The capital structure and variable compensation
Does the capital structure effect the variable executive compensation of
Dutch listed companies?
Name: F.M. Rootjes Student number: 10741917 Supervisor: dr E. Zhivotova Coordinator: dr P.J.P.M. Versijp
Study program: Bsc Economics and Business Specialization: Finance and Organization Amount of EC’s: 12
Date: 26-06-2018
Abstract
In this research the relation between leverage and variable compensation will be investigated. Prior research shows different explanations between these two components, which makes it interesting to investigate. Chemmenur et al (2013) found an increase in variable compensation when the leverage increases, while Ortiz-Molina (2007) found a negative causal effect. Both researches where done with US companies. No research has been done with respect to the Netherlands. This research intends to find out if there is a causal relationship between leverage and the variable compensation in the Netherlands. No conclusion has been found. The data provided insignificant results for the leverage ratio in the instrumental variable regressions. Further research is needed to test if there is a
Statement of Originality
This document is written by F.M. Rootjes who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document are 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.
Table of
Contents
Abstract 1 Introduction 4 Theoretical Framework 6 Hypothesis 13 Methodology 17 Results 21 Conclusion 28 Discussion 29 Appendix 30 References 40Introduction
In 2018 ING wanted to increase the salary of Ralf Hamers with 50%. However, due to social pressure, ING did not implement this increase of salary. Ralf Hamers has been the CEO of ING since October 2013. The salary increase would consist of shares on top of his base salary.
In the Netherlands, the bonus culture is seen as one of the reasons the financial debt crisis exacerbated (Fahlenbrach & Stulz, 2011). Due to incentive packages managers increase their risk-taking (Uhde, 2016). Excessive risk-taking led to risky loans provided by the banks and an increase in defaults. When the reserves were too small to back up the defaults, the banks needed help from the government to save the outstanding deposits. The question now is: what drives a company to an increase of salary of their CEO with 50% in one year? Research has been done with respect to the performance of a company related to the variable compensation of their managers (Core et all., 1999). The question now is: what other factors influence the CEO-pay. Current changing capital structures might explain the increasing variable compensation packages. To investigate this, the causal relationship needs to be detected. The issuance of debt increases for companies (Centrale Bureau voor de Statistiek, 2016). The choice of companies to issue debt might explain the increase in variable compensation. Chemmanur et al (2013) found a positive relationship between these two variables in the US.
Jensen and Meckling (1976) argue that debt reduces the cash flow and, therefore, reduces the agency costs. Debt and variable compensation are both used to reduce the agency conflict between managers and shareholders. Research in America showed a positive effect between variable compensation and leverage (Chemmanur et al, 2013). This research intends to find out if the interest-bearing debt of a company affects executive payments. In the Netherlands this affect has not been investigated before. Therefore, it is interesting to find out if this effect also holds in the Netherlands. The research will be done with the help of
structure is the leverage ratio. The regression model contains a variance of different control variables, like the age of the CEO and the size of the company. Furthermore, another suggestion for the compensation differences will be discussed.
The expectations of the research will be discussed under what is called ‘the hypothesis’. For this ‘hypothesis’, economic theories are used to explain what to expect. The economic theories that can be associated with the research are Managerial Power Theory, Optimal Contracting Theory, agency costs and bankruptcy costs. These theories will be further explained in the ‘theoretical framework’. After the’ hypothesis’ and the ‘theoretical framework’, the ‘method’ will be discussed. In the ‘method’, the regression model and the data that have been used will be explained. After the ‘method’ and results the ‘conclusion’ will follow. In the ‘conclusion’ the theory that corresponds with the outcome will be further discussed. The managerial power theory and the agency costs expect a negative relationship where the bankruptcy costs expect a positive relationship.
Theoretical Framework
Agency costs
Agency costs come from the separation of ownership and control (Fama and Jensen, 1983). The separation of ownership and control takes place when a firm becomes bigger. The owner cannot handle everything on his own and needs to hire a manager. The manager is then in control of the company and thus controls the assets of the owner. The owner wants the manager to act in his best interest because he is rational. This means he is only satisfied with the largest output. The manager also wants to act in his own interest. The differences between the interest of the owner and the manager are called the agency costs. These agency costs are mainly caused by information problems. Information problems can be divided in to two parts, asymmetric information problems and moral hazard. The asymmetric information problem arises before the event happens and moral hazard happens after the event took place. Compensation packages are mainly determined by the asymmetric information problem. The shareholders want to make sure that the manager is motivated enough before the event. In this case, the event is tasks the manager fulfills for the company.
