STOCKHOLDERS' VS BONDHOLDERS' INTERESTS: ROLE OF THE BOARD IN CREDIT RATINGS’ IMPROVEMENT

STOCKHOLDERS' VS BONDHOLDERS' INTERESTS: ROLE

OF THE BOARD IN CREDIT RATINGS’ IMPROVEMENT

Rozhitsyna Yelena - s1521373

Rijksuniversiteit Groningen Faculty of Economics

Master Thesis Prof.dr. H. van Ees

Yelena Rozhitsyna s1521373 2 This paper investigates a role of the BOD in regulating the conflict of interests between

INTRODUCTION ...4

THEORY AND HYPOTHESES DEVELOPMENT ...8

THEORY...8

CONCEPTUAL MODEL...10

HYPOTHESES...12

Size of the board...12

CEO power...13

CEO tenure. ... 14

Directors appointed after the CEO. ... 14

CEO duality. ... 15 BOD independence. ... 16 Age of directors... 18 Directors’ expertise...19 BOD education. ... 20 BOD nationality. ... 21 Directors’ networks... 21 Takeover defenses ...23

Staggered (classified) board... 23

Firm Liquidity ...24

METHODOLOGY ...25

DATA AND SAMPLE...25

VARIABLES...25

Dependent variables ...25

Independent variables ...26

Control variables ...29

OVERVIEW OF THE METHODOLOGIES...29

Credit rating...29

Firm liquidity ...32

MODELS SPECIFICATION...32

Credit rating...32

Firm liquidity ...33

ANALYSIS AND RESULTS ...35

DESCRIPTIVE STATISTICS...35

CORRELATIONS...36

REGRESSIONS...38

Credit rating...38

Firm liquidity ...43

SUMMARY AND CONCLUSIONS ...46

REFERENCES ...50

INTERNET SOURCES...58

FIGURES AND TABLES ...60

APPENDICES...73

APPENDIX A...73

APPENDIX B. ...76

INTRODUCTION

“Money must work”. This phrase became a byword, however, there is plenty of possible opportunities to make capital work while ensuring high profitability and low risk. Among the most popular investment opportunities in modern society are securities, both bond and equity, with the former being “the primary instrument of raising long-term capital in the United States” (Bhoraj and Sengupta, 2003). When taking an investment decision it is vital to consider many parameters of the target firm to reduce risks. Corporate governance (hereinafter CG), defined by Shleifer and Vishny (1997) as “the ways though which suppliers of capital to corporations assure themselves of getting a return on their investment” (p.737), is one of the factors considered by investment consultants and rating agencies. Nowadays institutional investors, money managers, hedge funds and others pay research firms (e.g. ISS, GMI and Corporate Library) dealing with CG ratings in order to obtain information about quality of CG structure of the firms they are investing or planning to invest in. Research by the Institutional Shareholder Services (hereinafter

ISS) indicates that “better corporate governance results in higher profit and lower risk” (ISS,

2005) and investors state that they would pay more for the shares of well-governed companies (e.g. Coombes and Watson, 2000).

to repossess some of the firm’s assets (collateral) or the opportunity to throw the firm into bankruptcy” (Shleifer and Vishny, 1997). Shareholders, in contrast, do not have any right to claim for dividends unless they are announced by the firm, but they may prefer “riskier actions to revive the company’s prospects” and increase the value of its equity (Fitch Ratings, 2004). Only shareholders have the right to vote for the BOD (Shleifer and Vishny, 1997) and thus may somehow influence its decisions “to obtain preferential treatment at the expense of other stakeholders” (Ashbaugh et al., 2004). It is said that BODs “see the company through the eyes of the shareholders” (Van den Berghe and Levrau, 2004). Moreover, present regulations and standards of CG (such as Sarbanes-Oxley 2002 and new listing standards proposed and endorsed by, e.g., NYSE and SEC among others) called to strengthen corporate governance are aimed, first of all, to protect shareholders’ interests and improve shareholder’s value by “addressing accountability of a firm’s senior managers and BOD to shareholders” (Bertsch and Watson,

Moody’s). Despite debtholders are also interested in proper accountability structures, as it was

illustrated above, there is a substantial stockholder-bondholder conflict which is “well documented in the literature” (see Klock, Manis and Maxwell (2005) for references). The source of the conflict lies in the different contractual structures and risk profiles of the two groups of principals (Fitch Ratings, 2004; Jensen and Meckling (1976) in citation by Klock et al, 2005). Thus, it would be logical to suggest that “good corporate governance” “with the primary focus on protecting shareholders’ interests is inappropriate and irrelevant to bondholders” (Fitch

Ratings, 2004). However, the results of the previous studies are ambiguous. For instance, Collins

and Tippie (2004) found that:

“…companies with stronger shareholder rights have lower credit ratings, so "governance mechanisms that benefit shareholders may do so at the expense of bondholders." Clearly, both groups of stakeholders prefer long-term prosperity. But some management moves, particularly ones that involve mergers, benefit shareholders at the expense of bondholders. However, when times get really tough, bondholders might prefer a bankruptcy in which they gain control of assets, leaving shareholders with nothing…”

and further:

company's takeover resistance and give more power to management as opposed to shareholders... that can be good for bondholders because buyouts often lead to bond losses, whether due to increased leverage or the introduction of other uncertainties” (Collins and Tippie (2004).

At the same time, recent findings by Low, Machija and Sanders (2006) support the somewhat opposite view of Gompers et al. (2003) (cited by Klock et al, 2005) that “stronger shareholder power, through superior monitoring of managers, improves asset and collateral values, which benefit bondholders.” This inference is further specified in the credit ratings’ guidelines for assessment of corporate governance from bondholders’ perspective stating that “a BOD which effectively promotes and protects the long-term interests of shareholders will, by and large, also mitigate risk for creditors, by assuring proper oversight of management” (Bertsch and Watson, Moody’s). Finally, according to Bhojraj and Sengupta (2003), "governance mechanisms can reduce default risk by mitigating agency costs and monitoring managerial performance and by reducing information asymmetry between the firm and the lenders.” Concluding the abovementioned, it can be argued that the interests of shareholders and bondholders “are generally aligned when better monitoring of management occurs” (Ashbaugh et al., 2004) but there are still some potential conflicts between the two groups of investors that require attention. These potential conflicts resulting in wealth transfer between shareholders and bondholders and corresponding risk shifting may be controlled by the same mechanisms which are used for addressing agency problem between stakeholders and management, namely by CG. For instance, Ashbaugh et al (2004) found that credit ratings improve under the condition of weaker shareholder rights with regard to corporate control. Put differently, companies’ risk profiles may be changed and assessed based on their CG practices.

If for institutional investors or lenders it may be possible, though costly, to assess credit risk, this task is practically unfeasible for individuals. In this case credit rating agencies (e.g.

Standard&Poor’s, Moody’s and Fitch) come to succour. They independently assign their

al. 2005; Bodt and Sherman. GMI), there is just a few academic papers (e.g. Bhojraj and Sengupta. 2003; Ashbaugh et al., 2004) in which authors look specifically at the relationship between corporate governance and the credit ratings.

correspond to larger net cash flows or operating cash flows?” Inasmuch as the relationship between board characteristics and corporate credit ratings may be contingent on company size, industries and countries, when analyzing this relationship I control for company size, company diversification index, industry and industry stability. Based on the availability of the databases I'll limit the scope of the study by searching for differences, if any, between large US corporations in ten industries1.

