Inside debt, CEO career horizon and financial
reporting quality
Name: Ashna Ramawadh Student number: 10363602
Thesis supervisor: Dhr. prof. dr. David Veenman Date: June 25, 2018
Word count: 13,480
MSc Accountancy & Control, specialization Control
Statement of Originality
This document is written by student Ashna Ramawadh 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.
Abstract
This study examines the effect of CEO inside debt holdings and CEO tenure on financial reporting quality. The theory predicts that inside debt holdings give CEOs incentives to take on less risky policies and choosing fewer risky investments. This would give CEOs less incentives to engage in more earnings management. Theory also predicts that as a CEO reaches retirement, they might engage in more earnings management. This is referred to the career horizon problem. In this thesis I test whether the career horizon problem might be mitigated with inside debt holdings. Consistent with my expectation, I find that inside debt holdings, when taking age into account, are negatively related to discretionary accruals, a proxy for financial reporting quality. Thus, inside debt holdings mitigate the career horizon problem and could therefore increase the financial reporting quality.
2 Literature review ... 7
2.1 Agency Theory, incentive compensation, earnings management and financial reporting quality ... 7
2.1.1 Equity incentives and earnings management ... 8
2.2 Inside debt ... 9
2.2.1 Inside debt and its effect on manager’s choices ... 10
2.2.2 Inside debt and Earnings Management ... 11
2.2.3 Inside debt and financial reporting quality ... 12
2.3 Career horizon problem ... 14
3 Hypothesis development ... 16
4 Research design ... 19
4.1 Measuring inside debt ... 19
4.2 Proxy for financial reporting quality ... 19
4.3 Control variables ... 21
4.4 The model ... 22
5 Data ... 23
5.1 Sample Selection ... 23
5.2 Constructing the variables ... 24
5.3 Descriptive Statistics ... 26
6 Findings ... 28
6.1 Results ... 28
6.2 Secondary regression ... 29
6.3 Discussion and implications ... 31
7 Conclusion ... 33
8 References ... 35
9 Appendix ... 38
9.1 Predictive validity framework ... 38
9.2 Summary of variables collected from Execucomp and Compustat ... 38
2
1 Introduction
Executive compensation schemes are widely discussed in the academic literature. In the accounting literature, the effect of these compensation schemes on accounting choices is largely researched. However, this is mostly done for equity compensation, disregarding other important forms of compensation. For example, Bebchuck and Jackson (2005) find that the ratio of executives’ pension value to executive total compensation during their service as CEO has a median of 34%. Ignoring such a component of executive compensation could result in less than optimal compensation schemes, which could in turn increase agency costs and incomplete conclusions in research.
Academic literature has so far placed its focus on equity compensation and its effect on the financial reporting quality (Cheng and Warfield, 2005; Erickson et al., 2006; McAnally et al., 2008; Baker et al., 2003). However, when looking at company proxy statements – a statement in which the company provides information that is required from them by the Securities and Exchange Commission - it becomes evident that executive compensation consists not only of equity or cash bonuses. It also consists of pensions and deferred compensation.
It is important to note that there are two types of pension schemes: defined contribution plans and defined benefit plans. Defined contribution plans refer to a pension plan where the company pays contributions into a separate entity (Picker et al.,2016). Furthermore, employers have no legal obligation to continue contributions if the plan does not have sufficient funds. Defined benefits plans refer to pension plans that are not defined contribution plans (Picker et al., 2016). It is a plan in which the employer pays contributions into a separate entity, however, the employer does have a legal obligation to continue contributions if the plan does not have sufficient funds (Picker et al., 2016)
Defined benefit pensions and deferred compensations can be seen as the company’s promise to pay its employees for their current service, in a future period. This is similar to debt financing in the sense that the company pays money it owes its employees in a future period. Furthermore, in the U.S. many of these deferred compensation and defined benefit pension plans are unsecured and unfunded (Sundaram and Yermack, 2008). This means that there is a risk that these components of compensation may not be paid if the firm defaults. This implies that the employees’ deferred compensation and defined benefit pensions are subject to the same risks as the firm (Sundaram and Yermack, 2008). As such, deferred compensation and defined
3 benefit pensions have debt-like features; in the case of bankruptcy the amount that can be paid is sensitive to the value of the company’s assets (Edmans and Liu, 2011).
It is then of interest to know how this inside debt can influence the accounting choices of managers. Particularly, if these components of compensation are unfunded, and contributions are based on the employee’s income, it might change managers’ tendencies to manage earnings, because managers would want the firm to have enough earnings to contribute towards their pension plans and deferred compensation. Earnings management in turn affects the financial reporting quality of the firm (Ge and Kim, 2014). The objective of financial reporting is to provide investors with information about the entity’s activities, so they can make informed decisions (Picker et al., 2016). It can then be said that high quality financial reports provide an accurate portrayal of the firm’s activities (Biddle et al., 2009).
Additionally, Dechow and Sloan (1991) found that CEOs in their final years of office tend to manage their earnings to improve short-term earnings performance. This is known as the CEO horizon problem, in which the CEO becomes short-sighted and focused on short-term performance instead of long-term performance. Due to this, CEOs might underinvest and cause a firm’s performance to decline. A point of interest would then be to see how inside debt influences the career horizon problem.
As such, the purpose of my thesis is to examine how inside debt holdings of CEOs affect financial reporting quality. In particular, I want to investigate how the inside debt holdings of CEOs in their final years influence the financial reporting quality. As such, I want to find an answer for the following research question:
RQ: Does having inside debt holdings for CEOs in their final years of office have an effect on financial reporting quality?
Providing an answer to this question is important because CEOs in their final years of office can be more inclined to act in their own interests instead of those of the firm (Dechow and Sloan, 1991). This means it could possibly lead to a decline in the financial reporting quality, because managers might manage their earnings, which affects the financial reporting quality. While CEOs do not directly influence the financial reports, if they have enough influence within the firm, they can use this to their advantage to influence the process of the financial reports through other managers in the firm. So, by providing an answer to the research question, more insight can be gained on how the inside debt holdings of these managers can affect the financial reporting quality of a firm.
4 The findings will therefore also be interesting from the agency conflict point of view. When designing a contract for a CEO, it is important that the interests of the CEO are aligned with those of the firm and/or shareholders. Insights into how inside debt influences the actions of the CEO, and how this affects the financial reporting quality, can thus lead to better understanding when designing the compensation for CEOs.
Furthermore, there is limited research on inside debt holdings such as deferred compensation and pensions compared to research on executive compensation components such as shares, stock options, bonuses, and other forms of compensation. As mentioned earlier, prior studies (Cheng and Warfield, 2005; Erickson et al., 2006; McAnally et al., 2008; Baker et al., 2003) have shown that the equity-based components of executive compensation can have an effect on financial reporting quality. However, as He (2015) shows, inside debt holdings also affect the financial reporting quality. Thus, the answer to this research question will extend the body of knowledge related to CEO inside debt holdings and their impact on financial reporting quality.
