Amsterdam Business School
The effect of a company’s joint CEO and CFO tenure on the quality of Financial Reporting
Final Version Name: Erik Verduin Student number: 5613310 Date: 10 December 2015
First supervisor: Dr. S.W. Bissessur Second supervisor: Dr. Ir. S.P. van Triest
MSc Accountancy & Control, specialization Accountancy Faculty of Economics and Business, University of Amsterdam
Abstract
This master thesis examines the influence of joint CEO and CFO tenure on the quality of financial reporting. Prior literature showed that the quality of both CEO and CFO is increasing during their tenure. To measure financial reporting quality, in this thesis SOx section 404 disclosures are used. In these disclosures, recorded in the annual accounts of the company, the executives of a firm have to give their opinion about the effectiveness of the internal control system. Material weaknesses have to be disclosed. Prior literature (Nagy, 2010) showed that these SOx section 404 disclosures are a solid proxy for financial reporting quality.
As CEOs and CFOs have to assess the effectiveness of the internal control system together, this assessment is one of joint responsibility. Therefore this research focusses on the influence of the length of their co-operation.
The outcome of the research shows that there is a positive relationship between the joint CEO and CFO tenure and financial reporting quality. This provides users of financial reports with guidance in determining the informativeness and quality of firms’ financial reports.
Acknowledgement
The final step to obtain the Master of Science degree in Accountancy at the University of Amsterdam was writing this Master Thesis. The whole process of writing this thesis has been a bumpy ride. Therefore I am very grateful for the support I experienced continuously.
First I want to thank my supervisor, dr. Sanjay Bissessur, for his patience and useful feedback. Next I want to thank, in alphabetical order, Dave, Esli, Leon, Simone and Wouter for editing. Finally I want to thank my employer, PwC, for giving me time and space to finish this thesis.
Table of contents
1 Introduction ... 6
1.1 Introduction of the subject ... 6
1.2 Problem definition ... 7
1.3 Research method ... 7
1.4 Contribution ... 7
1.5 Structure of the research ... 9
2 Literature review ... 10
2.1 Foundation theory: Principal-agent or agency theory ... 10
2.2 Sarbanes-Oxley Act and Material Weakness in Internal Control ... 11
2.3 Financial Reporting Quality ... 12
2.4 Tenure theory ... 14
2.4.1 CEO Tenure ... 15
2.4.2 CFO Tenure ... 16
2.5 Executive tenure and financial reporting quality ... 17
2.6 Joint responsibility of CEO and CFO on Financial Reporting Quality ... 18
2.7 Literature summary ... 19
3 Hypothesis development ... 20
4 Research Design ... 21
4.1 Sample selection method ... 21
4.2 Variables ... 22
4.2.1 Dependent variable: Financial Reporting Quality ... 22
4.2.2 Independent variable: Combined executive tenure ... 22
4.2.3 Control variables ... 22
4.2.4 Alternative dependent variable ... 24
4.3 Research Model ... 26
5 Results ... 27
5.1 Descriptive statistics ... 27
5.2 Results of hypothesis testing ... 31
6 Conclusion and limitations ... 34
6.1 Conclusion ... 34
6.2 Limitations ... 35
6.3 Suggestions for further research ... 35
1 Introduction
1.1 Introduction of the subject
In this thesis the purpose is to determine the relationship between the joint tenure of the Chief Executive Officer (hereafter: ‘CEO’) and Chief Financial Officer (hereafter: ‘CFO’), and the financial reporting quality of an organization. The tenure of the CEO and CFO is described as the number of years they are holding their position as executives together in the organization.
My motivation for the subject of this thesis comes from my personal experience. As observed during my years working as an auditor, it takes several years for a new company employee to become knowledgeable enough about the company to give valuable feedback and answers on an auditor’s questionnaire. It is an empirical question if the experience with employees working on a lower level in the organization, holds for companies executives as well. I will perform a quantitative research, based on available data from United States companies during the period 2004 to 2014. Ali & Zhang (2015) show in their paper that the quality of an executive has a positive correlation with tenure. The question is then if this quality, measured by Ali and Zhang in a broader perspective and focused on earning management, is also visible for financial reporting quality measured by SOx section 404 disclosures.
The motivation for investigation of the financial reporting quality comes from the fact that the users of the financial reports use the financial reports to monitor the performance of the company. This performance is an outcome of the performance of the executives. To monitor this, it is important that the users of the financial reports can rely on the financial reports (Bushman & Smith, 2001).
The quality of executives has been often the subject of research, which proves that society and science consider it as a subject of importance (Baik et al., 2011; Ge et al., 2011). This research is usually focusing on one or more characteristics of an executive after which they measure the impact of these characteristics on one or more proxies for quality. This has been done for decades wherein, due to changing perceptions of quality and favorable characteristics, both the characteristics used as well as the proxy’s for quality are constantly changing. However, more and more research is measuring executive’s quality as a proxy related to good behavior and reliability. Baik et al. (2011) investigated the ability of a company’s CEO and the accuracy of his forecasts. Their research indicates that the higher the ability of a CEO is, the more accurate their forecasts are. In other words, the higher the quality of a CEO, the higher the quality of the outcome of his work. When people base their decisions on financial reporting quality, higher quality will lead to better decisions. In additional testing they found that the industry specific experience as an executive, also known as the tenure, has the same outcome as ability has on the accuracy of forecasting. My research will extent to this by investigating the influence of tenure on financial reporting quality instead of forecast accuracy.
Contrary to prior research, which focused on the quality of a CEO and the relationship with the outcome of his work, Ge et al. (2011) investigated whether the CFOs’ style has impact on his or her accounting decisions. They conclude that the firms’ optimal financial reporting strategy could be negatively impacted by the individual CFOs’ style, unless the style is in line with the firms preferences. Hence, the style of the CFO is an important predictor of the company’s financial reporting. In these research, they’ve used different variables to measure style and characteristics. Their research only investigates the extent to which the CFOs’ style influences the financial reporting strategy and not per se the quality of this financial reporting. Does the style or characteristic of a CFO have influence on the quality of the financial reporting? That is where this thesis aims to contribute, by investigating this as part of the combined tenure with the CEO.
1.2 Problem definition
The objective of my thesis is to examine the influence of the combined CEO and CFO tenure on the financial reporting quality. Combined tenure reflects the number of years the CEO and CFO are working together in their current positions. To translate this objective into a research question, I have formulated the following question:
Does a relationship exist between the combined tenure of the company’s CEO and CFO, and the quality of financial reporting?
1.3 Research method
In this thesis data from public firms from the United States is used, collected from the Warton Research Data Service (WRDS) to create a sample of information about financial reporting quality and combined CEO and CFO tenure. Within the WRDS database, I use COMPUSTAT, AuditAnalytics and ExecuComp to collect the data. The sample consist of data of United States firms. The total sample comprise 11.116 firm years in the period 2005 until 2014.
The data will be used to find an answer to abovementioned research question. As proxy for financial reporting quality I use SOx section 404 disclosures, which are containing the CEO’s and CFO’s opinion about the effectiveness of the internal control system. This disclosures are collected from the AuditAnalytics database. Nagy (2010) showed in his research that these disclosures can be used as proxy for financial reporting quality.
