1
Master Thesis:
CEO inside debt and accounting
conservatism
Student Name: Yahui Ge Student Number: 10824790 Date: 17 August, 2015 Word count: 12,057 Supervisor: Dr. Bo Qin
MSc Accountancy & Control, specialization Accountancy Faculty of Economics and Business, University of Amsterdam
2
Abstract
Inside debt compensation is a substantial part of executive compensation but yet generates little concentration. Motivated by this situation, I adopt CEO-to-firm debt/equity ratio and investigate the relationship between CEO inside debt holdings and debt holder’s demand for accounting conservatism, measured by conditional and unconditional conservatism. Since CEOs with large inside debt holdings concern more about firm liquidation value and would act more like debt holders, they would report conservative accounting and are less likely to expropriate debt holders, thus decreases debt holder’s demand for accounting conservatism. Consistent with this notion, based on empirical results from 758 U.S. firms from 2007 to 2010, I find evidence of a significant negative association between CEO inside debt holdings and conditional conservatism but an insignificant relation regarding to unconditional conservatism. In addition, I find evidence that there’s a weakly significant relationship between the two elements when CEO personal debt-to-equity ratio exceeds firm leverage ratio, given that more than half of my sample CEOs have equity-based incentives rather than debt-based incentives. Overall, my study provides empirical evidence that inside debt is significantly negative related to conditional conservatism but not sensitive to unconditional conservatism.
3
Statement of Originality
This document is written by Yahui Ge, who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Acknowledgements
Firstly, I would like to thank my thesis supervisor Mr. Bo Qin, for all the academic and empirical advices he gave to me, and for always be so patient to resolve my thesis questions. Really appreciate his encourage and guidance through my thesis writing process.
Secondly, I want to express my gratitude to my classmates and friends, they gave me a lot of encourage and gave me some suggestions to improve my thesis quality.
4
Content
1.
Introduction ... 5
2. Literature review and hypothesis development ... 8
2.1 Agency cost of debt and inside debt ... 8
2.2Agency cost of debt and accounting conservatism ... 13
2.3. Hypothesis development ... 16
3. Data and variable measurement ... 17
3.1 Variable measurement ... 17
3.1.1 Inside debt measurement ... 17
3.1.2 Accounting conservatism ... 19
3.1.3 Empirical Model ... 23
3.2 Sample selection ... 26
4. Empirical results ... 27
4.1 Descriptive Statistics ... 27
4.2 Mul tivariate analysis ... 32
4.2.1 Conditional conservatism and CEO inside debt ... 32
4.2.2 Unconditional conservatism and CEO inside debt incentives .... 38
5. Conclusion and Discussion ... 42
5
1. Introduction
Substantial academic literatures research on the role of executive compensation and the implication of such research derives from the agency theory, which illustrates the conflict interests between principle and agent. In a corporate setting, agency problems arise due to the different incentives of shareholders and debt holders (principle) and managements (agent). As a result, executive compensations are designed in order to mitigate the agency cost arising due to management’s personal interests.
An executive compensation normally includes salary, bonus, restricted stocks and options, pensions and deferred compensations (Han and Pan, 2015). Researches about executive compensation heavily focus on equity-based compensations, which mainly are stock and option holdings. Recently, especially after the study of Sundaram and Yermack (2007) who take the first empirical step to investigate the role of CEO debt-based compensation, a small but growing stream of academic literatures has emerged on CEO inside debt compensations.
Inside debt holdings consist of defined benefit pensions and deferred compensations, and are widely used in executive compensation. Such compensation represent unfunded and unsecured liabilities firms owed to executives, rendering those executives at the default risk similar to that faced by outside creditors (Sundaram and Yermack, 2007; Edmans and Liu, 2011).
I investigate whether there’s a negative relationship between CEO inside debt holdings and firm accounting conservatism, in the form of asymmetric timeliness of earnings and unconditional conservatism.
My motivation for this study is as following: firstly, existing theories and literatures on executive compensation mainly focus on the use of cash and equity to maximize shareholder value. Yet, only a few literatures investigate the role of debt instruments in executive compensation. Secondly, even among the studies on executive inside debt, according to my knowledge, there’s scarce research on the direct relationship between CEO inside debt and accounting conservatism. Accounting conservatism is an efficient
6
mechanism that could mitigate conflicts between management and various contracting parties, and could reduce potential litigation by outside parties (Watts (2003a). To the extent that inside debt compensation shapes executive’s debt and equity incentives, it is meaningful to examine whether, and to what extent, such compensation would have impact on accounting conservatism. Thirdly, prior to 2006, there’s very limited information on executive compensation, especially on pension benefits and deferred compensations. In July, 2006, a comprehensive Securities and Exchange Commission (SEC) disclosure reform greatly improved the transparency of deferred compensations, thus giving a practical platform to sufficiently investigate the role of inside debt compensation. Overall speaking, there’s little research done in this field focus on the direct link between CEO inside debt and accounting conservatism, therefore my study is motivated by this to fill this gap in the literature by investigating the relationship between CEO inside debt holdings and conservatism measured by asymmetric timeliness of earnings and accrual-based unconditional conservatism.
By using data on CEO inside debt holdings from 2007 to 2010, I collect 3032 firm observations with complete financial and stock information and CEO compensation data. Then following the fundamental notion of Jensen and Meckling (1976), as well as two widely acknowledged empirical literatures (Sundaram and Yermack, 2007; Wei and Yermack, 2011), I adopt the CEO-to-firm debt/equity ratio as my variable of interest, and I also construct an dummy variable which is set equal to one when the ratio exceeds one, and zero otherwise. Accounting conservatism is estimated by two measurements: for conditional conservatism, I use the asymmetric timeliness of earnings based on Basu’s (1997) model; for unconditional conservatism, I use a composite variable which equals to the standardized value of market-to-book ratio minus accruals minus the skewness of earnings.
Prior literature documents that debt holder’s demand for conservatism is alleviated when there are less share holder-debt holder conflicts or less information asymmetry problems (Haw et al., 2014; Ahmed et al., 2002; Lafond and Roychowdhury, 2008). Given that higher CEO-to-firm debt/equity ratio indicates greater debt-based incentives,
7
I expect a negative relationship between this ratio and debt holder’s demand for accounting conservatism. As expected, I find strong evidence that CEO inside debt holdings are significantly negative with accounting conservatism measured by asymmetric timeliness of earnings, indicating that when CEOs hold large amount of inside debt, bad news is incorporated into earnings more timely than good news, that is, there’s greater asymmetric timeliness on recognition of gain and loss. When captured by accrual-based measurement of unconditional conservatism, accounting conservatism is insignificant with inside debt. But when CEO personal ratio exceeds firm leverage ratio, CEO inside debt becomes weakly significant and negative with unconditional conservatism. In addition, conditional conservatism is also negative but weakly significant with CEO inside debt when the CEO-to-firm debt/equity ratio exceeds one. This might because 26% of my sample firms’ CEO incentive ratio exceeds one, thus the majority of sample CEOs have equity-based incentives rather than debt-based incentives. In general, conditional conservatism is significant and negative with CEO inside debt holdings while unconditional conservatism is not. Such relationship become weakly significant when CEO-to-firm debt/equity ratio exceeds one.