The research done in the thesis will be based on the work of He (2015), but where his focus lies on the long-term inside debt holdings, my focus will be on the effect of CEO tenure. As such, the thesis will also attempt to expand on existing research regarding inside debt.
On December 15, 2006 the U.S. Securities and Exchange Commission (SEC) adopted changes to the disclosure requirements related to executive compensation. Among these changes, the SEC required public firms to disclose information with regards to the computation and value of executive pension benefits and deferred compensation (SEC, 2006). Based on this information, my sample consists of U.S. firms from 2006 to 2018. Using this sample, I test my prediction that the career horizon problem can be mitigated through the use of inside debt holdings.
Consistent with the literature (Dechow et al., 2010) I use discretionary accruals and the absolute value of the discretionary accruals as a proxy for financial reporting quality. Higher values of absolute discretionary accruals mean that managers might be engaging in earnings management. The presence of earnings management in turn could compromise the quality of the information provided by the firm’s financial reports.
My results show a negative association with between CEO age paired with inside debt holdings and earnings management. I also find that inside debt holdings, without taking into account the effect of age, are positively related with earnings management. Thus, my findings
5 suggest that, the career horizon problem can be mitigated through inside debt holdings which can lead to a higher financial reporting quality. However, when not controlling for CEO age, the effect of inside debt holdings on financial reporting quality is negative.
My research contributes to the literature in several ways. First, prior studies have focused on equity-based compensation and its effects. However, little attention has been given to inside debt, even though it is a significant part of executive compensation (Bebchuk and Jackson, 2005). Therefore, my findings expand the limited research that has been done on inside debt. The results show that inside debt may actually give managers motivation to engage in earnings management. These findings are contradictory with those of He (2015) who finds that inside debt holdings have a positive relation with financial reporting quality. Dhole et al. (2016) also find that inside debt holdings have a negative relation with earning management. My results indicate that the relation between inside debt and financial reporting quality is unclear at this point and that more research needs to be done on inside debt.
Second, my findings show that the effect of inside debt holdings on the career horizon problem is negative. In other words, the career horizon problem could be mitigated through the use of inside debt. This finding extends the research on the career horizon problem. Dechow and Sloan (1991) concluded that the career horizon problem could be mitigated through the use of equity-based compensation and a relay of succession. As my results show, inside debt holdings could mitigate the effect of the career horizon. This offers an opportunity for more research of the effect of inside debt holdings and the career horizon problem.
My findings also have implications for practice. Specifically, my findings can aid in the design of executive compensation contracts. When designing these contracts its important to consider all forms of compensation, including pension benefits and deferred compensation. These contracts can keep the interests of the firm and the CEO aligned. My results show that inside debt can be used as a tool to mitigate misalignment of interests due to the career horizon problem. My findings also suggest that when taking the CEO’s tenure into account, inside debt holdings can be used to decrease the likelihood of them engaging in earnings management. So, the results of this thesis could be taken into consideration when it comes to designing executive compensation contracts.
The remainder of the paper is structured as follows. Section 2 reviews the existing literature. The hypothesis is developed in section 3. Section 4 describes the research method
6 while section 5 discusses the data. The results are presented and discussed in section 6. Finally, the paper is concluded in section 7.
7
2 Literature review
2.1 Agency Theory, incentive compensation, earnings management and financial reporting quality
In a principal-agent relationship, the principal gives the agent some decision-making authority. The agent is expected to use this decision-making authority to act on behalf of the principal (Jensen and Meckling, 1976). However, if both parties are seeking to maximize their utility as economic theory suggests, then the agent might not always act in the interests of the principal, meaning there is a conflict of interests (Jensen and Meckling, 1976). This conflict of interests is referred to as the principal-agent problem in agency theory.
To mitigate the conflict of interests, the principal can monitor the agent. Another way to mitigate this problem is by having the agent prove that they will not act in their own interests, but instead in those of the principal (Jensen and Meckling, 1976). However, the conflict of interest cannot be mitigated completely, which means that there is a residual loss. The residual loss refers to the loss due to misalignment between the decisions made by the agent and the decisions in the interest of the principal (Jensen and Meckling, 1976). The cost of monitoring the agent, the cost of the agent proving their intentions and the residual loss are referred to as agency costs (Jensen and Meckling, 1976).
Depending on the firm’s capital structures, different agency costs may be incurred. These are the agency cost of equity and the agency cost of debt. Agency cost of equity refers to the agency costs between the shareholders and the managers (Jensen and Meckling, 1976). These agency costs occur when the interests of the shareholders and the interests of the managers are not aligned. This misalignment of interests could lead to suboptimal decisions by the managers. To prevent these costs, the shareholders can monitor the managers. As such, the agency costs of equity are the costs associated with the suboptimal decisions and the costs of monitoring the managers (Jensen and Meckling, 1976). The agency cost of debt refers to the agency costs associated with debtholders and the shareholders (Jensen and Meckling, 1976). Because firms use debt financing in addition to equity financing, there may be a conflict of interest between the shareholders and the debtholders. The conflict of interest between the shareholders and debtholders can be due to asset substitution in which the firm’s assets are substituted for riskier assets, which could increase shareholder value at the expense of the debtholders (Manso, 2008). Another conflict of interest between the shareholders and the debt holders is referred to as debt overhang (Manso, 2008). Debt overhang means that there is a lack
8 of cash when the company faces financial distress, which makes it more difficult for the company to get loans. Instead, the company would have to issue equity to fund further investments, however shareholders will be less inclined to fund such investments since the proceeds would go to the debtholders (Berk and DeMarzo, 2014). Debtholders take these conflicts of interest into account by writing strict debt covenants or require a high interest rate. As such, Jensen and Meckling (1976) show that the agency cost of debt consists of three elements: the loss of wealth as a result of the impact of debt on the investment decisions of the firm, the cost of monitoring, and costs associated with bankruptcy and reorganization. To address the agency costs Jensen and Meckling (1976) suggest managerial ownership to create more congruency between the interests of the principal and the agent.
The conflict of interests between the principal and the agent can be addressed through the use of incentive compensation schemes. These incentive compensation schemes provide incentives to the agent to encourage decision-making in such a way that it maximizes firm value (Dechow and Sloan, 1991). Built on Jensen and Meckling’s (1976) idea of managerial ownership, incentives are usually based on equity (such as shares or stock options). This is because the manager will then have a stake in the firm and this would align the interests of both principal and agent. Essentially, the incentives are based on firm performance – indicated by the firm’s stock price (Coles et al., 2006).
However, studies (Coles et al., 2006; Low, 2008) have found that incentive compensation based on equity could increase the likelihood of managers engaging in more risk-taking behaviour. This is opposite of the intention of incentive compensation, since this risk-taking behaviour could negatively impact the creation of firm value.
2.1.1 Equity incentives and earnings management
Cheng and Warfield (2005) find that managers whose equity incentives are high, are more likely to engage in earnings management. In their research they find that when managers have high equity incentives, the likelihood of these managers to report earnings that meet or just beat analysts’ forecasts increases. Even more so, if the equity incentives are consistently high, the managers are more sensitive to future stock performance. In order to avoid earnings disappointments in the future, these managers will engage in increased reserving of the current earnings (Cheng and Warfield, 2005).