A logistic regression model will be used to perform the empirical research.
Summarizing, this thesis investigates the influence of combined CEO and CFO tenure on financial reporting quality, and SOx section 404 disclosures are used as proxy for financial reporting quality.
1.4 Contribution
that it is measured according to the, since November 2004 mandatory, SOx section 404 disclosure upon discovered material weaknesses in the internal control system of the firm.
The relationship between the tenure of executives and financial reporting quality is an underexposed subject in the current scientific literature. Due to large accounting scandals in the past 20 years, the relationship between tenure and quality has been the subject of many researches when it comes to the tenure of auditors. The tenure of auditors in relation to financial reporting quality has been investigated several times (Johnson et al., 2002; Nagy, 2005) and even more articles are written about the relationship between auditor tenure and audit quality (a.o. Ghosh and Moon, 2005; Knechel and Vanstraelen, 2007). These researches all conclude that there is a relationship between auditor tenure and these quality measures.
Furthermore, this thesis contributes by extending the research mentioned earlier of Baik et al. (2011) and Ge et al. (2011). Where Baik et al. (2011) highlights the relationship between CEOs’ ability and forecast accuracy, this thesis will contribute by investigating this for the quality of financial reporting. And where Ge et al. (2011) have investigated whether CFOs’ style or characteristics influence financial reporting, this thesis will contribute to existing literature by examining the impact of one characteristic, joint CEO and CFO tenure, on financial reporting quality.
Empirical evidence concerning the relation between executive tenure and financial reporting quality is scarce. The only known research which examines the relationship between CFO experience, which not exclusively includes its tenure, and SOx section 404 disclosures is performed by Li, Sun & Ettredge (2010). They have focused primarily on CFO experience, while I use the combined CEO and CFO tenure for my research. Also, Ali and Zhang (2015) performed a research on the relationship between CEO tenure and financial reporting quality, where discretionary accruals are used as a proxy for financial reporting quality. They both see CEO and CFO as separate operating bodies, where my research investigates the combined tenure of the CEO and CFO. I use combined tenure because the CEO and CFO have to give their opinion about the effectiveness of internal control together and prior literature (Tulimieri and Banai, 2010) show that the role of the CEO and CFO is merging.
I will also use another proxy for financial reporting quality than Ali and Zhang (2015) did, although I am using discretionary accruals as an alternative proxy in order to verify the robustness of my empirical research. Therefore you can say that my research loosely extends the research of Li, Sun and Ettredge (2010) and Ali and Zhang (2015).
The final contribution of my research is that it helps stakeholders to enhance their decision making. For instance, when an auditor performs the risk assessment before and during the audit process, they always search for both internal and external information and evidence to support their conclusions. This research provides auditors with evidence of the influence of combined CEO and CFO tenure on the quality of the financial reporting. As this quality is measured by using internal control deficiencies, the evidence also tells something about the internal control of firms in
relation to the combined tenure of the executives. Auditors can use this information to enhance their risk assessment and therefore to establish a better and more specific audit approach.
Aforementioned proves the contribution of this thesis to science as well as professionals, and makes the subject interesting for research.
1.5 Structure of the research
This thesis will be structured in a logical order to guide the reader through the research. Therefore the next section of the thesis comprises the literature review. In this second section of the thesis the literature regarding the foundation theories and financial reporting quality, CEO tenure, CFO tenure and their joint responsibility will be discussed. The information set out in the second section will be used in the third section to develop the hypothesis. Afterward, in the fourth section, the empirical research will be discussed. This contains the research method, data and sample selection and the description of the metrics and (control) variables. In the following fifth section the results will be presented and discussed, consisting of descriptive statistics, correlations and regressions. In the final sixth section the results will be analyzed which will lead to a conclusion on the research question. This final section will also contain the suggestions for further research and a brief elaboration of the limitations of the research in this thesis.
2 Literature review
In this second section the theoretical framework on which this research is founded is presented. This theoretical framework starts with the foundation theory on which the research is build. A brief description regarding the basic elements of the principal-agency theory will be provided to enhance the understanding of the relationship induced.
To enhance the understanding and scope of the critical view, a quick description of the history of the Sarbanes-Oxley Act will be given first, which will be followed with the description of material weakness in internal control. Afterwards, I will continue with a critical view on the prior researches on financial reporting quality, CEO and CFO tenure, the joint responsibility of both executives in the organization and the link between their tenure and financial reporting quality of the company.
2.1 Foundation theory: Principal-agent or agency theory
Public companies are usually managed by other people than the owners of the company, which are in most cases several shareholders with small individual influence. This has already been described by Berle and Means in 1932. Jensen and Meckling (1976) describe the agency relationship as a contract where the principals (i.e. the shareholders) engage agents (i.e. Board of Directors, executives) to perform services on their behalf, in most cases performing the daily management of the company. This means that principals delegate certain decision-making authorities to the agents. Because ownership and decision making is delegated to different parties, it creates the problem of non-aligning targets and interests. The targets and interest of the principal are not always the same as those of the agent. This is called the principal-agent theory (Eisenhardt, 1989). Eisenhardt explains that due to information asymmetry between the principal and the agent, the principal cannot fully observe the behavior of the agent. Next to this information asymmetry, Berle and Means (1932) explain that because the ownership of listed companies is fragmented, the owners don’t have the power to pressure upon the agents. Their influence is therefore limited. As financial reporting is one of the most important ways for principals to observe and measure the agent’s performance, the quality of the financial reporting is of significant importance. This provides us with the link between the performance and quality of the CEO and CFO on one side, and the quality of financial reporting on the other.
Besides monitoring through financial reporting quality, the principal can influence the agent’s performance and decisions by setting up a compensation plan that aligns the objectives of the company with the goals of the agent. This stimulates desirable behavior of the management of the firm and addresses the principal agent conflict (Bebchuk and Fried, 2003). However, this compensation plan has to be well structured, as managers or executives with a variable compensation plan have the incentive to maximize their pay for the year, e.g. by using earnings management (Ali and Zhang, 2015). This will be discussed later on.
This thesis examines the relation between the combined tenure of these executives, which will be discussed later on, and the financial reporting quality.
2.2 Sarbanes-Oxley Act and Material Weakness in Internal Control
After years of turbulence, due to accounting scandals such as Enron and WorldCom, two United States senators, Paul Sarbanes (Democrat) and Michael Oxley (Republican), combined their pre-existing plans to enhance the corporate governance of United States companies into the Sarbanes-Oxley Act. Both senators were separately trying to enhance the regulations for corporate governance, but to other senators at that time this deemed unnecessary. This changed however, after among other cases the previously mentioned scandals.
In the summer of 2002 the Sarbanes-Oxley Act (hereafter: ‘SOx’) was introduced in the United States. This required the executives, in most cases the CEO and the CFO, of a SEC registrant to disclose the effectiveness of the internal control structure (SEC, 2003). SOx consist of multiple sections. Section 302 of SOx obligates the executives to evaluate the effectiveness of the internal control structure periodically. When a material weakness in the internal control is identified, it is mandatory to disclose (SEC, 2002). Section 404 of SOx, with the mandatory disclosure of material weaknesses since November 2004, goes a little deeper into the evaluation of the effectiveness of the internal control structure. In the annual report, the executives have to include an assessment of the effectiveness of the internal control structure. Not only the executives have to give their opinion about this effectiveness, also the firm’s public accountant has to attest this assessment (Doyle et al., 2007).