There are various types of agency costs. The three main agency conflicts are: shareholder-manager, manager-creditors and creditors-shareholder. Agency costs of debt arise between creditors and borrowers (Guay, 2008). Creditors want to make sure that their capital will be returned. Guay argues that managers not always act in the best interest of the creditors. For example, managers can decide to pay-out dividends to shareholders which increases firm leverages. Another problem that can arise is when managers can earn variable compensation. To earn a variable compensation, certain targets have to be met. The variable compensation increases the risk-taking of the managers (Uhde, 2016). Excessive risk-taking increases the risk for the creditors. If the creditor anticipates these potential events, he wants to get compensated for or get more insurance. The creditor can, for example, increase the amount of interest, demand more information of investment decisions or monitor the
Besides the agency costs of debt, there are the agency costs of equity. The agency costs of equity arise between the shareholders and managers. The managers might want to act in their own interest by spending the resources. One of these phenomena is called empire building. The manager spends the outstanding resources on overs while these take-overs are not in the best interest of the company. Humphery-Jenner (2012) provides evidence for managerial entrenchment. He argues that the directive encourages the growth of assets and cash which provides evidence for empire building. The shareholders want to anticipate for these hidden action problems by linking the managerial compensation to the firm performances. This is because the shareholders cannot monitor the managers constantly due to the information problems. The shareholders provide a compensation package with bonuses and shares if positive targets for the company will be reached. The costs the company makes to limit unwanted actions of the manager can be seen as agency costs of equity (Jensen and Meckling, 1976).
The total agency costs can be divided into three parts according to Jensen and Meckling (1976, p. 306). The costs to monitor the manager, the bonding expenditures by the manager and the residual loss. In this research, the main focus will be on the bonding expenditures the firm makes. The bonding costs are the costs the firm makes to the manager to ensure he acts in the interest of the company and thus the shareholders. Therefore, the agency costs are a reason for the shareholders to be compensated with high amounts of variable compensation.
Agency costs and debt
Zhang (2009) argues that both debt and incentive compensation have the potential to mitigate the agency costs. Zhang found evidence that firms adjust their capital structure and option compensation to reduce the cash flow and therefore reduce the potential agency costs. Also, the model provided significant results for a negative interaction between option compensation and the issuance of debt. This result suggests that higher debt levels and variable compensation can be seen as substitutes to reduce the agency costs of a firm. In addition, Brockman et al (2010) argued that short-term debt mitigates the agency costs of
Managerial power
Another theory which tries to explain the compensation packages of the managers is the managerial power theory. This theory implies that the high amounts of variable compensation of managers are caused by the managers themselves (Bebchuk et al, 2002). The managers can use their power to influence the compensation contracts. The managerial power theory contradicts the optimal contract. The optimal contract is the one which minimizes the agency costs. The limitation of this theory is that it does not encounter the influence the managers have on the board.
The Capital structure
This research focuses on the relationship between the agency costs, the capital structure and the indirect bankruptcy costs of a company. Modigliani & Miller (1976) argue that with perfect capital markets the firm value is not affected. This means the optimal capital structure is solely determined by market imperfections (Berk & DeMarzo, p. 605). Some market imperfections are taxes, agency costs, and financial distress costs. The bankruptcy costs are a form of financial distress costs. The bankruptcy costs which are investigated in this research are the human costs of bankruptcy. Berk & DeMarzo (2014) argue that the form of agency costs depends on the amount of debt. Figure 1 from Berk & DeMarzo (2014) below illustrates this in a graphical format where D* is the optimal debt level. For example, companies with high cash flows and low debt level may encourage managers to waste it. Contrary, companies with excessive debt levels can stimulate managers to take excessive risks or suffer from financial distress costs. This phenome is called the tradeoff between the optimal debt level. The optimal debt level is varying between companies due to the fact that firms have different characteristics and lays where the benefits minus costs of debt are maximized. The optimal value of debt is therefore difficult to determine.