Adding to this, Bergstresser and Philippon (2006) find that when a CEO’s potential total compensation depends on the value of stock and option holdings, they are more likely to
9 use discretionary accruals to manage, and therefore manipulate, the reported earnings. It is also found that when the accruals are high, the CEOs exercise larger numbers of options and shares (Bergstresser and Philippon, 2006).
Building on the previous, a study by Summers and Sweeney (1998) found that in firms where financial statement fraud was detected, the managers engaged in insider trade – selling their equity holdings based on their information about the firm’s performance. Another study by Burns and Kedia (2006) finds that a CEO’s option portfolio’s sensitivity to stock price is positively related to the inclination to misreport. Particularly, they find that compared to other components of compensation, stock options significantly affect the likelihood of misreporting (Burns and Kedia, 2006). This can be explained by the fact that CEO stock options have a limited downside risk when misreporting is detected (Burns and Kedia, 2006). In other words, the consequences of misreporting have a limited downside risk for the CEOs.
2.2 Inside debt
As mentioned, Jensen and Meckling (1976) argued that equity ownership would limit some of the agency problems. Jensen and Meckling (1976) also considered compensation based on equal ratios of debt-based and equity-based compensation. However, they did not elaborate on this, leaving it open for further research.
As previously mentioned, with defined benefit pensions the employer has a legal obligation to make payments to the pension plan, including when the plan does not have sufficient funds to make future payments to its employees (Picker et al., 2016). Therefore, it is owed the employees future payments for their current services. Defined contribution plans, however do not have a legal obligation to make further payments if the plan does not have sufficient funds to make future payments. This means that defined benefit pensions are similar to debt in the sense that the employer owes its employees future payments. Academic literature suggests that defined benefit pensions and deferred compensation can be viewed as forms of inside debt (Sundaram and Yermack, 2005; Edmans and Liu, 2011).
In 2006 the Securities and Exchange Commission required firms to disclose information regarding pension benefits in their proxy statements. This new requirement showed that in many of the defined benefit pensions and deferred compensation plans are unsecured (see Appendix). This implies that there is no guarantee of future payments of these compensation components. As such, these compensation components are subject to the same risks as those of the firm – in the case of bankruptcy, the amount that can be paid is sensitive
10 to the value of the assets. It can be said that defined benefit pensions and deferred compensation face a similar risk to that of the firm’s debt – default risk. However, these components of inside debt are only similar to debt if the pension plan or deferred compensation schemes are unfunded (Anantharaman et al., 2014). If the pension plans or deferred compensation schemes are funded, they lose their debt-like features.
Building on the proposition of Jensen and Meckling (1976) of an equal ratio of equity-based and debt-equity-based compensation, Sundaram and Yermack (2005) examine inside debt and its role in managerial compensation. The authors define inside debt as pensions and deferred compensation. It is usual that these compensation components are unfunded and are therefore exposed to the same risks as those of the firm. Sundaram and Yermack (2005) assert that when managers hold large pensions or deferred compensation that the agency costs of debt problems associated with risk-shifting (the situation in which managers make risky investment decisions) and excessive pay-outs should diminish. The authors conducted a case study and found support for their assertion; debt-based compensation has a role in reducing the agency costs of debt. This is because managers with large pensions will pursue strategies that will reduce the overall firm risk. They can do this by choosing fewer risky investments or increasing the length of the average maturity of outstanding debt (Sundaram and Yermack, 2005). The authors also find that as CEOs grow older, there is a shift towards inside debt instruments. Evidence is also provided that CEOs tend to manage firms more conservatively and commit to actions that reduce the probability of debt default (Sundaram and Yermack, 2005).
Similarly, Edmans and Liu (2011) provide a framework for the justification of the use of inside debt as part of efficient compensation. The authors find that the theoretical framework they developed provides evidence that inside debt decreases risk-shifting overinvestments and that inside debt induces managers to increase the liquidation value of the firm. This is because inside debts yield a positive payoff in bankruptcy proportional to the liquidation value (Edmans and Liu, 2011). This would mean that there is an alignment of interests between the CEO and the debtholders, because debt-based compensation can keep CEOs from risk taking and make CEOs focus on the long term.
2.2.1 Inside debt and its effect on managers’ choices
The findings of Cassell et al. (2012) extend the study of Sundaram and Yermack (2005). Cassell et al. (2012) examine the relation between CEO inside debt holdings and the riskiness of firm investment and financial policies. They hypothesize that CEOs with large inside debt holdings
11 will protect the value of the holdings against the risk of default and that CEOs do this by implementing less risky investments and financial policies. According to Cassell et al. (2012) equity-based compensation can lead to risky behaviour which negatively impacts debtholders because the excessive risk-taking behaviour can lead to an increase in the default risk of the firm. Cassell et al. (2012) find that there is a negative relation between their measure of CEO debt-to-equity ratio and future stock return volatility. This relation can be explained by the fact that CEOs engage in more conservative investments and financial policies, which is similar to the findings of Sundaram and Yermack (2008). Cassell et al. (2012) also find that there is a positive relationship between the CEO inside debt holdings and the firm’s working capital, which is a form of liquidity.
Dang and Phan (2016) examine the effect of CEO inside debt holdings on the choice of short-term vs. long-term debt. They test two hypotheses, one is the risk aversion hypothesis, which predicts a negative relationship between the variables of interest. This is because the inside debt holdings are exposed to default risk just like external creditors (Dang and Phan, 2016). The second hypothesis being an alternative to the first; a predicted positive relationship between the variables of interest. Dang and Phan (2016) find that there is a positive relationship between CEO inside debt holdings and short-term debt. There is also a negative relationship between CEO inside debt holdings and the cost of debt financing. The authors conclude that inside debt holdings are an efficient form of executive compensation. It gives managers incentives to shorten their debt maturity and thus save on the cost of debt financing and also improving firm performance.
In the same notion, Lee et al. (2018) investigate how and when CEO inside debt (debt-like compensation) affects debt contracting terms and corporate investment levels. Lee et al. (2018) find that inside debt can align management’s incentives with those of debtholders. In turn, this can reduce the agency costs of debt and increase the investment levels in firms facing constraints. Therefore, firms facing underinvestment can mitigate this problem through the use of inside debt. These findings extend the findings of Edmans and Liu (2011) by showing in which settings inside debt is more likely to be useful to mitigate agency costs.
2.2.2 Inside debt and Earnings Management
Earnings management also has consequences for the cost of new corporate bonds. Ge and Kim (2014) examine this effect. In particular, the authors examine the relation between real earnings management and the cost of new corporate bonds. Ge and Kim (2014) make a distinction
12 between two ways managers can manage earnings. The first way is accrual based and refers to the discretion managers exercise over accrual choices allowed under GAAP, such as asset impairment. The second way is by changing the timing and scale of the operating decisions (Ge and Kim, 2014). The authors test two hypotheses: one predicting a positive relation between the cost of new corporate bonds and the level of real earnings management and one predicting a negative relation between the cost of new corporate bonds and the level of real earnings management, arguing that earnings management increases information asymmetry. Ge and Kim (2014) find that, overall real earnings management is seen as a risk-increasing factor by credit rating agencies and bondholders and therefore require higher risk premiums. This can be interpreted as real earnings management increasing the cost of debt.
