The value of CEO and CFO financial expertise

(1)

1

The Value of CEO and CFO Financial Expertise

Andrea Schmidtová

Master’s Thesis

M.Sc. Business Economics: Finance

Thesis Supervisor: Dr Florian S. Peters

(2)

2

Abstract

The main objective of this thesis is to disentangle the long standing puzzle in the field of corporate governance related to the effect of financial expertise of key executives on firm performance. This topic is of utmost relevance in the hiring process, as the relevant findings could improve firm performance as well as mitigate the agency problem and information asymmetries present in the governance structures. The main hypothesis states that financial expertise among CEOs and CFOs in a company is a value enhancing factor, due to the ability of financial experts to overcome a number of biases in the decision making process as well as to better assess the conditions in the marketplace. Although the regression analysis provides some evidence against the null hypothesis, it is subject to endogeneity. On the other hand, the event study performed in this thesis leads to the conclusion that there are significant differences in the evolution and magnitude of stock market performance as well as corporate policies when comparing a financial expert CEO with a non-expert. The results show that financial experts are better able to manage operational performance, engage in more prudent hedging practices and also ensure a more robust rebound in terms of market returns.

Statement of Originality

This document is written by Andrea Schmidtová who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

Acknowledgements

The author expresses sincere appreciation to the Supervisor of this Master Thesis, Dr Florian S. Peters for his continuous support during the course of this work. Without his help and counsel, the

(3)

3

1. Introduction

There are many ways in which corporate governance is defined. Zingales (1998) portrays it as a set of constraints which balance out the bargaining power over the surplus funds generated by a firm. Gillan (2006) holds a similar view, in which corporate governance provides rules and laws which shape company’s operations. In contrast, Shleifer and Vishny (1997) present a view which is focused on the return on investment of the suppliers of finance. All in all, the main objective of studies in corporate governance is to analyse the organisational structures, such as executive compensation, the optimal structure of the board of directors, or an established code of ethics within companies. These analyses in turn determine, which characteristics of firms lead to a better alignment of interests of both the shareholders and managers, inducing a superior performance of a given organisation.

This thesis is motivated by the same objectives. It will examine the effect of a particular characteristic of a Chief Executive Officer (CEO) and also the Chief Financial Officer (CFO) that are hypothesised to have a positive impact on firm performance. The characteristic of interest is the financial expertise of the two key executives in a firm. The justification for this particular skill is provided by the Sarbanes Oxley Act of 2002, a key piece of legislation implemented after a series of corporate governance failures and accounting frauds. The act stipulates that there must be a financial expert on the audit committee, ensuring that the financial reporting of the firm is true and fair. The definition of a financial expert by the Sarbanes Oxley Act, as outlined by the U.S. Securities and Exchange Commission (2003) has been broadened to include the persons who have the understanding and are able to assess and apply the Generally Accepted Accounting Principles (GAAP), have the relevant experience in preparing and auditing of financial statements and also have an understanding of the internal controls with regards to financial reporting. This knowledge and skills could be acquired through education and/or experience in the relevant positions within a firm.

This thesis will take the argument of the U.S. Securities and Exchange Commission one step further, examining whether the presence of a financial expert not only in the audit committee but also among the top executives creates value for the firm. It will attempt to disentangle the persisting puzzle across related literature in this matter. Some authors argue that financial expertise leads to improved performance because such executives are able to better raise and access funds, put them to their most efficient use while overcoming a number of biases, confirming the advisory channel of value creation. Financial expertise also arguably improves outcomes through the monitoring channel, through which it increases investor trust and lowers the probability of earnings restatements, all pointing to the fact that such firm should perform better on the operational as well as the stock

(5)

5

market level. On the other hand, some authors argue that financial experts are more risk-loving which induces excessive risk taking especially in the times of a crisis. They may also be overconfident and overoptimistic, or potentially increase the agency problem within a firm by acting in the interest of creditors instead of shareholders.

It is obvious that previous studies in the related field have not arrived at a clear-cut answer to the question of the value of financial expertise. It is important, therefore, to explore the potential effect of financial expertise of CEOs and CFOs on firm performance further, not only because the previous findings were ambiguous but also due to the fact that a finding an effect would enable companies to choose their executives more efficiently to ensure a better return for their shareholders. It would also provide additional insights into the functioning of a firm, which is the main motivation behind this paper. The aim of this thesis is to formulate an answer to the following research question:

„Is financial expertise of CEOs and CFOs a value enhancing factor for the companies?“

In order to answer the proposed research question, this thesis uses a dataset merged from two distinct sources. The information with regards to the main variable of interest, financial expertise of executives, as well as a range of additional characteristics, was retrieved from the Institutional Shareholder Services (formerly RiskMetrics), namely their Directors database; and accounting and stock market indicators originate from the Compustat North America, Annual Fundamentals database. After a successful merger of the two datasets, a variety of descriptive statistics of the variables of interest was generated, in order to determine the potential left hand side variables for further analysis. The results presented by the summary statistics further confirm the hypothesis that there seem to be significant differences in firm and executive characteristics based on one’s level of financial expertise and thus this topic is worth investigating in greater detail. Subsequently, two methodological approaches are suggested. As is common across corporate governance literature, one section of this thesis will use regression analysis on the panel dataset, exploring the relationship between financial expertise of CEOs and CFOs and firm performance indicators. Firm performance has been further refined into two types: stock market performance, measured by gross returns and intra-year stock price volatility; and firm operational performance, proxied by cash flows, market leverage, cash holdings, investment and return on assets, which have been chosen based on the results of the summary statistics as well as what seems to be common practice within this research field, in order to capture a variety of performance indicators. In addition to the regression analysis, this thesis also introduces an alternative empirical approach, namely an event study based around

(6)

6

the year of CEO replacement and further depending on the level of financial expertise of the old and newly appointed CEO, in order to offer some new insights into the effects at play, as well as the evolution of performance measures around the time of the event. This way, some of the concerns over the regression analysis, especially with regards to endogeneity, can be addressed. Additionally, the use of different methodical approaches provides a robustness check of the overall analysis. With regards to the findings, the results of the regression analysis largely provided evidence to the contrary of the null hypothesis and the expectations that the relationship would be of a positive nature. Overall, financial expertise among CEOs and CFOs seemed to deteriorate firm performance wherever statistical significance was reported, with a negative coefficient on gross returns and a positive coefficient on volatility. Similarly, for the corporate policies, a negative impact on cash flows and cash holdings was recorded among CEOs and a negative effect on return on assets among CFOs. Although this was a surprising finding, it should be taken with a grain of salt, mainly because of the underlying characteristics and limitations of the regression methodology, which suffered from endogeneity and reverse causality, potentially biasing the coefficient estimates away from their true values. Therefore, it is essential to also consider the results of the event study, which were more in line with the hypotheses as well as the expectations formed by related literature. It was clearly shown that there are significant differences in the evolution of the performance indicators when comparing a company which exchanged their non-expert CEO with a financial expert and a company which did the exact opposite. It will be shown that financial expert CEOs are associated with a faster and more robust rebound in the stock price, they manage their cash flows and market leverage more effectively and they build up bigger buffers in the form of cash reserves to protect the companies from unexpected negative shocks to the marketplace. Additionally, the event study revealed an interesting phenomenon worth investigating further, whereby firm tolerance to CEO underperformance varied based on his or her level of financial expertise. It took longer for a financial expert CEO with suboptimal performance to be replaced, as opposed to a non-expert.