My study contributes to current literature in the following ways. Firstly, my study adds new insight into the components of CEO compensations. Up to now, the literature on CEO compensation primarily focuses on equity holdings. My study provides empirical evidence regarding to another CEO compensation component - CEO debt instruments, and find that CEO inside debt holdings can significantly affect firm’s accounting performance. Secondly, although there’s an emerging stream research on CEO debt-based compensation, they mainly focus on the impact of CEO inside debt on firm’s financial performance such as the reduced volatility of future stock returns, reduced cost of debt and less risky investment behavior (Wei and Yermack, 2011; Cassell et al., 2012). Different from these researches, my study focuses on agency cost of debt and investigate debt holder’s demand for accounting conservatism, which could have implications beyond prior work. Thirdly, my research contributes to the field of
8
accounting conservatism by estimating both the conditional conservatism and unconditional conservatism. The results indicate that CEO inside debt has substantial impact on both the two forms of accounting conservatism, and that when CEO-to-firm debt/equity ratio exceeds one, such relation becomes weak. Lastly, my research contributes to the agency theory which is the starting point of my hypothesis development. Specifically, large CEO inside debt holdings induce CEOs perform not only as share holders but also as debt holders. Under this circumstance, the interests of shareholders, debt holders and firm’s managements should be better aligned with each other, thus mitigating the agency problem.
The remaining part of this paper is organized as follows:
Section 2 represents the literature review and my hypothesis development; Section 3 introduces my variable measurement, empirical model and the sample selection procedure; Section 4 outlines the descriptive statistics and the multivariate analysis results; the final section concludes the paper.
2. Literature review and hypothesis development
2.1 Agency cost of debt and inside debt
Existing theories and literatures on executive compensation mainly focus on the use of cash and equity to maximize shareholder value. Yet, only a few literatures investigate the role of debt instruments in executive compensation.
Agency theory describes that conflict interests between the principle (i.e. shareholders or creditors) and the agent (i.e. management) due to the separation of ownership and control. For instance, conflict interests arise when shareholders and management have different tolerance towards risks and thus prefer different risky projects. Under such circumstances, agency cost of debt arises since managers would change firm investment policy or financial projects to transfer wealth to them. To protect own benefits, debt holders would require stricter covenants and ask for a
9
higher charge, which increases the agency cost of debt.
Inside debt compensation generally consists of two categories: defined benefit pension and deferred compensation. Defined benefit pension is a type of retirement plan where a specified monthly benefit on retirement is predetermined by a formula (such as the employee’s terminal earnings) based on the employee’s earnings history and duration of employment. Such plan is 'defined' since the benefit formula is defined and known in advance. Deferred compensation, in contrast, is an agreement in which a portion of the employee’s earned income is payable at a later date. By doing so, the employee actually makes an investment decision to lend money to his or her firm, for instance, at a fixed rate of return, or in the company’s stock. Given that inside debt is an unsecured and unfunded compensation, if the firm goes bankrupt, CEOs are less likely to receive future benefits, thereby CEOs with large inside debt holdings would concern more about firm liquidation value and their interests are more in line with debt holders. Therefore, inside debt compensation contributes to the decline of risk taking, thereby exposing managers to the same default risk and insolvency treatment as outside creditors (Sundaram and Yermack, 2007;Wei and Yermack, 2011; Cassell et al., 2012; Tung and Wang, 2012; Han and Pan, 2015).
Debt holders prefer firms to be more conservative due to the asymmetric payoff regarding to the net assets (Watts, 2003). That is, at the maturity of the loan, debt holders will not receive any additional benefit if the firm’s net assets exceed the face value of the debt. However, if the value of net assets is less than the loan contract payment, debt holders would suffer a loss. So, there exist agency conflicts between managers and debt holders, since equity-holding managers have incentives to reallocate benefits from debt holders to shareholders, at the expense of increasing the default risk of the firm. This is called “risk-shifting” or “asset substitution” problem, which refers to the agency cost of debt (Jensen and Meckling, 1976).
Similarly, a recent theoretical paper of Edmans and Liu (2011) indicates that inside debt is a superior remedy to the agency cost of debt than the equity-based compensations, because inside debt renders managers to consider the payoff in
10
bankruptcy, which mainly refers to the liquidation value, but not just the incidence of bankruptcy. Thus, they conclude that inside debt can improve executive effort as well as reduce the agency cost of debt.
Sundaram and Yermack (2007) take the first step to empirically investigate the role of debt-based compensation for CEOs and its implications. They show that pension benefits represent a significant component of overall CEO compensations; by using firm’s distance-to-default to measure firm risk, this paper conclude that CEOs with larger inside debt tend to manage their firms less risky, especially when their personal debt-to-equity ratio exceeds the firm debt-to-equity ratio, in order to reduce the likelihood of default and the risks to their own pension values. The study of Sundaram and Yermack (2007) is restricted to the analysis of pensions only, and they claim that other forms of inside debt such as deferred compensation is usually far less than the value of pensions, thus the omission of deferred compensation may not be serious. However, such claim could lead to an incomplete conclusion towards the role of inside debt.
A following study of Cen (2011) extends Sundaram and Yermack’s (2007) work by examining both pensions and deferred compensation using new disclosed information produced by a regulatory change of a comprehensive disclosure reform of Securities and Exchange Commission (SEC) in 2006. This study makes a contribution to the understanding of deferred compensation as prior literatures were not able to study this kind of inside debt due to lack of data. The study concludes that deferred compensation is about the same size as pensions in terms of its market value to CEO’s total compensation: the average deferred compensation represents 6.2% of a CEO’s total compensation, while pension represents 5.5%.
Wei and Yermack (2011) examine stockholder and debt holder reactions to the initial reports of CEO inside debts following 2006 SEC disclosure reform. They apply the CEO’s “relative incentive ratio” which captures how a unit increase in the value of the firm increases the value of CEO’s inside debt-to-equity ratio, scaled by a similar measure of how the same unit increase in the firm value affects the firm’s external debt
11
versus equity. This study documents that after the mandatory disclosure of CEO inside debt, bond price rise, equity prices fall, and the volatility of both securities drops for firms whose CEOs have sizeable inside debt holdings (i.e., CEO’s relative incentive ratio above 1). The results show that investors react significantly when a company discloses information that its CEO holds significant amount of inside debt relative to his equity claim. They conclude that these market reactions further imply that CEO’s inside debt holdings could reduce firm’s risk and transfer value from equity to debt.
Following the theoretical prediction of Jensen and Meckling (1976), who suggest that executive compensation with inside debt in a ratio similar to the firm’s capital structure could minimize debt holder losses, Cassell et al. (2012) adopt four measurements to test CEO inside debt holdings, and demonstrates that inside debt holdings can lead CEOs to take action less risky. Since a large inside debt holding might induce CEOs to keep firm value stably and reduce the likelihood of insolvency, as well as reduces the CEO’s risk-taking incentives derived from equity-based compensation, large CEO inside debt holdings elicits a reduction in firm riskiness. This is illustrated by the analysis of the riskiness of firm investment and financial policies. First of all, regarding to investment decisions, R&D expenditures are more risky compared to other investments, given that the future benefits of R&D investments are of highly uncertainty. Hence, there’s a negative association between CEO inside debt holdings and R&D expenditure, because CEOs with large inside debt holdings prefer less risky investments. Second, CEOs with large inside debt holdings tend to adopt diversified operations, hold more liquid assets and/or decrease the leverage of the firm, because such policies reduce CEOs’ exposure to bankruptcy. In summary, Cassell et al. (2012) concludes that CEOs with large inside debt holdings tend to chose investment and financial policies that are less risky.