In this thesis, I’ve focused on SOx section 404 disclosures as determinant of financial reporting quality. Reason behind this is that SOx section 404 has stricter documentation and more detailed disclosure requirements than SOx section 302 (Ge, Koester and McVay, 2014). This makes SOx section 404 disclosures more suitable to use as proxy for financial reporting quality.
While looking for examples of these misstatements and in order to clarify them, Ge and McVay found that there is a wide array of possible material weaknesses. However, the majority of the material weaknesses are caused by what they call ‘inadequate accounting resources’. An example of these inadequate accounting resources is the lack of qualified accounting personnel (Ge and McVay, 2005).
To clarify what is exactly meant with a material weakness in the internal control structure, the PCAOB described this weakness as follows in Auditing Standard No.5:
“A material weakness is a deficiency, or a combination of deficiencies, in internal control over financial reporting, such that there is a reasonable possibility that a material misstatement of the company's annual or interim financial statements will not be prevented or detected on a timely basis” (PCAOB, 2007a).
This suggests that the more effective the internal control structure is working, the lower the chance that company’s annual or interim financial statements contain material misstatements. This increases the chance of misstatements being prevented or detected on a timely basis. This assumption is endorsed by Costello (2011). He states that internal control reports (or disclosures) provide a strong assessment of the reliability, and thereby the quality, of the company’s financial reporting (Costello, 2011). This is consistent with the conclusion of Nagy’s (2010) research on the link between internal control deficiencies and financial reporting quality. He found that there is a negative correlation between SOX section 404 compliance and the likelihood of issuing material misstated financial statements. According to him, this suggests SOX section 404 is meeting the proposed objective of improving the quality of financial reporting.
Prior research also provide other determinants of financial reporting quality that are related to internal control effectiveness. Besides of containing less material misstatements, companies who comply to SOX section 404 also show higher accrual quality (Doyle et al., 2007 and Ashbaugh-Skaife et al., 2008) and higher earnings quality (Bedard, 2006) and more conservatism in their reporting (Goh and Li, 2011) .
2.3 Financial Reporting Quality
For this research, disclosing material weaknesses in internal control (hereafter: ‘MWIC’) is used as proxy for financial reporting quality.
Dechow et al (2010) performed an extensive literature study on earnings quality, one of the proxies for financial reporting quality. They describe earnings quality as follows:
Higher quality earnings provide more information about the features of a firm’s financial performance that are relevant to a specific decision made by a specific decision-maker.
As Bedard (2006) already showed, higher earnings quality is a proxy for higher reporting quality. This means that the proxies for higher earnings quality also say something about the financial reporting quality of a company. There are three main groups of proxies according to the research of Dechow et al (2010):
- Properties of earnings
- Investor responsiveness to earnings
- External indicators of earnings misstatements
Properties of earnings include several secondary proxies. In the research, earnings persistence, abnormal accruals derived from modeling accrual process, earnings smoothness, asymmetric timeliness and timely loss recognition and target beating are included in the properties of earnings. Earnings persistence is considered as a proxy for earnings quality because persistent earnings are generally more ‘sustainable’ then non-persistent earnings. Sloan (1996) proved this in his research by showing that persistent earnings have a higher cash flow component than non-persistent earnings which generally have a greater accrual component.
Cash flows are considered to be of a higher quality because these earnings are factually converted into cash, while accrual based earnings can be influenced by subjective management decisions which cause them to be misstated.
Abnormal accrual derived from modeling accrual processes are closely related to the abovementioned earnings persistence. Dechow et al. (2010) describe (extreme) accruals as low quality because they represent a component of earnings that has a lower persistence. The residual from accrual models is a proxy for earnings quality because they represent management discretion or the estimated error in their accruals. As both discretion as well as errors in the accruals are reducing the decision usefulness of the users of the financial reporting, their presence can be seen as an indication and proxy for lower earnings quality.
Dechow et al. (2010) describe earnings smoothness as a proxy for earnings quality. They first explain that it is hard to determine whether smoothness of earnings is caused by accounting rules or by intentional earnings manipulation. Earnings smoothness regarding accounting rules is caused by matching the earnings with their respective period. Intentional earnings manipulation is for instance used to show more persistence in earnings by spreading them over a longer period and to mitigate cash flow fluctuations. Despite this, Dechow et al. (2010) conclude that earnings smoothness can be used as proxy for earnings quality.
When the earnings smoothness is a result of managerial intervention to smooth the fluctuations in cash flows, it will lead to an adverse effect on earnings quality by making earnings and losses less timely and therefore less informative.
The timely recognition of losses has a positive impact on earnings quality because it opposes the natural optimism of managers. The disadvantage of this proxy is, according to Dechow et al. (2010), that the timely recognition of losses are leading to a lower earnings persistence in periods with relatively more bad news than good news because bad news, e.g. losses, are recognized earlier then good news.
The last proxy discussed by Dechow et al. within the properties of earnings is target beating. Target beating is a phenomenon which describes reported earnings around zero. Relatively a lot of companies are reporting a small profit compared to a small loss. Therefore a small profit can be an indication for earnings management, which is considered of having a negative impact on earnings quality.
When looking at investor responsiveness to earnings, Dechow et al. (2010) state that the earning response coefficient (hereafter: ‘ERC’) is a direct proxy for earnings informativeness and earnings quality. This coefficient is mainly measured using unexpected earnings (‘earnings surprise’) and the impact on returns. Dechow et al. (2010) emphasize that this proxy mainly indicates the level of informativeness of earnings and to a lesser extent about the quality. For instance, things as earnings management can lead to higher unexpected earnings and therefore to a higher ERC. This could lead to the conclusion that the earnings quality is higher, however, at the same time the
The final group of proxies Dechow et al. (2010) elaborate on are the external indicators of earnings misstatements. They provide us with three types of external indicators of earnings misstatements, namely (i) SEC Accounting and Auditing Enforcement Releases (AAERs), (ii) restatements and (iii) internal control procedure deficiencies which has to be disclosed under the Sarbanes-Oxley Act. These three indicators are all used in the prior literature investigated by Dechow et al. (2010) for measuring earnings misstatements and used as proxy for earnings quality. The advantage of using these indicators as proxy is that an external source, for instance the auditor or the SEC, has identified a problem with the quality. This makes it a relatively easy proxy to use, as misstatements are already provided by these external sources. With the proxies mentioned under properties of earnings and investor responsiveness to earnings, the researcher first has to derive the misstatement or undesirable actions from the data.
Dechow et al. (2010) further describe internal control procedure deficiencies that have to be disclosed under the Sarbanes-Oxley Act. They elaborate on both SOx section 302 as SOx section 404 disclosures and the positive association between internal control quality, where an adverse opinion means less internal control quality, and various earnings quality measures such as persistence and discretionary accruals.
Internal control effectiveness, which is assumed to be associated with a higher financial reporting quality, has been the subject of more prior researches. Goh and Li (2011) performed an empirical research on the relationship between internal control effectiveness and conditional conservatism. Conditional conservatism is, according to their research, also described as timely loss recognition. This timely loss recognition is one of the proxies for earnings quality as provided by Dechow et al. (2010) which has been touched upon earlier in this paragraph.