Similarly, Bharath et al. (2008) examined the role of accounting quality in debt contracting. Since financial statements provide information for debt providers, the quality of these reports is important. Bharat et al. (2008) predict that due to poor accounting quality, borrowers prefer private debt (loans from banks) over public debt (loans from government agencies). They also predict that poor accounting quality effects the price of debt. Bharat et al. (2008) find that indeed, when accounting quality is poor that private debt is preferred by borrowers. Additionally, the terms of debt contracting are more stringent for those with poor accounting quality (Bharath et al., 2008).
Dhole et al. (2016) examined the impact of CEO inside debt holdings on earnings management. They argue that CEO inside debt holdings have an effect on CEO motivation to engage in earnings manipulation to smooth earnings. As such, Dhole et al. (2016) predict that CEO inside debt holdings have a negative relation with accounting- or accrual-based earnings management. The results provide evidence for negative association between inside debt holdings and earnings management. Dhole et al. (2016) argue that the deterrence role of inside debt holdings is larger for firms with more volatile earnings, suggesting that earnings volatility affects the effect of inside debt on earnings management. Overall, Dhole et al. (2016) find that as the proportion of inside debt increases, the investment and reporting strategies change. Earnings management (both accounting- and accrual-based) decreases as inside debt holdings increase.
2.2.3 Inside debt and financial reporting quality
A review of the literature by Dechow et al. (2010) reveals several proxies for financial reporting quality. The authors distinguish the proxies in three broad categories. The first category is
13 based on the properties of earnings, such as accruals, earnings smoothness and timely loss recognition. The second category is investor responsiveness to earnings. The last category relates to external indicators of earnings misstatements, such as restatements and internal control weakness reported under the Sarbanes-Oxley act.
The incentive for financial misreporting should also be considered when discussing the effects of inside debt holdings. Karpoff et al. (2008) examine the penalties imposed on firms that have “cooked the books”. According to the authors, there is a belief that financial misreporting is punished lightly. However, as their research indicates the punishment from the market is nearly 7.5 times higher than the punishment imposed on firms by the legal and regulatory system (Karpoff et al., 2008). The reputational loss, which the authors define as “the expected loss in the present value of future cash flows due to lower sales and higher contracting and financing costs”, is high for firms. When looking at CEO inside debt holdings, as prior literature has stated, CEOs might manage in such a way that the firm’s default risk will decrease and not increase. Reputational losses could lead to a higher default risk and a lower firm value (He, 2015).
In his research, He (2015) focuses on the relationship between CEO inside debt holdings and financial reporting quality. Two competing hypotheses are presented. The hypothesis that CEO inside debt holdings are not related to financial reporting quality and the hypothesis that CEO inside debt holdings are related to financial reporting quality, whether this relation is positive, or negative is yet unclear. He (2015) comes to the conclusion that CEO inside debt holdings are positively related to financial reporting quality and thus concludes that CEO inside debt holdings improve the financial reporting quality.
When it comes to the relationship between accounting conservatism and inside debt holdings, Wang et al. (2017) hypothesized that there are two possible relationships. The first is a negative relationship, implying that higher inside debt holdings mean lower needs to adopt conservative policies, due to a decrease in risk-taking incentives. The second is a positive relationship, implying that accounting conservatism can increase the riskiness of inside debt holdings. Wang et al. (2017) find evidence for the first hypothesis. In particular, they find that firms with higher CEO inside debt holdings have lower accounting conservatism. This is especially the case for firms with high default risk and agency cost of debt. Wang et al. (2017) also provide moderate evidence that the negative relationship between inside debt holdings and accounting conservatism is more pronounced when a large part of the firm’s debt is unsecured, and managers can credibly commit to accounting conservatism ex ante.
14
2.3 Career horizon problem
The career horizon problem refers to the CEO’s career stage. Matta and Beamish (2008) examined the career horizon problem for CEOs. The career horizon problem can be described as follows: as the CEO gets closer to retirement, the career horizon gets shorter. It is therefore important what the implications of a shorter career horizon are on risk-taking (Matta and Beamish, 2008). The authors propose that CEOs have a growing risk aversion when their career horizon is shorter. Matta and Beamish (2008) believe that this has to do with the reputational loss and concerns for a legacy of success. They find that equity and in-the-money options do not mitigate concerns related to the career horizon problem. However, risk-taking is reduced for CEOs near retirement (Matta and Beamish, 2008). The findings in this paper would support what Gibbons and Murphy (1992) concluded in their paper; when the CEO is closer to retirement, the explicit incentives should be strongest. This means that in order to keep the CEO interests aligned with those of the firm, the incentives should be stronger closer to retirement.
An empirical analysis done by Antia et al. (2010) examined the effect of top managers’ near-sightedness on a firm’s market valuation. They do this by looking at the length of CEO decision horizon, measured by the expected CEO tenure. Antia et al. (2010) find that a shorter CEO horizon leads to more agency costs, a lower firm valuation and higher levels of information risk. The explanation for these findings is that CEOs will have a preference for investments that will pay them back faster at the expense of long-term value creation (Antia et al., 2010)
The findings of Antia et al. (2010) are supported by the research done by McClelland, Barker III and Oh (2012). McClelland et al. (2012) examined the relation between CEO career horizon and tenure and future performance of firms. They predicted that shorter career horizons lead to more risk-averse strategies, which in turn have negative influence on future firm performance. This relationship is intensified with high ownership. The findings of this study provide evidence for their hypothesis.
Another study done by Serfling (2014) investigates the relationship between CEO age and risk-taking behaviour. The link between CEO age and risk-taking behaviour is unclear; one strand of literature suggests that younger CEOs are more risk-averse due to career concerns, while another strand of literature suggests that younger CEOs are less risk-averse to signal their competence (Serfling, 2014). He hypothesizes that older (younger) CEOs prefer less (more)
15 risk. His study provides support for this hypothesis. The results show that older CEOs reduce firm risk by taking on less risky investment policies, investing less in R&D, more diversified management of operations and maintain a lower operating leverage (Serfling, 2014).
These findings support those of Dechow and Sloan (1991). Dechow and Sloan (1991) study the relation between executive incentives and the horizon problem. Specifically, the authors look at the relation between R&D expenditures and CEOs in their final years of office. Dechow and Sloan (1991) hypothesise that R&D expenditures decrease when CEOs are in their final years of office. They argue that CEOs do this to improve their short-term earnings, since the R&D expenditures may generate revenue past the CEO’s years of office. The findings of their research indicate that R&D expenditures decrease in the CEO’s final years of office, thus supporting their hypothesis. Dechow and Sloan (1991) find that stock ownership could mitigate this problem of underinvestment.