Overall, it could be concluded that financial expertise certainly plays a role in the performance of a firm, based on the results from both the methodological approaches. Although this thesis was not able to prove a causal relationship between executives’ financial expertise and firm performance, there is strong evidence pointing to a correlation between the two. Despite this not being the ideal result, it still offers a valuable insight into the functioning of firms, the findings of which could be applied in terms of credible signals which could be efficiently conveyed from firm leadership onto the shareholders in the situation of high information asymmetry. The results of this thesis also identify a number of weaknesses of related studies, especially in terms of the robustness of the results they

(7)

7

report, which seem to be largely dependent on the time period and dataset under consideration and the definition of a financial expert.

Clearly, one cannot arrive at new findings using the same datasets and methods as the previous studies. There is a number of characteristics that distinguish this thesis from the previous research, particularly referring to two papers by Custodio and Metzger (2014), and Guner, Malmendier and Tate (2008), forming the core of its original contribution. First of all, a completely different dataset is used for the purposes of the empirical analysis. It takes advantage of the fact that the Institutional Shareholder Services (ISS) data provider started to include information on the financial expertise in their Directors database in 2007. This particular variable is key for the analysis in this thesis and is based on a broader definition of financial expertise, as provided by the Sarbanes Oxley Act and outlined above. The use of such database allows for a twofold contribution of this research. First of all, it provides a robustness check to the previous studies, which only used a hand-collected sample of firms and the previous education and experience of its executives. Secondly, it allows us to perform the analysis on a much larger dataset, which increases the validity of this study. Not only is the sample larger, but it also covers a completely different time period. The two main related articles which this thesis builds upon work with datasets covering years up to the recent financial crisis, whereas this dataset works with years 2005 to 2015 and takes into account the effects of the recession. Another key differentiation point lies in the type of analysis used in this thesis. It will present an event study of the effect of CEO turnover on firm performance, distinguishing between different types of skills that the old and new CEOs possess, which has not been conducted in any previous academic paper. Additionally, the regression analysis distinguishes between CEOs and CFOs and allows for the comparison of the effects. Finally, this thesis will considers behavioural finance explanations as well as the signalling hypothesis, which could be used to analyse the implications of the findings in practice.

The remainder of this paper is organised as follows. Section 2 provides an overview of related literature, Section 3 will describe data selection and present descriptive statistics of the sample. Section 4 will propose a number of methodological approaches to data processing, Section 5 analyses the empirical results, discusses potential limitations of this study and suggests potential areas of further research, and Section 6 will conclude the thesis.

(8)

8

2. Literature Review

2.1. CEO skills and talent matter

The core of the discussion on the importance of managerial skills in terms of corporate policies and financial performance stems from the dichotomy of the understanding of a manager’s role in forming a business. Bertrand and Schoar (2003) contrast two opposing views of the influence of executives on firms. First of all, the neoclassical model assumes that executives are homogeneous and perfectly substitutable. The model also predicts that no single executive can exert enough power to control the whole organisation. On the other hand, the more recently developed agency models relax these arguably unrealistic assumptions. They build upon the hypothesis that top managers have a great discretion over the decisions being made in a firm, based on a variety of factors, such as seniority, accumulated experience or professional networks. Executives may also pursue their own objectives often misaligned with those of the shareholders. The heterogeneity and power of executives based on the agency view then provides the justification for further studies in terms of the effect of managers’ skills on corporate policies and firm performance.

First of all, the power hypothesis as a part of the agency model has been documented by Adams, Almeida and Ferreira (2005), who measured the overall influence of a CEO in a firm based on his or her insider holdings, formal positions and being a founder of the business. The authors found that firms with a more centralised governance structure are susceptible to the risk of judgement errors of the powerful CEO, which causes a more volatile performance due to the under-diversification of power structures. Therefore, this research provides evidence in favour of the agency model of managerial skills and power.

Secondly, the fact that various types of managerial talent and expertise have an effect on shareholder value and firm policies has been documented in a number of studies. Bennendsen, Perez-Gonzalez and Wolfenzon (2007) researched the effect of sudden deaths of CEOs, board members and their immediate family members on firm performance. Not only was the market reaction to such events negative, but also firm profitability, investment decisions and sales growth deteriorated. A study by Lin, Ma, Officer and Zou (2011) revealed that CEOs with previous experience in the military are able to better realise short-run and long-run synergies and experience positive abnormal returns on the announcement of acquisitions, which the authors attribute to the value system characteristic to the military structures. Such CEOs are also able to improve corporate governance and decrease excess cash holdings of the firms. Additionally, Nguyen, Hagendorff and Eshraghi (2015) argue that the announcement of appointing a director who is older, has prior executive experience and/or an Ivy League degree is accompanied by positive short-term abnormal

(9)

9

returns, based on their event study. Furthermore, Bertrand and Schoar (2003) found that earlier birth cohorts tend to act on the more conservative side, whereas younger CEOs and especially MBA graduates pursue more aggressive strategies. Among others, Malmendier and Tate (2005) also link CEO characteristics to corporate decision-making. The authors conclude that overconfident CEOs fail to diversify their insider holdings, tend to overestimate returns to investments and are reluctant to raise external funding due to the information asymmetries present on the capital markets.

