economist points out, “naive investors [] chose to follow stars--such as Anita Roddick of Body Shop and Asil Nadir of Polly Peck--without looking too closely at the numbers. Then, as now, it often ended in tears” (Economist, 2002). Although almost 5 years old, the story of Al Dunlap is fairly representative of the swinging moods of the American public toward charismatic CEOs. Worshipped by analysts, investors, and journalists alike, “Chainsaw” Al epitomized for a long time the charismatic CEO. A panoply of acclamations reiterated his “supernatural, superhuman, or at least specifically exceptional powers” (Weber, 1947: 358): a “pan-industrial leader,” a “pugnacious renegade,” “Nuclear Al,” “Equity Al, money in the bank.” At his peak, Dunlap was even praised as the source of "a new verb, ‘to dunlap,’ and a new gerund, ‘doing a dunlap’” (Carleen, 1997; Kadlec, 1998; Sellers, 1996: 69; 1


2 1998: 118; Wechsler, 1995: 44). Then, suddenly and abruptly, the charismatic hero fell, bringing a mass of disgruntled investors with him, and ultimately becoming “an incredible pig” (Palmer, 1999). A brief review of the business press of the last few years shows that the fascination of the public with charismatic CEOs all but disappeared since Al Dunlap’s catastrophic fall. According to Fortune, “like pornography, charisma is hard to define. But you know it when you see it” (Sellers, 1996: 69). In a more recent article, Fortune claims that ‘the skill most urgently needed in this economy [is] choosing the right CEO. [ . . . ] The fact is inescapable: these choices of single human beings exert enormous influence over entire enterprises. In the aggregate, they determine the prosperity of the nation’ (Charam & Colvin, 2000: 228). For the business magazine Across the Board, charisma “is life's gift to leaders who take daring action, win power, and use it” (Wareham, 1995). Particularly when it comes to the stock market, charismatic CEOs seem to represent the key to financial success for investors. As an example, the May 15 th , 2000 cover of Fortune shouted: ‘is John Chambers the best CEO on earth? Is it too late to buy his stock?’ (Serwer, 2000). The prophetic image attributed to some CEOs (Hegele & Kieser, 2001) is not simply a product of the business press, but also seems to be consistent with the perceptions of stock market actors. The analyst report from Bear & Sterns’ issued after Allen Questrom was nominated CEO at J.C. Penney’s is a good example of how stock market actors view charismatic CEOs: Allen Questrom, a proven winner. Allen Questrom assumed the role of CEO of J.C. Penney on September 15. He should bring new life to the company [ . . . ]. Without question, we think that Mr. Questrom is a powerful choice. [ . . . ] Mr. Questrom is both a leader and a visionary who is willing to take decisive steps forward. He


3 typically surrounds himself with the most capable team available so that he can continue to manage a company and not burden himself with every detail. (Bear & Sterns report, ‘In Allen (and Vanessa) We Trust’: Sept. 21st, 2000, pages 1, 4, 21. Vanessa Castagna is the firm’s COO). Despite such a widespread diffusion of similar attitudes toward charismatic CEOs among analysts and the popular business press, there’s only limited empirical evidence on the actual effects of CEO charisma on stock market performance (Tosi, Misangyi, Fanelli, Waldman, & Yammarino, 2002; Waldman, Ramirez, House, & Puranam, 2001). The purpose of this dissertation is to study how CEO charisma affects an important antecedent of stock market performance: the perceptions and judgments of securities analysts. Securities analysts were chosen because of their role as conveyors of information and investment suggestions to investors, fund managers, and the like (Zuckerman, 1999). Studying whether and how securities analysts are influenced by CEO charisma might shed some light on the effects of charisma on the stock market. Focusing on the effects of charisma on stock market actors constitutes a significant shift from the literature on charisma. So far, researchers have studied the effects of charisma on organizational members (e.g., House & Aditya, 1997; Lowe, Kroeck, & Sivasubramaniam, 1996). The effects on subjects lacking membership in the organization yet influencing significantly its long-term survival (e.g., stock market actors) remain largely unexplored. Such a lack of interest in the reactions of nonmembers to CEO charisma runs counter to the practical relevance of the issue. If anything, the current stock market turmoil underscores the importance of how and what CEOs communicate to the stock market. The emergence of a “social movement on corporate control” (Davis & McAdam, 2000) within the new “investor capitalism” (Useem, 1996) calls for firms to obtain commitment from powerful external audiences through the CEO’s vision and


4 persona (Arnold, 1988; Cheney & Christensen, 2001; Hambrick & Fukutomi, 1991). At a societal level, investor capitalism calls for the various audiences within the stock market (securities analysts, fund managers, and the like) to effectively search, exchange, and elaborate information in order to reach an accurate evaluation of firms. Understanding how CEO charisma influences these evaluations is a first step to avoid costly mistakes. The dissertation is organized as follows. Chapter 2 reviews Charismatic Leadership Theory (CLT) and discusses the concentration of the theory on internal members. The review is followed by an analysis of the features distinguishing organizational members from external stakeholders. As the analysis shows, these differences call for an extension and integration of CLT, so to be able to study the effects of CEO charisma on nonmember subjects. Chapter 2 then presents an analysis of the social and psychological dynamics of the stock market, followed by a discussion of the main construct of the study, CEO Charismatic Images (CCI). Such a redefinition of the construct of interest allows us to study charisma as an instance of symbolic management based on language and rhetoric. Finally, chapter 2 presents the hypotheses of the empirical study, focusing on three distinct dependent variables: favorability of analyst recommendations, uniformity of analyst recommendations, and forecast accuracy. Chapter 3 gives the sample, data, and methods of the study, while chapter 4 presents the results. Chapter 5 discusses the results, limitations, and conclusions of the study.


CHAPTER 2 THEORETICAL BACKGROUND Traditionally, the literature on charismatic leadership focused on the effects of charisma on organizational members, neglecting the consequences of charisma in the wider external environment of the firm (House & Aditya, 1997). The stock market is one of the contexts where, at least anecdotally, charisma seems to be having important effects. Yet, to this day, researchers have not addressed at all the issue of whether CEO charisma affects stock market performance. This study attempts to fill this gap, and focuses on an important antecedent of stock market performance, the reactions of securities analysts to organizational discourse projecting a charismatic image of the CEO. The need for a separate framework to study the external effects of CEO charisma emerges from the substantial differences between organizational members and stock market actors. The legacy of Charismatic Leadership Theory (CLT, from now on) is an extensive knowledge of how charisma operates among organizational members, yet this knowledge cannot be applied directly to the study of the effects of charisma on nonmember individuals, as shown by the limited number of (mostly qualitative) studies on leaders of nonbusiness organizations, such as political or religious leaders (e.g., Blasi, 1991; House, 1988). To study the external effects of CEO charisma it is necessary to acknowledge that the function of the executive is not just to manage organizational members; but also the external environment of the firm: executives represent the organization to outside stakeholders, rally support for the firm, obtain the resources necessary for the firm’s functioning, and engage in political dynamics with government 5


6 officials and other corporate actors (Pfeffer, 1981; Pfeffer & Salancik, 1978; Thompson, 1967). Assuming that CEO charisma generates organization-level effects (such as stock market reactions) only through its effects on organizational members overly reduces and simplifies the role of the executive to a mere coordinator of the inner workings of the organization. Focusing on what happens outside the boundaries of the firm requires then an extension and integration of CLT beyond its traditional concentration on internal members. The primary goals here are to understand how CEO charisma operates among stock market actors, what its effects are, and how these effects might differ from the effects on internal members. Indeed, outsiders lack three important features characterizing insiders: a task defined by the organization, a relationship of hierarchical dependence with the leader, and a communication network with the CEO at its center (House, 1977; House & Aditya, 1997). Once we take into consideration “followers” lacking these features, it becomes unclear what dependent variables should be used, as well as what are the processes involved in the generation of the external effects of CEO charisma. Hence, the need for a new theoretical framework. In developing the theoretical framework for this study, I first review briefly charismatic leadership theory and its main findings. Since CLT focuses mainly on internal actors, I then contrast analytically the former with external actors. After that, I discuss the major characteristics of the stock market, the segment of the external environment of interest in this study, as well as the role and main research findings on the task of the securities analyst. Such a specific analysis of the stock market and the securities analyst profession is required by the heterogeneity of the external environment


7 of the firm (Pfeffer & Salancik, 1978). Thus, to study the effects of CEO charisma on the external environment of the firm, it is necessary to specify which segment of the external environment is of interest, as well as its major characteristics. The next section then develops the theoretical framework of the study around the construct of CEO Charismatic Image (CCI from now on). The construct was developed by adapting the Conger and Kanungo model (1998) to the study of the external effects of charismatic leadership, and constitutes a re-definition of CEO charisma in symbolic terms. Last, I establish the bases for a symbolic theory of CEO charisma, presenting the hypotheses of the study. Charismatic Leadership Theory This section presents a brief review of the major definitions, consequences and processes addressed by CLT. As the review shows, one major characteristic of CLT is its concentration on internal organizational actors. Max Weber represents the unavoidable starting point for a review of the literature on charismatic leadership. Within the Weberian viewpoint (1947), charisma constitutes an ideal-typical form of authority distinguished from traditional and bureaucratic authority because its basis of legitimacy is neither in the status of the person who exercises authority nor in the rational/legal basis of authority. For Weber, charismatic authority rests on the “devotion to the specific and exceptional sanctity, heroism or exemplary character of an individual person, and of the normative patterns or order revealed or ordained by him” (Weber, 1947: 328). Hence, in the Weberian formulation, charisma is a form of authority, a characteristic of the social system stemming from the recognition by the followers of the exceptional qualities of the leader. Charismatic Leadership Theory (Conger & Kanungo, 1987, 1998; House, 1977; House & Aditya, 1997; House, Spangler, & Woycke, 1991; Klein & House, 1995;


8 Shamir, 1995; Shamir, Zakay, Breinin, & Popper, 1998; Waldman & Yammarino, 1999) modified Weber’s conceptualization of charisma, defining charismatic leadership as a relationship characterized by specific leader behaviors (such as articulating a vision of change, communicating high expectations from subordinates, making sacrifices, etc.) and subordinates’ responses (such as trust and admiration for the leader, unquestioning acceptance and respect, etc.). For instance, Waldman et al. (2001: 135) defined charisma as “a relationship between an individual (leader) and one or more followers based on leader behaviors combined with favorable attributions on the part of the followers [emphasis in the text].” It is important to understand that CLT embeds the charismatic relationship within an organizational context, whereby the leader and the followers share a common membership in the organization. This "organizational embeddedness" of charisma originated from the theory's attempt to apply to business organizations concepts developed in the political realm (Shamir 1995), and redefined charisma as a coordination mechanism operating within hierarchical authority, rather than in opposition to it, as the original formulation postulated (Weber, 1947). Empirical studies followed suit, focusing primarily on the effects of charisma on the members of the organization (Bycio, Hackett, & Allen, 1995; Hater & Bass, 1988; Howell & Avolio, 1993), or embedding experimental subjects within an organizational membership shared with the leader (e.g., Howell & Frost, 1989). This "organizational embeddedness" of charisma narrows the scope of research within the boundaries of the organization, as exemplified by Yukl & Van Fleet’s (1992: 174) definition of transformational leadership as “the process of


9 influencing major changes in the attitudes and assumptions of organization members [my emphasis]”. This “internal focus” of CLT led scholars to concentrate on the effects of charisma on task performance and the legitimacy enjoyed by the leader. Located at the individual and group level, these two dependent variables are present in most of the empirical research on the consequences of charisma (for a review of empirical results cf. DeGroot, Kiker, & Cross, 2000; Lowe et al., 1996). The theoretical argument is that charisma raises followers’ awareness of the importance of the collective goals, thereby increasing motivation, and ultimately task performance and the favorability with which subordinates and superiors view the leader (internal legitimacy). The effects of charisma on nonmember individuals, particularly on stock market actors, remain so far a completely uncharted territory. In terms of the processes involved in these effects, the precise dynamics of charisma are still subject to a wide discussion (House & Aditya, 1997; Judge & Ilies, 2002; Yukl & Van Fleet, 1992). The emerging, but still incomplete, consensus is that two processes link charisma to its consequences: affective processes and cognitive processes. That affect is involved is intuitive, given Weber’s (1947: 359-360) discussion of charismatic authority as “an emotional form of communal relationship.” Indeed, it would be impossible to conceive of behaviors such as casting aside one’s immediate self-interest and “making personal sacrifices in the interest of some [collective] mission” (House et al., 1991: 364) were it not for the operation of some type of emotional process. Linking leader and follower affective processes, Judge and Ilies (2002) argued that charismatic leaders experience positive emotions more often and more strongly, as well as higher


10 positive affect. This heightened affective profile, they argue, activates emotional contagion, whereby followers “catch” the leader’s affective states. Similarly, Kelly and Barsade (2001: 107) argue that “a good sender of emotions [in] an important, visible, or central position in a group may influence the degree to which a group experiences similar levels of affect.” This active emotional contagion going from the leader to the followers triggers heightened affective states in the latter, resulting in higher persistence, intensity of effort (Judge & Ilies, 2002), and liking. Like social similarity, affective similarity triggers compliance (Cialdini, 1995; Moscovici, 1985), that is loyalty, admiration, and task performance. Thus, it is clear that one necessary condition for these effects to occur is the sharing of leader and followers of a common organizational belonging (and therefore of a common goal and identity), as well as of a common and uninterrupted communication network. Conger and Kanungo (1998) identified three successive stages of a cognitive process involved in the emergence of charisma: evaluation of the status quo, formulation and articulation of organizational goals, and means to achieve the vision. The stages correspond to the leader’s “visionary behaviors” and operate as a cognitive framing that influences followers’ decisions (Fairhurst, 1993; Gardner & Avolio, 1998: 48). Although each of the three stages generates emotional reactions (Tommerup, 1990; Wasielewski, 1985), Conger and Kanungo emphasize their cognitive nature – as an assessment of the actual (evaluation of the status quo), an image of the possible (formulation and articulation of goals), and a clear connection between the two (means to achieve). In other words, through the vision, charismatic leaders influence follower decisions on a specific course, along a “conceptual roadmap”, a cognitive frame, or schema that orients


11 their behaviors and generates compliance (DiMaggio, 1997; Olsen, 1993; Tichy & Devanna, 1986). Again, a necessary condition for such a cognitive framing to occur is a common organizational goal, as well as a task defined by the organization. In conclusion, according to CLT charisma affects the task performance of the subordinates and internal legitimacy enjoyed by the leader by mobilizing followers’ effort through affective process as well as by directing decisions through a cognitive framing. Insiders and Outsiders This section lays the foundations for a theory of the external effects of CEO charisma by illustrating the differences between organizational members and stock market actors. Since the external environment is comprised of different populations, with different interests and cognitive frames (Powell & DiMaggio, 1991; Thompson, 1967), I will limit the analysis to a comparison of organizational and stock market actors, particularly securities analysts. As the next discussion shows, these two populations are qualitatively different. Therefore, research on the external effects of charisma cannot proceed by simply extending the existing theory to the external domain of the firm. As Table 2-1 suggests, insiders and outsiders differ along three major axes: their task (decision making vs. judgment), the relationship established with the top executive (dependence vs. interdependence) and the position of the CEO within the communication network (central vs. decentralized). As the first row of Table 2-1 shows, the first distinction between insiders and outsiders refers to their tasks, with organizational members primarily involved in performing a set of activities mandated by the organization (March & Simon, 1958) and


12 external actors primarily evaluating the firm’s performance potential according to external criteria. Table 2-1. Internal members vs. stock market actors Organizational members Stock market actors Task Decision making (performing on the job) Judgment (evaluating the firm) Relationship with the Leader Asymmetry (dependence) Symmetry (interdependence) Network position of the CEO Central (one-to-many) Decentralized (many-to-many) The two tasks correspond to the distinction between decision making and judgment tasks (Stevenson, Busemeyer, & Naylor, 1990). The former is a choice task where the alternatives constitute to a set of possible consequences considered valuable to the decision maker, while the latter is a choice task where the alternatives constitute categories for classifying stimuli. In other words, decision-making corresponds to choosing between different future consequences for the person who decides, while judgment entails classifying information into meaningful categories with little or no consequences for the person who is making the judgment. As an example of the decision vs. judgment distinction, an employee performing a task constitutes an instance of decision-making, while a superior evaluating his/her performance constitutes an instance of judgment. Similarly to performance evaluation, stock market evaluations are judgment tasks. As Scott (1998: 354) noted, “the same components are applicable whether the intent is to evaluate an individual performer or an entire organization.” Evaluating firms on the stock market is a judgment task that involves assigning a price to the firm’s stock according to the expected future income associated with ownership, adjusted for risk (Davis & McAdam, 2000: 198). Stock market actors such as investors, fund managers, and securities analysts evaluate the firm’s potential by classifying a given set of stimuli


13 (e.g., information about the firm) in meaningful categories (e.g., under-evaluated stock, over-evaluated, etc.). These differences are also evident in the actual activities of organizational and stock market actors. In their day-to-day life, organizational members execute a set of task activities in coordination with other organizational actors in order to achieve results that are valued by the organization as well as by the performer himself/herself. In such a context, “the great majority of people in lower echelons of the organization are not in the position to evaluate proposals for major organizational changes in detail or to judge the merits of organizational policies” (Shamir, 1995: 23). Indeed, such an evaluation does not even fall within the span of the legal and psychological contract linking employees to the organization (Simon, 1945). The very goal of stock market actors, in contrast, is to reach a judgment on the firm’s potential. Such a judgment is constructed by analyzing information about the firm that includes, among other things, the very organizational policies and strategies set forth by the CEO. Also, the judgment is constructed according to a classificatory scheme or evaluation model that is established outside the firm itself. Moreover, these task activities are conducted so as to achieve results (i.e., an evaluation of the firm) that are inherently conflicting with the interests of the organization that is being evaluated. These differences imply that the study of the external effects of CEO charisma must rely on dependent variables that are qualitatively different from task performance and hinge instead on the content of the stock market evaluations of the firm. The second row of Table 2-1 outlines a relational difference between organizational and stock market actors, with the former involved in an asymmetrical (authority)