Tung and Wang (2012) focus on the role of bank CEOs’ inside debt compensation in the global financial crisis. Banks are special institutions with high leverage and severe agency problems, which magnify bank manager’s incentives of increasing risk taking and debt-equity conflicts. Banks are subject to government regulation, and it is
12
government regulators who are responsible for constraining bank risk taking. However, after the global crisis, a number of commentators proposed to reform the structure of banker’s incentive compensation in order to contain the excessive risk taking that regulation failed to prevent. This paper finds that the presence of bank CEOs’ inside debt holdings preceding the Crisis is associated with better stock returns and accounting return on assets. The inclusion of inside debt into bank CEO’s personal portfolios would give them incentives to avoid excessive risk taking, since inside debt givers managers a stake in the firm’s liquidation value. The paper has important policy implications for banking regulation, suggesting that inside debt compensation for bank executives may offer an effective tool for risk regulation.
Han and Pan (2015) document a positive relation between relative CEO leverage ratio (CEO’s debt/equity ratio scaled by firm’s debt/equity ratio) and investment cash flow sensitivity, that is, the more that CEO’s debt-to-equity ratio exceeds firm’s debt-to-equity ratio, the higher investment cash flow sensitivity is. Since inside debt holdings affect CEO incentives and further affect CEO’s economic decisions, inside debt can be related to investment-cash flow sensitivity.
The paper formalizes two opposite hypothesis. The first one is risk aversion hypothesis: since managers with high risk aversion tend to rely more heavily on internal cash flow when they finance investments, CEO inside debt holdings induce managers to be more like debt holders than shareholders, thus they are averse to excess external debt to avoid default risk. Hence, inside debt decreases manager’s demand for external financing, and thus exaggerates investment-cash flow sensitivity. The other one is cost of debt hypothesis: when holding large inside debt, CEOs’ interests are more align with debt holders, thus mitigating debt holder’s concern of risk-shifting problem. This study suggests when CEOs hold large inside debt, there’re lower requirement of bank loan covenants and promised yields. Therefore, inside debt holdings decrease the cost of debt financing, thereby declines internal cash flow sensitivity. Overall, the empirical results of this paper are more in favor of the risk aversion hypothesis, which indicates that risk adverse CEOs mainly rely on internal cash flows when financing, even if the
13
cost of external financing is low. The reason behind this argument is that inside debt encourage managers to act more like bond holders who are risk-adverse to avoid default risk.
2.2Agency cost of debt and accounting conservatism
Accounting conservatism is defined as “the on average understatement of the book value of net assets relative to their market value” (Beaver and Ryan, 2005), and is divided into two types: the first type is unconditional conservatism, which is ex ante or news independent. Examples include immediate expensing of costs of most internally generated intangibles, accelerated depreciation of property, plant and equipment (Beaver and Ryan, 2005). The other one is conditional conservatism (ex post or news dependent), which is interpreted as “capturing accountants’ tendency to require a higher degree of verification for recognizing good news than bad news in financial statements” (Basu, 1997). Under such circumstances, economic loss is recognized more timely than gain in earnings. Hence, the different respond time in recognition gain and loss leads to differences between bad news and good news periods in the timeliness of earnings.
Basu (1997) points out the debt contracting role is the origin of accounting conservatism. Managers could take advantages of their informed position and have incentives to hide any negative information that would adversely affect their compensation. To avoid the decline of firm value if managers exploit firm wealth for their own sake, debt holders demand timely recognition of bad news relative to good news, because debt holders’ payoff is more sensitive to the decline than the increase of firm value (Basu, 1997; Watts, 2003a).
Several empirical literatures have examined accounting conservatism from the aspect of debt holder’s demand (Ahmed et al., 2002; Zhang, 2008; Aier et al., 2014; Haw et al., 2014).
Ahmed et al. (2002) investigate the role of accounting conservatism on bondholder-shareholder conflicts and find that accounting conservatism is efficient to mitigate such conflicts over dividend policy and in reducing firm’s debt of costs. The
14
conflicts emerge because conservatism accounting leads to lower reported earnings used in debt contracts, which would constrain dividends payout, so it is less likely that firms make excessive dividends payment to shareholders under higher accounting conservatism. Thus, to compensate for the reduced risk of excessive dividends payment, bondholders might require a lower rate of return, which means a lower cost of debt (i.e., better debt rating).
Zhang (2008) gives insight on direct contracting benefits of accounting conservatism to lenders and borrowers and finds that lenders offer lower interest rates to more conservative borrower. The intuition for this argument is that debt holders have downside risk but no upside potential in debt contracting, so they appreciate mechanisms that mitigate the downside risk. Meanwhile, accounting conservatism requires a more timely recognition of bad news than good news, providing debt holders more timely signals of default risk. Such accelerated signals of covenant violation benefit lenders ex post so that they can take protective actions in advance. At the same time, debt holders are likely to require a lower interest rate in exchange.
It is widely held that conservatism is largely affected by debt holder’s demand. However, there’s in general no causal link established between debt holder demand and conservatism. By examining changes following a 1991 Delaware ruling which expanded the scope of director’s fiduciary duties to include creditors for near insolvent firms, Aier et al. (2014) show that, these firms report more conservatively after this ruling. This is because that directors of near insolvent firms concern about firm insolvency status, and might take action to induce managers to meet the requirement of debt holders, so that mitigate the increased litigation risk due to director’s new fiduciary duties to creditors. Since this natural shock significantly mitigates endogeneity concerns and makes it easier to attribute changes in conservatism to debt holder’s demand, this study provides strong evidence to support the casual link between debt holders’ demand and accounting conservatism.
Haw et al. (2014) investigate the relationship between public debt financing and accounting conservatism. Because private firms are often with high information
15
asymmetry and high credit risk, their agency problems with bondholders are considered to be more severe. On the one hand, while bondholders only have limited access to bond issuer private information, it’s difficult for them to reduce the agency costs, so agency problems are more severe for private firms with public debts than those with private loans. Bondholders often rely on public information to protect their benefits. As a result, they require bond issuer to apply conservatism accounting to gain greater information benefits. On the other hand, bond issuers have the incentive to meet such conservatism since they can get a lower cost of debt in exchange. As information asymmetry problem is more serious for public debts than for private loans, conservatism accounting can reduce such problem to a greater degree. Therefore, private firms with public debt exhibit a higher degree of conditional conservatism, which suggests that firms take advantage of accounting conservatism to retain favorable terms in debt contracting. Further, this paper also shows that private firms with high information asymmetry and high credit risk show a greater increase in conservatism after the initial bond issue. Also, such level of conservatism is maintained even after the bond issue period, indicating that such firms have incentives to maintain a high degree of conservatism to establish reputation for high-quality financial reporting to get a lower interest rates in later period.
Apart from the association between cost of debt and accounting conservatism, the relationship between executive compensation and accounting conservatism has also been examined a lot. As mentioned above, conservatism accounting constrains management’s upward valuation of net assets (Watts, 2003). Based on such argument, O’connell (2006) examines the association between accounting conservatism and CEO compensations of UK earnings, and finds evidence that since accounting conservatism results in a lower correlation between earnings and returns in good news firm-years, it leads to an expanded compensation role for earnings in such years. Earnings are widely used to aggregate information in executive compensation, while earnings do not always reflect such information timely. Hence, such deficiency leads to the contracting role of earnings, and the contracting role of earnings is enhanced when the correlation between
16
contemporaneous earnings and returns is lower. This study supports the view that there’s a significant association between accounting conservatism and CEO compensation, and indicates that accounting conservatism affects CEO compensations, that is, CEO compensation exhibits a stronger (weaker) association with accounting earnings for good (bad) news. However, whether CEO compensations play a role in accounting conservatism remain unclear in this study.