There is also a number of experimental and innovative studies with regards to the effects of other CEO characteristics. For example, Halford and Hsu (2014) found that the returns around the appointment announcement of physically attractive directors are larger and positive. Such directors also enjoy higher acquirer returns in their transactions which persist beyond one year. The authors attributed this inconsistency to the fact that investors’ decisions are largely guided by the often unconscious first impressions of the directors in question. Additionally, a study by Bernile, Bhagwat and Rau (2015) shows that CEO characteristics can be traced back to childhood, where CEOs experiencing extremely adverse catastrophic events turn out to be more conservative and cautious than CEOs who experienced non-extreme disasters and became more desensitised and aggressive in their actions. Consequently, a clear link between attitudes toward risk and shareholder value or firm performance can be made.

To summarize, a number of studies have shown that key executives exert significant power over the functioning of the companies they lead. Due to the heterogeneity in terms of their skillsets and talents, executives’ actions in turn influence the profitability, operational and financial performance of the respective companies and are therefore worth exploring in greater detail.

2.2. Generalist versus specialist skills

Now that it has been established that CEO characteristics and skills have a significant influence over firm policies and stock performance, it is necessary to differentiate between the types of skills the CEOs might have. The most influential paper on this topic was presented by Custodio, Ferreira and Matos (2013). The authors differentiate between generalist skills which are transferrable across firms and industries, such as the breadth of past working experience, executive positions held or working at a conglomerate; and specialist skills, which are particular to a certain firm or industry. The value created by generalist skills, as measured by CEO surplus pay of professional managers which amounts to up to 19% (or $1mil. p.a.), was attributed to the recent changes in the markets in terms of deregulation, technological advances or changes in the product market. Relatedly, Murphy and Zabojnik (2004) indicated that the rising CEO compensation over the years cannot be explained by

(10)

10

the “fat-cat” theory of increasingly entrenched executives, but rather to the changing optimal mix of managerial skills sought for in the labour market. The authors argue that the value of transferrable skills such as the ability to manage a large firm or knowledge in the fields of economics, finance or management are more valuable than firm-specific knowledge about its clients, product or suppliers. This is attributed to the fact that the latter form of skill can be acquired easily and quickly due to the computerisation of the business environment. Murphy and Zabojnik (2007) also found that the pay of externally hired, presumably generalist managers, is higher than for internally promoted CEOs, further reinforcing the argument of the value of general, transferrable skills.

Related to that, Kaplan, Klebanov and Sorensen (2012) provide further evidence to the previous conclusions about the importance of management skills transferrable across firms and industries. The authors found that in the case of buyout and venture capital transactions, the future corporate performance is determined more by the execution, resoluteness and overconfidence of the acquirer CEO than his or her “soft” skills, such as communication, interpersonal skills and team-leadership. On the other hand, a study by Custodio and Metzger (2013) provides a counterargument and highlights the importance of industry expertise when it comes to creating value through diversifying acquisitions. The paper argues that CEOs with previous experience in the industry of the acquisition results in capturing of more of the transaction surplus through a better negotiating position as well as a lower premium paid. Similarly, the study by Masulis, Ruzzier, Xiao and Zhao (2012) concludes that the appointment of an independent expert director, determined by previous work experience in the same industry, leads to better firm performance, higher cash holdings and a lower frequency of earnings restatements. The authors attribute this finding to the fact that such directors suffer from less of a conflict of interests and better align own interests with those of the shareholders. Thus, in some cases, industry-specific expertise can be value enhancing despite the seemingly growing focus on generalist managerial skills.

Undoubtedly, there is evidence for the value of the transferable generalist skills when it comes to managing a firm, as can be demonstrated by director remuneration differentials and managerial labour market outcomes. However, one should be careful when assessing the overall value of generalist skills when put in contrast with firm- or industry-specific skills, which also prove valuable. Therefore, the overall effect of CEO skills based on the existing academic research is ambiguous and further study of this topic could bring valuable insights.

(11)

11

2.3. Financial expertise as a value-enhancing factor

If we accept the argument that generalist, transferrable skills are particularly valuable to the company, we may further analyse a specific skillset in greater detail. It could be argued that expert knowledge of finance, whether that is in terms of financial statements and reporting, the ability to raise funds efficiently or to assess the overall risk position of a company, is widely applicable to a range of industries and easily transferrable. More importantly, it is also valuable in the way that it takes time to acquire it and excel in applying it in practice. The following papers provide arguments for and against this proposition and offer examples of how CEOs and board members with financial expertise affect corporate policies and enhance company value.

First of all, Guner, Malmendier and Tate (2008) investigate the effect of directors with financial expertise on corporate decisions, such as funding and investment to cash flow sensitivity. The authors focus on the board memberships of bankers in non-financial institutions and find that their influence is indeed sizeable; however it varies to a great extent. The presence of commercial bankers on the board ensures more funding, especially for otherwise constrained firms and a lower investment to cash flow sensitivity. On the other hand, investment bankers among the directors improve the bond issuance of a firm. However, there is also a downside of bankers on the boards. A significant conflict of interests arises when the directors act more in the interest of creditors than the shareholders. Also, the flow of funds may be wasteful, especially when less profitable investments are pursued. The authors conclude that while the monitoring function of financial experts on the board is beneficial, the advisory role may be detrimental to shareholder value. It should be noted that this thesis is closely related to this research and adopts some of the methodology. Additionally, the conclusions with regards to bankers on boards of directors have also been supported by Morck and Nakamura (1999) who found in their sample of Japanese firms, that bankers tend to act in the interest of creditors and the positive stock market reaction to their appointment is driven by the potential improvements in liquidity (e.g. access to funding) rather than due to better oversight. Furthermore, Kroszner and Strahan (2001a) also point out the room for the conflict of interests of bankers on the boards in the interplay of directors’ fiduciary duty to the shareholders versus lender interests and lender liability. Thus, bankers tend to be present on the boards of safe firms rather than in companies where the additional oversight is actually needed. On the other hand, Kroszner and Strahan (2001b) argue that such board linkages mitigate information asymmetry and improve the flow of funds, which in turn enhances shareholder value.

Secondly, Custodio and Metzger (2014) further analyse the effect of financially-savvy directors on corporate decisions, providing inspiration and a comparison tool for this thesis. The authors find that the presence of directors with a career background in finance leads to more financially sophisticated

(12)

12

policies and the ability to find better financing options for the companies even under adverse circumstances, thus making value-enhancing decisions on behalf of the shareholders. Such directors are able to overcome the weighted average cost of capital (WACC) fallacy, where they distinguish between firm-wide and investment-specific WACC. The companies they govern also hold less cash, more debt and engage in more share repurchases, overall pointing to a more dynamic style of management, responding faster to policy changes, such as dividend tax rate cuts. Similarly, Harris and Raviv (2008) argue in favour of the benefits of financial expertise within a company, stating that such skills decrease the costs of acquiring subject-specific information, recognising and assessing the risks and complexity of the business model and environment and allow for better management and monitoring of a company.