During the last years of CEO service, earnings management may also occur. Ali and Zhang (2015) examine the changes in CEO incentives to manage their firm’s reported earnings during their tenure. They test two hypotheses. The first hypothesis is that the authors expect that earnings overstatement is greater in the early years of CEO’s service. The second hypothesis is that the authors expect that the difference in earnings overstatement in the first early years and last years of the CEOs service is smaller for firms who monitor their CEOs stronger. Ali and Zhang (2015) find that in the early years of CEO’s service there is earnings overstatement. Additionally, consistent with the career horizon problem, earnings overstatement is greater in the last years of CEO’s service.
The career horizon problem implies that the CEO will engage in behaviour that is not aligned with the interests of stakeholders. As such, the effect of the career horizon problem and financial reporting quality is also of interest. Huang et al. (2012) examine this association. Particularly, they look at the relation between CEO age and the financial reporting quality of firms. Huang et al. (2012) examine the financial reporting quality as meeting or beating analysts’ earnings forecasts and financial restatements. The authors predict that firms with older CEOs will have a higher financial reporting quality. Huang et al. (2012) find that evidence for their predictions. There is a negative association between CEO age and meeting or beating analysists’ forecasts and financial restatements, therefore implying a positive association between CEO age and financial reporting quality (Huang et al., 2012).
16
3 Hypothesis development
Previous literature has shown that inside debt holdings could reduce the agency cost of debt, positively influence the financial reporting quality, and improve the firm’s performance (Sundaram and Yermack, 2008; He, 2015; Dang and Phan, 2016). Furthermore, CEO inside debt holdings could lead to managers implementing strategies more conservatively and work to lower the default risk of the firm, since their pensions and deferred compensations are exposed to the same risks as those of the firm (Cassell et al., 2012; Sundaram and Yermack, 2008). Because the CEO inside debt holdings give the CEOs incentives to shorten the debt maturity, firms can save on the cost of debt financing (Dang and Phan, 2016). This is a result from the reduced agency cost of debt. Additionally, firms facing constraints can mitigate an underinvestment problem through the use of inside debt (Lee et al., 2018).
The literature review also shows that earnings management has a negative effect on the financial reporting quality. The reason earnings management negatively impacts the quality of the financial reports is because earnings can be misstated through earnings management. In turn, the earnings reporting in the financial statements will not be the true, accurate numbers. This would mean that the financial reports do not adequately provide information to the firm’s stakeholders and therefore the quality of the financial reports decreases. The consequences of low financial reporting quality are higher costs of new corporate bonds and more stringent terms for debt contracts (Ge and Kim, 2014; Bharath et al., 2008). This is because earnings management is seen as risk increasing, which increases the cost of debt as a consequence. Inside debt holding could be used to mitigate the incentives to engage in earnings management. As Dhole et al. (2016) find, as the proportion of inside debt increases, earnings management decreases.
As mentioned, CEO inside debt holdings will give CEOs incentives to engage in less risky behaviour. This means that CEO inside debt holdings would be positively related with financial reporting quality, since CEOs would be less likely to engage in earnings management or misreporting, due to reputational losses described in Karpoff et al. (2008). If the CEOs suffer reputational losses, their inside debt holdings become more exposed to higher risks. Therefore, inside debt holdings would have a positive effect on the financial reporting quality. This is supported by He (2015). Furthermore, Wang et al. (2017) find that firms with higher inside debt holdings have lower accounting conservatism. This effect is more pronounced for firms where debt is, for a large part, unsecured. This would imply that debt holders trust the quality
17 of the financial reports. Furthermore, as Ge and Kim (2014) show, real earnings management increases the cost of debt because manipulated numbers are not a reliable measure for bondholders or debt providers to assess the firm’s performance. As such, it exposes the unfunded pensions and deferred compensation to a higher risk, which is something the CEOs would like to avoid.
When it comes to the career horizon, previous literature tells us that CEOs become more risk-averse and that there are increased agency costs, higher levels of information risk and lower future firm performance (Matta and Beamish, 2008; Antia et al., 2010; McClelland et al., 2012; Serfling, 2014). In relation to earnings management, CEOs tend to overstate earnings in the last years of their service, consistent with the career horizon problem (Ali and Zhang, 2015). This would imply a negative impact on the financial reporting quality in the last years of CEO service. However, CEO age and financial reporting quality seem to be positively associated, implying that as CEOs get older (and the career horizon gets shorter) the financial reporting quality would also increase (Huang et al., 2012).
So, as mentioned, earnings management can increase the cost of debt since debt providers and debtholders may not be able to use the numbers in the firm’s financial reports to make decisions. This means that these debtholders and providers may see the company as riskier, since the numbers in the financial reports may not reflect the actual earnings. Therefore, the information asymmetry increases. In addition, in the last years of office, CEOs might have the tendency to engage in earnings management, lowering financial reporting quality. However, as previous literature has mentioned, inside debt could mitigate the effects of information asymmetry since these holdings are exposed to the same default risk as company debt. This means that in the case of bankruptcy, the amount of inside debt that can be paid is contingent on the value of the assets. So, it would be in the best interest of the CEO to protect its inside debt holdings from default risk, which in turn means that the company debt would also be protected from default risk. CEOs would do this by being more conservative and making less risky investments. Based on this discussion, I would expect that the inside debt holdings do have an effect on earnings management in the CEO’s final years of office. In particular I expect that the inside debt holdings can be used to mitigate the effect of the career horizon problem because CEOs would want to protect their inside debt holdings from default risk and therefore would engage less in earnings management
Thus, based on the literature, I expect that the inside debt holdings of managers in their final years as CEOs will be positively related with the financial reporting quality of the firm.
18 In hypothesis form:
H1: There is a positive relationship between the inside debt holdings of managers in their final years as CEOs and the financial reporting quality of a firm.
19
4 Research design
This thesis aims to look at the relationship between the inside debt holdings of CEOs and their tenure and the financial reporting quality of the firm. The predictive validity framework to show how the conceptual relation is operationalised is presented in the Appendix.
4.1 Measuring inside debt
As mentioned before, inside debt consists of defined benefit pensions and deferred compensation. To see if inside debt holdings of CEOs in their final years of office influence the financial reporting quality of a firm, a dummy variable will be constructed. The variable will equal 1 for firms where CEOs have inside debt holdings, and 0 otherwise. As such, it can be stated as follows:
ID = 1 if firm has CEO inside debt ID = 0 otherwise
Here inside debt is defined as the sum of the pension value and deferred compensation. If the sum of the pension value and deferred compensation is larger than 0, it means that the firm has inside debt holdings. If this is the case, then ID takes on a value of 1 to indicate that the firm has inside debt holdings. If the sum of the pension value and deferred compensation is 0 then it means that there is no inside debt in the firm and that ID will take on the value of 0 to indicate this.