Lafond and Roychowdhury (2008) document a negative association between managerial ownership and accounting conservatism measured by asymmetric timeliness of earnings. They claim that the separation of ownership and control gives rise to agency problems, under which circumstance the manager would have incentives to transfer wealth to themselves rather than making efforts to optimally manage the firm. Therefore, tying compensation with conservative reporting penalizes managers for their value-reducing actions, and defers the compensation until firm’s targets are achieved. Thus as managerial ownership increases, there’s less asymmetry for recognizing good news as gains than bad news as losses, thus the demand for accounting conservatism decreases.
2.3. Hypothesis development
The inside debt literature mainly focuses on the role of inside debt on executive’s risk behaviors, and the conclusion can be drawn from prior literature is that executives with large inside debt holdings consider more about firm’s liquidation value and act more like debt holders than equity holders, thus they are more likely to be less risky, and to reduce firm’s default risk.
Regarding to accounting conservatism, as can be drawn from relevant literature mentioned above, when there’s prominent contracting role of accounting, debt holders demand less conservatism as they know managers have less incentive to transfer wealth from debt holders to themselves, along with reduced cost of debt.
To summarize, accounting conservatism help debt holders to protect their investments and to detect the signals of firm’s covenant violation as conservative
17
reporting exhibits a more timely recognition of bad news than good news. Meanwhile, according to the theoretical and empirical evidences, because managers with significant inside debt compensations have large unfunded and unsecured claims against firms, they are more likely to concern about debt holders and less likely to expropriate debt holders’ wealth. Under such circumstance, debt holders would require a lower charge in exchange, such as a lower interest rate, therefore the cost of debt would decrease. In a word, debt holders’ demand for accounting conservatism decreases when CEOs have large inside debt holdings.
Thus, I make the following hypothesis:
H1: There is a negative association between CEO inside debt holdings and conditional accounting conservatism.
3. Data and variable measurement
3.1 Variable measurement
3.1.1 Inside debt measurement
According to Jensen and Meckling (1976), when CEO’s debt-to-equity ratio resemble firm’s debt-to-equity ratio, managers would act more like debt holders, and make economic decisions that mitigate the agency cost of debt. Following recent empirical studies (Sundaram and Yermack, 2007; Wei and Yermack, 2011; Cassell et al., 2012), I adopt two proxies to measure CEO inside debt incentives (INCENTIVES).
My first measurement is the CEO-to-firm debt/equity ratio. CEO inside debt compensation is an unsecured and unfunded obligation of the firm, thereby exposing CEOs to the same default risk as outside creditors. When CEOs’ debt and equity structure resembles or exceeds firms’ capital structure, debt holders demand less conservative reporting. Prior literature normally adopts the CEO to firm debt-to-equity ratio as one of the CEO inside debt measurement, which equals the CEO’s inside debt/equity ratio scaled by firm’s debt/equity ratio:
18
CEO to firm debt-to-equity ratio (CFR) = (CEO inside debt/ CEO equity) / (firm debt/ firm equity)
Following prior literature (Sundaram and Yermack, 2007; Wei and Yermack, 2011; Cassell et al., 2012), I define the variable measurement as following: CEO inside debt is the sum of the present value defined benefits plan (PENSION_VALUE_TOT) and deferred compensation (DEFER_BALANCE) which can be found in Execucomp. The value of CEO equity includes stock and stock options. CEO stock holdings are calculated by multiplying the number of shares held by CEO (SHROWN_EXCL_OPTS) by the stock price at the firm’s fiscal year end (PRCC_F), and the value of CEO stock options equals to the sum of Estimated Value of In-the-Money Unexercised Exercisable Options (OPT_UNEX_EXER_EST_VAL) and Estimated Value Of In-the-Money Unexercised Unexercisable Options (OPT_UNEX_UNEXER_EST_VAL), which represent the options outstanding, and the data are all available in Execucomp. Moreover, the firm’s debt is measured as the book value of its total liabilities (LT), and the firm’s equity is measured as the market value of total equity (calculated by multiplying the number of common shares outstanding by the stock price at firm’s fiscal year end).
Then, following Wei and Yermack (2011), Cassell et al.(2012) and Han and Pan (2015), I adopt an indicator variable, CEO-to-firm debt/equity ratio >1 (CFR>1), which equals to one if this ratio exceeds one and zero otherwise. The underlying idea is that the incentive effects of CEO inside debt holdings are particularly acute when the CEO’s debt-to-equity ratio exceeds that of the firm (Jensen and Meckling, 1975). Thereby when the CEO-to-firm debt/equity ratio equals 1, the CEO holds the mix compensation that is in line with the firm’s capital structure, so he has no incentive to transfer value from debt to equity or vice versa. However, if the ratio exceeds 1, the CEO would have incentive in line with the debt holders, decreasing debt holders’ demand for accounting conservatism. If the ratio is below 1, CEOs are more likely to expropriate debt holders, so debt holders’ demand for accounting conservatism would be higher.
19
3.1.2 Accounting conservatism
3.1.2.1 Conditional conservatism
I assess the degree of conditional accounting conservatism by applying the asymmetric timeliness measurement of Basu (1997) model, which focuses on the asymmetric recognition of accounting gains and losses. Under conditional conservatism, accounting losses are incorporated into earnings in a timelier manner than economic gains because accountants tend to recognize gains more conservatively, that is, only when the requirements are completely met and when there’s enough evidence to show an accounting gain.
I begin with Basu’s (1997) model, which is:
NIt = α0 + α1 DRt + α2 RETt + α3 RETt*DRt + 𝜀t (1)
In model (1), the dependent variable NIt is net income before extraordinary items (IB) reported in year t, deflated by the book value of total assets (AT) of this fiscal year; RETt represents the 12-month stock return of current fiscal year, and is calculated by the product of twelve months’ monthly return minus one; DRt is a dummy variable equal to 1 if RET is negative and 0, otherwise. Therefore, DRt = 1 represents negative return and is a proxy for bad news, while DRt= 0 indicates positive returns and is a proxy for good news. Moreover, the coefficient α3 of the interaction variable Rt*DRt captures the asymmetric timeliness.
Then combined with my study on the relationship between CEO inside debt and conditional conservatism, I expand Basu’s (1997) model with my test variable and some control variables which have impact on a firm’s conditional conservatism. The complete equation to test conditional accounting conservatism is as following:
20
+ α5RET*INCENTIVES + α6DRt*INCENTIVES
+ α7RET*DRt*INCENTIVES + α8SIZE + α9LEV + α10PROF + α11GROWTH + α12LITIG + 𝜀t, (2)
NIt: Net income before extraordinary items reported in year t, deflated by the book value of total assets of this fiscal year;
RETt: The 12-month stock return ending three months after the fiscal year end; DRt: A dummy variable equal to 1 if RET is negative and 0, otherwise; INCENTIVES: One of two variables: CEO-to-firm debt/equity ratio (CFR) or
CEO-to-firm debt/equity ratio>1 (CFR>1); SIZE: Firm size, which is the natural log of total assets;
LEV: Leverage, which is total long term liabilities divided by total assets; PROF: Profitability, which is net cash flow from operations divided by total
assets;
GROWTH: Sales growth, which is the ratio of total sales in year t to total sales in year t-1;
LITIG: Litigation risk, which equals one if a firm belongs to high-litigation industries(SIC code 2833–2836, 3570–3577, 7370–7374, 3600–3674, and 5200–5961), and zero otherwise.