Another avenue for the reasoning that financial expertise leads to value creation was proposed by Davidson, Xie and Xu (2004). The authors study a sample of voluntary appointments of financial experts on the audit committees of firms before it was made mandatory by the Sarbanes-Oxley Act of 2002. They find positive abnormal returns upon such announcements, especially for the appointees who were previously auditors themselves. The effect is smaller for directors with previous corporate financial management experience and financial statement analysis experience. Thus, the market rewards more finance-related skills in companies even if it is not a result of regulatory efforts. Along the same lines, DeFond, Hann and Hu (2005) performed a similar study of 3-day cumulative abnormal returns of appointing financial experts on the audit committee and conclude that investors especially value previous financial reporting experience, particularly in companies with already strong corporate government systems. The authors hypothesise that corporate governance mechanisms and financial experts on the boards may be complementary and that one facilitates the efficient use of the other.

Finally, additional benefits of financial experts on company boards include investor trust in the financial reporting. Agrawal and Chadha (2005) studied the probability of earnings restatements in relation to the presence of financial experts on the audit committee or the board of directors in general. They find that such probability decreases in line with the existence of the valuable oversight of the reporting practices. Additionally, Anderson, Deli and Gillan (2003) also find that the investors regard the earnings announcements more informative when the board contains an independent financial expert who is also active in voicing his or her concerns.

On the other hand, Minton, Taillard and Williamson (2014) provide a counterargument to the benefits of a financial expert on the board. The authors studied the effect of appointing an independent director with financial expertise on the boards of U.S. banks and concluded that such

(13)

13

director engages in more risk-taking activities, adversely affecting the bank’s performance, especially in the times of a crisis. Thus, financially-savvy directors are also characterised by a more benevolent attitude towards risk, which can be dangerous in some situations. It is unclear, however, if these conclusions can be extrapolated to industries outside of the banking sector. Financial institutions are by their nature heavily reliant on debt finance. The relatively small equity share of capital structure, often less than 5% of the balance sheet induces risk-taking by the shareholders, who are protected from downside risk by the option of a bailout in case of a bank failure, as argued by Eisenbach, Keister, McAndrews and Yorulmazer (2014).

Another line of reasoning for the detrimental effects of financial expertise on firm performance can be drawn from the insights of Behavioural Finance. Their arguments rely on the theory of non-standard beliefs in the form of overconfidence and overoptimism. Overconfidence, based on the definition by Malmendier and Tate (2005), relates to the inappropriate assessment of one’s abilities and judgements. Following a series of past successful decisions, overconfident managers tend to overestimate their own abilities and make suboptimal decisions. Therefore, it could be argued that executives who are also financial experts are more likely to exhibit this bias and negatively affect firm performance. The authors show that overconfidence has detrimental effects on firm policies and performance. This phenomenon was also shown in a study by Ben-David, Graham and Harvey (2010) who measured the miscalibration of the stock market returns volatility by finance professionals whose predicted distributions were too tight. The authors also showed that overconfident financial expert executives were more aggressive, a finding which is in line with the increased risk appetite of such directors. Similar conclusions have been drawn by Hilary and Hsu (2011). Additionally, Malmendier and Tate (2008) also draw the link between managerial overconfidence and their perceived ability to generate abnormal returns and conduct more acquisitions, which are value-destroying on average. Overoptimism, on the other hand, is defined by the misperception of the expected value of a given variable. As concluded by Schrand and Zechman (2012), executives under the influence of such bias are prone to financial reporting misstatements because they believe that the negative performance will be reverted in the short run. This study is further supported by a paper by Ahmed and Duellman (2012) who find that overoptimism affects reporting behaviour and such executives are more likely to put off loss recognition.

In conclusion, the evidence of financial experts on the boards of directors or in the form of either CEOs or CFOs largely points to operational and financial improvements, efficiency gains, better access to funding, mitigation of information asymmetry as well as increased investor confidence in the business. One should, however, be vary of the potential downside risks of such appointments, as they increase the risk-taking capacity of companies, or some of the biases described by behavioural

(14)

14

finance theories, which could deteriorate the company’s position, especially during crisis years. Clearly, the related literature yields a persisting puzzle regarding the overall effect of financial expertise on firm performance and it is important to investigate the topic further.

2.4. CEO turnover

Finally, it is relevant for the purposes of this thesis to make a connection between firm performance and CEO turnover due to the alternative methodology used in the empirical analysis. Helwege, Intintoli and Zhang (2012) analyse the historical evolution of CEO turnover rates and conclude an upward trend in the likelihood of CEO replacement as well as a decline in CEO tenures. The authors attribute this trend to the increase in investor activism, as opposed to voting with their feet and a more prominent presence of institutional investors among shareholder, which in turn increases the size of block-holdings. The authors find that the shareholders are better able to punish an unsuccessful CEO for his or her suboptimal performance. Additionally, Kaplan and Minton (2012) argued that one of the main driving factors behind CEO turnover is the underperformance of a given firm.

The fact that CEOs are being punished for inferior performance, as shown by previous research, justifies the use of CEO replacement as one of the methodological approaches of this thesis and will be built upon in further sections of this paper.

3. Data

3.1. Sample

The dataset used in this study is collated from a number of sources accessed via the Wharton Research Data Services. The information regarding the main variable of interest – executives’ financial expertise as well as additional director characteristics (demographic, professional and past experience) originates from the Institutional Shareholder Services (ISS) database using the Directors dataset. The sample used in the analysis covers all the available years for which the financial expertise information has been collected, ranging from 2007 to 2014. Subsequently, a list of all unique firms covered by the Directors sample was compiled, based on their cusip codes. For the list of firms based on their cusip codes, accounting information was retrieved from the Compustat North America Annual Fundamentals database, starting in 2004 and using all the available years up to 2015. These variables include proxies for corporate policies, such as investment, cash flows and leverage;

(15)

15

as well as additional control variables to be used in the regression analysis, such as firm size. Yearly closing, high and low stock prices were also retrieved from this dataset, for the purposes of an evaluation of stock market performance. The time period covered was extended beyond that of the Directors database in order to allow for the use of lagged variables as well as for the purposes of an event study and its extended event window. Additionally, the inclusion of pre-crisis fiscal years was necessary due to the external validity concerns of this paper.