14 relationship with the leader and the latter involved in a symmetrical one (interdependence). As Simon (1945: 177) put it, “of all the modes of influence, authority is the one that chiefly distinguishes the behavior of individuals as participants of organizations from their behavior outside such organizations.” In this this context, organizational members share a common goal and a common organizational identity with other members, as well as with the leaders of the organization. It is well known that such a sharing of goals and identity informs the behaviors of organizational actors and constitutes a major coordination mechanism (Simon, 1945). Indeed, many of the theoretical propositions and empirical findings of CLT hinge upon such mechanism. As an example, House (1977) argued that charismatic leadership affects follower behaviors by developing an affective influence on followers, leading to their identification with the leader and, in turn, to loyalty and commitment to the organizational goal. On the stock market, in contrast, the relationship is symmetrical – that is, external actors and firm executives are “well informed on and influenced by the other” (Useem, 1996: 207). As exemplified by CEOs and securities analysts, their relationship is one of reciprocal interdependence (Thompson, 1967). On one hand, CEOs depend on financial analysts, for they channel to the investor public recommendations affecting stock prices (Beneish, 1991; Branson, Guffey, & Pagach, 1998; Francis & Soffer, 1997) and, ultimately, CEOs’ reputations, careers, and compensation schemes (Davis & McAdam, 2000; Puffer & Weintrop, 1991). On the other, analysts depend on CEOs for critical resources that have a remarkable influence on analysts’ reputation, career and compensation (Hong, Kubik, & Solomon, 2000; Stickel, 1992): information (Angwin & Peers, 2001; Clemente, 1988;


15 Pulliam, 2002; Talley & Munk, 2002) and investment banking business (Hayward & Boeker, 1998). This relational difference has an important implication. Internal members share with the leader a goal and an identity, as well as directly bear the very consequences of their actions. Stock market actors, in contrast, do not share a goal or an identity with the leader, and their judgments are informed by the anticipation of consequences that are in direct contrast of interests with the leader and the organization itself (Davis & McAdam, 2000). In other words, when deciding whether to invest a firm, stock market actors do not take into account the welfare of the organization or of the leader him/herself but, rather, their own financial interest. Charismatic invocations of identity and commonality of purpose, in this last case, do not operate as they do with organizational members, where “the charismatic leader uses language to create meaning and a shared identity and community vocabulary (Conger & Kanungo, 1998: 172).” Thus, from the viewpoint of CLT, charisma cannot affect stock market actors because there is no common bond of identity linking the leader with his/her external followers. Research, in this case, must focus on alternative processes linking charisma to “follower” responses, unless the sole mechanism linking charisma to stock market performance is the internal performance of the firm. This link, however, is supported by research only with limited evidence (Waldman et al., 2001). The last feature to consider refers to the communication networks of internal and external actors. Indeed, communication is central to charisma (e.g., Conger & Kanungo, 1998; Gardner & Avolio, 1998), so we should take into consideration the characteristics of the communication networks established by the leader with his/her external


16 “followers” as opposed to the networks established with organizational members. As the last row of Table 2-1 shows, from a communication standpoint the major difference between organizational and stock market actors refers to the position of the charismatic leader within the communication network. Within the organization, the CEO enjoys a central position, so that all followers are essentially in contact with him/her and their attention is, at all times, focused on the leader, regardless of the “social distance” separating them from the leader (Shamir, 1995). In contrast, such a condition is not replicated within the volatile communication networks of the stock market, where at any point in time multiple actors evaluate multiple firms and multiple leaders (Thompson, 1967; Zuckerman, 2000). In such situations, the CLT paradigm is at a loss: follower attention is neither conceptualized theoretically, nor operationalized. Furthermore, we can legitimately expect different processes at work as well as different outcomes. In conclusion, internal and stock market actors differ in their relationships with the leader because the environments in which they operate are different. In order to study how charismatic leadership might affect external actors, it is necessary to take into account the characteristics of the specific segment of the external environment that is of interest.. In order to specify the segment of the external environment of interest for this study, the following section addresses the characteristics of the stock market and on the securities analyst profession. Social and Psychological Dynamics of the Stock Market This section focuses on the stock market and the securities analyst profession. From the viewpoint of the present study, the major feature of the stock market is its social nature. As White (1981: 518) argued, “markets are self-reproducing social structures among specific cliques of firms and other actors who evolve roles from observations of


17 each other’s behaviors.” The view of the stock market as a social structure assumes that actors (executives, analysts, and investors) are boundedly rational and has three important implications: evaluation occurs along institutionalized cognitive categories, actors engage in political and symbolic behaviors, and competition occurs along social as well as technical criteria. First, this view of the stock market suggests that interactions occur within and across socially meaningful categories, such as those defined by industry boundaries (Porac, Wade, & Pollack, 1999; Zuckerman, 1999) or type of security (Baker, 1984). In other words, before an evaluation can even be constructed, the judging actor must decide to which category the evaluation object (e.g., firm) belongs. Such a categorization conditions the volatility and absolute value of stock prices even before an evaluation on the firm is formed. Being aware of this phenomenon, executives typically attempt to manipulate the categories applied to their firms. For instance, in a study of industry categories Porac et al. (1999) showed that executives justify CEO pay by strategically expanding or contracting the categories used to compare their pay allocations to ‘peer companies.’ One route through which CEO charisma might therefore influence the categories which stock market actors apply to the firm is by defining the firm’s identity in ideological or moral terms (Shamir et al., 1998). An example of this type of effects is Mary Kay Ash’s vision for Mary Kay cosmetics: “we’re not in the cosmetics business, we’re in the people business. [ . . . ] In other words, our whole reason for existence is to give people the opportunity to enrich their lives” (Biggart, 1989: 113). Second, the social character of the stock market implies that executives and stock market actors engage in political behaviors and form political coalitions organized around


18 common ideologies. As Davis & Thompson (1994) showed in their historical analysis, the US stock market witnessed an increased activism in the s on the part of fund managers and advisors. This collective movement induced a constant struggle within the market and the legislative arenas about corporate control and governance. In such a context, executives have for a long time tried to reassure and manage shareholder perceptions. Westphal and Zajac (1998), for example, found that executives may declare, but not actually implement, the adoption of Long Term Incentive Plans so as to suggest that corporate governance issues are under control. Symbolic management of this type, they concluded, is a powerful way to influence the perceptions and evaluations of stock market actors: “firms can also influence market reactions and thus change their underlying market value through the use of symbolic action. Market reactions thus should perhaps be viewed more in terms of ‘soft’ numbers that reflect the subjective perceptions of a heterogeneous audience, neatly quantified and aggregated [ . . . ], reacting to changes in formal policy that may be independent of substantive practices” (Westphal & Zajac, 1998: 130). Third, competition on the stock market occurs along social, rather than merely technical criteria. “By adhering to the standards of the most substantial market participants” (Davis & McAdam, 2000: 198), executives acquire a favorable status, with weak links to the underlying quality of the financial asset (Podolny, 1993; Westphal, Gulati, & Shortell, 1997; Westphal & Zajac, 1994). Such a weak coupling of social and technical criteria derives from the boundedly rational nature of investors, that drives them to base their decisions on signals, rather than on the actual quality of a stock, which is unobservable, ambiguous and contradictory (Meyer & Gupta, 1994). Thus, the status


19 ordering of firms in the market determines who, and to what extent, gains access to market rewards such as higher stock prices (Zuckerman, 1999). The result is that financial goods themselves have a social nature: the value of a stock does not depend exclusively on underlying exogenous and independent qualities such as firm performance or employee effort and motivation, but rather on how actors view and respond to each other’s beliefs about these qualities. In this context, anecdotal evidence suggests that CEO charisma might act as a signal of the status of a firm, affecting how investors and other stock market actors perceive the firm, particularly when a succession is imminent. As a money manager investing in Lucent stated, “people aren’t really paying that much attention to second-quarter numbers. They want to know who the next CEO is going to be” (quoted by Rosenbush & Borrus, 2001: 104). These three characteristics of the stock market are also observable within the securities analyst profession. Securities analysts are organized and specialize along industry boundaries. As Zuckerman suggested, the industry specialization of securities analysts had a role in triggering the wave of de-diversification occurred in the US in the nineties: in a world of analysts specialized by industry, conglomerate firms ‘contradict the logic of the accepted structure of valuation’ (Zuckerman, 2000: 595). Second, competition among analysts assumes a social character: analysts initiate and terminate coverage in cascades, imitating each other (Rao, Greve, & Davis, 2001), and frame their recommendations so as to protect their reputation (Hong et al., 2000). Last, political behaviors occur on both sides of the relationship between securities analysts and corporate executives: when the latter are subject to negative reports, they ‘lash back at the


20 analysts themselves’ (Pulliam, 2002). Analysts, on their part, often ‘teach [their] clients a lesson’ by withdrawing coverage (Smith & Anand, 2002). In conclusion, what is commonly seen as an eminently rational and technical task, securities analysis, is instead a highly social domain, heightening, rather than reducing, the social nature of financial goods. In such a context, it is plausible that CEO charisma might affect securities analysts’ perceptions and, ultimately, evaluations. As the previous discussion suggested, symbolic management by firm executives affects stock market actors because the latter must rely on signals in order to evaluate the underlying, unobservable, quality of the firm’s stock – a consequence of the information asymmetry that exists between internal and external actors (Fama, 1980). This process entails attending to various signals originated by the firm and selecting among them, those that are thought to be correlated with the future performance of the firm. That CEO charisma constitutes one of these signals is a major assumption of this study. In order to test if and how this might happen, however, it is necessary first of all to expand charismatic leadership theory’s definition of charisma in ways that make it applicable to the stock market and, second, to identify the relevant dependent variables. CEO Charismatic Images (CCI) CEO Charismatic Images (CCI) are the main construct of interest of this study, and refer to the degree to which organizational discourse describes the Chief Executive Officer in charismatic terms. Four major components of CCI are of interest: statements directly describing the CEO’s persona that is, a) his/her charismatic qualities, achievements or behaviors; and three types of statements describing CEO charisma indirectly, through illustrations of his/her CEO’s vision: b) charismatic evaluation of the


21 status quo; c) charismatic formulation and articulation of organizational goals; d) charismatic means to achieve the goals. Two main reasons justify the development of CCI as a distinct construct specifically measuring the charisma of the CEO as it is communicated to stock market audiences: the type of medium through which charismatic language is conveyed to stock market actors and the actual content of the message. As discussed in the first section, stock market actors differ from internal ones on both aspects; they have access to different media and their interest focuses on different content. First, stock market actors rarely have direct access to and experience of the actual CEO’s behaviors. In the case of securities analysts, for example, there might be occasions in which they experience a direct contact with the CEO, such as road shows or conference calls. Mostly, however, their evaluation of multiple firms forces them to rely mainly on organizational documents and communications (Clemente, 1988). Second, rather than the mere behaviors of the CEO, stock market audiences (and securities analysts in particular) are interested in evaluating how these qualities affect the future performance of the firm, as well as in evaluating the CEO’s vision. As Useem (1996: 154) put it, “to the outside world [ . . . ], the persona of the chief executive plays a role akin to that of the architecture of company headquarters. The image of the CEO and of the building’s faade speaks of the quality of the organization on the inside.” To use a metaphor, stock market actors are not interested in the architecture in and of itself, but in what the architecture means for the future of their investment in the firm. It follows that if the focus is on stock market actors, research must shift its attention from the leader’s actual behaviors and vision to how those behaviors and vision are


22 symbolically represented. One way in which this occurs is through organizational discourse about the CEO’s behaviors and vision. It is this discursive material that should be collected and analyzed in order to measure CEO charisma as it is conveyed outside the firm. Although this has been done quite infrequently throughout CLT, several scholars call for such an endeavor (Bass, Avolio, & Goodheim, 1987; House et al., 1991). As an example, Waldman and Yammarino (Waldman & Yammarino, 1999: 281) argued that “distant CEO charismatic behaviors may be better assessed by content analysis of speeches and other communications, as well as by verification of consistency in stories about the CEO as told by distant organizational members.” Conceptualizing CEO charisma in terms of the degree to which the CEO is described by discourse in charismatic terms is a first step in the direction suggested by Waldman and Yammarino. In terms of content, CCI incorporates facets of CEO charisma dealing with both the persona and vision of the CEO. The construct hinges upon Conger and Kanungo’s (1998) theory of charismatic leadership. The theory was chosen because it puts a specific focus on the leader’s vision and symbolic communication, and was adapted in order to apply it to the external environment of the firm. As presented above, CCI includes direct and indirect descriptors, that is, symbolic representations of the persona and vision of the CEO. Drawing from Conger and Kanungo, direct descriptors are defined as the degree to which organizational discourse about the CEO includes accounts of his or her exceptional behaviors or describes his/her exceptional qualities. Several scholars noted the role of charismatic qualities and exemplary behaviors. Conger and Kanungo (1998: 56) maintained that charismatic leaders “engage in exemplary acts that are perceived by followers as involving great


23 personal risk, cost and energy.” Biggart (1989: 132) argued that “charismatic leaders typically must demonstrate their abilities to followers through miracles, the continued success of a mission, or other proofs.” Jacobsen and House (2001: 79) suggested that charismatic leaders “make public demonstrations of their dedication to the cause. Such demonstrations typically involve significant personal sacrifice or even danger, by which the leader projects an image of courage, dedication, and commitment to the interests of the collectivity.” Since these type of narratives are generally very appealing to the popular press, it is relatively easy to find examples such as Financial World’s description of John Reed, CEO of Citicorp in the late s (Crystal, 1992: 126-7): “Reed’s biggest feat of heroism has been to cut his pay in line with his company’s declining performance. [ . . . ] All in all, Reed is a compensation hero.” While direct descriptors refer to the CEO’s persona, indirect descriptors refer to the CEO’s vision, and are defined as the degree to which organizational discourse portrays the CEO’s vision in charismatic terms. Indirect descriptors include three main components: evaluation of the status quo, formulation and articulation of organizational goals, and means to achieve the vision. These components overlap with Conger & Kanungo’s (1998) model, differing from it solely because CCI explicitly taps into official discourse communicating these elements to the outside environment. Charismatic evaluation of the status quo refers to the degree to which discourse attributable to the CEO de-legitimizes the past and emphasizes the need for radical change. As Conger & Kanungo (1998: 52) noted, “what distinguishes charismatic from noncharismatic leaders is the charismatic leaders’ ability to recognize deficiencies in the present context.” Although CLT scholars recognize the role of crisis almost unanimously,


24 they traditionally focused on followers’ perceptions of crisis rather than on the leader’s descriptions of the situation in critical terms (Conger & Kanungo, 1998; House, 1977; Shamir et al., 1998; Waldman & Yammarino, 1999), as the present study does. Charismatic formulation and evaluation of organizational goals is defined as the degree to which discourse attributable to the CEO incorporates ideologicallyand morally-laden statements emphasizing the positive aspects of the vision and its contrast with the status quo. For Conger & Kanungo (1998: 53), “charismatic leaders can be distinguished from others by the nature of their goals and by the manner in which they articulate them” (also, cf. House, 1977: 197; Shamir, Arthur, & House, 1994). Charismatic means to achieve the goals is defined as the degree to which the CEO expresses concern for his/her internal or external constituents. Conger & Kanungo’s (1998: 55) original formulation is tailored on internal members: “in the use of rhetoric, words are selected to reflect [the charismatic leader’s] concern for followers’ needs.” In the present study, however, the definition has been extended so to incorporate shareholders and external stakeholders in general. In conclusion, CCI diverges from traditional conceptualizations of charisma in several ways. First, it emphasizes CEO charisma as an instance of symbolic management operating within an asymmetric relationship between the leader and the followers, and affecting stock market judgments rather than decision making and task performance. Second, it is based on language attributable to the CEO and conveyed through organizational discourse, and it is independent from “follower” perceptions. Last, in terms of content, CCI emphasizes both the persona and vision of the CEO, so as to capture those aspects of charisma that are relevant for stock market audiences. The


25 following section incorporates the re-defined construct of CEO Charismatic Image into a set of hypotheses about the effects of CEO charisma on stock market actors. Hypotheses of the study Some general issues warrant consideration before I set forth specific propositions. First, while there are many groups and audiences within the stock market (Davis & McAdam, 2000), my approach focuses exclusively on securities analysts. Although limiting, this choice helps to understand one of the central actors of the stock market, which is in turn a critical segment of the firm’s external environment. Second, the model assumes that securities analyst evaluations reflect subjective perceptions that might not be connected to substantive practices (Westphal & Zajac, 1998), as well as beliefs held about other analysts’ beliefs (Podolny, 1993; Rao et al., 2001; Zuckerman, 1999). The dependent variables of interest in this study are the favorability and uniformity of analyst recommendations, and the forecast accuracy of individual analysts. Most scholars consider recommendations and forecasts the main mechanism through which analysts affect the price of a stock on the market (Beneish, 1991; Francis & Soffer, 1997; O'Brien & Bhushan, 1990). Hence, these dependent variables are of substantive interest for firms, investors, and researchers. Analyst Recommendations: Favorability Figure 2-1 summarizes the hypotheses about the effects of CEO Charismatic Image on the Favorability of analyst recommendations. Hypothesis 1 refers to a direct effect of CCI on the dependent variable, so that higher CCIs will be associated with more favorable recommendations. Hypotheses 1a, 1b, and 1c refer to moderating effects, so that the relationship between CCI and Favorability will be stronger for firms with lower


26 past performance (Hypothesis 1a), for outsider CEOs (Hypothesis 1b), and for CEOs with stronger reputations (Hypothesis 1c). Analyst recommendations constitute a summary judgment on the future prospects of a firm and incorporate a suggestion for investors (e.g., “buy”, “hold”, “sell”). As such, they normally appear in the first lines of analyst reports. The report itself presents the reader with the reasoning underlying the recommendation. The line of reasoning contains both quantitative and qualitative assessments, and is founded, among other things, on interpretations of arguments presented by the CEO. As an example, Fahnestock & Co. Inc., a research firm, issued on March 5th, 2001 a report on J.C. Penney – right after Allen Questrom was appointed to the CEO position. The example presents anecdotal evidence that the charismatic image of the CEO affects analyst recommendations. Figure 2-1. Favorability of analyst recommendations CEO CharismaticImage Pastperformance of the firm Outsider status H1b: + CEO reputationH1c: + Favorability of analyst recommendations H1: +H1a: = direct effect= moderating effect The title, “On the Right Track” was directly followed by a “Buy” rating. The first paragraph of the report stated:


27 “For the first time in many years, we believe that J.C.Penney has the potential to significantly improve its operating performance. Penney’s new CEO cautions that several years of work will be required to strengthen the company sufficiently. His sober assessment indicates to us that he knows the extent of the problem and the degree of change needed to correct the problems. The company’s former leadership did not inspire our confidence. Moreover, having low expectations about the speed with which a turnaround might develop is beneficial because it leaves room for positive surprises”. As the report clarifies, the “buy” recommendation is supported by several aspects that are strongly related to the charismatic image of the CEO: the identity of the CEO as opposed to the identity of his predecessor, his negative assessment of the past and his cautious, but decisive, prediction of the time required to achieve a turnaround. Notable, among other things, is that the “low expectations” triggered by the CEO’s cautious and decisive stance open up a space for speculation on J.C. Penney’s stock. From this viewpoint, charismatic language such as “we had an unacceptable performance and now we’re going to work hard to turn around the company” is a resource for analysts, for its ambiguity opens up possibilities for speculation. In conclusion, the analyst report is analogous to any media: its success and acceptance among its audience hinges upon the presentation of a story, a consistent narrative of what might happen in the future (Chen & Meindl, 1991). This narrative, summarized by the recommendation, is founded, among other things, on the narrative elements provided by the CEO – both his persona and his vision. Hypothesis 1 thus follows: Hypothesis 1: CCI is positively related to the favorability of analyst recommendations, so that CEOs described as more charismatic will receive more favorable recommendations as compared to CEOs described as less charismatic. Besides from the anecdotal evidence, hypothesis 1 is supported by several empirical findings in the literature. Stock market studies suggest that analysts and


28 investors base their judgments on the market identity of the firm. According to Zuckerman, analysts convey to investors the market identity of the firm, and “when this identity fails to match the firm’s self definition, the firm’s stock performance should be impaired” (Zuckerman, 1999: 1413). Although Zuckerman focuses on analyst coverage decisions as proxies for the market identity of the firm, his arguments can be extended to recommendations, which are also based on judgments on the market identity of the firm. The importance of such a cognitive framing is evident from the words of an Investor Relations manager for Columbia Foods: “You must have access to information, both financial and market-type data. You must have also access to people because it is the perspective that you get that goes beyond the numbers that is important in describing the business and giving investors the confidence that they understand the company and what it’s trying to do” (reported by Useem, 1996: 187). As Davis and McAdam argued, an important phenomenon of investor capitalism is represented by the strategic framing and other ‘representational’ practices of both movement and economic entrepreneurs. In a post-industrial service economy, what is ‘produced’ is often not material products per se, but perceptions and identities. [ . . . ] The value added, in short, is perceptual, flowing from the creation of distinctive and desirable identities. The management of perceptions is aimed not only at consumers of products, of course, but also [ . . . ] employees and (actual and potential) investors, often using rather different messages. Because the nature of the product is perceptual, ‘external evaluations in such contexts are based largely on social rather than technical criteria (Davis & McAdam, 2000: 228). Indeed, one of the distinctive features of charismatic leadership is to provide followers with such a cognitive framing, or “perspective”, or “strategic framing”: “what effective leaders do is provide an understanding to their followers of why they are doing what they are doing [ . . . ]. Through choice of words and portrayals of future organizational outcomes, the charismatic leader uses language to create meaning and a shared identity and community vocabulary” (Conger & Kanungo, 1998: 172). Thus, since


29 CCI can be seen as the degree to which organizational discourse presents internal and external audiences with a coherent and consistent picture of the market identity of the company, and since securities analysts base, at least in part, their recommendations on the degree to which the company presents them with a coherent and consistent picture, it is reasonable to expect that CCI might positively affect analyst recommendations, as hypothesis 1 suggests. The rationale for the three moderating hypotheses follows the same line of reasoning: the effects of CCI on analyst recommendations will be modified by information increasing or decreasing the credibility, coherence and consistency of the CEO’s message. The first element to take into consideration is the past performance of the firm. This information is, intuitively, an important element that analysts and investors take into consideration when drafting their judgments of the firm’s potential. As an example, the following Business Week article describes the attitudes of investors after Questrom’s appointment at J.C. Penney. At this point, the firm’s stock had dropped 80% in 2 years, clearly a state of crisis. Invited by a group of the firm’s retirees holding shares in the company, Questrom “gave what was by most accounts a passionate speech about rejuvenating Penney, the nation’s larges department store chain. Those 40 minutes of encouragement earned plenty of goodwill with the frustrated retirees. ‘We left pretty confident that he will bring us into the 21st century right on target’, says Wallace J. Paprocki, head of the 2,000 member group” (Forest, 2001: 56). Charismatic leadership theory has long recognized the role of crises in triggering charismatic attributions and, consequently, in reinforcing charismatic language: “when an organization is


30 dysfunctional or when it faces a crisis, leaders may find it to their advantage to advocate radical changes, thereby increasing the probability of fostering a charismatic image for themselves” (Conger & Kanungo, 1998: 52). The idea that crises increase perceptions of charisma is also supported by most scholars, regardless of their theoretical perspective (House, 1977; Jacobsen & House, 2001; Meindl, 1990; Pfeffer, 1977; Shamir et al., 1998; Steyrer, 1998; Waldman & Yammarino, 1999). It is reasonable, then, to expect that in conditions of acceptable performance, the effectiveness of the charismatic language of the CEO in triggering positive evaluations might be diminished, as hypothesis 1a suggests: Hypothesis 1a: Past performance of the firm moderates the relationship between CCI and the favorability of analyst recommendations, so that in conditions of acceptable performance such relationship will be weaker, as compared to a situation where past performance is not acceptable. The second potential moderator is represented by the outsider status of the CEO. Agency Theory suggests that outsiders, as compared to insiders, “are perceived to be more able to initiate and implement strategic change” (Cannella & Lubatkin, 1993: 763). As Westphal and Zajac (1998) showed, stock market actors respond positively to changes in corporate governance consistent with more effective governance mechanisms. It is plausible that financial analysts would respond to such changes similarly to other stock market actors, and attribute more credibility to outsider CEOs. An example of how this might happen is contained in the following quote from a vice president of a money-management firm: “when you find good management, [ . . . ] you go for it. [ . . . ] You look to see a management working for the shareholders” (reported by Useem, 1996: 158).


31 One explanation for the higher credibility of outsider CEOs is that, unlike insiders, they escape attributions for the previous performance of the firm. For insiders, on the other hand, the past performance of the firm represents an obstacle to their credibility: an acceptable performance might still be attributed to their predecessor, while an unacceptable one casts doubts on an insider’s capability to turn around the company (Clapham & Schwenk, 1991). It follows that the outsider status of the CEO might reinforce his/her charismatic language, making it more credible in the eyes of securities analysts. Hypothesis 1b thus follows: Hypothesis 1b: The outsider status of the CEO moderates the relationship between CCI and the favorability of analyst recommendations, so that such relationship will be stronger for outsider CEOs, as compared to insiders. The last moderator variable included in the study is the reputation of the CEO. Meindl, Erlich and Dukerich (1985) found that the media are an important factor in attributing results to the CEO and diffusing romantic images of CEO heroism that tend to continue over time despite performance misfortunes (Chen & Meindl, 1991). Hayward and Hambrick argued that media praise for the CEO “not only crystallize and solidify attributions of organizational members, but they diffuse the CEO’s prestige across wider audiences” (Hayward & Hambrick, 1997: 108). It is therefore plausible that the reputation of the CEO in the popular press generates a multiplicative effect for CCI, since it increases the credibility of charismatic descriptions of the newly appointed CEO. Hypothesis 1c thus follows: Hypothesis 1c: The reputation of the CEO moderates the relationship between CCI and the favorability of analyst recommendations, so that such relationship will be


32 stronger for CEOs that enjoy a higher reputation, as opposed to CEOs with less reputation. Analyst Recommendations: Uniformity Figure 2-2 summarizes the hypotheses about the effects of CEO Charismatic Image on the second dependent variable of interest, the uniformity of analyst recommendations. Hypothesis 2 refers to a direct effect of CCI on the dependent variable, while hypotheses 2a, 2b, and 2c refer to moderating effects. The major difference between the hypotheses on favorability and the ones presented herein is that the former refer to the level, or mean, of recommendations, while the latter refer to the variance in recommendations. Figure 2-2. Uniformity of analyst recommendations CEO CharismaticImage Pastperformance of the firm Outsider status H2b: + CEO reputationH2c: + Uniformityof analyst recommendations H2: +H2a: = direct effect= moderating effect Moreover, while the favorability of analyst recommendations is an individual level construct (i.e., refers to the favorability of the recommendations issued by individual analysts), the uniformity of analyst recommendations refers to group-level effects – that is, to the amount of variance in recommendations across all analysts following a specific firm. The “consensus recommendation” (or the average recommendation across multiple


33 analysts) is relevant for investors and executives alike because it represents the collective perception of the firm’s value, and therefore it influences stock prices. The same average recommendation, however, can logically be associated with different levels of variance. From the viewpoint of an executive, the variance in recommendations is important because it is a proxy for the degree to which analysts as a group agree with his/her decisions, while for investors the variance in recommendations constitutes an indication of the reliability of the mean recommendation. From the viewpoint of this study, the variance of analyst recommendations is important because it tests the presence of social contagion effects (Meindl, 1990) among securities analysts, a population of followers of charismatic leaders never studied before. It follows that we can hypothesize that charismatic language, as expressed by CCI, triggers homogeneous evaluations among the group of securities analysts following the firm: Hypothesis 2: CCI is positively related to the uniformity of analyst recommendations, so that CEOs described as more charismatic will receive, across all analysts following the firm, more uniform recommendations as compared to CEOs described as less charismatic. The accounting and finance literature have long recognized that analysts are prone to “herding behaviors”, that is, to “ignore private information and copy the actions of others” (Hong et al., 2000: 121). To the extent that are favorable to the firm, these behaviors constitute a valuable consequence, since they trigger positive stock market reactions. Studies of charismatic leaders on the other hand, also suggest that charisma has group-level herding effects of the type described by the finance literature. For example, Klein and House (Klein & House, 1995) argued that the homogeneity of attributions of


34 charisma (that is, the variance in attributions among followers) is an important variable that should be included in studies of charisma as well as an important outcome in itself, distinct from and independent of the mean level of perceived charisma. According to Klein and House, several conditions lead to more homogeneous perceptions of charisma: the uniformity of treatment of the followers by the leader, the uniformity of the followers’ values, orientation to work, and social values, and the presence of attraction-selection-attrition dynamics that is, when both the leader and the followers choose independently to remain in the same group. The relationship between CEOs and securities analysts presents all three conditions: CEOs are compelled by Fair Disclosure laws to release the same information to all analysts and investors (Riepe, 2000); the profession of the securities analysts is characterized by relatively homogeneous socialization processes, career experiences, membership in and certification by the Financial Analyst Federation (Hayward & Boeker, 1998; Morley, 1988; Useem, 1996); and the choice of issuing recommendations for a firm is carefully made by each individual analyst based on his/her own career concerns and expectations of performance from the firm (Hong et al., 2000), to the point that it is not uncommon for an analyst to refer to the companies he/she follows as “my companies” (Balog, 1991). The basic foundation for the moderating hypotheses is that herding is more likely when analysts face a higher level of “correlated signal”, that is, several sources of information pointing to the same conclusion (Graham, 1999). Hypotheses 2a, 2b, and 2c thus follow: Hypothesis 2a: Past performance of the firm moderates the relationship between CCI and the uniformity of analyst recommendations, so that in conditions of acceptable


35 performance such relationship will be weaker, as compared to a situation where past performance is not acceptable. Hypothesis 2b: The outsider status of the CEO moderates the relationship between CCI and the uniformity of analyst recommendations, so that such relationship will be stronger for outsider CEOs, as compared to insiders. Hypothesis 2c: The reputation of the CEO moderates the relationship between CCI and the uniformity of analyst recommendations, so that such relationship will be stronger for CEOs that enjoy a higher reputation, as opposed to CEOs with less reputation. Analyst Forecast Accuracy Figure 2-3 summarizes the hypotheses about the effects of CEO Charismatic Image on analyst forecast accuracy, the third dependent variable of interest. Forecast accuracy is an individual level construct, constituted by the absolute difference between the performance forecasted by an individual analyst and the actual performance of the firm. In this dissertation, hypothesis 3 refers to a direct, positive, effect of CCI on the dependent variable. Since the hypothesis has an exploratory nature, no moderator hypotheses are added.: Hypothesis 3: CCI is positively related to analyst forecast accuracy so that CEOs described as more charismatic will receive more accurate forecasts as compared to CEOs described as less charismatic. Figure 2-3. Analyst forecast accuracy CEO CharismaticImage Analyst forecast accuracyH3: +


36 Hypothesis 3 has a substantive importance for investors because it is a measure of the analyst’s ability to correctly inform their decisions (Hunton & McEwen, 1997). The main goal of this hypothesis is to test whether charismatic language helps securities analysts’ accuracy in their predictions that is, whether charismatic language has a correspondence in the reality of business or not. Hypothesis 3 is justified by charismatic leadership theory (Conger & Kanungo, 1998; House et al., 1991; Waldman et al., 2001) as well as by the adaptationist perspective to executive succession (Friedman & Singh, 1989). Both perspectives argue that charismatic language reflects actual changes in organizational practices implemented by the charismatic CEO and leading in turn to better results for the firm (Waldman et al., 2001). Under this perspective, analysts correctly and rationally use CEO charisma as a signal estimating the CEO’s inherent ability to stimulate the employees and lead the firm (Chung & Jo, 1996). Its inclusion among the information sources used by analysts to evaluate the firm should therefore increase their capability to accurately predict the future performance of the firm. CCI should therefore be positively associated with forecast accuracy. In conclusion, the three sets of hypotheses should allow us a deeper understanding of the external effects of CEO charisma onto the perceptions and evaluations of securities analysts. With respect to the purpose of this research, the three hypotheses are built on the assumption that charismatic language is actually observed and used by securities analysts to construct their recommendations and forecasts. By testing whether the initial assumption is true (i.e., that charismatic language is an element within the informational space of securities analysts), the study sheds some light on the effects of charisma on the


37 stock market and ultimately constitute a first step toward an understanding of the effects of CEO charisma on society at large.


CHAPTER 3 METHODS This chapter illustrates the sample, data, and methods used to test the hypotheses discussed in the previous chapter. The first methodological issue of concern here regards the identification of organizational discourse containing references to the Charismatic Image of the CEO. A major choice of the study, from this viewpoint, was to select documents issued by the firm in the time period around CEO succession events. Several considerations led to this decision. As House et al. (1991) argued, a major problem in studying the relationship between charisma and performance is the fact that attributions of charisma might be stemming from the previous performance of the leader. In such a case, any inference on the direction of causality becomes impossible. This study adopted the solution proposed by House et al. (1991), that is, to focus on the first period in which the CEO is in place or, in other words, on charismatic language right before, during, and right after a CEO succession. Since the CEO who is being evaluated by stock market actors was not in charge in the previous period, we can safely assume that attributions of charisma were not tainted by previous CEO performance within the same firm. Moreover, CEO successions present a second advantage: during succession events the image of the CEO becomes central to corporate impression management strategies (Hambrick & Fukutomi, 1991; Pfeffer, 1981; Useem, 1996), salient to internal and external actors (Cannella & Lubatkin, 1993; Cannella & Shen, 2001), and ultimately predominant within organizational discourse. Hence, it is more likely that documents released around and about succession events will discuss the CEO and his/her vision. 38


39 Sample and Data The sample consists of all CEO succession events occurred between 1990 and 1999 within a stratified sample of 800 US publicly traded corporations randomly selected from across 30 industries (as defined by their 4-digits SIC code). In order to select the sample, I proceeded with a top-down approach, selecting first 30 industries (as defined by their 4 digit SIC code) chosen to represent conditions of low, medium, or high managerial discretion. Managerial discretion scores were obtained from a study by Hambrick and Abrahamson (1995), and the sample of industries was identified by selecting nine industries from each of the upper and lower quartile of the distribution of the discretion scores, and six industries each from the second and third quartiles. Such a procedure was aimed at obtaining a sample of industries that would vary in the degree to which managers enjoy latitude in pursuing different courses of action. After selecting the industries, a random sample of firms having at least $10 million in total assets was selected from each of the 30 industries. The central point in the time period of interest, 1995, was used to collect size data. Data from ExecuComp, a database of executive compensation, allowed the identification of 725 CEO succession events occurred during the period within the initial sample of 800 firms. Excluding cases where the succession occurred in correspondence with extraordinary events (e.g., mergers and acquisitions, bankruptcy, etc) and extraordinary appointments (e.g., co-CEO, interim CEO, etc.), the sample was reduced to 419 events. Availability of public relations documents about the succession event brought the final sample to 367 CEO succession events. In order to gather data on CEO discourse, I collected for each of the 367 firms included in the sample the press releases announcing the succession event and the first


40 letter to the shareholder released by the company and signed by the newly appointed CEO (press releases about succession events and the letters to the shareholders were obtained from ABI-Inform, Lexis-Nexis, and Compact Disclosure). Press releases constitute the first official announcement by a firm of the appointment of a new CEO, and their structure and goals are relatively homogeneous across firms. Their purpose is to inform of the appointment external observers such as investors and the general public. Their structure generally provides a brief profile of the new CEO, his or her professional background and past achievements, as well as a brief comment from the new CEO and/or another executive of the firm. Such press releases are the first occasion for corporate public relations to frame how external audiences look at the newly appointed CEO. The first letter to the shareholders issued after the appointment represents the first occasion in which the CEO communicates directly and officially with the shareholders. Hence, the vision of the new CEO can be expected to be extremely salient in this document (Hambrick & Fukutomi, 1991). The letter to the shareholders is widely used for the study of organizational symbolism (Abrahamson & Park, 1994; Arndt & Bigelow, 2000; D'Aveni & MacMillan, 1990; Porac et al., 1999; Westphal & Zajac, 1994). As Abrahamson and Park observed, “president’s letters [ . . . ] are important vehicles for communicating information to shareholders and other interested stakeholders” (Abrahamson & Park, 1994: 1307). Staw et al. noted that the letter to the shareholders is a means of impression management, and its structure and content reflects the “art of presenting good and bad news” (Staw, McKechnie, & Puffer, 1983: 596). The letter to the shareholder is the most widely read section of the annual report, and presents several