THEORY AND HYPOTHESES DEVELOPMENT

Theory

Credit ratings are essentially based on “rating agencies’ assessment of the probability distribution of future cash flows to bondholders, which in turn, depends on the future cash flows to the firm” (Ashbaugh et al., 2004). These factors will depend upon: 1) overall long-term profitability of the firm, and outcome of 2) the agency conflict between management and investors, and 3) conflict of interests between two groups of investors, shareholders and debtholders. Based on these conditions, investors into debt securities face default risk which may be split into the agency risk, information risk, and event risk. The agency risk represents the probability that management of the firm will act either opportunistically or in its self-interest (Bhojraj and Sengupta, 2003). The information risk concerns the situations when management of the firm possesses some firm-specific information which may adversely affect firm’s default risk but do not reveal it to investors in a timely manner (Bhojraj and Sengupta, 2003). The event risk occurs in situations when shareholders, empowered by certain CG features, “encourage management to undertake risky investments or engage in ownership changes,” such as takeovers or leveraged buyouts (LBO), “that can harm bondholders’ interests” and result in wealth transfer from them to shareholders (Ashbaugh et al. 2004). It is obvious, that decreased default risk is associated with higher credit ratings.

As it was mentioned above, BODs are called to monitor managers. To some extent this may be considered as a mechanism of the default risk mitigation. They should be able to address the agency risk by monitoring the management behavior and decisions; the information risk by establishing of sound accountability practices; and the event risk by regulation of conflicts

between stakeholders. There is a substantial empirical literature that attempts to relate BOD characteristics to company performance. These abilities of the BOD result from the roles it plays in CG. BODs appoint the top management of the companies (Useem, 2003), monitor executives’ behavior and performance, fulfill legal responsibilities as overseers, guard against any infringement of the law, establish and amend Bylaws, provide strategic advice, approve major contracts or mergers, make key decisions regarding real estate owned or managed by the corporation, promote the reputation of the corporation, and form a network to obtain key resources, such as capital, alliances and partnerships (Lawer, 2002; www.allbusiness.com). In general the roles of the BOD are categorized as monitoring, relational and possibly strategic management (Dallas, L.L., 2003). Here establishing and amendment of bylaws address the information risk while strategic advice and approval of contracts or mergers pertain to the event risk. Monitoring, or supervising, role relates to the BOD obligation to select competent executives and prevent them from opportunistic behavior and actions decreasing firm value. This allows the BOD to mitigate agency and information risks. Relational role of the BOD based on the BOD’s ability to form a network for obtaining key resources and may influence both agency and event risks. Finally, possible strategic management role of the BODs relates to the composition of expertise on the BOD and access to outside and inside information sources.

while Bhagat and Black (2002) documented that “firms with more independent boards do not perform better than other firms”. This may be explained also from the stewardship theory perspective implying that insiders are better aware of their firm’s assets and capability and may make a better use of this knowledge. At the same time Millstein and MacAvoy (1998) write: “Although the results do not prove causation, corporations with active and independent boards appear to have performed much better … than those with passive, non-independent boards.” Relational role of the BOD is explained from the resource dependence perspective. According to this theory, BODs represent a concentration of ties or channels to needed resources and information, both outside and inside. This combination of knowledge and resources enables corporations “to mediate its relationships with various stakeholders” while directors with different demographic characteristics “provide advice, support, enhanced status, and legitimacy to the corporations’ operations” (Dallas, 2003). Finally, having access to outside and inside information (through outside and inside directors respectively) BODs are able to engage in effective “strategic monitoring” (Dallas, 2003) and prevent management from making uninformed, opportunistic or self-interested strategic decisions which may entail event and default risks and result in credit rating degradation.

In general, BOD should be able to ensure stable future cash flows (interpret as “long-term performance”) of the company and their legitimate distribution, what in turn should promote company’s credit rating. Efficiency of BOD in performing its roles (affecting the overall default risk and credit ratings accordingly) is claimed to be dependent upon BOD characteristics, such as size, composition, heterogeneity (along gender, age, expertise, nationality, etc), distribution of power, and so on (Seidel and Westphal, 2002; Feng, Ghosh and Sirmans, 2005; Van der Berghe and Levrau, 2004; Nadler, 2004; Useem, 2003). These characteristics are used as the basis for hypotheses development presented below.

Conceptual Model

Prior to the hypotheses development it is useful to outline the conceptual model they will be based upon. This should help identify the major drivers of the relationship between the BOD characteristics and the credit ratings clearer understand the logic underlying the hypotheses’ reasoning.

1) and it must be said that credit ratings represent “an opinion of credit analysts on an issuer’s overall capacity to pay its financial obligations—i.e., its fundamental creditworthiness” (S&P’s, 2005). The opinion is formed based on the results of credit analysis which includes quantitative, legal and qualitative parts. While the quantitative and legal, the largest, parts of the analysis include investigation of financial and legal documents of the company, the qualitative part comprises analyses of the company’s competitiveness and quality of its management. There is no any established and widely accepted framework for qualitative analysis and international and national credit rating agencies use different in-house approaches. Inasmuch as I use credit ratings as the dependent variable and to ensure absence of a tautologous argument, it is wise to describe in short what Standard & Poor’s Rating Agency, data of which I use, takes into account when assigning a company with a credit rating (see the citations from the Standard & Poor’s ‘Corporate Rating Criteria 2005’ book in APPENDIX A. Reference to the book is given in the List of References at the end of the present paper) and emphasize that “a component of corporate governance that historically has not figured prominently in the rating process is board structure” (S&P’s).

Insert Figure 1 about here

Insert Figure 2 about here

Hypotheses

Size of the board

Although there are a number of theories predicting a relationship between BOD size and firm performance, the literature provides no consensus about the direction of that relationship (Dalton, Daily, Johnson, and Ellstrand, 1999). Thus, resource dependence perspective holds that bigger BODs will cause better firm performance thanks to greater number of links to necessary and critical resources (e.g. Hillman, Cannella, and Paetzold, 2000; Hillman, Dalziel, 2003; Pfeffer, 1972, 1973). On the other hand, studies based on the theory of social loafing (saying that with increase of the total number of people in the group the effort exerted by individuals in a group decreases (e.g., Latane, Williams, and Harkins, 1979)) show that smaller BODs associated with better performance (e.g. Yermack, 1996). Lipton and Lorsch (1992) proposed that for better control there is a need for a board of a specific size, namely composed of ideally 8 and a maximum of 10 directors. Results of Johnson, Daily and Ellstrand (1996) provide some support to this idea. On the other hand, some authors argue that small boards are more "manageable" from the CEO's perspective (e.g. Daily and Dalton, 1993; Chaganti et al., 1985).