Goh and Li (2011) found empirical evidence that firms which are having a higher degree of conditional conservatism have less material weaknesses in their internal controls. They also found that firms that are disclosing a material weakness in a certain year but manage to remediate their weakness in the next firms year, have a higher degree of conditional conservatism than firms that failed to repair the weakness that was disclosed in accordance with SOx section 404.
Summarizing, multiple researches on financial reporting quality have been performed. Various proxies for financial reporting quality are used, such as earnings quality, earnings persistence, accrual quality and conditional conservatism. There is also empirical evidence from different researches that SOx section 404 disclosures are associated with financial reporting quality. This is in line with the purpose of implementing the Sarbanes-Oxley Act as one of the purposes was to enhance the quality of financial reporting.
2.4 Tenure theory
As the purpose of this thesis is to investigate the influence of CEO and CFO tenure on the financial reporting quality, below will follow an overview of the existing literature regarding CEO and CFO tenure.
First the CEO tenure will be discussed, followed by the CFO tenure, finishing with the literature about the correlation between their tenure on financial reporting quality. So far, there has been a lack of literature focusing on the direct relation between CEO and CFO tenure and material weakness in the internal control structure. Therefore, the literature that is available for the other measures of financial reporting quality will be discussed as well, such as earnings quality and conservatism and their relationship with executive tenure.
2.4.1 CEO Tenure
Quality of work delivered depends for a significant part on the quality of the person performing the work. Therefore the question arises, how can you determine the quality of the CEO? Ali and Zhang (2015) document in their research that the markets view on the quality of a CEO is related to its tenure. They found that the markets perception is that a CEO with a short stay in its position is less talented than a CEO with a longer stay.
As stated by Hambrick and Fukutomi (1991), there are five seasons in the CEOs performance. These five seasons are:
1. Response to mandate
In this season a CEO is newly appointed to the job. He shows relatively high commitment to the job but he has relatively low task knowledge, he uses different information sources, has low power and a high task interest. In general you can say that the CEO in this stage of his tenure is working hard to master the job and keeps loyal to the paradigms he had when accepting the job.
2. Experimentation
Following the response to the mandate season, a CEO may continue to the experimentation season. Not all CEOs will make this step, as some CEOs functioning is good enough by just doing the job as mastered during the response to mandate season. In this season, usually after two or three years, the CEO deviates from the original tasks and attempts new approaches and broader gauged methods. This can only been done when a CEO has had some successes in the first season and gained some credibility.
3. Selection of an Enduring Theme
In this stage the CEO reflects on the things he has tried in the first two seasons and contemplates the way he want to lead and improve the company on the long-term. The CEO evaluates the initial paradigms he and further crystalize these with the experience he gained from the experimentation phase.
4. Convergence
In this fourth season the CEO has great knowledge of the organization and the power to change i.e. processes in the organization. However, Hambrick and Fukutomi see a turning point in this so called season. Although his commitments to his paradigms is strong, his task knowledge reached its
plateau, the information flow is getting narrower, more filtered and his task interest is starting to wane.
5. Dysfunction
Even though the power of the CEO is at peak level, he is starting to function worse. This is, in accordance to Hambrick and Fukutomi, caused by psychological causes such as boredom. The quality of the work doesn’t necessarily has to deteriorate, but the CEO is less likely to come up with new initiatives and his response time on things that occur will be slower.
It seems that in the convergence phase the CEO understands the organization well enough to decide where improvements can be made, i.e. in the internal control structure. Therefore, during his tenure, this should be the phase were the CEO reaches his optimal point in providing financial reporting quality. Based on the Hambrick and Fukutomi research, the graphic line of the expected financial reporting quality provided by a CEO doesn’t necessarily has to show a downwards slope in the fifth and final season of his tenure, but due to the slower response time on both internal as external developments this can have impact on the internal control framework and the timeliness of detecting misstatements or preventing them.
Finally, Dikolli et al. (2014) argue that the chance of being replaced as a CEO is changing during his tenure. Unexperienced CEO’s, with a tenure below the median, are replaced when the firm performs poorly more often than experienced CEO’s, with a tenure above the median. They call this the performance-turnover relation. Hence, the longer the tenure of a CEO, the signal he send’s to the firms’ owners is that his abilities are higher. This is called the signaling theory where executives send signals, to e.g. firm’s owners or the market with their actions and characteristics. Abovementioned shows that the interpretation of this signals changes along tenure develops.
2.4.2 CFO Tenure
This section will discuss the CFO tenure. The literature implies that there are there reasons to believe CFO tenure is different from CEO tenure. Although Mian (2001) suggests that there is a relation between CEO replacement and CFO replacement, the reasons for the leave of a CFO might not be the same. Mians research shows that after the departure of the CEO, the chance that a CFO departs is significantly higher. This is causes by two main reasons. First, the CEO and CFO are seen as a team and therefore their performance is considered to be a team effort. When this performance is below expectation from e.g. the market, they are both held responsible for this performance. Second, the CFO can make a promotion to the ultimate executive function, CEO. This function becomes available when a CEO is leaving. However, this brings forth a limitation for my thesis. The performances that executives are accounted for, are mostly financial performance and not directly related to the financial reporting quality.
Nevertheless, an adverse SOX section 404 opinion can lead to replacement of a CFO. Li, Sun & Ettredge (2010) show that companies with an adverse SOX section 404 opinion have less qualified CFOs. Therefore, their research provides us with the information that companies tend to have a
better chance of receiving a clean opinion after hiring a better qualified CFO. This suggest that the a CFO can have an positive impact on financial reporting quality already in the first years of their tenure.
One of the key indicators of CFO qualification is their experience (Li, Sun & Ettredge, 2010). They show that experience in holding position as CFO is negatively associated with the receipt of an adverse SOX section 404 opinion. It is important to keep in mind that the experience of the CFO does not necessarily has to be obtained within the company in their research.
Hoitash et al. (2012) find that overall board quality is associated with higher levels of internal control quality. Overall board quality is measured by different indicators such as board size, board independence, and board tenure. Here again, a longer tenure can be seen as positive for the quality of an executive, both CEO as CFO.
2.5 Executive tenure and financial reporting quality
In this thesis, SOX section 404 disclosures are used as proxy for financial reporting quality because the underlying idea is that a strong internal control structure detects and prevents material misstatements. Prior paragraphs on CEO and CFO tenure show us that prior literature expects the quality of an executive to increase during his tenure. However, one of the reasons for the termination of the tenure can be an adverse opinion on the effectiveness of the internal control structure by the external auditor.
The relationship between tenure of executives and financial reporting quality has been the subject of several researches. As earlier mentioned, this relationship has been proved present for earnings quality and accrual quality, which are also proxies for financial reporting quality according to Doyle et al. (2007), Ashbaugh-Skaife et al. (2008) and Bedard (2006). They showed us the link between SOX section 404 disclosures, earnings quality, accrual quality and financial reporting quality. To make an additional link with executives tenure, amongst others Zhang (2009) provide us with evidence. He suggests that there is correlation between CEO tenure and earnings quality. CEOs with a long tenure tend to report earnings less aggressively than CEOs with shorter tenure, which can be seen a positive relationship between tenure and earnings quality. This is in line with the earlier findings regarding tenure. The same can be said about tenure and accrual quality. Antia et al. (2010) conclude, supported by a research on accrual quality by Francis et al. in 2005, that there is a positive relation between accrual quality and CEO tenure, again in line with earlier findings. Antia et al. (2010) prove that there is a significant negative relationship between the CEO’s decision horizon and agency costs.