41 characteristics that make it suitable to study symbolic management: it is relatively free from legal restrictions about its form or content (Abrahamson & Park, 1994), communicates both facts and beliefs in a form that is directly approved by the CEO (D'Aveni & MacMillan, 1990; Fiol, 1989), reflects managerial attributions, locus of attention and framing strategies (Clapham & Schwenk, 1991; D'Aveni & MacMillan, 1990; Porac et al., 1999; Staw et al., 1983). Dependent Variables Favorability of Analyst Recommendations The first dependent variable is Favorability of Analyst Recommendations. Typically, different analysts issuing recommendations for any specific firm follow different formats, dictated by the brokerage house where they’re employed. Both I/B/E/S and FirstCall, however, map all the recommendations received to a standard 5-point scale (1=strong buy; 2=buy; 3=hold; 4=underperform; 5= sell). As such, recommendations can be treated just as any other set of ratings on a 5-point scale, with 1 = very favorable, and 5 = very unfavorable. Since analysts issue multiple recommendations over time, for each succession event included in the sample I collected from I/B/E/S and FirstCall (two databases of analyst recommendations and forecasts) all recommendations issued by securities analysts in the first six months following the release of the letter of the shareholders and averaged them. Uniformity of Analyst Recommendations The second dependent variable is Uniformity of Analyst Recommendations, that is, to the standard deviation of recommendations across all analysts following the firm. This variable is available on I/B/E/S, and it is calculated as the monthly dispersion of


42 recommendations around the mean (consensus) recommendation. For all the firms in the sample, I collected data on the standard deviation of recommendations for a year before the succession to a year after the filing of the letter to the shareholders. Since these data are already aggregated across all analysts following the firm, it is actually possible to study their evolution over time through a 2-level model where the first level refers to time, rather than to the individual analyst. Moreover, there is no need to aggregate the different observations into time windows. Forecast Accuracy The third dependent variable is Forecast Accuracy of individual analysts. Forecasts constitute an important output of the process of stock evaluation, and constitute a prediction, on the part of the analyst, of the future state of several performance variables, such as sales, return on investments, growth, or earnings per share. For the purposes of this study, I chose to focus on Earnings per Share. Following the accounting literature (Butler & Lang, 1991; Sheikholeslami, Wilson, & Selin, 1998; Stickel, 1992), forecast accuracy is measured by the average annual forecast error (E a,y,t ) of an individual analyst and is expressed by the following equation: E a,i,y(t) = |A i,y – F a,i,t |, where E a,i,y(t) is the absolute (unsigned) average forecast error of analyst a following firm i for fiscal year t; A i,y is the actual earnings per share for firm i in year t, and F a,i,t represents the forecast of analyst a of Earnings Per Share of firm i for fiscal year t. Data on analyst forecasted Earning Per Share (EPS) and actual EPS were collected from the I/B/E/S Detail + History database. For each of the firms in the sample, I collected all analyst forecasts and actuals from the release of the letter to the shareholders to six months after. Then, I calculated the forecast accuracy for each individual forecast


43 following the formula reported above, and averaged all the values for each analyst over the six months period, so as to obtain a single measure of forecast accuracy for each analyst. Independent Variables CEO Charismatic Image (CCI) CEO Charismatic Image was measured through a thematic text analysis of the letter to the shareholders. As discussed above, for each of the 367 companies of the sample, the first letter to the shareholders signed by the newly appointed CEO was available, and a content analysis of this document was used to measure CCI 1 . Content analysis is defined as “any systematic reduction of a flow of text (or other symbols) to a standard set of statistically manipulable symbols representing the presence, the intensity, or the frequency of some characteristic relevant to the social sciences” (Shapiro & Markoff, 1997: 14). Content analysis was chosen because it presents several advantages over other methods (such as surveys, ethnographies, etc.): first, it is a nonintrusive method (Popping, 2000) and therefore avoids the consistency, priming and implicit theory 2 effects frequently associated with surveys (Salancik & Pfeffer, 1977, 1978). Second, communications are measured directly, rather than indirectly (i.e., as perceived by an audience). Third, it allows the combination of quantitative and 1 A second measure of CEO charisma, not based on a text analysis but rather on expert ratings, was also used in the analysis. Its construction is described in the following section. 2 Consistency refers to the tendency of individuals to organize their responses to survey instruments so that their attitudes and beliefs do not contradict each other. Priming refers to a process whereby the survey instrument focuses the respondent’s attention on certain aspects of the situation, making certain items stored in memory more salient and inducing spurious correlations. Implicit leadership theories (Meindl et al. 1985) refer to the tendency of respondents to follow their implicit theories while filling surveys, thereby inflating correlations.


44 qualitative methods, making it particularly suited for the quantitative study of large samples through specialized software (Popping, 2000: 11). Specifically, this study relies on Thematic Text Analysis, a technique in which variables indicate the occurrence (or frequency of occurrence) of particular concepts in a given unit of text, such as a sentence, a paragraph, or a document (Popping, 2000). The result of the analysis is a data matrix where rows represent different text units (lines, sentences, or paragraphs) and columns distinct themes (Roberts, 2000: 260). In other words, Thematic Text Analysis measures the degree to which each document within a sample reiterates a specific theme or topic determined a-priori by the investigator. The topic of interest in this study, CEO Charismatic Images, was measured with dictionaries that were developed a-priori, based on theoretical considerations. As an example of the uses of thematic text analysis, Abrahamson and Park (1994) used the frequency of negative statements as a measure of the predominance of the “concealment of negative outcomes” theme within a sample of over 1,000 president’s letters. The thematic text analysis of the 367 letters to the shareholders was conducted according to the four steps defined by Popping (2000). Table 2-2 summarizes each step involved in the development of a measure of CCI: the conceptual nodes/CCI dimensions, the criteria for assignment of sentences to the nodes, the search dictionaries used to measure the specific CCI component of interest, and the measure used to obtain a score for each CEO on that component. Identification of conceptual nodes and textual units. In the first step, all the sentences within the 367 letters to the shareholders were coded with the QSR N6 software, which is particularly suited for such an operation since it allows the


45 manipulation of very large samples of documents. The coding involved assigning each sentence within each letter to one of three “conceptual nodes”: “assessment of the past”, “plans for the future”, and “shareholders, employees, and organizational capabilities” (sentences not referring to any of these categories were left uncoded: e.g., references to products or services). Table 2-2. Construction of CCI scores Conceptual node/ CCI dimension Coding criteria Concept category / Search dictionaries Measure Assessment of the past / Evaluation of the status quo Sentences referring to events initiated in the past, concluded at the time of the letter, and containing an evaluative posture. Charismatic evaluation of the status quo (184 terms): 93 “Negative feelings” terms, 56 “Negative and crisis-related terms”, 35 “negative evaluation terms” Relative frequency of negative and crisis-related terms Plans for the future / formulation and articulation of goals Sentences referring to the CEO’s strategy, vision, mission, or to actions initiated in the past and still ongoing in the present; sentences containing a tangible or intangible exhortation or a prediction. Charismatic vision (1136 terms): 146 Moral or ideological terms 709 Emotion and feeling terms Relative freq. of moral terms Relative freq. of emotion and feeling terms Shareholders, employees, and organizational capabilities / means to achieve the vision Sentences referring to employees, shareholders, and organizational capabilities. Charismatic implementation (605 terms): 383 Affiliation and Positive Affect terms 168 Internal and external constituents terms Relative freq. of affiliation and positive affect terms Relative freq. of references to internal and external constituents A conceptual node is a subset of the document that includes all text units (sentences, in this case) sharing a common topic according to the judgment of one or more coders. The three nodes used in this study correspond, theoretically, to Conger and Kanungo’s (1998) definition of charismatic leadership (as well as to the three


46 components of CCI described in the first chapter) and, empirically, to specific topics found to be generally present in the letter to the shareholders (Arndt & Bigelow, 2000; Westley & Mintzberg, 1989). In terms of the CEO’s “evaluation of the status quo”, Conger and Kanungo (1998) argued that charismatic CEOs tend to use language delegitimizing the past, emphasizing the crisis, if present, and invoking the need for radical change. The coding instructions for the first conceptual node were therefore developed so as to capture the section of the letter dealing with the CEO’s assessment of the past (Appendix A illustrates the three conceptual nodes and their construction in detail). In terms of the CEO’s “formulation and articulation of organizational goals”, Conger and Kanungo argued that charismatic CEOs articulate goals characterized by ideological and moral overtones, so the coding instructions for the second conceptual node (“Plans for the future”) were developed accordingly. The last conceptual node (“Shareholders, employees, and organizational capabilities”) was constructed so as to capture the sections of the letter addressing the issue of the implementation of the CEO’s vision. As Conger and Kanungo (1998) as well as several others (e.g., House et al., 1991; Shamir et al., 1994) argued, charismatic CEOs face the challenge of obtaining commitment and enthusiasm from their followers, so as to realize the vision and achieve a superior performance. Although the literature on charismatic leadership focuses exclusively on the members of the organization, in the context of this study it is the commitment of and support from external constituents that is important. This is particularly true for the shareholders, whose commitment to the firm in the initial phases of the CEO’s tenure constitutes, in itself, performance on the stock market (Hambrick &


47 Fukutomi, 1991). Therefore, the third node included all references to employees, organizational capabilities, and shareholders. The construction and identification of three separate conceptual nodes was made necessary by the presence of multiple topics within the letter to the shareholders. By assigning each sentence to one of the three nodes based on the topic of the sentence, subsequent analyses based on word counts could be performed separately on each node with different search dictionaries, rather than on the whole document, thereby increasing the internal validity of the final measure of CCI. In other words, a separate text analysis for each conceptual node increases the likelihood that the words used refer, in fact, to the CCI dimension under study (e.g., the use of the word “unsatisfactory” within the node “assessment of the past” reflects a negative assessment of the status quo by the newly appointed CEO). Concept categories and search dictionaries. The second step involved the construction of concept categories and specific search dictionaries for each of the three conceptual nodes (dictionaries were developed prior to the actual text-analysis following theoretical considerations). A concept category is equivalent to a construct or component of a construct of interest (e.g., “charismatic evaluation of the status quo”), and a dictionary is a set of words, or phrases (“search entries”) that are a concrete representation of the underlying theory (Popping, 2000: 44). In other words, a dictionary includes all the terms whose appearance within a document or conceptual node indicates the insistence of the speaker on a theoretically relevant theme – in this case, the three components of the CEO’s charismatic image. To use an analogy, a dictionary is equivalent to a scale measuring a particular construct, and the search entries included in it


48 are analogous to the individual survey items belonging to the scale. As an example, the words “unsatisfactory” and “unacceptable” were included in the dictionary used to analyze the first conceptual node (“assessment of the past”) because their utterance in the context of the CEO’s assessment of what happened before he was appointed represents, most likely, a negative evaluation of the status quo – a characteristic of charismatic communication according to the charismatic theory literature (e.g., Conger & Kanungo, 1998; Shamir et al., 1994). Five specific dictionaries were developed in order to measure the different components of CCI across the three conceptual nodes. For the first component/conceptual node, a single dictionary included 184 terms measuring the degree to which the CEO uses a negative language about the firm’s past performance and events. The terms were obtained from the Lasswell Value Dictionary, from Abrahamson and Park (1994), and from a list of 35 terms obtained inductively by reading the letters of the shareholders. For the second component/conceptual node, two separate dictionaries were used, one containing 146 moral and ideological terms, the other containing 709 emotion and feeling terms. For the third component/conceptual node, two dictionaries contained, respectively, 383 affiliation and positive affect terms, and 168 internal and external terms referring to internal and external constituents such as employees, shareholders, customers and suppliers, society and the government, and managerial fashions. All the terms in these four dictionaries were obtained from the LVD, from the Harvard IV-4 dictionary, and inductively from the letters to the shareholders. Although the analysis could have been performed with one dictionary for each conceptual node, the use of more two or more dictionaries allows to measure separately the different components of CCI (e.g.,


49 emotional and moral terms), and eventually allows the development of a construct validation study. Appendix B illustrates the dictionaries in detail and their construction. Measurement. The third step is the actual measurement and construction of a single CCI score for each CEO under study. This step was performed with Diction, a text analysis software. For each of the three conceptual nodes, I ran a separate analysis of all the 367 documents, generating separate raw word counts. I then divided the raw counts by the total number of words within each Letter to the Shareholders, in order to obtain a measure of the frequency of use of a charismatic language throughout the whole document. I then summed the scores to obtain a single measure of CCI for each CEO/firm. The mean CCI score for all the 367 firms included in the sample was .042 (i.e., on average, 4.2% of the words in the letter to the shareholders were of a charismatic nature), from a minimum of 0 to a maximum of .144 (std deviation = .022). Charisma (factor scores) Since the CCI measure was not validated in previous studies, it was important to understand what facets of charisma were responsible for the text analysis scores. I therefore collected a questionnaire-based measure of CEO charisma and factor-analyzed it with the text-based measure. Charisma ratings were first obtained by administering the Conger-Kanungo (1998) scale (reported in the appendix C) to a group of 3943 undergraduate students enrolled in a management course at the University of Florida in Fall 2002 and Spring 2003. Raters were presented with a scenario in which they were asked to assume the role of a shareholder receiving an announcement of a succession at the top position in the firm. Actual press releases and Letters to the shareholders issued by the 367 firms included in the sample were used. After reading the press release and the first letter to the shareholders issued by the newly appointed CEO, subjects were asked to


50 rate the charisma of the CEO with 28 items from the Conger-Kanungo (1998) questionnaire: 7 items from the “environmental sensitivity” subscale, 6 items from the “vision and articulation” subscale, two items from the “Status Quo” scale, 3 items from the “sensitivity toward organizational members”, 4 items from the “Personal Risk” scale, and 3 items from the “Unconventional Behavior” scale. In order to gauge the CEO’s sensitivity toward the shareholders, 3 items not included in the original Conger-Kanungo scale were added. These items were constructed by simply substituting “shareholders” for “members” in the original 3 items from the “sensitivity towards members” subscale. Appendix C reports the items of the C-K scale. Multiple subjects rated each of the 367 press releases, for a total of 3943 ratings. On average, 10.7 students rated each CEO (min = 3, max = 21). Once the ratings were obtained, I proceeded to correlate the CCI scores with the 7 scales from Conger and Kanungo (Environmental Sensitivity, Personal Risk, Sensitivity to Members, Status Quo, Unconventional Behavior, Vision, and the added Sensitivity to Shareholders scale). Table 3-6 presents the correlation matrix. As Table 3-6 shows, CCI was correlated with Personal Risk (r=.149, p<.001), the two Sensitivity subscales (sensitivity to members, r=.215, p<.001 and sensitivity to shareholders, r=.174, p<.001) and with Unconventional Behavior (r=.134, p<.05). I then performed a factor analysis with varimax rotation including CCI and the four correlated Conger-Kanungo subscales. The factor analysis produced two distinct factors with eigenvalues greater than 1, with CCI and the Sensitivity to Members and Sensitivity to Shareholders subscales all loading on the first factor (.414, .852, and .864 respectively), and Personal Risk and


51 Unconventional Behavior loading on the second factor (.799 and .856 respectively). I then performed an un-rotated factor analysis, including only CCI and the two sensitivity subscales with CCI. The analysis resulted in the extraction of one factor with engenvalues > 1, responsible for 60.9% of the total variance, with both the six CK items and CCI loading .781 on the factor. I then used the scores for the first factor to re-run the analyses substituting the factor score for CCI. Table 3-1. Correlation Matrix, CCI and Conger-Kanungo subscales 1 2 3 4 5 6 7 8 1. CCI 1.00 2. Environmental Sens. .054 1.00 3. Personal Risk .149** .290** 1.00 4. Sensitivity Members .215** .233** .254** 1.00 5. Sensitivity Shrhold. .174** .160** .196** .566** 1.00 6. Status Quo -.036 -.009 -.027 -.171** -.196** 1.00 7. Unconv. Behavior .134* .261** .436** .180** .101 .127* 1.00 8. Vision .092 .135** .209** .516** .424** .041 .182** 1.00 Notes : ** p < .01, * p < .05 Past Performance Consistent with much prior research (e.g., Rumelt, 1991; Westphal & Zajac, 1994), past performance was measured with return on assets, which is the firm’s net income divided by total assets for each year. All data were collected from Compustat for 3 years prior the succession and the 2 years following it. I then conducted a separate search on all the firms included in the Compustat database in order to obtain the industry means and standard deviations, with industries defined at the 3 digit SIC code. Following Zuckerman (1999), I reasoned that the 3-digit level would be more appropriate than the 4-digit level, in terms of the reliability of the means and standard deviations obtained, as well as in terms of the level of analyst specialization (Zuckerman, 1999). The search resulted in a total of 238,982 firm/year observations across all the 272 3-digit industries


52 represented in the Compustat database, with an average number of 437 observations per industry in the period between 1989 and 2001. These data were used to standardize performance by the primary industry in which the firm operates. In order to calculate the pre-succession performance, I then calculated the change in industry-standardized ROA from the year (t-4) to the year before the succession by subtracting year t-1 from year t-4 and then dividing by year t-4. As Zajac and Westphal (1996) argued, considering 3 years before the succession is more appropriate then considering simply the previous year. Outsider Status Consistent with the succession literature (Ocasio, 1999), the outsider status of the new CEO was measured by whether the newly appointed CEO was employed by the company for less than 2 years prior to his/her appointment (outsider = 1, otherwise =0). Given the large sample, I decided to rely for a first screening on the press releases of the succession event. These documents generally mention the background of the newly appointed CEO. In those cases where the background of the CEO referred to a position within a different company, the CEO was considered an outsider. In all cases where the background referred to an executive position within the company (e.g., Chief Operating Officer, Chief Financial Officer, Chairman of the board), the CEO was coded as insider. Unclear assignments were resolved by accessing the Compact Disclosure records about the managers and directors of the company. Given the aims of the study, I reasoned that this procedure was acceptable although not necessarily as precise as using the official records all along. Indeed, since the purpose of the research is to study the effects of organizational discourse on the perceptions of stock market actors, it is important to capture whether the press release introduces the new CEO to the investor public as an outsider to the firm, regardless of whether s/he is formally an outsider or not.