2003) and therefore able to provide sound strategic monitoring. Summarizing the discussion above, I conclude that smaller boards are less effective in performing monitoring function and mitigating default risk. On the other hand, if the size of the BOD is too big it may be difficult to reach expedient decisions due to vast divergence of points of view and taking into account the theory of social loafing. That is, too large BODs are not able to monitor effectively and reduce default risk. However, the theory does not provide any indication of an optimal size of the BOD (Van den Berghe and Levrau, 2004). Therefore, it is impossible to form any reasoned hypothesis. At the same time I may formulate a proposition describing relationship between BOD size and its ability to mitigate default risk of the company. Sign of the relation will be interpreted in support of one or another theory used herein above.

Proposition1: BOD’s size is related to the company’s credit rating

CEO power

CEO tenure. Making references to a number of previous studies Westphal and Zajac (e.g.

1994, 1995) write that it’s typically argued that as CEO tenure increases board control over the CEO diminishes. Furthermore, the stronger is the CEO’s influence on the board the more possibilities will be there for the CEO to practise symbolic management (Zajac and Westphal, 1995) and misinform the directors. Another perspective holds that, as time passes, simultaneously with CEO tenure his/her familiarity with the firm’s resources increases (Westphal and Zajac, 1994). Increased amount of firm specific information being at the CEO’s disposal, in turn, may either lead to a stable long-term performance of the firm or give the CEO more opportunities to act in self-interests and elevate agency and information risk for investors. Therefore, to protect interests of creditors and provide sound monitoring of management an effective BOD is needed. BOD efficiency here may be improved through the same increased familiarity with the firm and the CEO personality which is build up along with the directors’ term on the BOD. Taking this belief as a starting point and engaging the agency theory I may hypothesize that lowering board monitoring efficiency resulting from the increased CEO tenure (power) will hinder BOD ability to mitigate agency and information risks and consequently lead to an increased risk of fraud. Therefore:

H1: The longer is CEO’s tenure vis-à-vis the BOD’s average tenure the lower is the credit rating of the company

Directors appointed after the CEO. The next hypothesis is a logical extension of the

previous one. Specifically, Westphal and Zajac (1994, 1995) claim that the longer a person serves as a CEO, “the more personal power he/she acquires by populating boards with supporters (1994) … or sympathetic outsiders (1995).” Moreover, according to the attraction-selection-attrition (ASA) framework, CEO will tend to choose directors similar to them, sharing the same values and points of view (Schneider et al., 1995). From a creditor’s point of view, individuals that are beholden to management (in our case, chosen and recommended to the board by the CEO) are unlikely to question the company’s decisions (Fitch Ratings, 2004), because “reducing CEO influence is considered to be a proxy for the board willingness and ability to oversee

management2_{” (MIT SMR, 2004). On the other hand, “lack of reasonable director turnover,}

which may indicate the absence of fresh perspective” is considered by the credit rating agencies

2 _{Though discussion down to this point is relevant only for the boards without formal nomination procedures, I do}

as a “red flag” indicating poor corporate governance from a creditor’s perspective (Bertsch and Watson, Moody’s; Duffy, aicpa.org). However, term “reasonable” is not defined by the credit rating guidelines. Existing research, in turn, shows that there is no confidence in the actual positive outcome of directors’ turnover and inflow of fresh perspectives onto the BOD. Specifically, BOD composed of directors with similar terms may be viewed as a cohort that does not welcome newcomers; appearance of a “fresh” director on the BOD may lead to a conflict within the group and hinder its efficiency (McNeil and Thompson (1971) and Pfeffer (1981b) in citation by Hambrick and Mason, 1984). Finally, recalling the above discussion, BOD efficiency in management monitoring may be improved through increased familiarity the CEO personality which should developed the better the longer directors and the CEO serve together. Recapitulating the foregoing and drawing parallels to the topic of the study I may assume that the more directors are appointed after the CEO the lower is the risk of fraud and, thus, default by the company. This relationship will be more pronounced for companies with longer average BOD tenure. Therefore, the hypothesis will be as follows:

H2: The larger is the number of directors appointed after the CEO the higher will be the credit rating of the company. This relationship should improve for companies with longer average BOD tenure

CEO duality. Existing literature considers issue of CEO duality in the contexts of agency

1991, Rhoades et al., 2001). Results of the study by Ashbaugh et al. (2004), on the contrary, suggest that “it is costly for firms, in terms of default risk, to cede the CEO with too much board control.” These results are further confirmed by the rating agencies’ guidelines regarding assessment of CG quality from a creditor’s perspective which state that creditworthiness improves for companies with separate CEO and chairperson positions (Bertsch and Watson,

Moody’s; MIT SMR, 2004; Duffy, aicpa.org). This may be caused by the fact that “the board

chairman frequently sets the board’s agenda and, therefore, controls issues brought before the board” (Ashbaugh et al., 2004). However, according to the credit rating criteria (S&P’s, 2005) BOD structure represents just a minor fraction of the credit analysis and cannot influence it substantially. At the same time, CEO duality decreases shareholders’ control over the corporation and should be potentially favorable for bondholders as may mitigate event risk. In particular, in accordance with the stewardship theory predictions, the CEO may be reluctant to engage in risky projects which may hinder long-term profitability of the firm or put it under the risk of bankruptcy. The above conclusion is supported by empirical results of Ashbaugh et al. (2004) who found that credit ratings improve under the condition of weaker shareholder rights with regard to corporate control. Therefore, founding in large on the results of the previous studies and theory predictions (neglecting the credit rating guidelines’ requirements), and bearing in mind that both stable long-term performance (future cash flows) and the default risk are important elements of credit analysis while CEO duality does not figure prominently in the credit rating but can have an indirect effect on the default risk, I hypothesize that:

H3: Corporations with CEO duality will have higher credit ratings

BOD independence. The requirement of having a majority of independent directors on the

(Bhojraj and Sengupta. 2003). At the same time, Beasley (1996) (cited by Bhojraj and Sengupta (2003)) “documented a negative association between the proportion of the board that is composed of outsiders and the probability of financial statement fraud”. This evidence is especially important in the light of the fact that credit analysts vastly base conclusions regarding companies’ creditworthiness on the financial statements (S&P’s). Further, Bhojraj and Sengupta (2003) obtained empirical evidence that “firms that have… stronger outside control of the board enjoy… higher ratings on their new bond issues”. This result was confirmed by Ashbaugh et al. (2004) who found that “the greater the board’s ability to provide independent oversight of management the better the credit rating.” Dallas (2003) also provides evidence in favor of independent BODs: he argues that “higher proportion of outsiders on the BOD is a viable way of co-opting the environment and reducing uncertainty surrounding strategy development and execution.” Finally, Hillman and Dalziel (2003) operating by the agency theory maintain that “board dependence will negatively affect the relationship between a board’s ability to monitor and actual monitoring” (p.391). While, on the contrary, in the context of the resource dependence perspective, “when it comes to the provision of resources … a board’s dependence may be desirable ... and beneficial” (Hillman and Dalziel, 2003 p.391-392). For all that, Westphal and Bednar (2005) describe a phenomenon of pluralistic ignorance when outsiders being unsure of other directors’ opinion are reluctant to express their strategy concerns even if they think that the strategy in action is unsound. This situation becomes even more possible when directors are very demographically different from each other. This may lead to a loss of board effectiveness. Moreover, “greater board independence may … prompt the CEO to use more interpersonal influence” and symbolic management in attempts to adapt and regain the power (Westphal, 1998).