4.2 Proxy for financial reporting quality
Several proxies can be used for financial reporting quality. The common proxies used for financial reporting quality relate to properties of earnings such as accruals quality, abnormal accruals (Dechow et al., 2010). External indicators of earnings misstatements such as restatements and internal control weaknesses reported under the Sarbanes-Oxley act are also used as proxies for financial reporting quality (Dechow et al., 2010). The reason for this is that section 404 of the Sarbanes-Oxley Act requires that management reports on the effectiveness of internal control over financial reporting in their annual report and that an auditor is required to state a separate opinion on the management’s evaluation.
Based on the literature review relating to earnings management and financial reporting quality, the proxy of abnormal accruals will be chosen. Accruals can be used as a proxy for
20 financial reporting quality since the absolute value of accruals, in particular discretionary accruals, can indicate that managers might be engaging in earnings management. Earnings management can affect the quality of the financial reports of a firm. To estimate this proxy, the modified Jones Model from Dechow et al. (1995) will be used. The modified Jones Model tries to control for the effect of changes in the firm’s economic circumstances on non-discretionary accruals. The modification eliminates the tendency of the original Jones model to measure discretionary accruals with error when discretion is exercised over revenues (Dechow et al., 1995). Additionally, the modified Jones model assumes that all changes in credit sales are outcomes of earnings management (Dechow et al., 1995). The argument for this assumption is that is it easier to manage earnings by exercising discretion over the recognition of revenue on credit sales than it is to manage earnings by exercising discretion over the revenue on cash sales (Dechow et al., 1995)
The Modified Jones Model is written as: 𝑁𝐷𝐴̂𝑡 = α̂1( 1
𝐴𝑡−1) + α̂2(ΔREV𝑡 − ΔREC𝑡) + α̂3(PPE𝑡) Where,
ΔREV𝑡 = Revenues in year t less revenues in year t-1 scaled by the total assets at t-1
ΔREC𝑡 = Net receivables in year t less revenues in year t-1 scaled by the total assets at t-1 PPE𝑡 = gross PPE in year t scaled by the total assets at t-1
1
𝐴𝑡−1 = Total assets at t-1
α
̂1, α̂2, α̂3 = industry- years specific parameters, estimated by OLS using𝑇𝐴𝑡 = a1( 1 𝐴𝑡−1) +
a2(ΔREV𝑡 − ΔREC𝑡) + a3(PPE𝑡) + ε
Based on this information I will estimate the discretionary accruals using the following model 𝑇𝐴𝑡 = a0 + a1( 1
𝐴𝑡−1) + a2(ΔREV𝑡 − ΔREC𝑡) + a3(PPE𝑡)+ ε.
To estimate the parameters of this model, I use industry-year specific observations. I do this by creating a variable in Stata for the two-digit SIC code. This will indicate in which industry the firms in my sample are active in. Next, for each year and two-year SIC code, I drop the cases where there are less than ten firms available. The distribution of the two-digit SIC codes after
21 this process can be found in Panel B in table 1. Then, the regression is run for each combination of industry and year to obtain the parameters of the above regression.
Furthermore, this regression estimates the total accruals, which is measured as the income before extraordinary items less cashflow from operation. Total accruals include both non-discretionary accruals and discretionary accruals, therefore in the regression that I estimate, the error term represents the difference between the total accruals and the non-discretionary accruals. Higher values of non-discretionary accruals indicate that managers might be engaging in earnings management.
4.3 Control variables
In the regression analysis, control variables should be included to account for factors that are correlated with inside debt holdings of CEOs, financial reporting quality, or both.
Based on previous literature, the factors that will be controlled for are CEO equity ownership (EQUITY), firm size (SIZE), debt-to-asset ratio (DAR) and CEO age (AGE). I control for CEO equity ownership because prior studies have found that equity-based compensation can increase the likelihood of managers engaging in earnings management through discretionary accruals (Cheng and Warfield, 2005; Bergstresser and Philippon, 2006) but also to misreport (Burns and Kedia, 2006). Firm size is also controlled for because the different sized firms have different capital structures, which could influence the effect of inside debt on investment levels (Lee et al., 2018). Finally, the debt-to-asset ratio is included as a control variable to reflect the influence of debt over CEO’s decisions when they have inside debt holdings since their inside debt holdings are exposed to the same default risk as the firm’s debt (Dang and Phan, 2016). It can be seen as controlling for the firm’s leverage. The debt-to-asset ratio can also be seen as an indicator of the trade-off of which choices to make in the interest of the shareholders or debtholders. These variables are included since they may give CEOs incentives to engage in earnings management.
CEO equity ownership will be measured as the CEO’s equity holdings as a percentage of the shares outstanding at the end of the fiscal year. To measure firm size, the common measures are total assets, total sales or market capitalization (Dang et al., 2018). Debt-to-asset ratio is measured as the ratio of debt to assets of the firm at the end of the fiscal year. CEO age corresponds to the age of the firm’s CEO at the end of the fiscal year.
22
4.4 The model
The regression model that will be tested is the following:
𝐹𝑅𝑄𝑖𝑡 = α + β1𝐼𝐷𝑖𝑡+ β2𝐴𝐺𝐸𝑖𝑡+ β3𝐴𝐺𝐸𝑖𝑡∗ 𝐼𝐷𝑖𝑡+ β4𝐸𝑄𝑈𝐼𝑇𝑌𝑖𝑡+ β5𝑆𝐼𝑍𝐸𝑖𝑡+ β6𝐷𝐴𝑅𝑖𝑡+ ε𝑖𝑡
Where:
FRQ is estimated using discretionary accruals (DA) AGE is the age of the CEO
ID is inside debt, where 1 if there is inside debt and 0 otherwise AGE*ID is the interaction term of CEO_AGE and inside debt
EQUITY is the percentage of shares owned by the CEO
SIZE is the size of the firm measured as the firm’s market value DAR is the debt-to-asset ratio
ε is the error term
My hypothesis predicts a negative sign on the coefficients for inside debt and age. Therefore, it also expects a negative sign on the coefficient for the interaction term between age and inside debt.
Additionally, I will run the same regression but will use the absolute value of the discretionary accruals as a proxy for financial reporting quality. The model that will be tested will be the same as the model above, the only difference is that the FRQ will be estimated using the absolute value of the discretionary accruals.
Furthermore, I would like to compare the results with those of He (2015). The reason for this is that the author found a positive relationship between the variables of interest. It would then be interesting to know (if the hypothesis not rejected) how strong this positive effect is. A possible way I would do this is by looking at the difference between the coefficient in He (2015) results on the inside debt measure and the results of my inside debt measure and see if this is statistically significant.
23
5 Data
The data have been gathered from the databases available within the Wharton Research Data Services system. Data needed to estimate discretionary accruals are gathered from Compustat. Executive compensation data is gathered from Execucomp.
The sample period starts in 2006 and goes up to the most recent year, 2018. The reason the sample starts in 2006 is because starting from December 15, 2006 the SEC required public firms to disclose information with regards to the computation and value of executive pension benefits and deferred compensation (He, 2015). Furthermore, data for the fiscal year of 2005 is gathered for the variables in the modified jones model, since it needs data from the previous period. If I did not include the fiscal year of 2005, I would not be able to calculate the variables for 2006, thus missing out on observations in the sample. The observations will be in fiscal-years, since the data is available annually.