Ali and Zhang (2015) performed a research on the changes in CEOs’ incentives in using earnings management on their firms’ reported earnings in relation with the tenure of the CEO. As elaborated on earlier, earnings management considered to be negatively associated with earnings quality and reporting quality. Ali and Zhang (2015) use discretionary accruals as a proxy for earnings
paragraph 2.3, discretionary accruals are described as being negatively associated with earnings quality because it reduces the informativeness of the financial reporting and together with that, the quality of financial reporting. In their research, Ali and Zhang (2015) showed that CEO’s tend to overstate earning more in the earlier years of their tenure then in the later years of their tenure. In other words, CEO’s are making more use of earnings management in their earlier years serving as a CEO than in the following years. They found this using several proxies for earnings management, such as discretionary accruals and discretionary expenses. The outcome using these proxies was identical.
To summarize the information above, there seems to be a positive relationship between executives tenure and financial reporting quality. This will be discussed further on in with the hypothesis.
2.6 Joint responsibility of CEO and CFO on Financial Reporting Quality
According to the Sarbanes Oxley Act (2002), it is the responsibility of management to establish and maintain an adequate internal control structure and procedures for financial reporting. This shows that it is the joint responsibility of the management.
Tulimieri and Banai (2010) conclude in their article that the role of the CFO changed since the introduction of SOx. They mention a ‘metamorphosis unlike any other role in business history’. This is based on the fact that with the introduction of SOx, both CEO and CFO have to give their statements about the accuracy of financial statements and disclosures (SOx section 302) and about the effectiveness of the internal control over financial reporting (SOx section 404). According to Tilimieri and Banai this results in the CEO and CFO carrying the same level of responsibly. Not only did their roles change after introduction of SOx, their roles within the organization are even merging.
Mian (2001) agrees with the fact that CEO and CFO operate as a team. Even in the pre-SOx era their actions, including financial and operational results and reporting, are mainly a result from co-operation. This is based on the fact that he found an abnormally high CEO replacement in the period prior to CFO replacement, suggesting that these two persons are connected.
In his research, Vermeer (2005) investigates the relationship between CEO/CFO certification (SOx 302 and SOx 404) and the signal of credibility of the financial reporting. As they both sign the certification, he treats the CEO and CFO as one in his research.
Finally, Tulimieri and Banai (2010) expect the quality of what they call the ‘Chief Executive Partners’ (hereafter: CEP), containing the CEO and CFO, increases over time. Because upon completion of a project, they have to review results and revise where necessary. They state that this model of CEP will permit each of the partners to deploy their unique skills and ensure maximal involvement of the other.
Summarizing the information provided from prior literature, the executive management of an organization, for the purpose of SOx disclosures represented by the CEO and CFO, can be seen as one operating body.
2.7 Literature summary
The literature regarding the foundation theories is described in this section. I’ve discussed financial reporting quality, Sox section 404 disclosures, CEO and CFO tenure and other items such as accrual quality and earnings quality which will support the research in this thesis.
As a result of the literature analysis, the I’ve found the following results:
- SOx section 404 disclosures are a good proxy for financial reporting quality;
- CEOs and CFOs tend to become a higher quality along with the increase in tenure, this also applies for their joint tenure;
- Relationship between tenure, accrual quality, earnings quality and financial reporting quality has already been proven by using a wide variety of proxies;
- The internal control over financial reporting is the responsibility of both the CEO and the CFO, who are acting together as one.
3 Hypothesis development
In the second section of this thesis the foundation theory and the existing literature on Financial Reporting Quality, CEO tenure and CFO tenure have been elaborated on. Based on this prior literature, in this section the hypothesis will be developed.
From the principal-agent theory it appeared that executives will tend to maximize their own benefits rather than acting in the best interest of the company. To monitor this, the principals need financial reporting in order to decrease the information disadvantage they have in comparison to the executives. For these principals it’s important to know to which extend they can rely on the financial reporting that is provided by the executives for corporate control purposes (Bushman & Smith, 2001).
Ali & Zhang (2015) show that less accrual earnings management, one of the other determinants of financial reporting quality, is found when executives have a higher tenure. This is consistent with the other views on the relationship between the quality of an executive and his or her tenure which has been discussed in the literature review. Therefore the expectation is that along with tenure, the financial reporting quality is increasing.
Furthermore, there seems to be a joint responsibility of the two main executives, namely the CEO and the CFO (Tulimieri and Banai, 2010). On an annual basis, the CEO and CFO have to attest the effectiveness of the internal control system together in the annual reports. Prior literature shows that the quality and understanding of the firm of both the CEO and the CFO is increasing over time. I expect that the more experience they have together, the lesser the chance is that they have to disclose an adverse SOx section 404 opinion which states that they have found a material weakness in their internal control system as a result from their internal control assessment.
As a result from the information above, the hypothesis for the empirical research will be the following:
H1: The longer the joint CEO and CFO tenure, the lesser the chance that a company is having a material weakness in their internal control system, disclosed as an adverse SOx section 404 opinion.
4 Research Design
To investigate the relationship between the combined CEO and CFO tenure and financial reporting quality, this thesis use one empirical proxy for financial reporting quality and one alternative proxy in order to test the robustness of the research. This section will discuss the method of sample selection, the proxy used for financial reporting quality, the measurement of the combined CEO and CFO tenure and the control variables used.
4.1 Sample selection method
The data for this research will be obtained from available public sources. The timeframe of the data used will be 2005 to 2014. The starting point of January 2005 is the same as in the researches of Li, Sun & Ettredge (2010), Nagy (2010) and Eldridge et al. (2005). As the disclosure of SOx section 404 opinions is mandatory as from November 2004, 2005 is the first complete fiscal year since implementation.
The data will be collected through the Wharton Research Data Services (better known as ‘WRDS’), a database developed by the University of Pennsylvania (United States). In this database, two sub-databases are used to collect the data. The sub-databases used are Audit Analytics and COMPUSTAT. The data for SOx section 404 will be collected from the Audit Analytics database, where all SEC registrants are recorded in. This results in 118.459 unique opinions regarding the effectiveness of the internal control system for the period 2005 until 2014. The results contain both the conclusion that the internal control system is operating effectively as the conclusion that there are material weaknesses discovered during testing the effectiveness. However, there are also some lines with ‘ND’, which means that there is no recorded disclosure for the concerning company in that fiscal year. These items were removed from the final dataset later on.
For the tenure of the executives, both CEO as CFO, the data were derived from the Execucomp database, which is a separate section within the COMPUSTAT database. This section contains information about executive compensation, including their characteristics. I found a total of 18.663 unique records with information about the tenure of at least one of the company’s executives in a financial year. When deleted the lines in which or the CEO tenure or the CFO tenure was missing, 16.532 unique items were left to use in my research.