3.1.2.2 Unconditional conservatism
At present, there are various measurements of accounting unconditional conservatism. For instance, Zhang (2008) and Francis et.al (2014) use the market-to-book ratio as one of the proxies of unconditional conservatism; Givoly and Hayn (2000) establish the cumulative non-operating accruals and the time-series skewness of earnings, which have been widely used in the accounting literature. Consequently, based on prior studies (Givoly and Hayn, 2000; Zhang, 2008; Ahmed and Duellman, 2013; Francis et.al, 2014), I adopt three components of my unconditional conservatism measurement: market-to-book ratio (MTB), accrual-based
21
measurement (ACCRUALS) and earnings skewness (SKEWNESS).
The first component is the market-to-book ratio, which captures the understatement of net assets relative to the market value of firm equity. This measurement derives from a theoretical framework developed by Feltham and Ohlson (1995) in which they argue that accounting is conservative when the market value of a firm’s equity is greater than that of book value. Thus, the higher the ratio, the more conservative is the firm. Market-to-book ratio is widely used as a proxy for conservatism, and it is normally easy to calculate. However, one thing should be noted that the changes of market-to-book ratio might be caused by the changes in the market expectations of a firm’s sales growth rather than the changes of accounting conservatism (Givoly and Hayn, 2000). Thus, I include the sales growth as my control variable in the regressions, which will be discussed in detail in the Section 3.1.3 Empirical Model part.
The second component, Accruals, is the income before extra-ordinary items less net cash flows from operations plus depreciation expense deflated by total assets, and averaged over a 3-year period centered on year t, multiplying by negative one(Ahmed and Duellman, 2013). Hence, higher (positive) accruals indicate greater unconditional conservatism of the firm. The underlying idea is that conservatism accounting always recognize bad news quickly while delay the recognition of good news, thus resulting in persistently negative accruals. Meanwhile, accruals tend to reverse within one or two year period. Thus, persistently negative accruals over a long period indicate accounting conservatism (Givoly and Hayn, 2000).
The third component, Skewness, is the difference between skewness of cash flow from operation and skewness of earnings (income before extraordinary items) over last five years, and all variables are deflated by total assets of the fiscal year end (Ahmed and Duellman, 2013). According to Givoly and Hayn(2000), conservatism leads to a timely and complete recognition of bad news and a delayed and gradual recognition of good news, resulting in a negatively skewed earnings distribution, so that the higher the Skewness, the more conservative is the firm.
22
the standard deviation of one, and then construct a composite variable UNCON (MTB – ACCRUALS – SKEWNESS) as the overall measurement of unconditional conservatism. The higher the value of UNCON, the greater unconditional conservatism is the firm.
To test the relationship between CEO inside debt holdings and unconditional conservatism, I use the following empirical model:
UNCON = 𝛽0 + 𝛽1INCENTIVES + 𝛽2 SIZE + 𝛽3 LEV
+ 𝛽4PROF + 𝛽5 GROWTH + 𝛽6LITIG + 𝜀t (3)
UNCON: Equals to MTB – ACCRUALS – SKEWNESS, and all three components are standardized with the mean equals to zero and the standard deviation equals to one;
INCENTIVES: one of two variables: CEO-to-firm debt/equity ratio (CFR) or CEO-to-firm debt/equity ratio>1 (CFR>1);
SIZE: Firm size, which is the natural log of total assets;
LEV: Leverage, which is total long term liabilities divided by total assets; PROF: Profitability, which is net cash flow from operations divided by total
assets;
GROWTH: Sales growth, which is the ratio of total sales in year t to total sales in year t-1;
LITIG: Litigation risk, which equals one if a firm belongs to high-litigation industries(SIC code 2833–2836, 3570–3577, 7370–7374, 3600–3674, and 5200–5961), and zero otherwise.
To summarize, my study uses two measurements to test accounting conservatism, including one for conditional conservatism and one composite variable for unconditional conservatism:
1. The asymmetric timeliness measurement of Basu’s (1997) model to capture conditional conservatism;
23
2. The composite variable UNCON = MTB – ACCRUALS – SKEWNESS, where MTB is market-to-book ratio, ACCRUALS is an accrual-based measurement which is the three-year average calculated as the income before extraordinary items less net cash flows from operations plus depreciation expense deflated by total assets and multiply by negative one, and SKEWNESS is the difference between skewness of cash flow from operation and skewness of earnings, deflated by total assets. All three components are standardized with the mean equals to zero and the standard deviation equals to one.
3.1.3 Empirical Model
My study investigates the relationship between CEO inside debt holdings and firm accounting conservatism, in the form of conditional conservatism and unconditional conservatism. Specifically, I estimate the following empirical models using ordinary least regression (OLS):
To test the relationship between CEO inside debt holdings and conditional conservatism, I apply the Basu’s (1997) model which has been outlined in Section 2.1.2.1 Conditional conservatism part:
NIt = α0 + α1RETt + α2DRt + α3INCENTIVES + α4RETt*DRt + α5RET*INCENTIVES + α6DRt*INCENTIVES
+ α7RET*DRt*INCENTIVES + α8SIZE + α9LEV + α10PROF + α11GROWTH + α12LITIG + 𝜀t, (2)
To test the relationship between CEO inside debt holdings and unconditional conservatism, I use the following empirical model:
UNCON = 𝛽0 + 𝛽1INCENTIVES + 𝛽2 SIZE + 𝛽3 LEV
+ 𝛽4PROF + 𝛽5 GROWTH + 𝛽6LITIG + 𝜀t (3)
24
NIt: Net income before extraordinary items reported in year t, deflated by the book value of total assets of this fiscal year;
RETt: The 12-month stock return ending three months after the fiscal year end; DRt: A dummy variable equal to 1 if RET is negative and 0, otherwise; UNCON: Equals to MTB – ACCRUALS – SKEWNESS, and all three components
are standardized with the mean equals to zero and the standard deviation equals to one;
INCENTIVES: One of two variables: CEO-to-firm debt/equity ratio (CFR) or CEO-to-firm debt/equity ratio>1 (CFR>1);
SIZE: Firm size, which is the natural log of total assets;
LEV: Leverage, which is total long term liabilities divided by total assets; PROF: Profitability, which is net cash flow from operations divided by total
assets;
GROWTH: Sales growth, which is the ratio of total sales in year t to total sales in year t-1;
LITIG: Litigation risk, which equals one if a firm belongs to high-litigation industries(SIC code 2833–2836, 3570–3577, 7370–7374, 3600–3674, and 5200–5961), and zero otherwise.
My variable of interest is the CEO-to-firm debt/equity ratio. Since the inside debtis an unsecured and unfunded compensation that exposes CEOs to the same default risk and insolvency treatment as outside creditors, debt holders’ demand for accounting conservatism is likely to be lower when CEOs own large inside debt holdings.