53 CEO Reputation CEO reputation consists of a simple count of articles published in the period between 2 years before the succession took place to 1 year after by 7 nationally renowned newspapers (Atlanta Constitution, Boston Globe, Chicago Tribune, Los Angeles Times, New York Times, Wall Street Journal, Washington Post) and on business periodicals. Newspaper articles were obtained from the Lexis-Nexis database, and magazine articles were obtained from the ABI-Inform database. This procedure was used by Hayward and Hambrick (1997) to measure media praise for the CEO. In their analysis of CEO hubris, the authors searched for articles about the CEO The authors content analyzed the articles using a 5-point scale measuring the degree of attributions of positive performance to the CEO, defining an aggregate measure of media praise as the sum of all scores for all the articles about a CEO. They also reported, however, that this measure was highly correlated with simpler article counts about the CEO, so in order to reduce the complexity of the data collection, I used this last measure. Control Variables CEO Controls For each CEO, the following control variables were collected through an analysis of the press releases of the succession events and complemented with a search on Compact Disclosure: CEO Age, CEO duality (dummy variable with dual position of CEO and Chairman of the board = 1), Predecessor disposition (dummy variable with predecessor remaining with the firm in any role = 1), and tenure at the time of the release of the annual report. The age of the CEO was generally reported by the press releases of the succession event, which also mentioned whether the new CEO was also nominated to the position of Chairman of the board, or President where a Chairman position is absent


54 (CEO duality), and whether the predecessor remained in the firm in any role such as member of the board, consultant, chairman, or honorary chairman (predecessor disposition). In cases where the press release did not report all the information, the relevant information was found in Compact Disclosure, which reports the composition of the board of directors, as well as of the management team of the firm, and the respective ages. Last, tenure at the time of the release of the annual report (measured in number of days) was included in order to control for potential changes in the CEO’s language at different stages of his/her mandate. As an example, one would reasonably expect a negative evaluation of the status quo to be more justifiable by a CEO that was appointed one month before the issuing of the letter to the shareholders, as compared to a CEO who has been in charge of the company for 11 months. Or, alternatively, one could expect stronger negative evaluations from a CEO who has had the time to assess the situation, as compared to a CEO who was appointed one month before the issuing of the letter to the shareholders. Regardless of the direction of the effect, I felt it necessary to include this variable to control for it. Firm Controls For each firm, I collected from Compustat firm size (log sales and log number of employees), and Fortune Rank (dummy variable with 1 = firm is included in the Fortune ranking. Both variables referred to the year before the succession. Analyst Controls At the individual analyst level, the following control variables were collected: previous average recommendation and previous forecast accuracy, forecast horizon, and monthly change in recommendations. Previous recommendations (forecast accuracy) were obtained by repeating the procedure used for the dependent variable, that is, by


55 obtaining all the recommendations (forecast accuracy) issued by each analyst in the six months before the release of the letter to the shareholders and then averaging for each analyst in order to obtain a single measure. Forecast horizon refers to the time horizon for which the analyst is constructing his or her predictions. It is logical to assume that forecasts referring to time farther into the future might be less accurate then short-term forecasts, so I calculated this variable as the number of days between the estimate date and the forecast period end date. Both values are available on I/B/E/S. Analyst age was obtained by the Nelson’s Directory of Investment Research. Finally, as a measure of the inter-temporal variability of recommendations, I included change in recommendations as a control variable in the uniformity models. This variable was constructed as the sum of the number of recommendations increased and decreased from the previous month divided by the total number of recommendations in the current month. All the three values are supplied by I/B/E/S. Analytical Methods This study includes variables collected at different levels of analysis. At the lower level, the first nesting is represented by time: individual analysts issue multiple recommendations and forecasts. At a higher level, multiple observations are nested in individual analysts, which are in turn nested in firms/succession events. From a methodological viewpoint, this multiple nesting (which configures at least 3 levels: time, analysts, and firms) violates the assumptions of Ordinary Least Squares (OLS) regression, and therefore calls for the use of appropriate statistical techniques such as Hierarchical Linear Modeling (Bryk & Raudenbusch, 1992). Briefly, HLM involves a separate estimation of the Level 1 coefficients (i.e., those expressing the individual effects) for each firm/succession event. These estimates are then simultaneously treated at


56 Level 2 as dependent variables, so as to observe the effects the language contained in the letter to the shareholders on the average evaluations produced by each analyst, controlling for the individual-level factors. In order to reduce the level of complexity of the models involved, I eliminated the need for a separate level of analysis for the time factor by averaging by analyst over the specific time window of interest. Thus, for favorability and forecast accuracy, level 1 refers to individual analysts’ data averaged over a time window (6 months after the filing of the letter to the shareholders). For the uniformity of recommendations (which constitutes a group-level variable, since it refers to the uniformity across a group of analysts), I instead included at level 1 the different values assumed by the standard deviation of the recommendations over time. It follows that the structure of the HLM models changes depending on whether level 1 refers to analysts (favorability of recommendations and forecast accuracy), or to time (uniformity of recommendations). In the first case, the model is as follows (for simplicity, I include only the model testing the CCIxReputation moderator hypothesis): Level 1 (analyst): Y ji = 0j + 1j (Y(t-1) ji ) + 2j (Analyst_ctrl ji ) + r jt (3-1) Level 2 (firm): 0j = 00 + 01 (CCI j ) + 02 (REP j ) + 03 (CCIxREP j ) + 04 (CV j ) +u 0j (3-2) 1j = 10 + u 1j (3-3) 2j = 20 + u 2j (3-4) where: Y ji = Average recommendation (or forecast accuracy) of analyst i for firm j within the time window of interest;


57 Analyst ctrl ji = Analyst-level control variables (age, reputation); CCI j = CEO Charismatic Image score for CEO j; REP j = Reputation score for CEO j; CV j = Firm-level and CEO-level control variables (time invariant); Equation 3-1 models recommendations (forecast accuracy) at time t as a function of recommendations (forecast accuracy) at time t-1 (pre-succession), and controls for antecedents of recommendations located at the analyst level (age, reputation, etc.). Equation 3-2 models the average recommendation across all analysts following firm j as a function of CCI, CEO reputation, their interaction, and a set of firm-level control variables. Equations 3-3 and 3-4 are included without predictors, since at this stage the effect of CCI on the change in the dependent variable from before to after the succession, and on the relationship between analyst controls and the dependent variable are not of substantive interest. In the case of the uniformity of analyst recommendations, the model is as follows (for simplicity, I include only the model testing the CCIxReputation moderator hypothesis): Level 1 (time): Y ji = 0j + 1j (TIME) 2j (TIME) 2 j (Time-varying covariates ji ) + r jt (3-5) Level 2 (firm): 0j = 00 + 01 (CCI j ) + 02 (REP j ) + 03 (CCIxREP j ) + 04 (CV j ) +u 0j (3-6) 1j = 10 + u 0j (3-7) 2j = 20 + u 0j (3-8) 3j = 30 + u 3j (3-9) where: Y ji = Standard deviation of recommendation across all analysts firm j at time i;


58 TIME andTIME 2 = linear and square growth terms. Time-varying covariates ij= Analyst-level control variables (age, reputation); CCI j = CEO Charismatic Image score for CEO j; REP j = Reputation score for CEO j; CV j = Firm-level control variables (time invariant); Equation 3-5 models the standard deviation of recommendations at time i as a function of an average standard deviation across all observations, a linear and a squared time vector, and a set of time-varying control variables (average recommendation). Equation 3-6 introduces the firm-level predictors (CCI, reputation and the interaction terms) of the standard deviation of recommendations during the first time period after the succession took place (effect coding were constructed so as to code as time=0 the first period after the succession event). Equations 3-7, through 3-9 were left without predictors.


CHAPTER 4 RESULTS Table 4-1 presents the parameter estimates and variance components for the null models for the three dependent variables. The null model analyses indicated that there was significant between-firm variance in all the three dependent variables (p <.01), suggesting that hierarchical modeling of these data is appropriate. Means, standard deviations, and inter-correlations for all the level 2 variables measured in the study are presented in Table 4-2. The proportion of variance 3 accounted for by firms/CEOs was respectively 15.6% for favorability of recommendations, 77.2% for the uniformity of recommendations, and 84.3% for Forecast accuracy. Table 4-1. Parameter estimates and variance components for the null model Sample/Parameters 00 2 00 Favorability of recommendations 2.163** .628 .116** Uniformity of recommendations .703** .045 .152** Forecast Accuracy .300** .039 .226** Notes : N =146, 247, and 313, for Favorability, Uniformity, and Forecast Accuracy, respectively. ** p < .01. The magnitude of the variance accounted for by the firm-level for uniformity and forecast accuracy (77.2% and 84.3%, respectively) is not surprising, given the characteristics of the two dependent variables. 3 These proportions were calculated as 00/(2+00) 59


Table 4-2. Descriptives and correlations, level2 Mean Std dev 1 2 3 4 5 6 7 8 9 10 11 1. CEO DUALITY .305 .46 2. CEO AGE 50.63 6.98 .228** 3. OUTSIDER .335 .47 .006 .052 4. PREDECESSOR PREDISP. .684 .47 -.389** -.189** .011 5. CHARISMA (RATINGS) 3.46 .14 -.025 .048 .169** .001 6. REPUTATION 1.18 3.42 .087 -.041 .026 -.037 .154** 7. PAST PERFORMANCE (3 yrs change in Std. ROA) .99 10.76 -.004 .072 .065 .020 -.009 .027 8. FORTUNE RANKING .24 .43 .131* .133* -.123* -.089 .031 .254** .000 9. FIRM SIZE (Ln Sales) 6.11 2.21 .088 .103* -.191** -.042 .001 .275** .043 .645** 10. CEO CHARIS. IMAGE (CCI) .04 .02 -.067 -.016 -.035 .108* .159** -.005 -.023 .014 .082 11. CEO TENURE 212.25 140.7 .073 -.019 .046 -.024 .105* .056 .061 .057 .044 -.043 12. CHARISMA (Factor Score) 0.00 1.00 -.078 -.001 .004 .072 .592** .040 -.021 .063 .104* .781** .007 60 Notes : N =367. ** p < .01 (two-tailed), * p < .05 (two-tailed).


61 Indeed, uniformity of recommendations was operationalized as the standard deviation in recommendations among all analysts following a firm, and therefore it is reasonable to observe that a large part of this standard deviation is actually explained by characteristics of the firm itself. Forecast accuracy, on the other hand, is also the result of the difference between forecasts and actual values, which are logically influenced by firm level factors. The parameter estimates for Favorability of recommendations are presented in Table 4-3, while the following tables present the results for Uniformity (Table 4-4) and Forecast Accuracy (Table 4-5). The tables present standardized and unstandardized regression coefficients. To obtain standardized coefficients, I used the standard deviations of the criterion variables (i.e., the square root of the 2 and 00 values presented in Table 4-2) and the standard deviations of the predictor variables (presented in Table 4-2). All of the tables are organized as follows: model 0 constitutes the baseline, and corresponds to a one-way Anova with no predictors. Model 1 adds the independent variables at level 1 (e.g., recommendations for the same analyst averaged over the 6 months before the filing of the letter), model 2 adds the independent and control variables at level 2, while model 3 drops those variables that appear to be insignificant in model 2 (for Favorability, CEO Age was kept throughout the analysis since it presents an almost significant coefficient in model 2: = .322, p = .054). Finally, model 4 adds the interaction terms. Tests for the heterogeneity of variance at level 1 were rejected, so all the models assume a homogeneous 2 . The results for Favorability of Recommendations are presented in Table 4-3, with unstandardized coefficients in parentheses.


62 Table 4-3. Parameter estimates for Favorability of Recommendations (unstandardized coefficients in parentheses) Variables\models 0 1 2 3 4 Intercept 00 2.163** 1.754** 2.161** 2.162** 2.162** Previous recomm. .072** (. 1 73 ** ) .058** (. 1 39 ** ) .058** (. 14 0 ** ) .058** (. 14 0 ** ) Level 2 p redictors : Outsider (dummy) -.151 (-.125) -.136 (-.113) -.137 (-.114) Reputation -.114 (-.013) -.114 (-.013) -.123 (-.014) Change in Zroa (3 yrs) -1.601** (-.058**) -1.739** (-.063**) -1.711** (-.062**) CEO Char Images (CCI) -.234* (-4.567*) -.243* (-4.734*) -.255 (-4.969) CEO Age .322 (.018) .251 (.014) .233 (.013) CEO tenure .257 (.001) CEO Duality (dummy) -.045 (-.038) Fortune Ranking (dummy) -.116 (-.105) Firm Size (LnSales) .079 (.014) Outsider x CCI -.007 (-.003) Change in Zroa x CCI .079 (.042) Reputation x CCI .001 (-.001) Notes : N =146. Avg. n j = 7.3 analysts per firm (min=3, max=19, std dev.=4.2). ** p < .01, * p < .05, p <.1 The results for models 2 and 3 show support for the first hypothesis (H1; see Table 4-3). The coefficient for CEO Charismatic Image (CCI) was negative 4 and significant at 4 The coefficient is negative because of the coding of the dependent variable, where 1=strong buy and 5=strong sell.


63 the p < .05 level in both models ( = .234, p < .018; = .243, p < .018). In model 4, the standardized coefficient for CCI was = .255, with p < .068. These results suggest that the higher the use of a charismatic language, the more favorable the recommendations, as hypothesized. Moreover, the magnitude of this effect is far from being trivial: as the standardized coefficient shows, one standard deviation change in CCI leads to one quarter of a standard deviation change in the dependent variable, controlling for the pre-succession change in firm performance. In order to get an approximate understanding of the magnitudes involved, it is also interesting to note that the variables included in Model 3 explain collectively more than 48% of the variance in recommendations 5 . These results hold controlling for the recommendations in the 6 months before the filing of the letter, and for the change in industry-standardized ROA from 4 years before to 1 year before the succession. Previous recommendations were significant predictors of recommendations in the 6 months following the filing of the letter ( = .072, p < .001 for model 1; = .058, p < .001 for all other models), and the standardized 3 years pre-succession change in ROA was also strongly related to the favorability of analyst recommendations ( = -1.601, p = .008; = -1.739, p = .008; = -1.711, p = .008, for models 2, 3, and 4 respectively). In terms of standardized coefficients, pre-succession change in performance was the strongest predictor of recommendations after the succession, followed by CEO charismatic images. Finally, the results presented in Table 4-3 (model 4) do not support hypotheses 1a, 1b, and 1c, which predicted that past performance (H1a), outsider status (H1b), and CEO 5 This measure, which is analogous to an R2 measure of variance explained, was obtained as follows: % variance explained = [(00mod0 – 00mod3) / 00mod0].


64 reputation (H1c) would moderate the relationship between CCI and the favorability of recommendations. Although the main effects for past performance, outsider status and reputation showed the predicted sign ( = -1.711, p = .008; = -.137, p = .366; = -.123, p = .323, respectively), only past performance reached significance. Moreover, the interaction terms presented weak and insignificant effects, and thus provided no support for the three moderating hypotheses. Table 4-4 presents the parameter estimates for Uniformity of recommendations (standardization of the coefficients was obtained with the same method discussed for the previous dependent variable). In terms of models, the main difference with Favorability consists in the group-level nature of the dependent variable: while favorability of recommendations is a measure of an individual analyst’s perceptions (averaged over 6 months), uniformity of recommendation was operationalized as the standard deviation of recommendations across all analysts following the firm, and calculated once a month. Thus, level 1 includes multiple observations within the same firm nested in time. In order to analyze such data, I adopted an effect coding, where I coded each measure of the dependent variable with an integer expressing the number of months before or after the filing of the letter to the shareholders (from months to +11 months, with 0 = standard deviation of recommendations in the first month after the filing of the letter to the shareholders). It follows that the results presented in Table 4-4 refer to the effect of the level 2 independent variables on the standard deviation of recommendations in the first month after the filing of the letter (i.e., when both the linear and squared growth terms are equal to zero), controlling for the change in recommendations from the previous month.