For the tenure of the CEO I could use the data of appointment and the date of their leave to calculate the tenure of the CEO in the specific financial year. For the CFO tenure I used the option ‘Annual CFO Flag’. This item indicates which person was serving as CFO for the specific company during a financial year. When a CEO or CFO was replaced during the year, I used the tenure of the last CEO or CFO (sometimes there were several replacements during one year) for my research as this person is more likely to file the SOx section 404 disclosure. This means that a CEO or CFO that is appointed during the year has the same influence in this research as a CEO or CFO that was
The final part of my dataset was collected from the COMPUSTAT database, where I used the section ‘North America’ under the ‘Compustat Monthly Updates’. From this database I was able to collect the data of 15.408 companies with 115.362 lines of information for the period 2004-2014. I used this extended period (starting in 2004 instead of 2005) in order to be able to determine lagged variables and calculate the increase or decrease of a certain variable in 2005, the first year in scope.
I combined the data of the three different data queries in one final dataset. The data from a firm year is only usable when complete. Therefore all different items needed to be present for a firm year to be recorded in the final dataset. After deleting the lines with missing items, a dataset of 12.786 was left to use for testing.
The last step before the final sample was established, was excluding industries which are regulated and banks and financial institutions. This is based on the research of Roychowdhury (2006). Therefore I excluded all firms in these industries based on their SIC codes. SIC codes between 4400 and 5000 (regulated industries) and between 6000 and 6500 (banks and financial institutions) were removed from the data. I want to examine the impact of the combined tenure on the financial reporting quality, measured by using SOx section 404 disclosures. As regulated industries have more prescribed obligations, even as banking and financial institutions, these are less comparable with the other companies.
After processing the abovementioned, the final sample consist of 11.116 firm years.
4.2 Variables
4.2.1 Dependent variable: Financial Reporting Quality
As set out in the literature review earlier on, I will use SOx section 404 disclosures as proxy for the quality of financial reporting. Since the implementation of SOx, companies are mandatory to disclose material deficiencies in their financial reporting system as from November 2004. Nagy (2010) shows in his research that SOx section 404 disclosures are a strong proxy for financial reporting quality. He also argues that this is in line with the objective of the Sarbanes-Oxley Act, which is increasing the reliability and quality of the financial reporting.
4.2.2 Independent variable: Combined executive tenure
The combined tenure is the independent variable used in this research. It will be calculated by taking the tenure of the CEO and the tenure of the CFO at the end of the financial year together. This means that if a CEO is currently working for 10 years as CEO at a firm but the CFO joined 4 years later, the combined tenure is 6 years in the current year.
4.2.3 Control variables
The control variables are derived from the research of Li et al (2010). This contains the following control variables:
CEO and CFO Combined Experience
To examine the impact of total combined experience, I use this as a control variable. The combined experience represents the experience of the CEO and the CFO at a specific company combined for the year. This means that is a CEO is working for 2 years and the CFO for 4 years in their current position, the combined experience will be 6 years.
Growth
Firms that are growing rapidly are likely to grow quicker than their internal control system can be adapted to. This increases the opportunity for material weaknesses in the internal control system (Doyle, Ge and McVay, 2005). Also Dechow et al. (2010) show that companies with a higher growth have a lower earnings persistence. This indicates that companies with a higher growth tend to have a lower earnings quality, which also is a proxy for financial reporting quality. Pan et al (2013) find in their research on the relationship between CEO tenure and investments cycles that CEOs disinvest more in the early stages of their tenure and postpone investments. Along with the lengthening of the CEOs’ tenure, they found that the disinvestment rate declines and opposite to that, the investment rate increases. Therefore tenure is expected to be negatively associated with firms’ growth.
Loss
Nagy (2010) shows in his paper that companies that are filing a material weakness according to SOx section 404 are generally less profitable then companies who don’t. This is supported by Dechow et al. (2010), they’ve found in prior research that firms that are performing poorly are more likely to use accounting tactics to increase their reported earnings. Hence, a poor performance, often leading to a loss, is expected to be negatively associated with financial reporting quality. Ali and Zhang (2015) use loss as a control variable as well in their research on the relationship between CEO tenure and earnings management, which also is a proxy for financial reporting quality.
Firm size
The last control variable I will use is the firm size, measured as the log total of assets. I will use this variable because on the one hand firms with a bigger size tend to report a material weakness in internal control more often (Nagy, 2010). On the other hand, bigger firms are considered to have better internal control systems because they have to control more assets and have more employees and resources to spend on internal control (Ge and McVay, 2005 and Doyle, Ge and McVay, 2007). Dechow et al. (2010) supports both assumptions based on their investigation of prior literature. They found that firm size would be negatively associated with earnings quality, another proxy for financial reporting quality, because for instance bigger companies make more use of income decreasing accounting methods because of greater political or regulatory scrutiny. They also found more recent researches which conclude the opposite and say that firm size is positively associated with earnings quality. This mainly because the earlier mentioned reason that bigger firms have
4.2.4 Alternative dependent variable
To test the robustness of my empirical research, I make use of an alternative dependent variable. To make it a valid robustness test, the alternative dependent variable should be another proxy for the same objective as tested in the original model. Therefore I use discretionary accruals as alternative proxy for financial reporting quality. As set out earlier in the literature review, Dechow et al. (2010) describe discretionary accruals as a proxy for earnings quality. Bedard (2006) showed in his research that earnings quality are a solid proxy for financial reporting quality.
To estimate discretionary accruals, there are several models. The first model is de Jones-model (Jones, 1991). This model is as follow:
𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝑇𝑇𝑇𝑇𝑇𝑇−1= 𝑎𝑎1 � 1 𝑇𝑇𝑇𝑇𝑇𝑇−1� + 𝑎𝑎2 � ∆𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑇𝑇 𝑇𝑇𝑇𝑇𝑇𝑇−1� + 𝑎𝑎3 � 𝑃𝑃𝑃𝑃𝑅𝑅𝑅𝑅𝑇𝑇 𝑇𝑇𝑇𝑇𝑇𝑇−1�+ 𝜖𝜖 Where:
𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 = The total accruals for the year 𝑇𝑇 𝑇𝑇𝑇𝑇𝑇𝑇 = The total assets for the year 𝑇𝑇
∆𝑅𝑅𝑅𝑅𝑅𝑅𝑇𝑇 = The total change in revenue compared to prior year for the year 𝑇𝑇 𝑃𝑃𝑃𝑃𝑅𝑅𝑇𝑇 = The total Property, Plant and Equipment for the year 𝑇𝑇
In this equation, the change in revenue and PPE are included to control for changes in non-discretionary accruals due to changing conditions. The change in revenue represents the economic performance of the firm before managers’ manipulations and PPE is used to control for the portion of total assets in relation to the non-discretionary depreciation expense. The depreciation expense is included in the calculation of total accruals. Refer to the calculation later in this section. All items are scaled by lagged total assets to reduce heteroscedasticity (Jones, 1991).
The second model is derived from this model but modified by Dechow et al. (1995) to increase the detecting power of earnings management. They added the change in receivables in this model because they assume that all changes in receivables in the event period are a result of earnings management (Dechow et al., 1995). This second model is called the Modified Jones Model.