The characteristics of this dataset form one of the main contributions of this research in the form of using more recent data, which also cover the 2008 financial crisis. Additionally, no other research uses the Directors database from ISS, specifically for the case of the main independent variable (Financial Expertise), as is reported in this dataset. Other academic papers tend to use a hand-collected database with regards to executives’ previous education and experience background, which largely restrict their sample size and introduce the potential for external validity concerns. For illustration, Guner et al. (2008) work with 282 firms and 2,910 firm-year observations, whereas this thesis uses a multiple times larger final dataset, depending on the type of analysis conducted.

3.2. Summary Statistics

The initial Directors database contains 110,744 unique observations, out of which there are 2,051 firms governed by 17,743 directors. The sample includes 2,226 Chief Executive Officers (CEOs) and 14,828 CEO-year observations; and 124 Chief Financial Officers (CFOs) and 1,228 CFO-year observations.

Please refer to Table 1 for the summary statistics of executives’ characteristics. The first section of Table 1 provides an overview of the demographic characteristics of the whole sample of directors. The average director is 64 years old, after correcting for the apparent outliers and data entry errors. 13.4% of the directors are female, whereas a large majority are males. Caucasian descent is the most prevalent, encompassing 80% of the sample, with African-American, Asian and Hispanic ethnicities represented to a significantly smaller extent. All the other ethnic types occur in less than 1% of the sample and have therefore been categorised as “Other”. The next two sections of Table 1 present summary statistics specific to the two executive positions of interest, CEOs and CFOs and their professional characteristics. It is surprising that only 0.5% of the CEOs in the sample qualify as financial experts, based on the definition in the Sarbanes Oxley Act. This value is higher for CFOs, reaching 14.5% of the sample. Although the proportion is higher for CFO, which was expected, the value still seems quite low considering that one would expect there to be more CFOs with financial

(16)

16

expertise. This inconsistency could be attributed to the rigid criteria and potential judgement calls which had to be made when assembling the ISS Directors database. The relatively low proportion of financial experts in the sample will have to be taken into consideration when conducting and interpreting the results of the empirical analysis. With regards to other professional characteristics, the average tenure of a CEO is slightly above 11 years, whereas CFOs serve for almost 12 years on average. CFOs are more likely to be independent, whereas CEOs seem to be more related to their respective companies. On the other hand, CEOs hold more outside directorships than CFOs, having a broader network of professional connections.

The right side of Table 1 presents the results of a t-test analysis of the difference of means of the variables of interest based on the financial expertise of the executives. In this case, the t-test of means is used because for the continuous variables in the sample, means and medians are similar in magnitude; and a test of medians would be irrelevant for dummy variables. In the sample, there are 4,298 financial expert directors, as opposed to 17,445 non-financial experts. As we can see, all but three of the differences in means are statistically significant at the 1% level. On average, a financially-savvy executive is older and is more likely to be of Caucasian descent. More importantly, there seem to be statistically significant differences in professional characteristics of the two key executives. Financial expert CEOs are more likely to hold additional directorships as well as be independent, when compared to their non-expert counterparts. The same is the case for CFOs, based on their financial expertise. Overall, we may conclude that there are significant differences in the characteristics of executives, which, based on the arguments presented in Section 2 of this thesis, are expected to have an impact on firms’ operational and financial performance, further justifying this field of study, as well as the use of executive characteristics as control variables in the regression analysis. Finally, the results obtained in Table 1 are largely consistent with those presented by Custodio and Metzger (2014), further validating their relevance.

(17)

17

Variable Observations Mean Median Std. Dev. Financial Expert Nonfinancial Expert Std. Error t-value

Directors Demographic Characteristics (n=4,298) (n=17,445)

Age 21,721 63.678 64 8.687 64.589 63.453 -1.136 *** 0.148 -7.685 Female 21,743 0.134 0 0.341 0.139 0.133 -0.006 0.006 -0.999 Caucasian 21,743 0.800 1 0.400 0.857 0.786 -0.071 *** 0.007 -10.389 African-American 21,743 0.037 0 0.189 0.033 0.038 0.005 0.003 1.505 Asian 21,743 0.026 0 0.160 0.019 0.028 0.009 *** 0.003 3.358 Hispanic 21,743 0.016 0 0.127 0.019 0.016 -0.003 0.002 -1.236

CEO Professional Characteristics

Financial Expertise 2,226 0.005 0 0.073 Tenure 2,225 11.053 9 8.950 13.167 11.041 -2.126 2.591 -0.820 Insider 2,226 0.952 1 0.214 0.000 0.957 0.957 *** 0.059 16.353 Independent 2,226 0.044 0 0.206 1.000 0.039 -0.961 *** 0.056 -17.121 Linked 2,226 0.003 0 0.056 0.000 0.003 0.003 0.016 0.195 Other Directorships 14,828 0.679 0 0.823 1.462 0.648 -0.814 *** 0.035 -23.486

CFO Professional Characteristics

Financial Expertise 124 0.145 0 0.354 Tenure 124 11.855 10 7.593 8.444 12.434 3.990 *** 1.910 2.089 Insider 124 0.790 1 0.409 0.000 0.925 0.925 *** 0.063 14.729 Independent 124 0.202 0 0.403 1.000 0.660 -0.934 *** 0.059 -15.826 Linked 124 0.008 0 0.090 0.000 0.009 0.009 0.023 0.411 Other Directorships 1,228 0.376 0 0.647 0.422 0.355 -0.066 * 0.040 -1.668 Table 1

Directors Summary Statistics (2007-2014)

***p<0.01, **p<0.05, *p<0.1

Notes: This table reports the descriptive statistics of the Directors dataset from the ISS database, covering years from 2007 to 2014. The variables Age, Tenure and Other Directorships are continuous, whereas the remaining variables are binary. Variables were winsorized at the 1% level where appropriate. The right side of the table reports the results of a t-test on the difference of means between two groups, determined by the financial expertise of the directors.

(18)

18

In the next step, the Directors dataset was merged with the Compustat Fundamentals dataset. This process yielded the Master dataset, which contains 121,810 unique observations. The merger was performed based on company identifier (ticker) and year of observation. Please refer to Table 2 below for a detailed description and formulas of the different measures used in further analysis.