65 As Table 4-4 shows, results for models 2 and 3 show partial support for the second hypothesis (H2). Indeed, in Model 2 the coefficient for CEO Charismatic Image (CCI) was negative and significant at the p < .05 level ( = -.098, p =.044). In model 3, the standardized coefficient for CCI was = -.084, with p < .098. These results suggest that the higher the use of a charismatic language, the smaller the standard deviation across all recommendations issued by analysts one month after the filing of the letter to the shareholders, as hypothesized. These results hold controlling for the change in standardized ROA from 4 years to 1 year before the succession, which was a significant and negative predictor of the standard deviation of recommendations across all models ( = -.165, p < .001 for model 2; = -.176, p < .001 for model 3; and = -.138, p < .001 for model 4). This result suggests that the higher the increase in performance preceding the succession, the smaller the standard deviation across all recommendations issued by analysts one month after the filing of the letter to the shareholders. In other words, when firm performance increases over time, analysts tend to engage in herding behaviors and their recommendations collectively converge toward the consensus recommendation. In terms of the other control variables at level 1 and level 2, as Table 4-4 shows, none of them did significantly predict the uniformity of recommendations. Finally, the results presented in Table 4-4 (model 4) do not support hypotheses 2a, 2b, and 2c, which predicted that past performance (H2a), outsider status (H2b), and CEO reputation (H2c) would moderate the relationship between CCI and the uniformity of recommendations. Although the main effects for past performance and outsider status showed the predicted


66 sign, they did not reach significance ( = -138, p = .359; = -.094, p = .135, respectively). Table 4-4. Parameter estimates for Uniformity of Recommendations (unstandardized coefficients in parentheses) Variables\models 0 1 2 3 4 Intercept 00 .703** .644** .644** .647** .648** Time (effect code) -.339 (-.019) -.339 (-.019) -.334 (-.019) -.339 (-.019) Time 2 (effect code) .001 (.001) .001 (.001) .001 (.001) .001 (.001) Change in recommendations -.011 (-.044) -.010 (-.043) .011 (-.044) -.011 (-.045) Level 2 predictors: Outsider (dummy) -.049 (-.059) -.087 (-.072) -.094 (-.078) Reputation .001 (.002) .003* (.007*) .002 (.006) Change in Zroa (3 yrs) -.165** (-.006**) -.176** (-.006**) -.138 (-.005) CEO Char Images (CCI) -.098* (-1.918) -.084 (-1.632) -.067 (-1.302) CEO Age .018 (.001) CEO tenure .001 (.001) CEO Duality (dummy) -.031 (-.026) Fortune Ranking (dummy) .100 (.091) Firm Size (LnSales) .147 (.026) Outsider x CCI .098 (.040) Change in Zroa x CCI .017 (.009) Reputation x CCI .035 (.019) Notes : N =247. Avg. n j =22.3 firm/month obs. (min=11, max=25, std dev.=3.2). ** p < .01, * p < .05, p <.1


67 Moreover, and contrary to hypothesis H2c, Reputation showed a positive sign ( = .002) significant at the p = .067 level. Finally, all the interaction terms presented weak and insignificant effects, and thus provided no support for the three moderating hypotheses. In order to test for the robustness of these results, I repeated the analysis including all the level 2 independent variables simultaneously as predictors of the intercept, time slopes, and change in recommendation slope. The results were essentially replicated: in model 3, the standardized coefficient for the effect of CCI on the standard deviation of recommendations in the first month after the release of the letter to the shareholders was = -.190, with p < .017 (other results not reported here). Table 4-5 presents the parameter estimates for Forecast Accuracy (standardization of the coefficients was obtained with the same method discussed for the previous dependent variable). In terms of models presented in the table, the main difference with the previous results consists in the absence of interaction terms. The results presented in Table 4-5 refer to the effects of the level 2 independent variables on the forecast accuracy of individual securities analysts averaged over the first 6 months after the filing of the letter to the shareholders, and control for the forecast accuracy in the previous period, as well as for the horizon of the forecasts issued. Briefly, hypothesis 3 stated that the charismatic image of the CEO would increase the accuracy of securities analysts because it charismatic language conveys informative data on the future state of the firm. As Table 4-5 shows, the results for models 2 and 3 indicate no support for hypothesis 3. Indeed, the coefficient for CCI was = .016, p = 442 and = .018, p=359 for models 2 and 3 respectively, suggesting the absence of a significant relationship.


68 Table 4-5. Parameter estimates for Forecast Accuracy (unstandardized coefficients in parentheses) Variables\models 0 1 2 3 Intercept 00 .298** -.048* -.049* .050* Previous forecast accuracy .240** (.240**) .239** (.239**) .243** (.242**) Forecast Horizon .465** (.002**) .310** (.001**) .311** (.001**) Level 2 predictors: Outsider (dummy) -.027 (-.027) -.027 (-.027) Reputation -.001 (-.001) -.001 (-.001) Change in Zroa (3 yrs) -.210 (-.009) -.133 (-.006) CEO Char Images (CCI) .016 (.368) .018 (.432) CEO Age -.011 (-.001) CEO tenure -.011 (-.001) CEO Duality (dummy) .032 (.033) Fortune Ranking (dummy) -.010 (-.011) Firm Size (LnSales) .006 (.001) Notes : N =313. Avg. n j = 10.7 analysts per firm (min=3, max=50, std dev.=11.6). ** p < .01, * p < .05 In terms of the control variables, the results presented in Table 4-5 show that both forecast accuracy in the previous 6 months ( = .240, p < .001; = .239, p < .001; and = .243, p < .001 for models 1,2 and 3, respectively), as well as the horizon of the forecasts ( = .465, p < .001 for model 1, = .310, p < .001 for model 2, and = .311, p < .001 for model 3) are significantly related to the post-succession accuracy of securities analysts. In particular, these results show that the longer the horizons over which the analyst draws his/her forecast, the higher the absolute magnitude of the error in his/her forecast. None of the other control variables was significantly related to the dependent variable. In order to perform a test of the robustness of these results, I repeated the analysis with a different definition of the dependent variable (raw difference between forecast and


69 actual value, instead of absolute difference), with a different time frame (one year instead of 6 months after the filing of the letter to the shareholders), with a different centering of the level 1 predictors (group-mean centered instead of uncentered predictors), as well as I re-ran the analysis testing for the effects of CCI on the relationships between the dependent variable and the level 1 predictors (intercept-and-slopes-as-outcomes model: Bryk & Raudenbusch, 1992). Across all analyses (results are not reported here), I obtained the same pattern of results, with CCI having no significant effect on the dependent variable. The last set of analyses was dictated by methodological considerations concerning the validity of the text-based measure of charisma. Using the factor scores described in Chapter 3 as a second measure of charisma, alternative to the text-based one, I repeated the analyses presented above. Table 4-6 presents the results for all the dependent variables (to save space, the Table presents results only for models 3 and 4 for each dependent variable of interest, omitting models 0, 1, and 2). As Table 4-6 shows, the results are somewhat similar to those presented in the previous Tables, save for the effects of CEO charisma on the uniformity of recommendations. Similarly to CCI, the charisma factor shows a significant effect on the Favorability of recommendations ( = -.109, p < .05; = -.136, p < .05 for models 3 and 4 in Table 4-6), as well as no significant effect on Forecast Accuracy ( = -.006, p = .878; = -.007, p = .857 for models 3 and 4 in Table 4-6). The main difference with the previous set of results is the absence of an effect of the charisma factor on the uniformity of


70 recommendations, despite the sign of the relationship is consistent with the previous set of results ( = -.081, p = .147; = -.065, p = .249 for models 3 and 4 in Table 4-6). Table 4-6. Charisma factor, parameter estimates for all dependent variables (unstandardized coefficients in parentheses) Favorability Uniformity Forecast Accuracy Variables \ Models 3 4 3 4 3 4 Intercept 00 2.160** 2.156** .647** .647** .300** .300** Previous recommendation .149* (.139*) .149* (.139*) Time (effect code) -.142 (-.019) -.142 (-.019) Time 2 (effect code) -.081 (-.001) -.082 (-.001) Change in Recommendations -.011 (-.044) -.011 (-.047) Previous Forecast Accuracy .154** (.154**) .154** (.154**) Forecast Horizon .313** ( .001** ) .313** ( .001** ) Level 2 predictors: Outsider (dummy) -.057 (-.105) -.078 (-.143) -.076 (-.068) -.084 (-.075) -.075* (-.069*) -.078* (-.071*) Reputation -.060 (-.013) -.014 (-.003) .067* (.007*) .063 (.007) .014 (.001) .012 (.001) Change in Zroa (3 yrs) -.437** (-.057**) -.399** (-.052**) -.110** (-.007**) -.012 (-.001) -.192 (-.013) -.200 (-.013) Charisma Factor (CCI, Ratings) -.109* (-.093) -.136* (-.116*) -.081 (-.034) -.065 (-.027) -.006 (-.003) -.007 (-.003) Tenure .092 ( .001 ) .103* ( .001* ) CEO Age .115* ( .015* ) .122* ( .016* ) Outsider x Charisma factor -.077 (-.070) .036 (.016) .015 (.007) Change in Zroa x Charisma factor .080 (.075) .094 (.043) -.029 (-.014) Reputation x Charisma factor .051 (.043) .049 (.020) .009 (.004) Notes : N =146, 247, and 313 for favorability, uniformity, and forecast accuracy, respectively. Avg. n j = 2.6, 7, and 10.8. ** p < .01, * p < .05, p <.1


71 Moreover, as Table 4-6 shows, the outsider status of the CEO was a significant, and negative, predictor of the forecast accuracy of securities analysts ( = -.075, p < .05; and = -.078, p < .05 for models 3 and 4 in Table 4-6). In other words, the forecasts issued by analysts following a firm led by an outsider are more accurate that those issued by analysts following a firm led by an insider. Similarly, reputation was a significant, and positive, predictor of the uniformity of recommendations ( = .067, p < .05; and = .063, p = .073 for models 3 and 4 in Table 4-6), so that CEOs with a stronger reputation result in larger standard deviations among analysts following the firm. In other words, more reputed CEOs trigger a lower uniformity of judgments among analysts. Regarding the other predictor and control variables, the pattern of results presented in Table 4-6 substantially reproduces the results presented in the previous Tables. The results presented in Table 4-6 suggest several considerations. In terms of the construct validity of the text analysis, the results suggest that sensitivity to members and shareholders might constitute the facets of charisma captured by CCI. Further studies should better clarify such a relationship between the sensitivity to members and shareholders, as perceived by student raters, and the textual utterances contained in official corporate documents. Second, the results presented in Table 4-6 constitute an ulterior test of the robustness of the results, suggesting that CEO charisma, whether measured through text or through ratings, affects the recommendations of securities analysts without significantly increasing their accuracy. All in all, the main conclusions of the study are confirmed.


CHAPTER 5 DISCUSSION AND CONCLUSIONS Summarizing the results of the previous section, the analysis shows that the main variable of interest of this study, CEO Charismatic Images (CCI), has important effects on the individual and group judgments of securities analysts following the firm. First, stronger emphases on charismatic language result in more favorable recommendations by individual analysts. This result alone tells us that the use of charismatic language affects securities analysts in ways that lead them to present the firm to the investor public in more favorable ways. Further, the magnitude of this effect is far from being irrelevant. Among all the predictors, only past performance has a greater effect than charismatic language. Given that analyst recommendations are an important antecedent of stock market reactions (Beneish, 1991; Francis & Soffer, 1997), the symbolic representation of CEO charisma has the potential to affect, at least in the short run, the stock market value of the firm through securities analyst judgments. Future studies testing different time windows should allow a better understanding of the long run effects. This study did not test directly whether a charismatic language has an effect on investor decisions to buy or sell stocks of firms led by charismatic CEOs. However, it is well known that investor decisions, and therefore stock prices, are strongly influenced by securities analyst recommendations (Barber, Lehavy, McNichols, & Trueman, 2001; Beneish, 1991; Desai, Liang, & Singh, 2000; Francis & Soffer, 1997; Palmon, Sun, & Tang, 1994; Scott, Sudip, Iskandar-Datta, & Mai, 1995). Further research should be conducted on the relationship between charismatic language, investor decisions, and stock prices. 72


73 Second, the results presented in Table 4-4 suggest that the effect of charismatic language is not limited to the individual reactions of securities analysts, but extends to collective analysts’ perceptions of the firm. That is, investors in a firm led by a charismatic CEO will receive more uniform opinions from securities analysts, in addition to more favorable recommendations. Since it is known that securities analysts are not immune from processes of imitation, social contagion, and herding (Cote & Goodstein, 1999; Hayward & Boeker, 1998; Hong et al., 2000; Rao et al., 2001), this study suggests that CEO charisma might be one factor triggering such collective reactions. What is new for the study of analyst herding behaviors is that while herding occurs as a rational strategy employed by analysts to protect their reputation (Hong et al., 2000), it might be exacerbated as a result of a psychological reaction to charismatic language. Thus, it may be triggered by symbolic management tactics employed by corporate executives. The third finding of the study is the possibility that the rosy portrait of firms led by charismatic CEOs provided by securities analysts is an emotional reaction not corresponding to the actual prospects of the firm. Indeed, charismatic language does not seem to have tangible effects on forecast accuracy. As the results in Table 4-5 suggest, securities analysts issuing forecasts for firms led by charismatic CEOs are not any more accurate that those following less charismatic CEOs. Thus, because of the more favorable recommendations, investors are more likely to be misled. It is important to note that the results for Forecast Accuracy cannot be explained either by the lack of statistical power or by the specific analytical model used. Indeed, both at level 1 and at level 2, the sample available for Forecast Accuracy was much larger in magnitude than the samples for Favorability and Uniformity (313, 146, and 247, respectively). Moreover, the results


74 presented are robust to multiple specifications of the model, multiple specifications of the time frame, of the centering of the variables as well as to different definitions of the dependent variable. The absence of a relationship between the charismatic language and Forecast Accuracy might be explained by the so-called “consistency motive”, as demonstrated in Chen and Meindl’s (1991) description of the role of the news media in portraying Donald Burr of People Express. Despite his performance failures, there was a media persistence in supporting Burr that was attributed, among other things, to schematic knowledge through which performance cues would be selectively processed. This argument can be extended to securities analysts. Financial analysts probably have similar incentives to maintain a consistency of attributions, and therefore credibility, among the public. Once analysts form a charismatic attribution, they might be motivated to maintain their forecasts firm in the face of contradicting information so to maintain consistency with previous attributions (Salancik & Pfeffer, 1977) and, ultimately, credibility within the investor public. In other words, once analysts form a charismatic attribution, this will color successive attributions and lead to overestimates of the performance potential of the firm. Performance evaluation scholars have long recognized the importance of automatic categorization of individuals: once an employee performing acceptably is assigned to a category, further memory-based judgments of that employee are colored by the category prototype (Feldman, 1981). Finding that the Favorability and Uniformity of recommendations are not accompanied by increased accuracy constitutes evidence that charismatic language reflects a decoupling of the actual performance of the firm from what is communicated outside.


75 Last, as Table 4-6 suggests, a different operationalization of CEO charisma provides some supports for the results obtained with the text analysis. When assessed using a variation of the Conger-Kanungo scale, I found that CEO Charisma results in more Favorable recommendations, but has no significant effects on the Forecast Accuracy of individual analysts or on Uniformity. Theoretical Implications These results have several implications for charismatic Leadership Theory. First, CEO charisma does not affect only the internal members of the organization such as employees and board members. It also affects an entirely different population of “followers”, external stakeholders, in ways that could possibly be just as relevant as those occurring within the firm. External constituents are qualitatively different from internal followers, and these differences should be taken into account by developing theoretical frameworks and measures capable of capturing how the charismatic relationship unfolds in external contexts. For example, researchers could extend their reach to the perceptions and judgments of important external stakeholders such as stock market actors, investors, government regulators, by identifying appropriate dependent variables. That might produce some interesting insights about charismatic leadership theory. Moreover, the results for Uniformity of recommendations warrant further study on the collective reactions of external stakeholders to CEO charisma. Several authors pointed to the variance in perceptions of charisma among internal followers as an important research topic (Klein & House, 1995; Meindl, 1990). Studying how collective perceptions of charisma operate outside of the firm might have a higher practical relevance than studying the same phenomenon within the organization. Within the firm, behaviors, and therefore job performance, are bound to a degree by regulations, technologies, work


76 roles, etc. (Kerr & Jermier, 1978), so the social contagion of charisma has also limited effects on work behaviors and job performance. In the external context, in contrast, stakeholder reactions are not bound by these structural constraints, and social contagion effects, as collective perceptions, drive individual investment decisions that have far-ranging consequences for corporations and society (Davis & McAdam, 2000). Second, looking at these external stakeholders, as well as to internal members, might provide some understanding of the different effects of CEO charisma on different audiences. For example, this study found that CEO charisma might affect stock prices through the investment recommendations of securities analysts (Beneish, 1991; Francis & Soffer, 1997). These results might explain Tosi and colleagues’ (2002) finding that CEO charisma affects the stock market performance (i.e., change in the market value of the firm) but not the internal performance of the firm. It is entirely possible that the effects of CEO charisma on these different audiences (e.g., internal vs. stock market audiences) might be mediated by different factors. That is, the relationship between CEO charisma and internal performance might be mediated by employee effort and board dynamics, while the relationship between CEO charisma and stock market performance might be mediated by stock market perceptions, particularly those of securities analysts. It is entirely plausible that research might find significant relationships in one context and nonsignificant ones on the other, as Tosi and colleagues (2002) did. Ultimately, to extend the reach of CLT into the external environment of the firm researchers need to extend the assumption that “charisma is particularly important at the top executive level as a means of mobilizing an organization to meet the demands of its environment” (Waldman et al., 2001: 135).