dependence perspective, and taking into account results of the previous studies, I hypothesize that:

H4: The proportion of independent directors is positively correlated with credit rating

Age of directors. The phenomenon of change in expectations depending on the age of a

target person has been extensively studied in sociology. “Nine commonly held assumptions about work and age -- ranging from the belief that age determines health, physical strength, cognitive capacity and productivity, through to the assumption that the older we are, the more difficult we find adapting to change” (Platman, K., 2006) have been revealed. For instance, “many UK employers appear to apply a "common knowledge" argument, which says that "every sensible person knows" that as people get older… there is a decline in cognitive functions and reaction times” (Smethurst, S., 2006). However, Smethurst (2006) makes reference to findings of a research professor from the Finnish Institute of Occupational Health and one of the leading experts in this field, who says that

“some cognitive functions, such as control of use of language or the ability to

process complex problems, improve with age… In most work tasks, speed and

precision can be substituted by the high motivation of ageing workers and the experience and wisdom they have assembled throughout their working life. Even though the speed of learning may slow with age, strong motivation to learn can compensate for slower learning speed."

ISS has “Age of directors” item in the list of the rating variables for assessment of

time Fairchild and Li (2005) assume that it is “plausible that older directors may have more experience, and therefore, are considered to be of higher quality.” However, they found no significant confirmation of their assumption.

Speculating on the above I may say in short that aging implies less flexibility and creativity but greater experience and ability to decide complex problems (Hambrick and Mason, 1984). Older, hence more experienced, individuals may be better able to identify and prevent fraudulent attempts of the management as well as provide necessary strategic advice while are associated with lower BOD power. Therefore, due to contradictory theory predictions, it is impossible to form any reasoned hypothesis. At the same time I may formulate a proposition describing relationship between BOD’s age and its ability to mitigate default risk of the company. Sign of the relation will be interpreted in support of one or another theory used herein above.

Proposition2: The average age of the BOD is correlated with the credit rating

Directors’ expertise

and Milton, 2000; Zajac and Westphal, 1995). This notion contrasts the resource dependence perspective implying greater effectiveness of the board composed of directors with different experience and backgrounds who are supposed to provide more valuable contributions into the decision making process and ensure efficiency of board decisions (e.g. Schneider et al., 1995). On the other hand, higher levels of diversity among the directors may lead to pluralistic ignorance and consequently affect the process of board decision making (Westphal and Bednar, 2005).

Finally, Carpenter and Westphal (2001) make a distinction between demographic and expert composition preferences for companies depending on the level of their environmental stability. Thus, they argue, that for firms with a stable environment homogeneous directors’ expertise, providing it is industry-specific, is more valuable as they “not only have more information at their disposal but also have more efficiently structured information, and this leads to faster and more accurate information processing.” On the other hand, if the environment is unstable, directors’ ability to contribute to the success of the firm will greatly depend on their possession of a broader range of knowledge and skills (Carpenter and Westphal, 2001).

Therefore, the best variant of expertise and demography composition in the board for a company will depend on the company’s environment or industry it operates in. Moreover, while organizational outcomes can be partially predicted from managerial backgrounds, industry environment, in its turn, may condition composition of the managerial expertise (Hambrick and Manson, 2001). The same maxims may be expanded to include the board of directors.

BOD education. Credit rating agencies pay a great attention to “the ability of directors to

understand the business and financial implications of the focal company’s strategic options, to ask relevant and analytical follow-up questions and to make well-reasoned decisions” (Fitch

Ratings, 2004; Duffy, M. N., aicpa.org; Bertsch and Watson, Moody’s). The above is impossible

without an appropriate background. “Appropriate” here “depends on the company’s sector or industry, the complexity of its operations, and the mix of skills that the other board members bring to the table” (Fitch Ratings, 2004; Duffy, M. N., aicpa.org; Bertsch and Watson,

Moody’s). With respect to the company creditworthiness and agency and information risks the

real state of affairs in the eyes of the board. Secondly, more homogeneous boards (under the abovementioned condition) will be more effective in monitoring and decision making, and, thus, mitigating default risk what, consequently, should favourably affect the credit rating of the firm.

H5: The more homogeneous is the BOD education (providing it is industry-specific and the level of stability in the industry is high) the higher will be the credit rating of the company. Or alternatively, the more heterogeneous is the BOD education (providing the level of instability in the industry is high) the higher will be the credit rating of the company

BOD nationality. Bearing in mind the above discussion and introducing conditions of

diversification I may hypothesize that strategies of diversified firms will be the more complex the higher their diversification index is and thus will require a broader pool of knowledge and skills to cope with it (Boone et al., 2004). Whereas I consider geographical diversification here, I may argue that greater national diversity of directors will allow provision of necessary resources and make it easier for the board to assess quality and correctness of managerial decisions (Van den Berghe and Levrau, 2004). In short, for geographically diversified firms greater national heterogeneity of the board will mitigate information and agency risks. Therefore:

H6: The more heterogeneous is the BOD nationality (providing the firm’s diversification index is high) the higher will be the credit rating of the company

Directors’ networks. The impact of board interlocks has been vastly studied by

the firm’s performance (Davis, 2005). Directors network ties also serve to spread knowledge and awareness about specific organizational changes, facilitate diffusion of particular organizational practices and even “more fundamental belief systems about corporate strategy, organizational structure, or the boards’ role in the organization” (Zajac and Westphal, 1996). The process of popular practices adoption, also called ‘institutionalization’, as well may be accelerated by organizational network ties (Westphal, 1997; Davis, 2005), in our case, by directorship interlocks. Moreover, new practices are adopted though ‘vicarious’ learning by “direct communication between managers and directors and raising awareness about potential benefits … and drawbacks” which may be encountered from the adoption (Westphal and Zajac, 2001).

When considering board advisory function, both strategically related and unrelated ties help directors “acquire relevant knowledge through social interaction with other directors in board and committees, as board members evaluate management and raise ideas and suggestions for better strategy implementation” (Carpenter and Westphal, 2001) and thus improve the board’s effectiveness (Westphal, 2002 (b)). Such ties also improve monitoring capability of the boards broadening their “schemata or knowledge structures” and “should facilitate adaptation to environmental changes” (Carpenter and Westphal, 2001).