The model is as follow:
𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝑇𝑇𝑇𝑇𝑇𝑇−1= 𝑎𝑎1 � 1 𝑇𝑇𝑇𝑇𝑇𝑇−1� + 𝑎𝑎2 � ∆𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑇𝑇− ∆𝑅𝑅𝑅𝑅𝑇𝑇𝑅𝑅𝑇𝑇 𝑇𝑇𝑇𝑇𝑇𝑇−1 � + 𝑎𝑎3 � 𝑃𝑃𝑃𝑃𝑅𝑅𝑅𝑅𝑇𝑇 𝑇𝑇𝑇𝑇𝑇𝑇−1�+ 𝜖𝜖 Where:
𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 = The total accruals for the year 𝑇𝑇 𝑇𝑇𝑇𝑇𝑇𝑇 = The total assets for the year 𝑇𝑇
∆𝑅𝑅𝑅𝑅𝑅𝑅𝑇𝑇 = The total change in revenue compared to prior year for the year 𝑇𝑇 ∆𝑅𝑅𝑅𝑅𝑇𝑇𝑇𝑇 = The total change in receivables compared to prior year for the year 𝑇𝑇 𝑃𝑃𝑃𝑃𝑅𝑅𝑇𝑇 = The total Property, Plant and Equipment for the year 𝑇𝑇
After this model, several researches added variables to optimize the estimation of discretionary accruals. McNichols (2002) added cash flows to the 1991 Jones, Kothari et al. (2005) added return on assets (ROA) to control for unusual performance and Dechow et al. (2012) added reversals of discretionary accruals. All this additions had the purpose to increase the explanatory power of the model and to enhance the estimation of discretionary accruals. However, Keung Shih (2014) explained that adding ROA causes the model to underestimate discretionary accruals and that the measurement error will be negatively correlated with the true discretionary accruals.
In recent literature, the Modified Jones model from Dechow et al. (1995) is still commonly used (Dechow, 2010; Badertscher et al., 2011). Therefore I will use the Modified Jones model to estimate discretionary accruals.
First total accruals have to be calculated for each firms’ year. Therefore I use the balance sheet approach of provided by Jones (1991).
𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 = ∆𝑇𝑇𝑇𝑇𝑇𝑇 − ∆𝑇𝑇𝑇𝑇𝐶𝐶𝐶𝐶𝑇𝑇 − ∆𝑇𝑇𝐶𝐶𝑇𝑇 + ∆𝐷𝐷𝑇𝑇𝐶𝐶𝑇𝑇 − 𝐷𝐷𝑅𝑅𝑃𝑃𝑇𝑇 Where:
𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 = The total accruals for the year 𝑇𝑇 ∆𝑇𝑇𝑇𝑇𝑇𝑇 = The change in current assets in year 𝑇𝑇 ∆𝑇𝑇𝑇𝑇𝐶𝐶𝐶𝐶𝑇𝑇 = The change in current assets in year 𝑇𝑇 ∆𝑇𝑇𝐶𝐶𝑇𝑇 = The change in current liabilities in year 𝑇𝑇
∆𝐷𝐷𝑇𝑇𝐶𝐶𝑇𝑇 = The change in debt included in current liabilities in year 𝑇𝑇 𝐷𝐷𝑅𝑅𝑃𝑃𝑇𝑇 = The depreciation and amortization amount in year 𝑇𝑇
4.3 Research Model
The research model to calculate the effect of combined CEO and CFO tenure on financial reporting quality is loosely derived from the paper of Li, Sun & Ettredge (2010). When testing their first hypothesis they only investigate the effect of CFO quality. One of the measures of this quality is the number of years the CFO is holding position as a CFO in 2004. I will extend this by replacing the number of years the CFO is holding position as CFO for the combined tenure of the CEO and CFO. I also will extend the period under research from just 2004 and 2005 in the Li, Sun & Ettredge research to the period 2005 until 2014 in this thesis. The model will investigate the relationship between the combined tenure of the CEO and CFO and the chance that a company is filing an adverse SOx section 404 opinion. According to Ge and McVay (2005) and to Li, Sun & Ettredge (2010), this thesis will use a logistic regression. Lammers (2007) explains in his book that when you are performing a research with a dependent variable with just two options, 1 or 0, you have to use a logistic regression.
The logistic regression model will be the following:
𝑇𝑇𝐷𝐷𝑅𝑅𝑅𝑅𝑅𝑅𝐶𝐶𝑅𝑅𝑇𝑇 = 𝑎𝑎 + 𝐵𝐵1 𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 + 𝐵𝐵3 𝐺𝐺𝑅𝑅𝑇𝑇𝐺𝐺𝑇𝑇𝐶𝐶𝑇𝑇 + 𝐵𝐵4 𝐶𝐶𝑇𝑇𝐶𝐶𝐶𝐶𝑇𝑇 + 𝐵𝐵5 𝐹𝐹𝐹𝐹𝑅𝑅𝑇𝑇𝐶𝐶𝐹𝐹𝐹𝐹𝑅𝑅𝑇𝑇 + 𝜖𝜖 The variables are defined as follows:
Variable Description of the variable
Dependent variable
ADVERSE Dummy variable for the disclosed SOx section
404 opinion for the current year, with a 1 for an adverse opinion and a 0 otherwise
Independent variable
TenureCOM The combined number of years of CEOs and
CFOs tenure in the current year
Control variables
GROWTH The growth of the company, measured in the
percentage asset growth over the year.
LOSS Dummy variable that equals 1 if a firm reports a
loss for the year and a 0 otherwise.
FIRMSIZE Reflects the size of the firm, measured as the
5 Results
In this section, the descriptive statistics and the empirical results of the analyses will be described. This will start with the descriptive statistics and follow with the outcomes of the analyses conducted. There I will discuss the implications of the outcome on the hypothesis.
5.1 Descriptive statistics
In this section the descriptive statistics will be described, together with the correlations. In section 4.1 a description about the sample selection method has been provided. There I also described how I came from the initial dataset to the final sample. A final sample of 11.116 firm years is used for the empirical research.
Tabel 1 provides the descriptive statistics for the SOx section 404 model. As a dummy is used for the dependent variable ADVERSE, the minimum value is 0 (no adverse SOx section 404 opinion) and the maximum value is 1 (adverse SOx section opinion). The mean is close to 0 due to the fact that from the 11.116 firm years, 419 of these years have an adverse opinion and 10.697 don’t have an adverse opinion.
The mean of combined tenure is 3,3 years with minimum of 0 and ad maximum of 22. This means that the average time a CEO and a CFO have together is relatively short. The tenure of the both executives is not solely included in the analysis. When I look at these tenures in the dataset that was used to calculate the combined tenure, the average tenure of a CEO is 7,0 years and the average tenure of a CFO 5,1 years. In his research, Zhang (2009) found an average tenure of 6,7 years for CEOs and in their research in 2011, Ge et al. (2011) found an average CFO tenure of 3.4 years. This could be caused because of using different sources or by looking at different timeframes.
For the control variables, the average growth of the companies in the sample is a little under 10 percent on average. The loss factor is 19,1%, which means that a little under 1 out of 5 firm years are containing a reported loss.
Tabel 2 provides the descriptive statistics for the alternative model with discretionary accruals as dependent variable.