The descriptive statistics of the variables of interest to be used as dependent or control variables in the regression analysis are presented in Table 3. The analysis uses the accounting measures presented by Guner et al. (2008), Custodio and Metzger (2014), as well as is common practice across related literature. For detailed formulas of each measure including the Compustat abbreviations,

Variable Definition

Table 3 Firm Summary Statistics

Assets Book value of assets (AT)

Asset Growth Book value of assets (AT) in year t minus book value of assets in year t-1, divided by book value of assets in year t-1

Capital Property, plant and equipment (PPENT) Investment Capital expenditure (CAPX)

Cash Flow Income before extraordinary items (IB) plus depreciation and amortization (DP) Cash Ratio of cash and short-term investments (CHE) and book value of assets (AT) Retained Earnings Retained earnings (RE) divided by common equity (CEQ)

R&D Research & development expenditures (XRD) divided by book value of assets (AT) Investment/Lagged Capital Capital expenditure (CAPX) divided by lagged property, plant and equipment (PPENT) Cash Flow/Lagged Capital Income before extraordinary items (IB) plus depreciation (DP), divided by lagged property,

plant and equipment (PPENT)

Tobin's Q Book value of assets (AT) plus market value of equity (CSHO*PRCC_C) minus book value of equity (AT-LT-PSTKL+TXDITC+LO), all divided by total value of assets (AT)

ROA Earnings before extraordinary items (IB) plus interest expense (XINT) plus income statement deferred taxes (TXDI) plus investment tax credit (ITCI), all divided by lagged book value of assets (AT)

ROE Net income (NI) divided by lagged book value of equity (BKVLPS*CSHO)

Altman's Z-score 3.3 times the difference between operating income before depreciation (OIBDP) and depreciation and amortization (DP), plus sales (SALE), plus 1.4 times retained earnings (RE), plus 1.2 times working capital (WCAP), all divided by book value of assets (AT)

Book Leverage Long-term debt (DLTT) plus current liabilities (DLC) divided by the sum of numerator plus stockholder equity (SEQ)

Market Leverage Long-term debt (DLTT) plus current liabilities (DLC) divided by the sum of numerator plus market value of equity (CSHO*PRCC_C)

Net Book Leverage Long-term debt (DLTT) plus current liabilities (DLC) minus cash and short-term investments (CHE), all divided by the book value of assets (AT)

Table 4 Stock Return Summary Statistics

Gross Return Closing share price (PRCC_C) in year t minus closing share price in year t-1, divided by closing share price in year t-1

Volatility Annual price high (PRCH_C) minus annual price low (PRCL_F), divided by the average of the two values

Table 2

Variable Definitions

Notes: This table reports the definitions and formulas used in the calculation of the accounting measures reported in Table 3 and Table 4 and also in the empirical analysis. The abbreviations reported in the brackets refer to the respective Compustat items.

(19)

19

please refer to Table 2 above. In a short summary, assets represent the book value of total assets, asset growth is the estimated annual change in assets, capital is defined as property, plant and equipment. Investment is characterised as capital expenditures, and cash flow is the sum of earnings before extraordinary items and depreciation. The latter two measures are also normalised by the value of lagged capital for more appropriate comparison purposes. Cash is the level of cash and short-term investments normalised by total assets, retained earnings are normalised by the value of common equity. Research & development expenses are further divided by total assets. Tobin’s Q is defined as the sum of total assets and market equity minus the difference of total assets and liabilities, divided by total assets. Return on Assets (ROA) is measured as the ratio of income and lagged total assets. Return on Equity (ROE) is net income normalised by lagged book equity. Altman’s Z-score is calculated according to the widely accepted formula, described in Table 2. Book leverage equals the sum of long-term debt and debt in current liabilities, scaled by the sum of the numerator and stockholders’ equity. Market leverage is calculated as long-term debt plus debt in current liabilities scaled by the sum of the numerator and market equity. Net book leverage compares the indebtedness of a company relative to its liquid holdings and is calculated as a ratio of long-term debt plus current liabilities minus cash, relative to the book value of assets.

(20)

20

Variable Observations Mean Median Std. Dev. Difference

Financial Expert Nonfinancial Expert

Assets ($Million) 111,282 20723.060 3564.100 66196.880 3705.200 3760.389 -55.189 1.498

Asset Growth 111,155 0.068 0.037 0.164 0.038 0.039 -0.001 1.930 *

Capital ($Million) 107,400 2778.882 383.983 6395.127 444.743 418.832 25.911 -1.940 *

Investment ($Million) 99,431 449.689 74.746 1001.651 83.500 79.587 3.913 -0.861

Cash Flow ($Million) 106,312 969.553 219.300 2158.076 242.714 235.240 7.474 -1.308

Cash 99,583 0.140 0.086 0.150 0.090 0.085 0.005 -3.745 ***

Retained Earnings 111,064 0.598 0.594 8.199 0.597 0.609 -0.012 1.625

R&D 55,828 0.040 0.020 0.052 0.019 0.019 0.000 -1.075

Investment/Lagged Capital 96,207 0.344 0.191 4.497 0.191 0.191 0.000 0.431

Cash Flow/Lagged Capital 103,066 1.255 0.568 7.340 0.559 0.567 -0.008 1.578

Tobin's Q 86,922 1.728 1.399 1.131 1.374 1.391 -0.016 1.882 * ROA 54,482 0.079 0.075 0.083 0.074 0.076 -0.002 2.549 ** ROE 87,962 0.145 0.130 0.318 0.127 0.132 -0.005 3.766 *** Altman's Z-score 86,590 2.000 1.919 1.182 1.912 1.925 -0.013 1.246 Book Leverage 110,150 0.349 0.345 0.244 0.354 0.345 0.009 -2.879 *** Market Leverage 98,704 0.222 0.180 0.195 0.186 0.181 0.004 -1.619 Wilcoxon z-value Table 3

Firm Summary Statistics (2004-2015)

***p<0.01, **p<0.05, *p<0.1

Median

Notes: This table reports the descriptive statistics of the Compustat North America Annual Fundamentals dataset, covering years from 2004 to 2015. Variables were winsorized at the 1% level where appropriate. All the variables are expressed as ratios, unless otherwise indicated, including a unit of measurement. The right side of the table reports the results of a Wilcoxon rank-sum test on the difference of medians (distribution) between two groups, determined by the financial expertise of the directors. Please refer to Table 2 of this thesis for detailed formulas.