Specifically, in Model (2), the coefficient α1 on RET reflects the timeliness of earnings to good news, α4 on RET*DR captures the incremental timeliness of earnings to recognize bad news. In addition, the coefficient α5 reflect the impact of CEO inside debt holdings on how quickly earnings report good news, and the most important coefficient α7 on RET*DR*INCENTIVES measures how CEO inside debt holdings impact the incremental timeliness of earnings respond to bad news. To the extent that CEOs with large inside debt compensations prefer delaying the recognition of good
25
news and inducing timelier recognition of bad news, CEO inside debt holdings would lead to the decrease of debt holders’ demand for conditional conservatism. Accordingly, I expect α5 to be negative due to delaying recognition of good news, and α7 to be negative due to greater asymmetry in timely earnings recognition. Adversely, if CEO inside debt increases demand for conditional accounting conservatism, then α5 and α7 are expected to be positive.
In Model (3), the coefficient 𝛽1 on INCENTIVES indicates how CEO inside debt holdings has impact on unconditional accounting conservatism. H1 predicts the demand for accounting conservatism decreases when CEOs own inside debt compensation, which anticipates 𝛽1 to be negative.
In estimating the empirical models, I use five control variables which are found to be associated with my dependent variable accounting conservatism.
Firstly, I include firm size (SIZE) to control for the effect of firm size on accounting conservatism. On the one hand, according to information asymmetry hypothesis, because larger firms have less information asymmetry issues, there’s less demand for conservative accounting for larger firms (LaFond and Watts, 2008). On the other hand, political cost hypothesis suggests that larger firms are subject to greater government scrutiny (Watts and Zimmerman, 1978), thus there should be a positive relation between firm size and accounting conservatism. Therefore I do not predict a sign on the coefficient between firm size and accounting conservatism.
Secondly, I include Leverage as a control variable because leverage has a positive impact on conservatism. Prior literature suggests there’s higher demand for conservatism in firms with high leverage, since such firms have severe conflicts between debt holder and equity holder (Ahmed et al., 2002).
Thirdly, I control for profitability as Ahmed et al. (2002) find evidence that due to costs occurred when applying conservatism accounting, profitable firms tend to use more conservative accounting than low-profitability firms. Thus I expect a positive relationship here.
26
level of accruals and in turn affect measurements such as ACCRUALS and SKEWNESS (Ahmed and Duellman, 2007).Sales growth is positively related to mart-to-book ratio, given that market has increasing expectation on high growth firms. However, sales growth is expected to be negatively related to accruals and skewness, since high growth firms are likely to be less conservative to pursue better accounting and stock performance, which affects the level of accruals such as inventory and receivables. Thus, I do not predict a sign on the coefficient for GROWTH.
Fifthly, I control for litigation risk (LITIG) and expect a positive coefficient to capture the greater demand for accounting conservatism for firms in higher litigation risk industries (Watts, 2003).
3.2 Sample selection
Prior 2006, firms were not required to disclose executive deferred compensation plans, however nearly every firm has a deferred compensation plan for its executives, so the data on this kind of inside debt compensation was of great importance but extremely limited (Sundaram and Yermack, 2007). As earlier noted, in July 2006, the Securities and Exchange Commission (SEC) adopted new regulation regarding to more complete and transparent disclosure of executives and director compensations, since when the deferred compensation data became available. Therefore I choose the sample period from 2007 to 2010.
I begin by search the universe database in Execucomp database, and identify firms with complete and consecutive data on the four-year CEO inside debt compensation. These firms cover the current S&P 1500 plus companies that were part of the S&P 1500 are still trading. Because I focus on the study of CEO inside debt holdings, I excluded firm observations for which both the CEO pension benefits and CEO deferred compensation data are missing, and this procedure yields 8177 firm observations. Then I deleted firms without consecutive four-year data on CEO inside debt compensation, and then I merged these firm codes with Compustat and Center for Research on Security Prices (CRSP) database to generate relevant financial and stock information
27
data, and excluded firms without continuous financial or stock data from 2007 to 2010. After that, I got 3588 observations. I further excluded firms that have missing data on operating cash flow and depreciation and amortization expense. After these procedures, my final sample consists of 758 firms with 3032 firm observations.
4. Empirical results
4.1 Descriptive Statistics
Panel A of Table 1 describes my sample selected procedures, Panel B of Table 1 outlines the industry classification based on SIC codes across my sample firms and indicates that the sample firms are from a wide range of industries. Among these industries, manufacturing industry represents 46.57% of the total firm observations.
Table 1
Sample selection and sample distribution
Panel A: Sample reconciliation
Observations with no missing data on CEO inside debt data (2007-2010) 8,177 less: Observations that both pension benefits and deferred compensation are zero (3391) less: Observations without consecutive data from 2007 to 2010 (638) less: Observations with missing data to calculate accounting conservatism (616) less: Observations without complete data on Cash Flow from Operations and Depreciation (500) Final sample of firm observations from 2007 to 2010 3032 Numbers of firms in the final sample 758
Panel B: Sample Industry Distribution
Industry SIC Observations Percent
Agriculture, Forestry and Fishing 0100-0999 1 0.13% Agricultural, Production- Livestock and Animal 1000-1999 63 8.31%
28
Specialties
Manufacturing 2000-3999 353 46.57% Transportation, Communication, Electric, Gas and
Sanitary
4000-4999 111 14.64%
Wholesale and Retail Trade 5000-5999 93 12.27% Finance, Insurance and Real Estate 6000-6799 60 7.92%
Services 7000-8999 76 10.03%
Public administration 9000-9999 1 0.13%
Total 758 100.00%
Table 2 show the sample distribution by year according to two main variables: stock return and CEO-to-firm debt/equity ratio. Specifically, the number of observations with negative stock return is 1313 out of 3032, representing 43.30% of the total firm observations, and is mainly distributed in year 2007 and 2008 under the general background of 2008 financial crisis. Meanwhile, regarding to CEO-to-firm ratio, most CEOs’ personal leverage ratio doesn’t exceed firm leverage ratio, and observations with CEO-to-firm debt/equity ratio>1 account for nearly30% (795 out of 3032) of total CEO-to-firm debt/equity ratios, and are evenly distributed from 2007 to 2010.
Table 2
Sample distribution by Stock Return and CEO-to-firm debt/equity ratio by year
Observations Percent
Positive Stock Return
2007 323 18.79%
2008 73 4.25%
2009 694 40.37%
2010 629 36.59%
Total Observations with Positive Stock Return 1719 100%
Negative Stock Return
29
2008 685 52.17%
2009 64 4.87%
2010 129 9.82%
Total Observations with Negative Stock Return 1313 100% Total Stock Return Observations 3032
CEO-to-firm debt/equity ratio>1
2007 178 22.39%
2008 206 25.91%
2009 211 26.54%
2010 200 25.16%
Total Observations with CEO-to-firm debt/equity ratio>1 795 100% Total Stock Return Observations 3032
Table 3 provides descriptive statistics for the variables used in my analysis. Specifically, the mean CEO inside debt holdings is 8.7 million, suggesting CEOs hold substantial amount of inside debt compensations. The average CEO stock and option holdings in my sample are 115 million, which are far higher than the average amount of CEO inside debt holdings. Also, the mean (median) CEO debt/equity ratio is 0.52 (0.14), which means for the majority of my sample firms, CEO inside debt is less than CEO equity holdings. This result is consistent with recent empirical literatures, for instance, Cassell et al. (2012) document an average (median) CEO debt-to-equity ratio of 0.40 (0.15). In terms of my variable of interest, the mean (median) of CEO-to-firm debt/equity ratio and CEO-to-firm debt/equity ratio >1 is 1.10 (0.21) and 0.13 (0.00) respectively. The results also indicates the CEO-to-firm debt/equity ratio is highly right-skewed.The results also show that many of my sample CEOs has equity-based incentives. Moreover, my sample firms have an average (median) value of assets of 16.34 (3.45) million, market-to-book ratio of 2.32 (1.88), leverage ratio of 0.21 (0.19), profitability of 0.11 (0.10) and sales growth of 1.07 (1.05).