77 A third implication for Charismatic Leadership Theory is methodological. This study focused specifically on verbal and symbolic behavior, presenting evidence that measuring directly charismatic language is not only possible, but also a potentially useful methodological solution to the problem of separating the measurement of charisma from the perceptions of the audiences. As Shamir et al. argued, most studies of charisma “have often blurred the distinction between the behaviors of a leader and their effects on followers” (Shamir et al., 1998: 404). Separating these two elements allows not only a clearer distinction between the predictor and the outcomes, but also allows studying specific classes of behaviors, such as verbal behaviors, that are of interest in and of themselves. This implies that the association between the message and the interpretation should be studied empirically, rather than embedded in the definition and/or measurement of charisma (e.g., as in House, 1977). Moreover, external audiences rely mostly on written documents to reach a judgment about the CEO. Studying language, as expressed in written documents, is therefore a viable strategy to gauge the level of charisma of the CEO projected toward external audiences. Researchers studying the external effects of CEO charisma should focus on language and its processes, because those are the vehicles through which charismatic leadership exerts its effects on external actors. Fourth, Charismatic Leadership Theory should modify its assumptions on the nature of executive leadership. Top executives are not simply lower level managers with more power. Rather, they constitute a qualitatively distinct category, dealing with and managing the external environment above and beyond the internal one (Pfeffer & Salancik, 1978). It is not surprising that “the organizational and environmental context in which leadership is enacted has been almost completely ignored [by leadership research]”


78 (House & Aditya, 1997: 445), given that most CLT research placed an excessive focus on “superior-subordinate relationships” (House & Aditya, 1997: 465). In order to overcome these limitations, Charismatic Leadership Theory might profit from a more complex view of the role of the executive as part of an institutional level, distinct from the managerial one for its specific concentration on actively managing resource dependencies with the external environment (Pfeffer & Salancik, 1978; Selznick, 1957; Thompson, 1967). Fifth, there are some implications for assessing organizational performance. I chose to focus on an important antecedent of stock market performance, the individual and collective judgments of securities analysts. Hence, the results of the study do not allow drawing generalizations on the relationship between CEO charisma and most conventional measures of firm performance. However, given the relevance of analyst recommendations and forecasts as antecedents of stock market performance it is possible to reflect, in general, on issues related to the study of the charisma-firm performance relationship. Just as for job performance, an important “criterion problem” underlies firm performance (Austin & Villanova, 1992): firm performance is not a simple algebraic sum of the individual and group task performances of the members of the organization. Rather, and this is particularly critical when dealing with the stock market, it is a complex system of loosely connected components, where substantive performance is neither easily distinguished from, nor necessarily correlated with, symbolic performance (Meyer & Gupta, 1994; Westphal & Zajac, 1998). This study suggests that research on the external effects of CEO charisma should specify carefully the component of firm performance of interest, its relationship with other components, and its relevance for both the organization and external stakeholders – for the two don’t necessarily overlap. On the


79 stock market, for example, investor perception is performance, and those perceptions are not necessarily accurate judgments of what happens within the firm. Finally, the results of this study have implications for symbolic management and the social scientific study of the stock market. This study illustrates one symbolic way through which organizations affect their environments – by constructing a charismatic image around their CEOs and then projecting such an image onto the stock market. Thus, it suggests a renewed and deeper understanding of the discourse, language and rhetoric of charisma and ultimately points to the external environment of organizations as a legitimate and important object of study. It is a step toward the rehabilitation of “a line of inquiry set adrift by organization theory in its formative years: the study of how organizations affect the social systems in which they are embedded (Stern & Barley, 1996: 146).” While other studies have shown that the stock market reacts to symbolic management (Westphal & Zajac, 1994), this study suggests that the social contagion effects of CEO charisma can be triggered by the careful use of words and text as a symbolic management tactic hinging upon the projection of a charismatic persona and vision. This may be valuable for firms, but may not be as valuable for analysts and investors. Implications for Practice This study raises several practical implications. From the viewpoint of corporations, a charismatic CEO should be seen as something more than just an effective means of obtaining commitment and performance from employees (Mohrman & Cohen, 1995). Rather, a charismatic CEO can provide for corporations what today matters a great deal: coordinating with and obtaining trust (and therefore resources) from external institutions and actors (Davis & McAdam, 2000; Elsbach, 1994; Hirsch, 1986;


80 Zuckerman, 1999, 2000). This study provides preliminary evidence that to the extent that (and as long as) external stakeholders believe in the charismatic CEO, performance lapses may be overlooked, and mistakes forgiven, at least for a period of time. A second practical implication of this study is that what is desirable for executives is not necessarily desirable for securities analysts and investors alike. From the viewpoint of securities analysts, “earnings forecast accuracy is one measurable performance characteristic that establishes analyst reputation (Stickel, 1992: 1813)”. As the results of the study show, analysts who attend to the charismatic language of the CEO to craft their recommendations do not derive from it an increased earnings forecast accuracy. Thus, paying attention to CEO charisma might be a less than desirable strategy for analysts. From the viewpoint of investors, a charismatic CEO might not necessarily be “money in the bank (Kadlec, 1998),” as the story of Al Dunlap shows. In a society increasingly preoccupied with accountability, corporate governance, and shareholder value, a charismatic CEO might not be the best bet, if all that stems from it are more favorable and more uniform perceptions but not more accurate ones. Some Caveats for Future Research Finally, some general issues for future research warrant consideration. First, since the study focuses explicitly on securities analysts, the results might not generalize to different audiences (e.g., fund managers, individual investors, etc.). Their reactions to CEO charisma should be studied after a careful specification of the model underlying their respective judgment tasks, but the study of the reactions of those subjects seems very promising at this point. Second, since the study was conducted on US firms, its conclusions might not generalize outside Anglo-Saxon contexts (Chen & Meindl, 1991). Third, both CEOs and securities analysts operate in predominantly male environments.


81 The results of this study might therefore be a depiction of the particular male cultures of those environments (Cals & Smircich, 1991). Fourth, the fact that analysts respond to charismatic language does not imply the absence of other symbolic management tactics operating within the information environment of securities analysts and affecting their perceptions (Rao et al., 2001). Last, generalizations concerning the Uniformity of Recommendations should be drafted carefully since the results were not replicated across the two different definitions of the Charisma variable (text-based and factor scores). Moreover, the results for uniformity of recommendations could stem from a statistical artifact: to the extent that CEOs are rated more favorably by analysts, average ratings will tend to be closer to the upper end of the scale, and therefore, show lower variance which could account for the results presented. Future research should focus on the relationship between the mean recommendation (i.e., Favorability) and its variance (i.e., Uniformity of Recommendations).


APPENDIX A CONCEPTUAL NODES As described in the first chapter, CCI is comprised of three additive components: evaluation of the status quo, formulation and articulation of goals, and means to achieve the vision. These components correspond to separate conceptual nodes developed to analyze the letters to the shareholders: “Assessment of the past”, “Plans for the future”, and “Shareholders, employees and organizational capabilities.” The names of the nodes are intentionally neutral, so as to avoid cuing the coders about the underlying theory and constructs of interest. As far as the CEO’s assessment of the past is concerned, coders were instructed to assign to this conceptual node all sentences in the letter that were consistent with two criteria: a. the sentence describes some event initiated in the past and concluded at the time of the letter; b. the sentence includes an evaluation of the event. Of particular interest was the CEO’s assessment of the firm’s past performance. An important issue here has to do with the timing of the succession event and of the letter to the shareholders. The letter to the shareholders is generally issued and available to the public a couple of months after the end of the fiscal year. At this time, the newly appointed CEO has already spent several months in his/her new role, so his assessment of the past might also include actions initiated during his tenure and according to his/her decisions. Since it is unlikely, to say the least, that a CEO in the first year of his/her tenure would use less than positive expressions when referring to the results his/her own decisions, the inclusion in the first node of references to these type of actions would bias 82


83 the analysis, underestimating the degree to which the CEO expresses a negative evaluation of the status quo. Since the letter to the shareholder does not indicate the decision maker who initiated a specific action in the past, it was assumed that all actions initiated in the past and still ongoing at the time of the letter could be either attributed directly to decisions of the newly appointed CEO or, indirectly, to decisions made by the predecessor and subsequently endorsed by the new CEO. Therefore, all the references to past actions still ongoing had to be excluded from the first conceptual node and included in the second. As far as the second conceptual node (“Plans for the future”) is concerned, coders were instructed to assign to this node all sentences that: a. referred to the CEO’s strategy, vision, mission, for the years to come, or to actions initiated in the past and still ongoing in the present; b. contained an exhortation (in the form of “we should, ought, must”) or a prediction of the future state of the firm, either tangible (e.g., “we will reduce debt by 12% within the end of the year”) or intangible (e.g., “What we're doing, really, is building a new IBM -new culture, new directions, new spirit”). The third and last conceptual node was constructed so as to tap into the language rallying the support of internal and external constituents by referring to employees, shareholders, and the organization as a whole.


APPENDIX B CONCEPT CATEGORIES AND SEARCH DICTIONARIES As illustrated above, the three conceptual nodes were analyzed with five separate search dictionaries, one for the charismatic assessment of the status quo, two for the charismatic vision, and two for the charismatic implementation. The first dictionary, measuring the charismatic evaluation of the status quo, includes 184 terms and expressions from three different sources. From the NegAff category of the Lasswell Value Dictionary (Weber, 1988), I included 93 words of negative affect "denoting negative feelings and emotional rejection” (e.g., “awful”, “collapse”, “detrimental”, etc.). From Abrahamson and Park (1994), I included 56 negative words denoting a negative organizational outcome (e.g., “sluggish”, “disappointing”, “downturn”, “inability”, “worst”, etc.). To these terms, I added a list of 35 words and expressions not included in the previous two lists but frequently recurring in the letters to the shareholders in conjunction with negative assessments of the status quo (e.g., “bureaucratic”, “unacceptable”, “terrible”, “lags”, etc.). The charismatic vision was measured with two separate dictionaries, one focusing on the use of moral or ideological language (Conger & Kanungo, 1998), the other on emotion and image-based language (Emrich, Brower, Feldman, & Garland, 2001). The use of two separate dictionaries, as opposed to their merging into one single dictionary, allows to test whether the two types of language are actually correlated in the language of chief executive officers. The dictionary focusing on moral and ideological terms included 146 terms from three different sources. From the Rectitude category of the Lasswell 84


85 Value Dictionary, I included 98 terms concerned with moral values (e.g., “believe”, “discipline”, “duty”, “sincere”, “trust”, “pledge”, etc.). From the Ought category of the Harvard IV Psychosocial Dictionary (Weber, 1988), I included 18 terms indicating a moral imperative within a general cognitive orientation (e.g., “must”, “should”, “ought”, “imperative”, etc.). To these terms, I added a list of 30 words and expressions not included in the previous two lists but frequently recurring in the letters to the shareholders with an ideological or moral connotation (e.g., “leadership”, “vision”, “transformation”, “tough”, “turnaround”, etc.). The second dictionary measuring the charismatic vision focused on emotional language and included 709 terms and expressions from three different sources (Harvard IV-4, Lasswell, and expressions from the letters to the shareholders). From the Ovrst category of the Harvard IV-4 Dictionary, I included 313 overstatement words reflecting presence of emotional expressiveness and “indicating emphasis in realms of speed, frequency, causality, inclusiveness, quantity or quasi-quantity, accuracy, validity, scope, size, clarity, exceptionality, intensity, likelihood, certainty and extremity” (from the General Inquirer homepage), such as “always”, “clear”, “coherent”, “decisive”, “indisputable”, “urgent”, etc. From the Emot category of the Harvard IV-4 Dictionary, I included 169 words related to emotions, such as “excited”, “enthusiasm”, “feel”, “faith”, “passion”, “regret”, etc. From the Arousal category of the Harvard IV-4 Dictionary, I included 53 words indicating arousal of affiliation and hostility, such as “challenge”, ”inspiration”, “motivate”, “optimism”, “ready”, etc. From the Feel category of the Harvard IV-4 Dictionary, I included 30 words describing particular feelings, including gratitude, apathy, and optimism, such as “fervor”, “resolute”, “vigilant”, etc. With respect


86 to the Lasswell Value Dictionary, I included, from the Afftot category, 111 terms indicating affection and friendship, such as “allegiance”, “care”, “loyalty”, “zeal”, etc. To these terms, I added a list of 33 words and expressions not included in the previous dictionaries but frequently recurring in the letters to the shareholders with an emotional overtone (e.g., “dramatic”, “exciting”, “milestone”, “record setting”, “spectacular”, etc.). As far as the last conceptual category is concerned (“charismatic implementation”), Conger and Kanungo (1998) argued that charismatic leaders express high expectations from followers, confidence in their abilities and concern for their needs. I therefore constructed two separate dictionaries. For the first one, I included 383 terms from two sources: 60 words of positive affect “denoting positive feelings, acceptance, appreciation and emotional support” (General Inquirer homepage), such as “bright”, “rejoice”, “reward”, etc from the PosAff category of the Lasswell Value Dictionary. To these, I added 323 terms indicating affiliation or supportiveness (e.g., “admire”, “affection”, “cohesion”, “passion”, etc.) from the Affil category of the Harvard IV Psychosocial Dictionary. The second dictionary included a list of 168 words and expressions not included in the previous two lists but frequently recurring in the letters to the shareholders with reference to shareholders (23 terms), employees (84 terms), customers and suppliers (18 terms), society and the government (27 terms), and managerial fashions (16 terms). Examples of these terms are: “accountability”, “committed”, “distributors”, “humanity”, and “Quality.” Maintaining the two dictionaries separate allows controlling the presence of different linguistic patterns across the documents. Last, to reduce the risk of inflating and/or biasing the scores, I proceeded to check manually and remove from each of the dictionaries terms that: a. are inconsistent with the


87 theory underlying the study; b. might assume different meanings in the context of the letter to the shareholders (e.g., technical terms specific to the industry, business terms, etc.); c. duplicated terms present more than once in the same dictionary. As an example, from the emotions and image-based dictionary I eliminated terms such as “above”, “billion”, “board”, “capital”, “exempt”, “just”, “strike”, “interest”, etc. Overall, this operation led to the elimination of 89 terms from the negative assessment dictionary (32% of the initial total), 88 terms from the morality/ideology (24% of the initial total), 623 terms from the image based/emotions dictionary (43% of the initial total), and 179 terms from the shareholders dictionary (21% of the initial total).


APPENDIX C ITEMS OF THE CONGER-KANUNGO SCALE Table C-1 reports the items from the Conger-Kanungo scale (Conger & Kanungo, 1998) used in the study to obtain ratings of CEO charisma, while Table C-2 reports the items included in the scale but not used in the study. Table C-1. Items from the C-K scale used in the study # Item: “The CEO portrayed in the documents you’ve just read” Scale 5 ....readily recognizes barriers/forces within the organization that may block or hinder achievement of his/her goals. 11 ...readily recognizes constraints in the physical environment (technological limitations, lack of resources, etc.) that may stand in the way of achieving organizational objectives. 14 ...readily recognizes constraints in the organization's social and cultural environment (cultural norms, lack of grassroots support, etc.) that may stand in the way of achieving organizational objectives. 22 ...recognizes the abilities and skills of other members in the organization. 26 ...readily recognizes new environmental opportunities (favorable physical and social conditions) that may facilitate achievement of organizational objectives. 27 ...recognizes the limitations of other members in the organization. 7 entrepreneurial: seizes new opportunities in order to achieve goals. Environmental sensitivity 13 ...provides inspiring strategic and organizational goals. 16 inspirational; able to motivate by articulating effectively the importance of what organizational members are doing. 17 ...consistently generates new ideas for the future of the organization. 18 an exciting public speaker. 24 ...appears to be a skillful performer when presenting to a group. 25 ...has vision; often brings up ideas about possibilities for the future. Vision and articulation 4 ...influences others by developing mutual liking and respect. 8 ...shows sensitivity for the needs and feelings of other members in the organization. 19 ...often expresses personal concern for the needs and feelings of other members of the organization. Sensitivity to members 29 ...shows sensitivity for the shareholders. 30 ...influences shareholders by developing mutual liking and respect. 31 ...often expresses personal concern for the shareholders. Sensitivity to Shareholders (adapted from CK scale) 12r ...advocates following non-risky, well-established courses of action to achieve organizational goals (reverse coded). 20r ...tries to maintain the status quo or the normal way of doing things (reverse coded). Status quo 88


89 Table C-2. Items from the C-K scale used in the study (continued) # Item Scale 10 pursuing organizational objectives, engages in activities involving considerable self-sacrifice. 15 ...takes high personal risks for the sake of the organization. 23 ...often incurs high personal costs for the good of the organization. 28 pursuing organizational objectives, engages in activities involving considerable personal risk. Personal Risk 6 ....engages in unconventional behavior in order to achieve organizational goals. 9 ...uses nontraditional means to achieve organizational goals. 21 ...often exhibits very unique behavior that surprises other members of the organization. Unconventional Behavior


LIST OF REFERENCES Abrahamson, E., & Park, C. 1994. Concealment of negative organizational outcomes: an agency theory perspective. Academy of Management Journal, 37(5): 1302--1334. Angwin, J., & Peers, M. 2001. Cold calls: AOL may be snubbing Merrill. The Wall Street Journal, Nov. 1st: C1,C2. Arndt, M., & Bigelow, N. 2000. Presenting structural innovation in an institutional environment: hospital's use of impression management. Administrative Science Quarterly, 45: 494--522. Arnold, J. E. 1988. Communications and strategy: the CEO gets (and gives) the message. Public Relations Quarterly, Summer: 5--13. Austin, J. T., & Villanova, P. 1992. The criterion problem: 1917-1992. Journal of Applied Psychology, 77(6): 836-874. Baker, W. E. 1984. The social structure of a national securities market. American Journal of Sociology(89): 777--811. Balog, S. J. 1991. What an analyst wants from you. Financial Executive, July-August: 47--52. Barber, B., Lehavy, R., McNichols, M., & Trueman, B. 2001. Can investors profit from the prophets? Security analyst recommendations and stock returns. The Journal of Finance, 56(2): 531. Bass, B. M., Avolio, B., & Goodheim, L. 1987. Biography and the assessment of transformational leadership at the world class level. Journal of Management, 13: 7---19. Beneish, M. 1991. Stock prices and the dissemination of analysts' recommendations. Journal of Business, 64: 393--416. Biggart, N. W. 1989. Charismatic Capitalism: Direct Selling Organizations in America. Chicago: University of Chicago Press. Blasi, A. J. 1991. Making Charisma: the Social Construction of Paul's Public Image. Brunswick, NJ: Transaction Publishers. Branson, B. C., Guffey, D. M., & Pagach, D. P. 1998. Information conveyed in announcements of analyst coverage. Contemporary Accounting Research, 15(2): 119--143. Bryk, A. S., & Raudenbusch, S. W. 1992. Hierarchical Linear Models: Applications and Data Analysis Methods. Oxford: SAGE. 90


91 Butler, K. C., & Lang, L. H. P. 1991. The forecast accuracy of individual analysts: evidence of systematic optimism and pessimism. Journal of Accounting Research, 29(1): 150-156. Bycio, P., Hackett, R. D., & Allen, J. S. 1995. Further assessment of Bass's (1985) conceptualization of transactional and transformational leadership. Journal of Applied Psychology, 80(4): 468--478. Cals, M., & Smircich, L. 1991. Voicing seduction to silence leadership. Organization Studies, 12(4): 567--602. Cannella, A. A., & Lubatkin, M. 1993. Succession as a sociopolitical process: internal impediments to outsider selection. Academy of Management Journal, 36(4): 763-793. Cannella, A. A., & Shen, W. 2001. So close and yet so far: promotion versus exit for CEO heirs apparent. Academy of Management Journal, 44(2): 252--270. Carleen, H. 1997. Pan-industrial leaders, Forbes, Vol. 159: 18--20. Charam, R., & Colvin, G. 2000. The right fit. Fortune, April 17: 226--239. Chen, C. C., & Meindl, J. R. 1991. The construction of leadership images in the popular press: the case of Donald Burr and People Express. Administrative Science Quarterly, 36: 521--551. Cheney, K. H., & Christensen, L. T. 2001. Organizational identity: linkages between internal and external communication. In F. M. Jablin, & L. L. Putman (Eds.), The New Handbook of Organizational Communication: 231--269. Thousand Oaks: CA: SAGE. Chung, K. H., & Jo, H. 1996. The impact of security analysts' monitoring and marketing functions on the market value of firms. Journal of Financial and Quantitative Analysis, 31(4): 493--512. Cialdini, R. B. 1995. Influence: the Psychology of Persuasion. New York: Quill. Clapham, S. E., & Schwenk, C. R. 1991. Self-serving attributions, managerial cognition, and company performance. Strategic Management Journal, 12(1991): 219--229. Clemente, H. A. 1988. How to understand security analysts -their needs, their motives. Financial Executive(Nov--Dec): 41--45. Conger, J. A., & Kanungo, R. N. 1987. Toward a behavioral theory of charismatic leadership in organizational settings. Academy of Management Review, 12(4): 637--647. Conger, J. A., & Kanungo, R. N. 1998. Charismatic Leadership in Organizations. Thousand Oaks:CA: SAGE. Cote, J., & Goodstein, J. 1999. A breed apart? Security analysts and herding behavior. Journal of Business Ethics, 18(3): 305. Crystal, G. S. 1992. Let's hear it for John Reed! Financial World, January 7: 126-127.