Finally, Ashbaugh et al. (2004) cited a work of Beasley (1996) who found that “as the number of outside directorships in other firms held by outside directors decreases, the likelihood of financial statement fraud decreases” (p. 15). This may be explained by the possibly reduced directors’ involvement in the BODs activities due to their excessive work load. Nevertheless, credit rating agencies normally consider directors’ appointments to other boards as a parameter of directors’ expertise level (MIT SMR, 2004; Fitch Ratings, 2004; Bertsch and Watson,

Moody’s; Duffy, aicpa.org). Generally greater directors’ expertise is considered to contribute to

the bondholders’ interests by virtue of increased ability of the board to monitor and evaluate decisions of the managers. But at the same time, a board will not be effective if the directors composing it cannot devote sufficient amount of time and energy to their responsibilities due to overcommitment arising from participation in a number of other boards. According to the Fitch

Ratings Credit Policy directors generally should not serve on more than five corporate boards at a time (Fitch Ratings, 2004). However, again, this directors’ expertise parameter does not figure

prominently in the credit rating but can have an indirect effect on the default risk.

monitoring and resources and advice provision activity and thus should be positively associated with credit ratings. This proposition is supported by findings of Ashbaugh et al (2004) and Collins and Tippie (2004). But if there are too many overcommitted directors on the board, board efficiency will suffer. However, there are no directors serving on more than five boards at a time in the sample I use. Therefore I hypothesize:

H7: The average number of directors’ appointments to other boards is positively correlated with credit rating

Takeover defenses

Staggered (classified) board. The final demographic dimension of interest for the present

study is existence of different classes of directors on the board. Classified or staggered boards are considered to be the mightiest of all takeover defenses (Bebchuk et al. 2002, 2004). If the directors’ terms are classified, board control cannot be achieved in a single annual meeting election, what allows maintaining the company’s independence. This practice often affects investment returns and often makes “the stockholders worse off with the corporation remaining independent than they would be if the hostile bid were accepted” (Bebchuk et al., 2002). However, Collins and Tippie (2004) state that:

“…factors such as staggered terms for directors … increase the company's takeover resistance and give more power to management as opposed to shareholders... that can be good for bondholders because buyouts often lead to bond losses, whether due to increased leverage or the introduction of other uncertainties” (Collins and Tippie (2004).

The above conclusions supported by empirical results of Ashbaugh et al (2004) who also found that credit ratings improve under the condition of weaker shareholder rights with regard to corporate control. Furthermore, credit rating agencies pay particular attention to companies’ takeover defense provisions (e.g. Bertsch and Watson, Moody’s; Fitch Ratings, 2004), and Cremers et al. (2005) have documented that “the rating agencies view shareholder control harmful for the firm’s bondholders.” Therefore, I hypothesize that:

Firm Liquidity

As it was mentioned above, credit ratings are essentially based on “rating agencies’ assessment of the probability distribution of future cash flows to bondholders, which in turn, depends on the future cash flows to the firm” (Ashbaugh et al., 2004). Therefore, credit ratings should coincide with the real liquidity positions of firms expressed through operating cash flows (OCF), showing companies’ ability to generate the resources needed to meet their current liabilities. However, it may happen that the credit rating is just a symbolic thing and does not correspond to liquidity of the firms.

H9: Larger operating cash flows of companies correspond to higher credit ratings

High credit ratings are generally perceived to reduce capital cost for rated firms, in other words, they should facilitate cash inflows into the rated companies. Anyway, if my above hypothesis is not correct and the credit rating is only a symbolic thing and does not lower the cost of capital, there must be substitute characteristics conditioning easier access to borrowing capital. Board interlock ties may be such a conducive factor. Hillman and Dalziel (2003), for example, adduce results of studies by Zald (1987) and Provan (1980) showing that “nonprofit agencies were better able to raise funds when important community members served on their boards.” Bierlen and Featherstone (1998) further argue that “the presence of credit constraints is attributed to asymmetry of information between borrowers and lenders.” These constraints may be removed with the help of director networks which, as discussed above, may channel information and required resources to the focal firms. Therefore:

H10: The greater the average number of directors’ appointments to other boards the larger will be the firm’s cash flow

Moreover, association with business groups or banks is widely considered as a factor diminishing liquidity- and consequently investment-constraints (Baum and Thies, 1999). Thus, “firms took more debt when they had bankers on their board” (Davis, 2005). Clearly:

H11: The more connections there are between a firm’s CEO and directors and financial institutions the larger will be the firm’s cash flow

At the same time, if a company has enough inside resources to meet its liabilities, it will be reluctant to increase its leverage:

connections to financial institutions.

METHODOLOGY

Data and Sample

To address the hypotheses I use a sample comprising 496 companies from Fortune 500 list

for 2004, the sample was collected under the Data Collection Project3 and based on the website

www.theyrule.net. Whereas all the data on credit ratings, BOD characteristics should be available, the final sample consists of 261 companies (3240 directors). Credit ratings data are taken from the Standard and Poor’s Rating Agency (S&P’s) Website and are long-term organizations’ (issuers’) credit ratings. A Standard & Poor's issuer credit rating is a current opinion of an obligor's overall financial capacity (its creditworthiness) to pay its financial obligations (S&P’s). The choice of the rating agency is conditioned by my confidence in that credit analysis employed by the S&P’s does not examine the characteristics of the BODs I make central to the study (see Appendix A). Choice of the long-term ratings is conditioned by the assumption that investors in debt securities are more interested in the long-term performance of the companies they invest in rather than in the short-term which may be subject to abrupt changes. Data on the BODs’ composition and directors and CEOs characteristics are taken from the proxy statements (SEC website among other sources), annual reports and investor information section on the companies’ websites. Information on directors’ education and expertise is found in proxy statements, International Directory of Business Biographies, Notable Names Database (NNDB) as well as with help of the Google. Financial information is taken from annual reports (SEC website and companies’ websites).

Variables

Dependent variables

operations for 2004 by current liabilities for 2004 (Mills & Vamamura, 1998). An alternative measure of firms’ liquidity is net cash flow for 2004 (LOGNC04), reflecting the real state of affairs of the companies in respect of short-term liquidity and measured as log of net cash flow of the companies in 2004. To capture information conveyed by the credit ratings in its fullest I use three credit rating variables: RATING, RAT_DISCR and RAT_DUM. The ratings range from AAA+ (highest possible rating indicating the highest creditworthiness) to R (lowest rating, assigned to an obligor under regulatory supervision) (see APPENDIX A). The credit rating variables are coded according to the schedule provided in Table 1. RATING is an ordinal variable representing issuer’s long-term credit rating for 2004 and taking values from 1 (for rating R) to 9 (rating AAA-, AAA, AAA+). Two other credit rating variables are developed on its basis. RAT_DISCR is a discrete variable taking into account position of a company within certain rating category and its tendency towards lower or higher credit rating. This variable takes values from 1 (for rating R) to 25 (rating category AAA+). Finally, to allow for a more general categorization of the firms in sample I use RAT_DUM, a dummy variable taking value of 1 if a company’s credit rating is of investment grade (BBB- of higher) and 0 otherwise.

Insert Table 1 about here

Independent variables

Board size (BOD_SIZE) was expressed as the total number of directors on the board.

Alternate directors are not taken into account. When a CEO was replaced by a newcomer (a person who had not been a director before he/she was appointed as a new CEO) and if this happened before or on June 30, 2004, I took into account the data on the newcomer, while if this event occurred after June 30, 2004, data on the retired CEO was considered.