30
Table 3 Descriptive statistics
MEAN Std.dev Q1 Median Q3
NI 0.0257 0.2478 0.0348 0.0588 0.0789 RET 0.1784 0.8387 -0.2280 0.0765 0.3833 DR 0.4330 0.4956 0 0 1.0000 MTB 2.3242 27.8812 1.2638 1.8751 2.9658 ACCRUALS 0.0239 0.0559 -0.0021 0.0128 0.0333 SKEWNESS 0.4261 1.4024 -0.5290 0.3971 1.3416 UNCON 0.0000 1.8753 -0.7492 0.2190 0.9839
CEO inside debt holdings (thousands)
8,697.3789 15,797.0384 938.0923 3,533.3800 9,910.9463
CEO stock holdings (thousands)
103,494.4568 1,091,756.1480 6,658.0929 18,506.0517 49,622.1727
CEO option holdings (thousands)
11,561.2106 32,810.6553 157.6618 2,496.5910 10,172.0280
CEO equity holdings (thousands)
115,055.6673 1,106,820.8480 8,197.5271 23,578.5070 63,046.7045
CEO D/E ratio 0.5260 3.2219 0.0451 0.1430 0.3933 CEO to firm D/E ratio
(CFR)
1.1013 22.7409 0.0693 0.2074 0.5413
CEO to firm D/E ratio>1 (CFR > 1) 0.1375 0.3445 0 0 1.0000 Assets(million) 16,344.4333 63,136.9670 1,229.2490 3,449.8405 11,510.7010 Log(Size) (SIZE) 3.5886 0.6827 3.0896 3.5378 4.0611 Leverage (LEV) 0.2067 0.1586 0.0913 0.1969 0.2954 Profitability (PROF) 0.1053 0.0735 0.0631 0.0985 0.1431
31 Sales Growth (GROWTH) 1.0749 1.6975 0.9481 1.0461 1.1299 Litigation Risk (LITIG) 0.1900 0.3923 0 0 0
Table 4 outlines the correlation relationship among CEO inside debt measures, firm-specific conservatism measures and control variables, over the period 2007 – 2010. In particular, the correlation between CFR and NI is significantly negative, and so is the correlation between CFR>1 and NI. Since lower earnings represents for higher conservatism due to the delay the recognition of positive returns and recognize negative returns more quickly, thus according to the correlation matrix, CFR and CFR > 1 are positively related with accounting conservatism, which is opposite with my hypothesis, thus remains to be investigated in following multivariate analysis part. Furthermore, as expected, CFR and CFR>1 is significantly negative with UNCON, indicating that CEO inside debt compensation decreases the demand for unconditional accounting conservatism. With respect to control variables, the correlations on SIZE and the two conservatism measures (NI and UNCON) are significantly positive at 1% level, supporting the political cost hypothesis which claims that larger firms are subject to greater political cost, thus greater demand for accounting conservatism. The correlation between NI and LEV is significantly negative, indicating higher leverage leads to lower earnings thus higher conservatism, which is consistent with Ahmed et al. (2002)’s claim that conservatism and leverage are positively related. However, the correlation between UNCON and LEV is significantly negative. As for profitability, it is significant and positive with NI, but negative with UNCON. In addition, the correlations on GROWTH and LIT are insignificant. NI and RET is significantly positive, which suggests earnings contain a part of information that is conveyed by stock returns. Overall, the control variables do have significant effect on conservatism measures, so it is important to control for these differences in the multivariate analyses.
32
Correlation matrix between variables
Variable RET NI UNCON CFR CFR>1 SIZE LEV PROF GROWTH LITIG
RET 1 NI 0.2732 *** 1 UNCON -0.0320 * 0.2229 *** 1 CFR -0.0219 ** -0.040 ** -0.054 ** 1 CFR>1 -0.0414 ** -0.0263 ** -0.0303 ** 0.0937 *** 1 SIZE -0.0587 *** 0.0475 *** 0.0545 *** -0.0448 ** -0.0433 ** 1 LEV -0.0180 -0.1370 *** -0.0551 *** -0.0501 *** -0.1133 *** 0.1573 *** 1 PROF 0.1311 *** 0.1814 *** -0.0960 *** -0.0031 0.0809 *** -0.0956 *** -0.0849 *** 1 GROWTH -0.0253 0.0156 0.0120 -0.0045 -0.0169 0.0115 -0.0182 -0.0248 1 LITIG 0.0736 -0.0492 -0.0428 -0.0323 0.0552 -0.0490 *** -0.1237 *** 0.1189 *** -0.0046 1
* indicates significance at 10% level; ** indicates significance at 5% level; *** indicates significance at 1% level.
4.2 Mul tivariate analysis
4.2.1 Conditional conservatism and CEO inside debt
33
asymmetric timeliness measurement based on Basu’s (1997) Model (Model (2)). I conducted two OLS regressions: column (1) is “CFR only” which represents the regression without control variables, and column (2) “CFR with controls” means regression with control variables. All p-values are based on two-tailed significance test. In column (1) “CFR only”, the coefficient on negative stock returns, RET*DR, is positive (0.2560) and significant at 1% level, indicating that bad news incorporates into earnings in a timelier way than good news. With respect to the main explanatory variable, the coefficient on RET*DR*CFR is negative (-0.0202) and significant at 1% level, providing strong evidence that CEO inside is negatively related to the asymmetric timeliness of gain and loss recognition when CEOs hold larger amount of inside debt compensation.
Column (2) “CFR with controls” shows the regression results after adding several firm-level control variables. Consistent with the results documents in column (1) in Table 5, the coefficient on RET*DR*CFR is negative (-0.0199) and significant at 1% level.
In context of control variables, for brevity, I only report the interaction results. As expected, the coefficient on RET*DR*SIZE (firm size) is positive (0.0357) and significant (p< 0.001), supporting the notion that large firms likely face large political costs that induces them to use more conservative accounting (Watts and Zimmerman, 1978). Consistent with prior literature (Ahmed et al., 2002), the coefficient on RET*DR*LEV (leverage) is positive (0.7755) and significant (p < 0.01). Besides, RET*DR* PROF (profitability) is significantly positive (0.4771, p < 0.01) with accounting conservatism, which provides evidence that low-profitability firms are less conservative due to costs occurred when applying conservatism accounting (Ahmed et al., 2002). I find no significant relation between sales growth (litigation risk) and the asymmetric timeliness of earnings, since the coefficients on RET*DR*GROWTH (RET*DR*LITIG) are insignificant. Overall, the R-squared increases from 0.1075 to 0.1467, suggesting there’s a stronger relationship with the conditional conservatism after adding control variables. Thus, my control variables add certain value into this
34 model.