Table 1
Descriptive statistics of the variables in the regression model Panel A: variables in the SOx section 404 opinion model
Variables Mean Minimum Maximum STD
ADVERSE 0,038 0 1 0,191 TenureCOMB 3,342 0 22 3,294 FIRMSIZE 7,396 1,272 12,764 1,604 GROWTH 0,098 -0,947 6,544 0,316 LOSS 0,191 0 1 0,393 Number of observations 11116
The sample period is 2005 to 2014. ADVERSE is 1 if a company receives an adverse initial SOx 404 opinion for the year, 0 otherwise. TenureCOMB is the combined number of years of CEOs and CFOs tenure in the current year. FIRMSIZE reflects the size of the firm, measured as the lagged total assets. GROWTH is the growth of the company, measured in the percentage asset growth over the year. LOSS is 1 if a company reported a loss for the year, 0 otherwise.
Table 2
Descriptive statistics of the variables in the regression model Panel B: variables in the discretionary accrual model
Variables Mean Minimum Maximum STD
DiscAcc ,000 -1,073 4,222 0,214 TenureCOMB 3,342 0 22 3,294 FIRMSIZE 7,396 -2,430 12,764 1,604 GROWTH ,098 -0,947 6,544 0,316 LOSS ,191 0 1 0,393 Number of observations 11.115
The sample period is 2005 to 2014. DiscAcc is discretionary accruals of a firm, estimated as the residual of the accruals model. TenureCOMB is the combined number of years of CEOs and CFOs tenure in the current year. FIRMSIZE reflects the size of the firm, measured as the lagged total assets. GROWTH is the growth of the company, measured in the percentage asset growth over the year. LOSS is 1 if a company reported a loss for the year, 0 otherwise.
The correlations coefficients in Table 3 and 4 indicate the degree of correlation between the variables uses in testing the hypothesis. In this matrix, the range of the coefficients for the Pearson Correlation can be between -1 and 1. When two variables are perfectly negatively correlated the coefficient will be -1 and when variables are perfect positively correlated the coefficient will be 1. An outcome of 0 will indicate that there is no correlation at al.
The correlations are important to test the model for multicollinearity between the variables used in the analysis. In common literature, a correlation of -0.700 and lower or 0.700 and higher affects the reliability of the model. Tabel 3 provide us with the information that none of the variables have a coefficient that exceeds that boundaries. The dependent variable, ADVERSE, is at a two-tailed 1% level significant negatively correlated with the independent variable, combined tenure
(TenureCOMB). This means that when tenure goes up, the chance of issuing an adverse SOX section 404 disclosure goes down.
Table 3
Correlations
Panel A: variables in the SOx section 404 opinion model
Variables ADVERSE TenureCOMB FIRMSIZE GROWTH LOSS
ADVERSE Pearson Correlation 1 Sig. (2-tailed) N 11116 TenureCOMB Pearson Correlation -,048 ** 1 Sig. (2-tailed) 0,000 N 11116 11116 FIRMSIZE Pearson Correlation -,080 ** ,021* 1 Sig. (2-tailed) 0,000 0,028 N 11116 11116 11116 GROWTH Pearson Correlation -0,015 0,013 ,033 ** 1 Sig. (2-tailed) 0,122 0,158 0,000 N 11116 11116 11116 11116 LOSS Pearson Correlation ,118 ** -,089** -,193** -,191** 1 Sig. (2-tailed) 0,000 0,000 0,000 0,000 N 11116 11116 11116 11116 11116
**. Correlation is significant at the 0.01 level (2-tailed). *. Correlation is significant at the 0.05 level (2-tailed).
Before running the final model, CEO tenure, CFO tenure and combined experience (as the sum of the experience of the CEO and of the CEO in a firm year), were included in the model. Due to a high correlation coefficient with combined tenure, these variables were excluded in order to assure the reliability of the model.
Table 4 shows the correlations for the alternative model with discretionary accruals as dependent variable. Also in this model, none of the variables exceed the boundaries that are mentioned earlier. The dependent variable in this model, DiscAcc, is not significant correlated with the independent variable, combined tenure (TenureCOMB).
Table 4
Correlations
Panel B: variables in the discretionary accrual model
DiscAcc TenureCOMB FIRMSIZE GROWTH LOSS
DiscAcc Pearson Correlation 1 Sig. (2-tailed) N 11115 TenureCOMB Pearson Correlation 0,012 1 Sig. (2-tailed) 0,221 N 11115 11116 FIRMSIZE Pearson Correlation ,075 ** ,021* 1 Sig. (2-tailed) 0,000 0,000 N 11115 11116 11116 GROWTH Pearson Correlation ,653 ** 0,013 ,033** 1 Sig. (2-tailed) 0 0,158 0,000 N 11115 11116 11116 11116 LOSS Pearson Correlation -,094 ** -,089** -,193** -,191** 1 Sig. (2-tailed) 0,000 0,000 0,000 0,000 N 11115 11116 11116 11116 11116
**. Correlation is significant at the 0.01 level (2-tailed). *. Correlation is significant at the 0.05 level (2-tailed).
5.2 Results of hypothesis testing
In this section, the results from the logic regression are presented and discussed. The outcome of the dependent variable in the model used is binary, which means that it is 1 when a firm disclosed an adverse SOx section 404 opinion and a 0 if otherwise. First, the Omnibus Test of Model coefficients will be discussed. This test shows the Chi² of the model, which provide information about the fit of the model.
Table 5
Omnibus Tests of Model Coefficients
Chi² df Sig. Step 182,128 4 ,000 Block 182,128 4 ,000 Model 182,128 4 ,000 -2 Log likelihood 3387,067 Nagelkerke R² ,059
The Chi²-test compares the likelihood of the estimated model with the likelihoods of the model with only one constant variable. The number of degrees of freedom (df) is 4, given the fact that 4 independent variables are used in the model. The Chi² of 182,128 is significant with 4 degrees of freedom. The significance values are p < ,001, which indicates the accuracy of the model improves when the explanatory variables are added. This implies that the model including the independent variables TenureCOMB, FIRMSIZE, GROWTH and LOSS predict the outcome better than a model without these variables. The Nagelkerke R² of the model is 5.9%. This suggest that the logistic regression model which is used explains 5.9% of the variation in outcome of the dependent variable. Table 6 provide the results from the logistic regression. This shows the outcome of the logistic regression model that is presented in section 4.3.
Table 6
Joint CEO and CFO tenure and SOx section 404 opinions Dependent variable = SOx section 404 opinoins
Variables Coefficient df Sig.
TenureCOMB -,070 1 ,000 FIRMSIZE -,194 1 ,000 GROWTH ,134 1 ,330 LOSS 1,025 1 ,000 Constant -1,967 1 ,000 Nagelkerke R² ,059 Number of observations 11.116
The sample period is 2005 to 2014. ADVERSE is 1 if a company receives an adverse initial SOx 404 opinion for the year, 0 otherwise. TenureCOMB is the combined number of years of CEOs and CFOs tenure in the current year. FIRMSIZE reflects the size of the firm, measured as the lagged total assets. GROWTH is the growth of the company, measured in the percentage asset growth over the year. LOSS is 1 if a company reported a loss for the year, 0 otherwise.