Figure 1
Developments
Ernst and Young (2017) released a report about executive remuneration in the Netherlands. It contains information about the payments the Dutch listed companies make to their top managers. It shows an increase in payments from 2014 to 2016. Their base salary increased approximately with 2%. Ernst and Young argue that this increase is caused by market adjustments. Therefore, the base salary increase is not something to investigate. The short-term incentives show growing numbers for CEOs from 2014 till 2016. The long-term incentives increased as well, especially for companies in the AMX. The CEO targets for long-term incentives increased with respect to the median and average level. Also, the use of shareholding requirements for CEOs have increased. Ernst and Young conclude with the results that the compensation has become more performance-based. This is mainly caused by the perception that the executive remuneration is excessive compared to the financial results. The increase in variable compensation can be caused by many factors. For example, the increase in company size or the increase in leverage. Therefore, it is interesting to determine the causal effect between the leverage and the variable compensation, to see if the leverage can be a potential reason.
Dong et al 2010) provided evidence that managers with higher volatility of equity-based compensation choose debt rather than equity to finance operating activities. The argument provided for this finding relates to the dilution effect. Managers do not like to finance with equity because this increases the amount shares. An increase in the number of shares reduces the earnings per share for the existing shareholders. Therefore, a manager who has a large number of shares prefers to finance with debt for their own wealth.
The Ernst & Young report shows an increase in long-term incentive compensation targets from the year 2015 to 2016 for CEOs. Dong et al (2010) also provide evidence which suggests that equity-based compensation increases the risk-taking of managers. The company and thus the shareholders are responsible for the drawbacks of the excessive risk-taking from the manager. Therefore, companies should take into account the potential drawbacks to find the optimal level of incentive compensation. They also suggest that the preference for debt due to the increase in equity-based compensation is related to the excessive risk-taking of the managers. Managers overleverage the company to increase the equity-based compensation and thus their own wealth. Lefebvre & Vieider (2014) provide further evidence with respect to stock options. They show that stock options rather than stocks do not mitigate the excessive risk-taking problem. The risk aversion did not increase when stock options were issued instead of stocks.
Chemmanur et al (2013) found out that leverage and compensation are positively related. They argue that the bankruptcy costs of high leverage ratios are implemented in the costs of human capital. Chemmanur et al (2013) argues that the direct bankruptcy costs cannot explain the low leverage ratios companies attain. Berk et al (2010) implies that a risk averse employee wants to be compensated for the expected bankruptcy costs because the employee needs to insure himself for potential bankruptcy. These costs are called indirect bankryptcy costs.
Cultural differences
The Dutch results could differ from the American research. Potential differences can be explained by the difference in culture. To analyze the country differences between the United States of America and the Netherlands, the Hofstede framework is used. It measures country aspects with the use of six dimensions. In table 1 the dimensions of both countries are showed in a histogram (Hofstede Insights, 2017). Power distance measures the rate of acceptance towards inequalities in the society. Both countries score low. Individualism is measured in the rate of interdependence a society contains. Both countries score high. The third column contains the results of the masculinity for both countries. A country which has a high level of masculinity is indicated with high levels of competitions and success. The US scores high and the Netherlands scores low. This difference can be an indication of the higher compensation levels in the US. The average variable compensation of the US sample is 5.095.309,64 dollar. In the Netherlands the average is lower: 819.841,50 euro. *1,05 (Dollar/Euro) = 860.833 dollar. The next dimension is uncertainty avoidance, it measures how the society reacts on the unknown events. For example, countries with high uncertainty avoidance usually have more rules. Both the US and the Netherlands score approximately in the middle. A high level of long-term orientation of a country show increased reactions on changes in traditions, education and saving for the future. The Netherlands score high while the US score low. The last dimension is indulgence. A country with a high level of indulgence is known for its way to allow people to enjoy themselves. Both countries score high. For further research it would be interesting to investigate if these cultural differences influence the wage gap. This research focusses on the relationship between leverage and variable compensation.