PAGE 100

92 D'Aveni, R. A., & MacMillan, I. C. 1990. Crisis and content of managerial communications: a study of the focus of attention of top managers in surviving and failing firms. Administrative Science Quarterly, 35(1990): 634--657. Davis, G. F., & McAdam, D. (Eds.). 2000. Corporations, classes and social movements after managerialism. (Vol. 22). Greenwich: CT: JAI Press. Davis, G. F., & Thompson, T. A. 1994. A social movement perspective on corporate control. Administrative Science Quarterly, 39(1994): 141--173. DeGroot, T., Kiker, D. S., & Cross, T. C. 2000. A meta-analysis to review organizational outcomes related to charismatic leadership. Revue Canadienne des Sciences de l'Administration -Canadian Journal of Administrative Sciences, 17(4): 356--371. Desai, H., Liang, B., & Singh, A. K. 2000. Do all-stars shine? Evaluation of analyst recommendations. Financial Analysts Journal, 56(3): 20. DiMaggio, P. 1997. Culture and cognition. Annual Review of Sociology, 23: 263--287. Economist. 2002. Fallen idols: the world is falling out of love with celebrity chief executives. The Economist, May 4th-10th: 11. Elsbach, K. D. 1994. Managing organizational legitimacy in the California cattle industry: the construction and effectiveness of verbal accounts. Administrative Science Quarterly, 39: 57--88. Emrich, C. G., Brower, H. H., Feldman, J. M., & Garland, H. 2001. Images in words: presidential rhetoric, charisma and greatness. Administrative Science Quarterly, 46: 527--557. Fairhurst, G. T. 1993. Echoes of the vision: when the rest of the organization talks Total Quality. Management Communication Quarterly, 6(4): 331--371. Fama, E. F. 1980. Agency problems and the theory of the firm. Journal of Political Economy, 88: 288-307. Feldman, J. M. 1981. Beyond attribution theory: cognitive processes in performance appraisal. Journal of Applied Psychology, 66(2): 127--148. Fiol, M. C. 1989. A semiotic analysis of corporate language: organizational boundaries and joint venturing. Administrative Science Quarterly, 34: 277--303. Forest, S. A. 2001. Can an outsider fix J.C. Penney? Business Week, February 12: 56-58. Francis, J., & Soffer, L. 1997. The relative informativeness of analyst's stock recommendations and earning forecast revisions. Journal of Accounting Research, 35(2): 193--211. Friedman, S. D., & Singh, H. 1989. Ceo Succession and Stockholder Reaction the Influence of Organizational Context and Event Content. Academy of Management Journal, 32(4): 718-744. Gardner, W. L., & Avolio, B. J. 1998. The charismatic relationship: a dramaturgical perspective. Academy of Management Review, 23(1): 32--58.

PAGE 101

93 Graham, J. R. 1999. Herding among investment newsletters: theory and evidence. Journal of Finance, 54(1): 237-268. Hambrick, D. C., & Abrahamson, E. 1995. Assessing managerial discretion across industries: A multimethod approach. Academy of Management Journal, 38(5): 14-27. Hambrick, D. C., & Fukutomi, G. D. S. 1991. The seasons of a CEO's tenure. Academy of Management Review, 16: 719--742. Hater, J. J., & Bass, B. M. 1988. Superiors' evaluations and subordinates' perceptions of transformational and transactional leadership. Journal of Applied Psychology, 73(4): 695--702. Hayward, M. L. A., & Boeker, W. 1998. Power and conflict of interest in professional firms: evidence from investment banking. Administrative Science Quarterly, 43(1998): 1--22. Hayward, M. L. A., & Hambrick, D. C. 1997. Explaining the premiums paid for large acquisitions: evidence of CEO hubris. Administrative Science Quarterly, 42: 103-127. Hegele, C., & Kieser, A. 2001. Control the construction of your legend or someone else will: an analysis of texts on Jack Welch. Journal of Management Inquiry, 10(4): 298-309. Hirsch, P. 1986. From ambushes to golden parachutes: corporate takeovers as an instance of cultural framing and institutional integration. American Journal of Sociology, 91(4): 800--837. Hong, H., Kubik, J. D., & Solomon, A. 2000. Security analyst's career concerns and herding of earnings forecasts. Rand Journal of Economics, 31(1): 121--144. House, E. R. 1988. Jesse Jackson & the politics of charisma : the rise and fall of the PUSH/Excel program. Boulder: Westview Press. House, R. J. 1977. A 1976 theory of charismatic leadership. In J. G. Hunt, & L. L. Larson (Eds.), Leadership: the Cutting Edge: 189--207. Carbondale: Southern Illinois University Press. House, R. J., & Aditya, R. N. 1997. The social scientific study of leadership: quo vadis? Leadership Quarterly, 23(2): 409--473. House, R. J., Spangler, W. D., & Woycke, J. 1991. Personality and charisma in the U.S. presidency: a psychological theory of leader effectiveness. Administrative Science Quarterly, 36: 364--396. Howell, J. M., & Avolio, B. 1993. Transformational leadership, transactional leadership, locus of control, and support for innovation: key predictors or consolidated-business-unit performance. Journal of Applied Psychology, 78(6): 891--902. Howell, J. M., & Frost, P. J. 1989. A laboratory study of charismatic leadership. Organizational Behavior and Human Decision Processes, 43: 243--269.

PAGE 102

94 Hunton, J. E., & McEwen, R. A. 1997. An assessment of the relation between analysts' earnings forecast accuracy, motivational incentives and cognitive information search strategy. The Accounting Review, 72(4): 497. Jacobsen, C., & House, R. J. 2001. Dynamics of charismatic leadership: a process theory, simulation model, and tests. Leadership Quarterly, 12(1): 75--113. Judge, T. A., & Ilies, R. 2002. The outer limits of work motivation: transformational leadership effects on employee motivation, Working Paper, University of Florida. Kadlec, D. 1998. Is that you, Al? Time, 151(10): 44. Kelly, J. R., & Barsade, S. G. 2001. Mood and emotions in small groups and work teams. Organizational Behavior and Human Decision Processes, 86(1): 99--130. Kerr, S., & Jermier, J. M. 1978. Substitutes for Leadership: Their Meaning and Measurement. Organizational Behavior and Human Performance, 22(3): 375. Klein, K. J., & House, R. J. 1995. On fire: charismatic leadership and levels of analysis. Leadership Quarterly, 6(2): 183--198. Lowe, K. B., Kroeck, K. G., & Sivasubramaniam, N. 1996. Effectiveness correlates of transformational and transactional leadership: a meta-analytic review of the MLQ literature. Leadership Quarterly, 7(3): 385-425. Meindl, J. R. 1990. On leadership: an alternative to the conventional wisdom. Research in Organizational Behavior, 12: 159--203. Meindl, J. R., Erlich, S. B., & Dukerich, J. M. 1985. The romance of leadership. Administrative Science Quarterly, 30: 78-102. Meyer, M. W., & Gupta, V. 1994. The performance paradox. Research in Organizational Behavior, 16: 309--369. Mohrman, S. A., & Cohen, S. G. 1995. When people get out of the box. In A. Howard (Ed.), The Changing Nature of Work: 365--410. Sal Francisco, CA: Jossey-Bass. Morley, A. 1988. Overview of financial analysis. In S. N. Levine (Ed.), The Financial Analyst Handbook: 3-33. Homewood, Ill.: Irwin. Moscovici, S. 1985. Social influence and conformity. In G. Lindzey, & E. Aronson (Eds.), Handbook of Social Psychology, Vol. 2: 347--413. New York: Random House. O'Brien, P. C., & Bhushan, R. 1990. Analyst following and institutional ownership. Journal of Accounting Research, 28(supplement): 55--76. Ocasio, W. 1999. Institutionalized action and corporate governance: The reliance on rules of CEO succession. Administrative Science Quarterly, 44(2): 384. Olsen, M. E. 1993. Forms and levels of power exertion. In M. E. Olsen, & M. N. Marger (Eds.), Power in Modern Societies: 29--37: Westview Press. Palmer, J. 1999. The Sunbeam Chainsaw Massacre: Al Dunlap and his crew cut down to size, Barron's: 62.

PAGE 103

95 Palmon, O., Sun, H.-L., & Tang, A. P. 1994. The impact of publication of analysts' recommendations on returns and trading volume. The Financial Review, 29(3): 395. Pfeffer, J. 1977. The ambiguity of leadership. Academy of Management Review, January: 104--112. Pfeffer, J. 1981. Management as symbolic action: the creation and maintenance of organizational paradigms. In L. L. Cummings, & B. M. Staw (Eds.), Research in Organizational Behavior, Vol. 3: 1--52. Greenwich:CT: JAI Press. Pfeffer, J., & Salancik, G. R. 1978. The External Control of Organizations: a Resource Dependence Perspective. New York: Harper & Row. Podolny, J. M. 1993. A status-based model of market competition. American Journal of Sociology, 98(4): 829--972. Popping, R. 2000. Computer-assisted Text Analysis. London: SAGE. Porac, J. F., Wade, J. B., & Pollack, T. G. 1999. Industry categories and the politics of comparable firm in CEO compensation. Administrative Science Quarterly, 44: 112--144. Powell, W. W., & DiMaggio, P. 1991. The New Institutionalism in Organizational Analysis. Chicago: University of Chicago Press. Puffer, S. M., & Weintrop, J. B. 1991. Corporate performance and CEO turnover: the role of performance expectations. Administrative Science Quarterly, 36(1991): 1--19. Pulliam, S. 2002. Analysts to tell congress that skepticism gets them abuse. The Wall Street Journal, March 19th: C1, C16. Rao, H., Greve, H. R., & Davis, G. F. 2001. Fool's gold: social proof in the initiation and abandonment of coverage by Wall Street analysts. Administrative Science Quarterly, 46(2001): 502--526. Riepe, M. W. 2000. New reg enhances 'interpretation' of investment news. Journal of Financial Planning, November: 46-47. Roberts, C. W. 2000. A conceptual framework for quantitative text analysis. Quality and Quantity, 34: 259-274. Rosenbush, S., & Borrus, A. 2001. Lucent's dark days, Business Week: 102-106. Rumelt, R. P. 1991. How much does industry matter? Strategic Management Journal, 12(1991): 167-185. Salancik, G. R., & Pfeffer, J. 1977. An examination of need-satisfaction models of job attitudes. Administrative Science Quarterly, 22: 475--498. Salancik, G. R., & Pfeffer, J. 1978. A social information processing approach to job attitudes and task design. Administrative Science Quarterly, 23: 224--251. Scott, B. W., Sudip, D., Iskandar-Datta, & Mai, E. 1995. Investment analyst recommendations: A test of "the announcement effect" and "the valuable information effect". Journal of Business Finance & Accounting, 22(5): 659.

PAGE 104

96 Scott, W. R. 1998. Organizations: Rational, Natural, and Open Systems (4th. ed.). Saddle River, NJ: Prentice Hall. Sellers, P. 1996. What exactly is charisma? It's real, it matters for your success. And it can be dangerous. Fortune, 133(1): 68--75. Sellers, P. 1998. Can Chainsaw be a builder? Fortune, 137(1): 118--120. Selznick, P. 1957. Leadership in Administration. New York: Harper & Row. Serwer, A. 2000. There's something about Cisco. Fortune, 141(10): 114. Shamir, B. 1995. Social distance and charisma: theoretical notes and an exploratory study. Leadership Quarterly, 6(1): 19--47. Shamir, B., Arthur, M. B., & House, R. J. 1994. The rhetoric of charismatic leadership: a theoretical extension, a case study, and implications for research. Leadership Quarterly, 5(1): 25--42. Shamir, B., Zakay, E., Breinin, E., & Popper, M. 1998. Correlates of charismatic leader behavior in military units: subordinates' attitudes, unit characteristics, and superiors' appraisals of leader performance. Academy of Management Journal, 41(4): 367--409. Shapiro, G., & Markoff, J. 1997. A matter of definition. In C. W. Roberts (Ed.), Text Analysis for the Social Sciences: 9-33. Mahvaw, NJ: Lawrence Erlbaum. Sheikholeslami, M., Wilson, M. D., & Selin, J. R. 1998. The impact of CEO turnover on security analysts' forecast accuracy. The Journal of Applied Business Research, 14(4): 71--75. Simon, H. A. 1945. Administrative Behavior (4th ed.). New York: The Free Press. Smith, R., & Anand, G. 2002. Stock research comes under fire in CEO's battle. The Wall Street Journal, April 16th.: A1, A12. Staw, B. M., McKechnie, P. I., & Puffer, S. M. 1983. The justification of organizational performance. Administrative Science Quarterly, 28: 582--600. Stern, R. N., & Barley, S. R. 1996. Organizations and social systems: organization theory's neglected mandate. Administrative Science Quarterly, 41(1996): 146--162. Stevenson, M. K., Busemeyer, J. R., & Naylor, J. C. 1990. Judgment and decision making theory. In M. D. Dunnette, & L. M. Hough (Eds.), Handbook of Industrial and Organizational Psychology, 2nd ed., Vol. 1: 283--374. Palo Alto, CA: Consulting Psychologists Press. Steyrer, J. 1998. Charisma and the archetypes of leadership. Organization Studies, 19(5): 807--828. Stickel, S. E. 1992. Reputation and performance among security analysts. Journal of Finance, XLVII(5): 1811--1836. Talley, K., & Munk, C. W. 2002. Call on Kmart reflects how Prudential has put piece of the rock into analysis. The Wall Street Journal, January 30th.: C12.

PAGE 105

97 Thompson, J. D. 1967. Organizations in Action. New York: Mc Graw Hill. Tichy, N. M., & Devanna, M. A. 1986. The Transformational Leader. New York: Wiley. Tommerup, P. 1990. Stories about an inspiring leader. American Behavioral Scientist, 33(3): 374--385. Tosi, H. L., Misangyi, V. F., Fanelli, A., Waldman, D. A., & Yammarino, F. J. 2002. CEO charisma, compensation and firm performance. Paper presented at the Academy of Management Conference, Denver, Colorado. Useem, M. 1996. Investor Capitalism: How Money Managers are Changing the Face of Corporate America. New York: Basic Books. Waldman, D. A., Ramirez, G. G., House, R. J., & Puranam, P. 2001. Does leadership matter? CEO leadership attributes and profitability under conditions of perceived environmental uncertainty. Academy of Management Journal, 44(1): 134--143. Waldman, D. A., & Yammarino, F. J. 1999. CEO charismatic leadership: levels of management and levels of analysis effects. Academy of Management Review, 24(2): 266-285. Wareham, J. 1995. Eight steps to charisma. Across the Board, 32(4): 49. Wasielewski, P. L. 1985. The emotional basis of charisma. Symbolic Interaction, 8(2): 207--222. Weber, M. 1947. The Theory of Social and Economic Organization (T. Parsons, Trans.) (1997, 10th ed.). New York: Free Press. Weber, R. P. 1988. Basic Content Analysis (3rd ed.). Beverly Hills: Sage. Wechsler, L. D. 1995. "you want somebody to like you, get a dog", Forbes, Vol. 156: 44-47. Westley, F., & Mintzberg, H. 1989. Visionary leadership and strategic management. Strategic Management Journal(10): 17--32. Westphal, J. D., Gulati, R., & Shortell, S. M. 1997. Customization or conformity: an institutional and network perspective on the content and consequences of TQM adoption. Administrative Science Quarterly, 42: 366--394. Westphal, J. D., & Zajac, E. J. 1994. Substance and symbolism in CEO's long-term incentive plans. Administrative Science Quarterly, 39(1994): 367--390. Westphal, J. D., & Zajac, E. J. 1998. The symbolic management of stockholders: corporate governance reforms and shareholder reactions. Administrative Science Quarterly, 43: 127--153. White, H. C. 1981. Where do markets come from? American Journal of Sociology, 87(3): 517--547. Yukl, G., & Van Fleet, D. D. 1992. Theory and research on leadership in organizations. In M. D. Dunnette, & L. M. Hough (Eds.), Handbook of Industrial and Organizational Psychology: 147--197. Chicago: IL: Rand McNally.

PAGE 106

98 Zajac, E. J., & Westphal, J. D. 1996. Who shall succeed? How CEO/board preferences and power affect the choice of new CEOs. Academy of Management Journal, 39(1): 64. Zuckerman, E. W. 1999. The categorical imperative: securities analysts and the illegitimacy discount. American Journal of Sociology, 104(5): 1398--1438. Zuckerman, E. W. 2000. Focusing the corporate product: securities analysis and de-diversification. Administrative Science Quarterly, 45: 591--619.

PAGE 107

BIOGRAPHICAL SKETCH Angelo Fanelli graduated from Universita’ Bocconi (Milan, Italy) in 1994 and obtained his first Ph.D. (Direzione Aziendale) from Universita’ di Bologna (Bologna, Italy) in 2002. 99