5.1 Sample Selection
A list of the variables gathered from the databases can be found in section 7.2 of the appendix. As mentioned, data was gathered from Compustat and Execucomp. This was done by searching the entire database for active firms. For the variables needed for CEO inside debt, the pension value and deferred compensation were gathered from Execucomp. Pension value yielded 116,632 observations, while deferred compensation yielded 116,650 observations. Additionally, I also gathered information about the executive’s age from Execucomp with a result of 121,466 observations. Next, data was gathered for the modified jones model as well as the data for firm size, which has three proxies that can be used: sales, total assets and market value of equity. In my thesis, I will use market value of equity as the proxy for firm size. This resulted in 60,437 observations. The result for the CEO equity measure is 84,272 observations. The initial sample selection and its descriptive statistics are presented in table 1. Panel A shows the descriptive statistics of the initial sample before any missing variables are dropped. Panel B shows the distribution of the sample after data was manipulated. The usable sample ended up being 7,299 observations.
24
Table 1 - Sample Selection and Distribution
Panel A: Initial Sample and descriptive statistics
Variable Observations Mean SD Min Max
% CEO share ownership 84,272 0.984 3.901 0.000 275 Pension value 116,632 1,019 3,842 0.000 115,822 Deferred compensation 116,650 922.308 4,965.825 -38,290 250,578 Age 121,466 52.823 7.565 26 96 Assets 79,317 23,440 160,118 160,118 3,771,200 Debt 58,476 2,981 45,429 0.000 3,296,298 Gross PPE 62,722 3,974 18,813 0.000 501,070 Receivables - total 67,619 5,648 62,325 -0.006 3,192,828 Receivables - estimated doubtful 45,825 51.369 810 0.000 285.296
Revenue 79,044 4,420 18,197 -15,009 496,785
Size (Market Value) 60,437 4,216 20,178 0.000 790,050 Panel B: Sample Distribution after data manipulation
Two-digit SIC code Frequency %
13: Oil and gas 347 4.8%
20: Food and Beverages 250 3.4%
28: Chemicals and related products 614 8.4%
34: Metal and Metal products 184 2.5%
35: Industrial Machines 531 7.3%
36: Electronics 634 8.7%
37: Transport (Vehicles etc) 302 4.1%
38: Instruments and related 535 7.30%
48: Communication Devices 223 3.1%
50: Wholesale 227 3.1%
73: Services 866 11. 9%
80: Services - Health 169 2.3%
Industries with < 2% representation 2,417 33.1%
Total 7,299 100%
Table 1 shows sample selection and distribution. Panel A shows the descriptive statistics for the initial sample. Panel B shows the distribution of the sample based on the two-digit SIC code, after data manipulation. The usable sample is 7,299 observations
5.2 Constructing the variables
The next step is to construct the variables. The first variable to be constructed is the CEO inside debt measure. To do this, the pension value and deferred compensation values were summed for each observation. After constructing the inside debt measure, the variables for the financial reporting quality proxy are constructed. These are the input variables for the modified Jones model. To start off the net receivables need to be calculated for the model. The net receivables are calculated by subtracting the estimated doubtful receivables from the total receivables. The variables in the modified jones model are scaled by the total assets. Therefore, the next step is to scale the variables. To get ΔREV, the difference in revenues in year t and year t-1 are calculated. After this is done, it is scaled by the assets in year t-1. ΔREC are calculated in a similar way, the difference between the net receivables in year t and year t-1 is calculated and
25 then scaled by the assets in year t-1. Lastly, the gross PPE are scaled by the total assets in year t-1.
After constructing the inside debt measure and the Jones model variables, the control variables are also constructed. CEO equity ownership will be measured as the CEO’s share holdings as a percentage of shares outstanding at the end of the fiscal year for each observation. This data is directly gathered from the Execucomp database. The debt-to-asset ratio for the firm will be constructed by measuring the ratio of debt to assets of the firm at the end of the fiscal year for each observation. CEO age is the age of the CEO in the observation year. This data is gathered from Execucomp and can be readily used. Finally, I’ve gathered data on the three common proxies for firm size. These are sales, total assets and market value of equity. The data is readily available from the database. In my thesis, I will be using market value of equity as a proxy for firm size.
When dealing with data, the effect of outliers needs to be considered. Therefore, I winsorize the data. Winsorizing the data means that the values that are considered outliers in the sample are converted so that those values are no longer considered outliers, these values will be limited to the either the smallest and/or largest value in the sample (Salkind, 2010). In my sample, I winsorize the variable at the top and bottom 1%. In doing so, the effect of outliers in the sample is mitigated.
26
5.3 Descriptive Statistics
Table 2 shows the descriptive statistics for the input variables for the modified Jones model (1995) which is used to obtain the discretionary accruals as a proxy for financial reporting quality. Total accruals were calculated by subtracting the cash flow from operations from the income before extraordinary items. The difference between ΔREVand ΔREC as well as the gross PPE are scaled by the assets of the previous observation year. The variables have been winsorized at the 1st and 99th percentile to reduce the effect of outliers in the sample.
Table 3 shows the descriptive statistics for the control variables. The percentage of shares owned by the CEO is the percentage of shares the CEO owns as a percentage of shares outstanding. In my sample the average of CEO share ownership is 0.925% and the median is 0.154%. The statistics show that the mean firm size – which is based on market value of the firm – is 8,925 and the median is 1,942. This indicates the sample is diverse when it comes to firm size. The mean debt-to-asset ratio is 0.219 and the median is 0.197. Furthermore, the statistics indicate a less diverse sample in terms of CEO age, the mean and median CEO age is 53. All variables in table 3 are winsorized at the 1st and 99th percentile to reduce the effect of outliers in the sample.
Table 2 - Descriptive Statistics Modified Jones Model
Obs Mean SD. Min Q1 Median Q3 Max
Total accruals 7,299 -0.058 0.085 -1.113 -0.088 -0.050 -0.022 0.529
Assets 7,299 0.002 0.022 0.000 0.000 0.001 0.002 1.869
ΔREV- ΔREC 7,299 0.054 0.218 -1.044 -0.020 0.036 0.117 1.639 Gross PPE 7,299 0.533 0.426 0.000 0.215 0.410 0.756 2.710 Table 2 shows the descriptive statistics for the input variables for the modified Jones Model (1995) used to estimate the total accruals. All variables are winsorized at the top and bottom 1% to reduce the effect of outliers.