Board education heterogeneity index, BOD_EDU, is calculated as the Theil index by the

following formula:

( ) ( )

(

)

[

]

∑

directors with education or expertise in a given education field divided by the total number of

directors in the BOD); and wi indicates the proportion that each education field represents in the

sample (calculated as entity divided by the total number of different areas of all directors’ education and/or expertise). Different areas of all directors’ education and/or expertise were coded by numbers from 1 to 8 depending on the area of person’s education (engineering, business, law, medicine, arts, social sciences, sciences and other). “The Theil index is a heterogeneity measure that provides a measure of the discrepancies between the distributions of a resource within a population” (Wilson and Lockshin, 2003). Put it differently, this is a measure of population heterogeneity weighted by the parameter of interest. In the purposes of the analysis I mean the BOD of each separate firm in the sample to be a population while directors sharing each education/expertise field presented in the BOD compose groups within the population. This index takes into account number of different education/expertise fields on the BOD and number of directors sharing the same education/expertise field and allows comparing between firms. The Theil index ranges from 0 for no heterogeneity to 1 if each director on the BOD has only one education/expertise field or higher if there are directors’ with two or more education/expertise fields. Higher value of the Theil index corresponds to higher heterogeneity.

To reflect its usefulness for the firm operating in a certain industry board expertise is further specified through a dummy variable EXPINDSP taking value of 1 if at least one of the board members possesses industry-specific expertise and 0 otherwise.

Board nationality heterogeneity index, BOD_NATION, is computed in the same way as for

BOD education heterogeneity by the same formula:

( ) ( )

(

)

[

]

∑

where n is a number of different countries of all directors’ residence; pi represents the proportion of total number of directors residing in country i (calculated as the number of directors residing

in a given country divided by the total number of directors in the BOD); and wi indicates the

proportion that each country of residence represents in the sample (calculated as entity divided by the total number of different countries of all directors’ residence). Different countries of all directors’ residence were coded in accordance with the list of the countries (APPENDIX C)).

to the boards. Firm connection to financial institutes (FIN_TIES) is measured by the number of CEO or directors appointments as a director or CEO in BODs of financial institutions. For financial institutions themselves this number equals total number of directors plus number of ties to other financial institutions, if any.

CEO power is measured with help of five different most commonly used variables. CEO’s

tenure vis-à-vis BOD’s tenure (CEO_BOD_TEN) is computed by dividing CEO’s tenure by

average tenure of the BOD (this average does not include tenure of the CEO). If there was a change of CEO in 2004, I took into account data attributed to the person who was the CEO for more than 6 month in 2004. When a CEO was replaced by a newcomer (a person who had not been a director before he/she was appointed as a new CEO) and if this happened before or on June 30, 2004, I took into account the data on the newcomer, while if this event occurred after June 30, 2004, data on the retired CEO was considered. Portion of the board appointed after the

CEO (BODAFTERCEO) is calculated as a number of directors appointed after or in the same

year as the CEO divided by the total number of directors without CEO (or both the predecessor CEO and the successor CEO when the latter one is a newcomer). CEO duality, DUALITY, is a dummy variable taking value of 1 when CEO serves as a chairman of the board and 0 otherwise.

Board independence index, BOD_IND, was calculated as a percentage of the BOD comprised of

outside independent directors. When talking about independent directors I mean outsiders without family or business ties to the focal company. Though this is only one of the usually used dimensions of independence, it is the best one corresponding to the data in my possession. In general, I rely on the information on directors provided in companies’ proxy statements and take into account conclusions by the companies’ BODs regarding directors’ independence when available. As a result of the mentioned limitation the definition of independence here may not fully correspond to one used in the Sarbanes-Oxley Act of 2002 or accepted by the NYSE or SEC. Average age of all directors represents another demographic variable, BOD_AGE. Average

BOD tenure (AVBODTEN) is measured as an average tenure of all board members and used

only as a control variable in models comprising CEO power.

Staggered board, BOD_STAG, is a dummy variable taking value of 1 if there is more than

Control variables

I use industry category and size of company as explanatory variables which allow controlling for any BODs’ demography characteristics endogenous for an industry or a company (Murray, 1989; Baysinger and Hoskisson. 1990). I consider this choice to be wise also grounding on characteristics of the credit analysis policies employed by the S&P’s which especially evaluate company’s competitiveness within its industry. This control variable is also introduced in the attempt to check if credit ratings of corporations depend on its pertaining to a specific industry.

INDUSTRY is a nominal variable taking values from 0 to 9 depending on the industry a

firm operates in (energy, materials, industrials, consumer discretionary, consumer staples, health care, financials, information technology, telecommunication service, utilities). Firm size (LOGASSET) measured as log of total assets (e.g. Ashbaugh et al, 2004; Aggarwal and Williamson, 2006). Firm diversification index (DIVERS) is calculated as a Herfindahl index (i.e. the sum of squared proportions) on the spread of firm sales in five different geographic segments (the US, Europe, Asia/Pacific, Latin America and the rest of the world); a positive measure of diversification is obtained by subtracting the Herfindahl index from 1 (Boone et al., 2004). Finally, industry stability (IND_INSTAB) is measured through the coefficient of variation of sales within the industry and calculated with help of the following formula:

100 * _ X S INSTAB IND ₌ x _,

where Sx is the standard deviation of sales in industry x and X represents average sales of all

firms in industry x. Greater value of the index associated with greater instability of the industry. If value of the index is greater than 150%, the industry is considered to be highly unstable.

Overview of the Methodologies

Credit rating

statistical models able, with different precision, to explain and predict the ratings (see Kaplan and Urwitz, 1979, for an extensive analysis and further references). However, these researchers used either regression analysis on a multivalued (discrete) dependent variable or multiple discriminant analysis. I fully agree with the critics of Kaplan and Urwitz (1979) who put that

“Ordinary least squares regression analysis assumes that bond ratings represent equal intervals on a bond risk scale (interval scale assumption). But MDA avoids the interval scale assumption by assuming that bond ratings are nominal scale measures. That is… AAA bonds are of a different risk category than AA”

In other words, these techniques do not allow explicit use of all information conveyed by the dependent variable being qualitative, ordinal, categorical but neither interval nor nominal and not necessarily linear. Further, “to handle this situation”, Kaplan and Urwitz (1979) proposed “to distinguish between the dependent variable of theoretical interest, Y, and the observed dependent variable, Z.” They assumed that in an ideal situation it would be possible to measure the default risk (Y) with the highest precision what would put the dependent variable on an interval scale and would made it suitable to a linear model, whereas, “due to inadequate measurement or observation techniques … we only observe an ordinal version of Y, namely Z” (Kaplan and Urwitz, 1979, p240-241). Taking this assumption into account I can suppose that inasmuch as my dependent variable, long-term issuer’s credit rating, is coded in a 25-point scale it is relatively closely approximated with the abovementioned ideal situation case. In other words, it is suitable to be explained by a linear model, e.g. a multiple regression. Inasmuch as the model assumes the constant variance of the observations, i.e. homoskedasticity, after controlling for compliance with this assumption and in case of heteroskedasticity existence it will be necessary to correct the model. This may be done with the help of application of White’s estimator to the variables’ error terms (Hill et al., 2001).

order to identify the common factors affecting countries’ credit ratings Mellios and Paget-Blanc (2006) employed a principal component analysis and the impact of the variables correlated with these factors on ratings was then assessed through an ordered logistic model. Finally, when testing the predicted relations between corporate governance components and credit ratings Ashbaugh et al. (2004) estimated a series of ordered logit models. They used this type of models “because the seven categories of credit ratings convey ordinal risk assessments; thus firms’ preferences across the rating categories can be rank ordered but it is impossible to assume uniform differences in benefits (costs) between the categories” (Ashbaugh et al., 2004).