(21)

21

With regards to the summary statistics obtained, the results presented in Table 3 are largely in line with the related literature, specifically the two key papers this thesis refers to, by Custodio and Metzger (2014) and Guner et al. (2008). The discrepancies usually only apply to the mean value of a variable, whereas the median is compatible with the expected results to a large extent. The underlying reason for the differences is the time range considered by this research. Related academic papers usually cover the years before the recent financial crisis, whereas this dataset is centred around it and includes the years of the financial crisis. This is due to the availability and collection constraints of the variable of interest, as well as the aim of this paper to use a more recent dataset to confirm the findings in the related literature. Therefore, some of the profitability measures are skewed downwards by the large outliers caused by the crisis. It is important to note, however, that the median values conform to the standard range predicted for these measures and can thus be considered as reliable. One can also notice that the standard deviations of the variables are of a very large magnitude compared to the means and medians, which is also attributable to the large dispersion and heterogeneity among the firms in the sample. Finally, it is important to point out that the time period covered by Table 3 is extended beyond that of Table 1 due to the fact that further analysis will require a number of lags for some of the variables. Also, the sample size for the event study may be extended if there is more past accounting information available in the dataset.

Unlike in the analysis in Table 1, the values of mean and median are vastly different for the balance sheet items and profitability measures, which is likely caused by the increased skewness of the distribution of these variables, including extreme values especially in the left tail. Therefore, the right side of Table 2 reports the results of a Wilcoxon rank-sum test of the equality of the medians, comparing the firms with financial expert executives to their counterparts, instead of a t-test on the difference in means presented in Table 1. Such approach to evaluating the equality of distributions among two groups with particular applications in the field of Finance have been previously used by, for example, Peiro (1999) or Doyle, Ge and McVay (2007). Although no definite conclusions with regards to a causal relationship can be drawn by this test, the results indicate that in many of the cases, the distribution of the variables of interest is statistically significantly different across the two groups, providing further justification for this field of study. Additionally, this table offers us insights into the potential dependent variables to be used in the empirical analysis, based on the statistical significance of their differences in distribution, for example cash holdings, return on assets or leverage.

To summarize, a median firm in the sample has $3.6bn in assets, which grow by 3.7% year on year. It holds 8.6% of total assets as cash and short-term investments and invests 2% of the book value of assets into innovation and development of new products. Its return on assets amounts to 7.5%

(22)

22

annually and its market leverage reaches 18%. The average value of the Altman’s Z-score indicates that it is in a healthy operating position and the Tobin’s Q of more than 1 indicates that its market value exceeds its book value, on average.

Finally, Table 4 below lays out the final element of the analysis, the stock market performance of the firms within the sample period, based on the methodology presented by, for example, Core, Holthausen, and Larcker (1999). Two main measures are used to characterise the distribution of the results. Gross return is calculated as the annual growth rate of year-end closing stock prices. Volatility is defined as the within-year differential between the stock price high and low, scaled down by the average of the two values. As we can see, the average gross stock return of all the firms in the sample was 15.9% and the mean intra-year price volatility was 54.3%. The magnitude of these values is largely affected by the sample period, which includes the financial crisis, as well as the fact that this is a raw measure and is not weighted in any way, therefore susceptible to outliers. The Wilcoxon rank-sum test suggests that there are statistically significant differences between companies with and without a financial expert executive, especially for the volatility of stock returns.

To illustrate the evolution of stock returns and volatility further, please refer to Figure 1, which traces out the average gross return and intra-year volatility across the years covered by the sample, with the dashed line representing the U.S. annual GDP growth rate (scale on the right axis), retrieved from the World Bank database. It is clear, how the financial crisis affected both of these measures, with volatility spiking in the period from 2007 to 2009 and returns hitting record lows over the same years. When we combine the findings presented in Section 2.4 with regards to the negative

Variable Observations Mean Median Std. Dev.

Gross Return 107,997 0.159 0.000 0.719 1.252

Volatility 110,803 0.543 0.463 0.295 -3.697 ***

Table 4

Stock Return Summary Statistics (2004-2015)

Wilcoxon z-value

***p<0.01, **p<0.05, *p<0.1

Notes: This table reports the descriptive statistics of the Compustat North America Annual Fundamentals dataset, covering years from 2004 to 2015. Variables were winsorized at the 1% level where appropriate. Gross return is approximated by an annual growth rate of the closing year stock price. Volatility is calculated as the differential between a yearly high and low stock price, scaled by the average of the two. Both variables are expressed as

percentages. The right side of the table reports the results of a Wilcoxon rank-sum test on the difference of medians (distribution) between two groups defined by the financial expertise of the directors.

(23)

23

relationship between firm performance and CEO turnover, it is obvious why this particular time period is of great interest for further analysis. The significant dispersion of both the stock returns and volatility allows for a more precise estimation in the regression analysis. Additionally, at the time of a crisis, CEO replacement is more likely, making for a bigger sample size in the event study. Further, companies may not only be replacing their CEOs but they may also think of financial expertise as a value enhancing factor, thereby increasing the proportion of financial experts hired as key executives, further justifying this field of study and time period considered.

4. Methodology

4.1. Regression Analysis

The first part of this thesis will analyse the effect of CEO and CFO financial expertise on the variables of interest using OLS regressions of the panel dataset. Before the regression analysis can be conducted, the merged Directors and Compustat dataset is split into two; one dataset covering CEOs and the other encompassing CFOs, in order for the combination of firm identifier (gvkey) and year to uniquely identify all observations.

(24)

24

The main regression equation of interest exploring the relationship between CEO financial expertise and firm performance is as follows:

Rit = β1*FINdummyit + γ*Controlsit + αi + λt + εit

The main coefficient of interest, β1, will capture the difference of returns for firms which have financial experts as CEOs versus non-financial experts. The control variables in this regression will be based on the standard practice in this field of study, holding constant the main accounting indicators such as the leverage or firm size, subsequently adding CEO demographic characteristics; as well as industry controls. Additionally, a number of suitable interaction terms will also be used in the regression, for example a dummy for financial expertise interacted with CEO independence, age or tenure, which could highlight additional key characteristics of a CEO. The regression equations will also use time and entity fixed effects to control for time-invariant and entity-invariant characteristics and clustering of standard errors on the firm level to correct for heteroskedasticity and correlation of errors within firms across time.

An alternative specification exploring the effect of financial expertise on corporate policies instead of stock market performance is as follows:

OperationalProxyit = β1*FINdummyit + γ*Controlsit + αi + λt + εit

In the alternative specifications, the dependent variable will be interchanged for proxies of firm value-enhancing decisions, namely the cash flows, market leverage, cash holdings, investment and return on assets. These variables have been chosen in line with the related literature, as they capture firm operating performance and underlying characteristics, as well as based on the findings presented in Table 3 in Section 3.2.

The null and alternative hypotheses can be summarized as follows:

H0: β1 >0 – controlling for other variables, financial expertise among CEOs and CFOs enhances both stock performance and operational performance of a given firm.