Table 3 - Descriptive Statistics Control Variables
Obs Mean SD Min Q1 Median Q3 Max
%CEO shares 7,299 0.952 2.854 0.001 0.045 0.154 0.509 21.059 Size 7,299 8,925 23,338 49.232 702.573 1,942 6,560 169,351
DAR 7,299 0.219 0.192 0.000 0.045 0.197 0.326 0.870
CEO's Age 7,299 53.253 7.271 37 48 53 58 73
Table 3 shows the descriptive statistics for the control variables. % of CEO shares refer to the shares the CEO owns as a percentage of shares outstanding. Size is measured by the firm’s market value. DAR is the debt-to-asset ratio, calculated as debt as a percentage of the debt-to-assets. CEO age is the age of the CEO in the observation year. All the variables are winsorized at the top and bottom 1% to reduce the effect of outliers
27 The descriptive statistics for the main variables are presented in table 4. The discretionary accruals (DA) were obtained from the modified jones model. The mean of the discretionary accruals is 0.079 and the median is 0.038. Pension value and deferred compensation are the components of inside debt. The mean and median of pension value is 953 and 0 respectively. The mean and median of deferred compensation 752 and 0 respectively. Inside Debt is the sum of pension value and deferred compensation. The mean and median inside debt holdings are 1,815 and 51. This indicates a diverse sample in terms of inside debt. Finally, there is a dummy variable ID. This variable indicates if a firm has CEO inside debt holdings. The variable takes the value of 1 if a firm has CEO inside debt holdings and 0 if there is no CEO inside debt. The statistics show that ID has a mean of 0.540, which implies that 54% of the firms in my sample have inside debt holdings. All the variables except for the dummy variable, ID, have been winsorized at the 1st and 99th percentile to reduce the effect of outliers in the sample.
The correlations among the inside debt variable, financial reporting quality variables and control variables are presented in table 5. The statistics indicate a positive correlation between ID and DA, suggesting that managers with inside debt are more likely to engage earnings management. The correlation between ID and |DA| suggest that managers with inside debt are less likely to engage in upwards or downwards earnings management. Again, all the variables have been winsorized at the 1st and 99th percentile to reduce the effect of outliers in the sample.
Obs Mean Std. Dev. Min Q1 Median Q3 Max
DA 7,299 0.079 0.176 -0.329 -0.020 0.038 0.142 0.723 |DA| 7,299 0.127 0.144 0.000 0.028 0.074 0.169 0.723 Pension Value 7,299 954 2,931 0.000 0.000 0.000 135. 19,309 Def. Comp 7,299 752 2,080 0.000 0.000 0.000 440 14,209 Inside Debt 7,299 1,816 4,680 0.000 0.000 51 1,176 31,857 ID (Dummy) 7,299 0.540 0.498 0.000 0.000 1.000 1.000 1.000 Table 4 shows the descriptive statistics for the main variables. Discretionary accruals are obtained from the Modified Jones model. The pension value represents the present value of the CEO's pensions. Inside Debt is calculated by summing pension value and deferred compensation. The inside debt dummy takes the value of 1 if a firm has CEO inside debt holdings and 0 if there is no CEO inside debt. All the variables except for the dummy have been winsorized at the top and bottom 1%.
Table 5 - Correlations N= 7,299 1 2 3 4 5 6 7 1 ID 1.000 2 DA 0.036 1.000 3 |DA| -0.016 0.765 1.000 4 CEO age 0.120 0.017 0.005 1.000 5 Size 0.165 0.044 0.026 0.022 1.000 6% CEO shares -0.113 -0.017 -0.004 0.204 -0.064 1.000 7 DAR 0.158 0.000 0.021 0.017 0.050 -0.076 1.000
Table 4 shows the correlations among the FRQ proxy variables, inside debt variables and the control variables. The variables have been winsorized at the top and bottom 1% to reduce the effect of outliers in the sample. Bolded numbers indicats significance at at least 10% level.
28
6 Findings
6.1 Results
Table 6 - Regression results
Panel A: Regression without taking CEO age into account Dependent variable: DA
Variable Coefficient t-statistic p-value
Intercept 0.073** 19.04 0.000 ID 0.010** 2.45 0.014 SIZE 0.000** -1.01 0.001 EQUITY -0.001 3.23 0.311 DAR -0.007 -0.63 0.530 adjusted R-square 0.002 number of obs 7,299 F-statistic 5.370
Panel B: Regression taking CEO age into account
Dependent variable: DA
Variable Coefficient t-statistic p-value
Intercept 0.024 1.18 0.239 ID 0.069** 2.35 0.019 CEO_AGE 0.001** 2.39 0.017 CEO_AGE*ID -0.001** -2.05 0.04 SIZE 0.000** 3.27 0.001 EQUITY -0.001 -1.36 0.173 DAR -0.007 -0.69 0.493 adjusted R-square 0.003 number of obs 7,299 F-statistic 4.600
Table 6 presents the results for the hypothesis test. Panel A shows the results of the regression without taking CEO age into account. Panel B shows the results of the regression results when taking CEO age into account. The dependent variable in both OLS regressions is financial reporting quality, measured as discretionary accruals. *, **, *** indicate significance of the coefficients at the 10%, 5% and 1% level.
Table 6 presents the results of the regression. Panel A shows the regression results without taking the age of the CEO into account. The model has an adjusted R-square of 0.2% and an F-statistic of 5.370. This indicates that the model is significant at the 5% level. In panel A the coefficient on CEO inside debt holdings is 0.010 and statistically significant at 5% level. This finding suggests that managers with inside debt holdings are 1% more likely to engage in earnings management, ceteris paribus. This finding is not consistent with He’s (2015) findings that inside debt holdings decrease the likelihood of managers engaging in earnings management. In other words, the results in panel A of table 6 imply that CEO inside debt
29 holdings are positively related with financial reporting quality, meaning that inside debt holdings increase the likelihood of earnings management.
Panel B of table 6 shows the regression results when taking into account the effect of the CEO’s age. The model has an adjusted R-squared of 0.003. The F-statistic is 4.60 indicating that the model is significant at the 5% level. The coefficient on ID is 0.069 and statistically significant at the 5%. This would again imply that managers with inside debt holdings are more likely to engage in earnings management by 6.9%, ceteris paribus. The coefficient on CEO age is 0.01 and is statistically significant at the 5% level. This result suggests that the likelihood of an older CEO engaging in earnings management increases by 1%, ceteris paribus. When looking at the interaction term of CEO inside debt holdings and CEO age, the coefficient is – 0.001 and is statistically significant at the 5% level. This finding suggests that the likelihood that a manager engages in earning management decreases by 0.1% if the CEO is older and has inside debt holdings in the firm. In other words, if you take the effect of age into account the likelihood of managers engaging in earnings management is smaller. This finding is consistent with my hypothesis that the financial reporting quality of a firm increases when the CEO is older and has inside debt holdings in the firm.
6.2 Secondary regression
Another regression was conducted to look at the relation between financial reporting quality, CEO age, and inside debt holdings. In this regression the proxy for financial reporting quality is the absolute value of the discretionary accruals. In this way, the effect of CEO inside debt holdings and CEO age are taken into account for both upwards and downwards earnings management.
The results of the secondary regression are presented in table 7. First the regression was run without taking into account the effect of CEO age and with the absolute value of the discretionary accruals as the dependent variable. These results are shown in Panel A of table 7. Next, the regression was run taking into account the effect of CEO age. Again, the dependent variable is the absolute value of the discretionary accruals. The results of this regression are presented in panel B of table 7.