Though it is generally agreed that probit and logit yield very similar substantive results, these models are based on different data assumptions. Thus, despite probit takes into account an ordinal categorical character of the dependent variable, it assumes that the dependent variable reflects an underlying normally distributed quantitative variable. This assumption I find to be violated when talking about credit rating as it possesses qualitative characteristic, that is, while we can say that investing in a firm with the highest rating AAA is less risky than investing in a firm with rating AA we cannot say how much less risky it is (Kaplan and Urwitz, 1979). On the other hand, logit model based on the assumption of equal (or evenly distributed) categories also assumes that the categorical dependent reflects an underlying qualitative variable and uses the binomial distribution. In general, logit does not assume linearity of relationship between the independent variables and the dependent, does not require normally distributed variables, does not assume homoskedasticity, and in general has less stringent requirements than OLS. It does, however, require that observations are independent and that the logit of the independent variables is linearly related to the dependent (www.columbia.edu). Log-linear analysis used in logit allows determining relation between variables, predicting the expected frequencies of dependent variable, understanding the relative importance of different independent variables in predicting a dependent, and confirming models with help of a goodness of fit test (www.columbia.edu).

Firm liquidity

Explaining companies’ cash flows represents a less complicated task since the dependent variable is continuous. In this case logistic regression can be used to predict a dependent variable on the basis of continuous and/or categorical independents and to determine the percent of variance in the dependent variable explained by the independents; to rank the relative importance of independents; to assess interaction effects; and to understand the impact of covariate control variables (www.columbia.edu). Alternatively, a multinomial linear regression may be used.

Models Specification

Credit rating

All empirical tests are based on the general model representing credit ratings as a function of people’s characteristics corrected for firm’s characteristics and environment effects

Credit rating = f(BOD characteristics, firm characteristics)

In its fullest form the regression equation looks like following:

Yi = β0 + β1* BOD_INDi + β2* BOD_SIZEi + β3* BOD_AGEi + β4* DUALITYi

+ β5* STAG_BODi + β6* BOD_EDUi + β7* BOD_NATIONi + β8* DIR_NETi +

+ β9* CEO_BOD_TENi + β10* BODAFTERCEOi + +β11*INDUSTRYi + β12*

LOGASSETi + β13*DIVERSi + β14*IND_STAB + β15* AV_BOD_TEN+ εi

(1)

where Y=1) RAT_DISCR, the discrete variable of credit rating coded on 25-points scale (see Table 1); 2) RATING, an ordinal variable coded on 9-point scale (see Table 1); and 3)

RAT-DUM, a dummy variable taking value of 1 if a company’s credit rating is of investment grade

(BBB- of higher) and 0 otherwise; β0, β1, …. β15 are coefficients of the constant and independent

variables respectively; εi is an error term. Signs of the coefficients indicate the direction of

Firm liquidity

To test the hypotheses regarding nature of the credit ratings I will run ordered logit (binary logistic) regressions based on the general model representing company’s credit rating as a function of its operating cash flow controlled for the firm’s characteristics and environment effects:

Credit rating = f(operating cash flow, firm characteristics, environment effects)

In its fullest form the regression equation looks like following:

Zi = b0 + b1*OCF2004i + b2* INDUSTRYi + b3* DIVERSi + b4* LOGASSETi +

+ b5*IND_INSTABi + b6*FIN_TIESi + b7* DIR_NETi + εi

(2)

Where Z is a categorical variable credit rating discrete 2004 (credit rating investment grade

2004); b0, b1, …. b7 are coefficients of the constant and independent and control variables

respectively; εi is an error term. Signs of the coefficients indicate the direction of correspondent variables in relation to the dependent variable. Thus, according to the hypotheses I expect sign of coefficient β5 to be negative, while all the rest coefficients are expected to be positive. Results of the regressions presented in Table 7 and discussed below in Section ANALYSIS AND RESULTS.

Further, I run a series of multinomial linear regressions to test hypothesis about company reluctance to increase its leverage in case of sufficiency of inside resources to meet its liabilities. These regressions will also help revealing whether BOD demography characteristics used in the study influence company ability to generate sufficient resources to meet its current liabilities. The analysis will help further clarify the nature of the credit ratings by comparative analysis of the results with the results of regressions based on equation (1).

Wi = β0 + β1* BOD_INDi + β2* BOD_SIZEi + β3* BOD_AGEi +

+ β4* DUALITYi + β5* STAG_BODi + β6* BOD_EDUi + β7* BOD_NATIONi +

+β8* DIR_NETi + + β9* CEO_BOD_TENi + β10* BODAFTERCEOi +

+β11*INDUSTRYi + β12* LOGASSETi + β13*DIVERSi + β14*IND_INSTABi +

+ β15* AV_BOD_TENi+ β16* FIN_TIESi + εi

(3)

hypothesis and assuming that credit ratings indicate real state of affairs regarding firms liquidity in each period, I expect signs of coefficients β9, β14 and β16 to be negative, while all the rest coefficients are expected to be positive. Results of the regressions presented in Table 8 and discussed below in Section ANALYSIS AND RESULTS.

Finally, hypotheses regarding alternative sources of financing are tested with help of the following general model:

Cash flow = f(credit rating, BOD characteristics, firm characteristics)

In its fullest form the multinomial linear regression equation looks like following:

Ci = b0 + b1*RAT_DISCRi + b2* INDUSTRYi + b3* DIVERSi + b4*

*LOGASSETi + + b5*IND_INSTABi + b6*FIN_TIESi + b7* DIR_NETi + εi

(4)

Where C is a continuous variable CASH_FLOW 2004; b0, b1, …. b7 are coefficients of the

constant and independent and control variables respectively; εi is an error term. Signs of the

coefficients indicate the direction of correspondent variables in relation to the dependent

variable. Thus, according to the hypotheses I expect sign of coefficient β5 to be negative, while

all the rest coefficients are expected to be positive. To further specify an effect of credit ratings on companies’ cash flows I rerun the regressions substituting variable RAT_DISCR for dummy variable RAT_DUM establishing fundamental differences between investment and speculative credit ratings. This procedure should allow capturing major direction and strength of credit ratings influence, if any, on companies’ credit and investment constraints. Results of the regressions presented in Table 9 and discussed below in Section ANALYSIS AND RESULTS.

ANALYSIS AND RESULTS

Descriptive Statistics

number of financial ties for an average focal board: while ranging from 0 to 25 this variable is distributed with the mean equal to 3.15, Skewness = 2.123 and std.error = .145. Kurtosis statistics equal to 3.484 with std. error = .29 says about a high concentration of the variable distribution around the mean value. Firms in the sample are predominantly large (log of total assets mean equals 7.22, skewness=.783, std.error=.151 with Kurtosis=.643 and std.error=.3) with a low diversification index (mean = 0.2048, median is 0, the upper quartile value is .44) and came from unstable industries (mean = 245.5, median = 184.12, the lower quartile = 127.66 and the upper quartile = 269.06). Net cash flows of the firms are heavily concentrated on the left side of the distribution indicating that most of the firms in the sample are in a weak liquidity position, though log of net cash flows (not existing for negative values) is normally distributed. Finally, firms in the sample have quite a poor ability to generate operating cash flows sufficient to meet their current liabilities (mean=0.447, median=0.405, skewness=1.005 [std.error=0.164], the upper quartile =0.636). Specifically, more than 75% of the firms are not able to produce operating cash flow sufficient to meet all their current liabilities.