Under the null hypothesis, it is expected that financial expertise of key executives within a company improves its performance because they are able to make superior decisions with regards to the long-term performance of the company. Having previous experience and/or education in the field of finance arguably removes some biases in decision-making, such as the WACC fallacy. These firms are

(25)

25

able to raise external funds more efficiently and better manage their cash flows. Additionally, previous research has shown that financial experts on boards lead to less restatements in the reporting and also induce investor confidence.

H1: β1 =0 – controlling for other variables, financial expertise among CEOs and CFOs does not affect stock performance and/or operational performance of a given firm.

There are two possible scenarios in the case that the null hypothesis is rejected by the regression analysis. Firstly, financial expertise of CEOs and CFOs could be irrelevant and not impact firm performance at all. This would be evidence against the agency model of firm management and support the neoclassical model, which assumes that no single executive exerts sufficient power over the company and that the whole board controls the company unanimously. Additionally, it could also be the case that financial expertise as a skill is not as valuable as other skills, for example managerial talent or industry expertise and professional network. There is also a third reason for getting statistically insignificant results, that being the fact that there are two opposing forces at play, financial expertise among key executives both enhances and deteriorates firm performance and these effects cancel each other out.

H2: β1 <0 – controlling for other variables, financial expertise among CEOs and CFOs is detrimental to stock performance and/or operational performance of a given firm.

Secondly, another explanation for the rejection of the null hypothesis is that financial expertise among the key executives worsens firm performance. Potential justifications of this finding include the fact that financially savvy executives are more prone to risk taking than their generalist counterparts. They may take on excessive risks, especially in the time of the crisis, which adversely affect firm performance. Additionally, previous research in behavioural finance has also shown that financial experts are subject to overconfidence and overoptimism. Such executives consistently overestimate their abilities, rely too much on their own judgements and also miscalibrate the probability distribution of the range of potential outcomes, thereby making suboptimal decisions, which may adversely affect firm performance.

Despite the use of advanced OLS regression methods and controlling for a variety of characteristics, the issue of endogeneity must be addressed. Bias to the findings may arise from two distinct avenues. First of all, financial expertise could be correlated with other director’s characteristics, such as inherent talent and education. This would constitute an omitted-variable bias and could be solved by adding the appropriate controls for CEO and CFO characteristics. Secondly, a bias may arise through the endogenous nature of firm-CEO/CFO matching process. This is based on the specific preferences of firms and CEOs and CFOs respectively. Boards of directors consciously choose a

(26)

26

candidate that best suits their objectives and key executives self-select themselves into firms that match their skills and preferences. Thus, an already well-performing and strongly governed firm may select a financial expert to keep the company on the right course; alternatively a financial expert may self-select himself/herself into a successful firm to uphold his/her reputation. The unobserved heterogeneity created by the matching process, particularly in the case of within-firm time-invariant characteristics, can be addressed by including firm fixed effects in the regression analysis, as was argued by Kaplan et al. (2012); however, arguably there still remains an element of endogeneity. For example, it could be the case that in the times of an economic downturn, more financial experts are hired overall. This would result in a negative effect of financial expertise on firm performance due to the bias, despite the fact that in the long run, financial expertise may be beneficial to the firm. This concern will partially be addressed by the use of an event study methodology, described in the next section.

4.2. Event Study

The second empirical approach to analysing the puzzle of the value of financial expertise uses a unique design, which was not implemented in any of the previous studies and therefore forms the main contribution of this thesis. The nature of the dataset used allows to conduct an event study based around the replacement of CEOs within firms. Two key scenarios are distinguished:

I. A non-financial expert CEO is replaced by a financial expert; II. A financial expert CEO is replaced by a non-financial expert.

There are two other possible combinations of financial expertise replacement; however in such instances the sample size decreases dramatically and the inference on such results is limited. Additionally, this analysis will only focus on the CEO replacements as the sample size for CFO replacements is not sufficiently large to draw any relevant conclusions. The event date is defined as the year in which such switch was conducted and the event window is determined as [-5, +5] years around the date of replacement. This window was chosen based on the availability of the data, as well as for practical reasons. Subsequently, the evolution of the main variables of interest will be recorded and analysed. Gross stock returns and volatility will be used as proxies for firm market performance and cash flows, market leverage, cash holdings, investment and return on assets will capture the operating performance of given firms. The core of the analysis will compare the direction and magnitude of the effects of appointing a financial expert CEO, as opposed to a non-financial expert CEO.

(27)

27

This methodology, albeit not endogeneity-free, offers valuable insights into the dynamic relationship between CEO replacements and firm performance and can provide further evidence to the findings of the regression analysis. It partially addresses the concern of self-selection and a potential higher probability of financial expert appointments during a crisis. The null and alternative hypotheses are similar to those outlined in Section 4.1, with the core expectation being that an appointment of a financial expert will lead to positive outcomes, and that such outcomes are greater in magnitude than the change in performance related to an appointment of a non-financial expert CEO.

Although the main source of endogeneity in the form of CEO-firm matching is still present and the results may be biased by the fact that CEOs and firms self-select, the findings of the analysis are still valuable. Even though this approach may not determine a causal relationship, a correlation between CEO financial expertise and firm performance could serve as a credible signal of the true value and commitment of the company to its long-term performance and provide novel insights to the investors, who suffer from the asymmetric information problem, not being able to truly assess the value of the firm. The credibility of such signal lies in the fact that, as previous studies pointed out, the compensation of independent executives possessing generalist skills is higher than the remuneration of insiders and specialists in a given field, in terms of industries. Therefore, a signal in the form of appointing a financial expert CEO by a company that assesses its true internal value as higher than market beliefs is credible because it is costly, determined by the pay differential of the two types of executives.

5. Results

5.1. Regression Analysis

The first section of the empirical analysis reports OLS regressions of CEO and CFO financial expertise on stock performance in terms of gross return and price volatility, gradually adding control variables in terms of firm and executives’ characteristics, as well as suitable interaction terms.

Table 5 jointly presents the regression results in the case of CEOs on the left side and of CFOs on the right side. As we can see, the variable of interest (Financial expert) is only significant in the most basic regression specification when other variables are not controlled for and when time and entity fixed effects are omitted. Such results are still interesting however because of the very nature of fixed effects methodology. When the difference in size of the two subgroups of the main independent variable (financial expertise) is big in magnitude and lacks the desirable variation, as was shown in Table 1 in Section 3.2, fixed effects regressions result in insignificant coefficients of

The value of CEO and CFO financial expertise