Table 5
Asymmetric timeliness and CEO-to-firm debt/equity ratio
Expected Sign CFR only (1) CFR with controls (2) RET + 0.0674*** (10.30) 0 .0664*** (10.30) DR ? 0.0800*** (6.18) 0 .0650*** (5.05) CFR - -0.0091*** (-3.21) -0.0091*** (-3.26) RET*DR + 0.2560*** (8.36) 0.1426*** (3.41) RET*CFR - 0.0203*** (3.96) 0.0204*** (4.06) DR*CFR + 0.0092*** (3.19) 0.0093*** (3.27) RET*DR*CFR - -0.0202*** (-3.78) -0.0199*** (-3.78) RET*DR*SIZE ? 0.0357*** (3.03) RET*DR*LEV + 0.7755*** (8.49) RET*DR*PROF + 0.4771*** (6.44) RET*DR*GROWTH ? -0.0008 (-0.25) RET*DR*LITIG + 0.0385
35 (0.41) R-squared F-statistics 0.1075 2.11*** 0.1429 2.08***
The sample consists of 758 firms and 3032 observations from 2007 to 2010. Column (1) represents the regression without control variables, and column (2) means regression with control variables. t-Values are in parentheses, and the significance is based on two-tailed test. */**/*** indicates significant at 1/5/10% level respectively.
This table represents the OLS regression results based on Basu’s (1997) model:
NIt=α0+α1RETt+α2DRt+α3INCENTIVES+α4RETt*DRt+α5RET*INCENTIVES+α6DRt*INCE
NTIVES+α7RET*DRt*INCENTIVES+α8SIZE+α9LEV+α10PROF+α11GROWTH+ α12LITIG
+ 𝜀t. The dependent variable is NI, which stands for net income before extraordinary items,
deflated by the book value of total assets; RETt is the 12-month stock return of current fiscal year; DR is a dummy variable equal to 1 if RET is negative and 0, otherwise; INCENTIVES is one of two variables: CEO-to-firm debt/equity ratio (CFR) or CEO-to-firm debt/equity ratio>1 (CFR>1); SIZE is the natural log of total assets; LEV is total long term liabilities divided by total assets; PROF is net cash flow from operations divided by total assets;
GROWTH is the ratio of total sales in year t to total sales in year t-1; LITIG equals one if a firm belongs to high-litigation industries(SIC code 2833–2836, 3570–3577, 7370–7374,
3600–3674, and 5200–5961), and zero otherwise.
Table 6 shows the regression results when CEO debt/equity ratio exceeds firm debt/equity ratio, in which circumstances CEOs are more likely to act as debt-holders and firms are more likely to report conservative accounting.
Column (1) is “CFR>1 only” which represents the regression without control variables, and column (2) “CFR>1 with controls” means regression with control variables. All p-values are based on two-tailed significance test. Specifically, the R-squared are 0.1145 and 0.1433 respectively. In column (1), the coefficient on RET*DR is positive (0.0607) but insignificant, but the coefficient on RET*DR*CFR>1 is negative (-0.1477) and moderately significant (p <0.05).
36
Column (2) reports results with control variables. After adding several control variables, the coefficient on the coefficient RET*DR is still insignificant, and the coefficient on RET*DR*CFR>1 is negative (-0.1359) and weakly significant (p < 0.10). Taken together, the regression results provide weak evidence that when CEO debt/equity ratio exceeds firm debt/equity ratio, bad news is recognized into earnings more quickly than good news. The relationship between CFR>1 and asymmetric timeliness measurement is not so strong since only 26% of my sample firms’ CEO-to-firm debt/equity ratio exceeds one. This means the majority of my sample CEOs have equity-based incentives rather than debt-based incentives, so lack of variance in CFR>1 may contribute to the insignificant results.
Regarding to the control variables, the correlations on firm size (RET*DR*SIZE), leverage (RET*DR*LEV) and profitability (RET*DR*PROF) are significant and positive, on sales growth (RET*DR*GROWTH) is significant and negative, but the correlations on litigation risk (LITIG) is insignificant. This is consistent with the results from Table 5 and the reasons are illustrated there.
To summarize regression results in Table 5 and Table 6, there’s significant evidence suggesting that CEO-to-firm debt/equity ratio is negatively related to asymmetric- timeliness-based conditional conservatism, and the relationship holds but becomes weak when the ratio exceeds one. Overall, the regression results support Hypothesis 1, that is, there’s a negative association between CEO inside debt holdings and accounting conservatism.
Table 6
Asymmetric timeliness and CEO-to-firm debt/equity ratio > 1
Expected Sign CFR>1 only (1) CFR>1 with controls (2) RET + 0.0605*** (8.88) 0.0585*** (8.72)
37 DR ? 0.0428*** (2.87) 0.0304** (2.05) CFR>1 - -0.0696*** (-3.77) -0.0669*** (-3.68) RET*DR + 0.0607 (1.64) -0.0094 (-0.21) RET*CFR>1 - 0.2017*** (11.38) 0.2032*** (11.65) DR*CFR>1 + 0.0780** (2.64) 0.0718** (2.47) RET*DR*CFR>1 - -0.1477 ** (-2.00) -0.1359* (-1.87) RET*DR*SIZE ? 0.0200* (1.66) RET*DR*LEV + 0.7702*** (7.61) RET*DR*PROF + 0.7173*** (3.69) RET*DR*GROWTH ? -0.0127*** (-3.73) RET*DR*LITIG + 0.0804 (1.51) R-squared F-statistics 0.1145 2.17*** 0.1433 2.14***
The sample consists of 758 firms and 3032 observations from 2007 to 2010. Column (1) represents the regression without control variables, and column (2) means regression with control variables. t-Values are in parentheses, and the significance is based on two-tailed test. */**/*** indicates significant at 1/5/10% level respectively.
38
NIt=α0+α1RETt+α2DRt+α3INCENTIVES+α4RETt*DRt+α5RET*INCENTIVES+α6DRt*INCE
NTIVES+α7RET*DRt*INCENTIVES+α8SIZE+α9LEV+α10PROF+α11GROWTH+ α12LITIG
+ 𝜀t. The dependent variable is NI, which stands for net income before extraordinary items,
deflated by the book value of total assets; RETt is the 12-month stock return of current fiscal year; DR is a dummy variable equal to 1 if RET is negative and 0, otherwise; INCENTIVES is one of two variables: CEO-to-firm debt/equity ratio (CFR) or CEO-to-firm debt/equity ratio>1 (CFR>1); SIZE is the natural log of total assets; LEV is total long term liabilities divided by total assets; PROF is net cash flow from operations divided by total assets;
GROWTH is the ratio of total sales in year t to total sales in year t-1; LITIG equals one if a firm belongs to high-litigation industries(SIC code 2833–2836, 3570–3577, 7370–7374,
3600–3674, and 5200–5961), and zero otherwise.
4.2.2 Unconditional conservatism and CEO inside debt
incentives
Besides conditional conservatism, I test the relationship between unconditional conservatism and CEO inside debt holdings. As explained before, I use accrual-based measurement UNCON which equals to the standardized value of MTB minus Accruals minus Skewness. Because there’s only one independent variable, I skip the “CFR only” test and directly show the results of CFR with control variables, presented in column (3) in Table 7.
The coefficient on CFR is negative (-0.0100) but insignificant with UNCON, which demonstrates that there’s no significant relation between CEO inside debt holdings and unconditional conservatism (for example, the understatement of net assets, measured by my accrual-based measurement.
With respect to control variables, the coefficient on firm size (SIZE) is positive (0.1395) but insignificant, and the coefficient on leverage (LEV) is significant (p < 0.01) and positive (0.9492). Contrary to the expectation, profitability (PROF) is significantly