Insert Table 2 about here

Correlations

significant (at 0.01 and 0.05 levels respectively); while correlation BOD nationality heterogeneity becomes insignificant. Finally, correlations for net cash flow 2004 and log of net cash flow 2004 turn to be less significant (decreasing down to 90% and 95% respectively). In general, eight (seven) of the thirteen BOD demography characteristics represent correlation with the credit rating 2004 and investment grade rating (rating discrete variable) that are significant at 0.1 or below (six of them significant at 0.01 level). These initial results indicate that BOD demography variables are significant determinants of credit ratings.

Correlation between various BOD demography variables are generally fall below 30% though for some of them ranges from 30% to 50%. Thus, BOD age appears to be correlated with average BOD tenure by 35%. Larger BODs tend to be composed of directors from different countries (correlation 32%) and have more ties to financial institutions (39%), though the latter may result from the fact that banks usually have large boards and I took into account all inside and outside financial ties. BOD independence and education heterogeneity are promoted by directors networks (correlations 33% and 50% respectively) which also provide financial ties to the companies (correlation 38%). Finally, as expected, CEO tenure vis-à-vis BOD tenure correlated with the portion of the BOD appointed after or in the same year with the CEO at 87%.

There are also some interesting results for correlations between BOD variables and firm and industry variables. Thus, firm size appears to be correlated with industry (30,3%), BOD size (50%), directors’ network (37%) and ties to financial institutions (59%). Log of net cash flow 2004 is correlated with BOD size (30,4%), directors’ network (41%), firm financial ties (32%) and firm size (53%). These results, again, may be dependent on the presence of financial institutions in the sample.

is significantly positively correlated with directors’ network and CEO duality. These results may be interpreted to show that increased board independence may indicate process of institutionalization channeled through directors’ network (e.g. Westphal, 1997; Davis, 2005) or an attempt of the board to compromise CEO power.

Firm diversification is significantly negatively associated with industry stability and negatively but insignificantly – with firm size. In other words, firms in unstable industries tend to be smaller and less diversified, what is quite logical for a high risk associated with industry instability. BOD size is significantly positively correlated with firm size what supports findings of Pfeffer (1972, 1973).

Due to the high intercorrelations between industry specific expertise and board expertise heterogeneity I will exclude the former variable from the regressions (Ashbaugh et al., 2004).

Insert Table 3 about here

Regressions

Credit rating

services and utilities industries4. In general, the benchmark model has an explanatory power of 31.2%.

The remaining columns of Table 4 report the results of the hypotheses testing (Models 1 to 12) and of the regression which includes all the independent and control variables (Model 13). In Models 12, 5, 2 and 8, containing individual variables of classified terms for directors, BOD independence, CEO tenure vis-à-vis BOD tenure and BOD education heterogeneity respectively, the variables’ relations have signs opposite to predicted. Specifically, variable ‘CEO tenure vis-à-vis BOD tenure’ is significantly (at 0.01) positively related to the credit rating while for the rest variables the relation is negative though insignificant. The result may be interpreted as indicating adverse though marginal influence of staggered BODs, BOD independence and BOD education heterogeneity on companies credit ratings. All the rest variables relate to the credit rating in predicted directions. BOD portion appointed after the CEO and directors network have significant positive influence on the credit rating (at the 0.05 and 0.01 levels respectively).

When regressed together in Model 7, variables representing CEO power characteristics (BOD independence, BOD age, CEO tenure vis-à-vis BOD tenure, BOD portion appointed after the CEO and CEO duality) show no significant change from the results produced by individual models, except for BOD independence and BOD age variables which change their signs indicating marginal positive (negative) influence on the firms credit ratings under the conditions of greater overall CEO power. Thus, with greater overall CEO power BOD independence positively influences credit rating of the company due to the greater BOD ability to provide independent monitoring of management. BOD average age, in turn, is negatively associated with the credit rating. This may be due the facts that “directors’ age is positively associated with subsequent appointments to low-control boards” and “older directors may be perceived in the market for corporate directors as being less likely to embrace newer perspectives reflecting more active board involvement in decision making” (Zajac and Westphal, 1996). Therefore, they should be less able to mitigate default risk of the company due to lower ability to provide effective management monitoring. The interaction terms CEO vis-à-vis BOD tenure*Average

BOD tenure, BOD portion after CEO*average BOD tenure, and BOD independence*BOD age* CEO vis-à-vis BOD tenure* BOD portion after CEO*Average BOD tenure are intended to

better directors’ familiarity with the firm and the CEO personality and may improve BOD monitoring and risk mitigating efficiency. All three interaction terms have negative and significant coefficients (at the 0.01, 0.05 and 0.01 levels respectively) indicating negative effect of greater CEO power on the credit rating.

Model 11 contains all variables of BOD expertise (BOD education heterogeneity, BOD nationality heterogeneity and average number of all directors’ appointments to the BODs). While being significantly positively associated with the credit rating (at 0.01), average number of directors’ appointments appears to negatively affect phenomena of BOD education and nationality heterogeneity deteriorating values of their estimates. The effect may be due to negative repercussion of excessive directors’ work-load on efficiency of heterogeneous BODs. The interaction terms BOD education heterogeneity*industry instability*BOD nationality

heterogeneity*firm diversification*directors’ network (BOD nationality heterogeneity*firm diversification, and BOD education heterogeneity*industry instability), called to capture BOD

‘effective’ expertise, do not produce statistically significant results and indicate only marginal positive (negative and positive respectively) influence on the credit rating. At the same time, when control variables are excluded from the interaction term and it includes only independent variables BOD education heterogeneity*industry instability*BOD nationality heterogeneity, the result of the regression shows positive and significant at the 0.01 level relation with the credit rating, indicating positive influence of BOD’s expertise on the firm’s default risk.

In Model 13, where I jointly tested all the BOD demography characteristics’ association with the firms’ credit ratings, seven of ten independent variables have coefficients with predicted signs, while for the other three (classified directors’ terms, BOD age and BOD education heterogeneity) the prediction has failed. At the same time, only two variables (size of the BOD and directors network) significantly related to the credit rating (at 0.1 and 0.01, both positive). Surprisingly, contrary to the hypothesis, variable indicating the “mightiest of all takeover defenses” negatively relates to the credit rating. Though the relation is insignificant, it becomes more pronounced with introduction of all BOD demography variables into the model. Explanatory power of the Models ranges from 24.2 % to 37.4%. The highest explanatory power corresponds to the Models containing CEO power variables CEO tenure vis-à-vis BOD tenure

and BOD portion appointed after the CEO: Model 2 (R2 = 0.341), Model 3 (R2 = 0.329), Model