Table 1
Blue = The Netherlands Purple = The United States
Hypothesis
In this research, the relation between the leverage ratio and the variable compensation will be investigated. Jensen and Meckling (1976) argue that the amount of debt brings discipline to the managers. The fixed payments of rent which need to be paid every month reduce the free cash flow and therefore reduce the opportunity for managers to waste the cash. Another explanation for the relationship between the amount of debt and variable compensation is the bankruptcy costs. The idea is that the threat of bankruptcy costs induces managers to act with more effort for the company (Grossman and Hart, 1982). The threat of bankruptcy incites creditors to monitor the firm more intensely and to do so because of a fear of failing and a loss of jobs. John and John (1993) and Ortiz-Molina (2007) provided evidence that a higher leverage ratio reduces the variable compensation sensitivity. In addition, Zhang (2009) found evidence that both debt and variable compensation reduce the agency costs and can therefore be seen as substitutes. A higher leverage ratio therefore reduces the agency costs which can result in a reduction of variable compensation. These relations imply a negative relationship between the leverage ratio and the variable compensation.
From the view of the managerial power theory, the variable compensation is explained due to the managerial entrenchment. The variable compensation is excessive compared to the agency costs. Therefore, the amount of debt is not a substitute for the agency costs. However, Berger et al (1997) found out that companies with high managerial entrenchment tend to avoid debt. They argue therefore that CEOs tend to avoid debt, which is due to the fact that leverage increased when the company faced a reduction of managerial entrenchment. From this point of view, the expectation will be negative. An increase in leverage will be therefore be associated with a reduction in variable compensation.
The indirect bankruptcy costs theory expects a positive relationship between the leverage and the variable compensation (Chemmanur et al, 2013. Due to the risk-aversion of the employee, the company has to compensate the manager for potential bankruptcy. Berk et al (2010) adds that the indirect bankruptcy costs of human capital are large enough to
increase the leverage ratio. Therefore, firms do not want to increase their leverage when the costs exceed the tax benefits.
Data
The data set consists of the listed companies of the AEX, AMX and AScX and the NYSE. These indexes are all part of the same company, namely NYSE Euronext. The datasets can therefore be compared better. For these companies not all the required data for the research are available on WRDS. The databases which acquire the information are Bureau van Dijk for the Dutch data and Compustat for the American Data. The datasets need the information about the CEO’s and financials. The total amount of companies counts to 1150 plus 75 is 1225. The database provided only 954 and 53 data points. The American data needed to be collected from different sources, namely American data and Execucomp-monthly data. These two lists did not complement each other and the missing data points needed to be sorted out. The remaining data points combined consisted of 872 points. Due to endogeneity problems an instrumental variable is needed for the regression model. The remaining 872 data points where therefore reduced to 720 companies due to extra information of the total market value. Since this research is conducted in early 2018, it is possible that some of the data for 2017 are not yet available. Therefore, 2016 has the latest available data and will be subject to this research. All the companies should have the Annual Report on their website, otherwise it can be obtained at the Chamber of Commerce. The information related to the specific company can be calculated with the use of the year reports or on WRDS.
Some of the Dutch companies had their financial values mentioned in euros or pounds. To make sure all the values are the same (in dollars), the amounts are recalculated in dollars by using the exchange rates of the European Central Bank.
The total remuneration package will be used, which includes: base salary, short-term compensation (cash bonus), long-term compensation (stock options), pension payments and other compensation. Some of the data are available on WRDS. This database includes Bureau van Dijk, which has data on companies in Europe that can be used. The descriptive statistics of the American and Dutch data are shown in appendix 7 and 9.
consolidated with the compensation packages, the remainder consisted of 5441 points. The data set was reduced due to the information about CEO bonusses and equity-based compensation values. The descriptive statistics of the sample are in appendix 2.
Methodology
Variables
For the regression model, the executive compensation is divided into two parts. The fixed and variable compensation. The fixed compensation consists of the base salary plus the pension payments and other fixed compensations like a company car. The variable compensation consists of STI and LTI. STI is usually an annual cash bonus based on performances of the company and the manager. LTI is meant for the long-term incentives. Usually it consists of performance shares and/or options. Only the incentive compensation is used for this research. The incentive compensation is variable during the year and depends on the results of the company and manager. The variable compensation is determined before the start of the year and the control variables are therefore also taken one year before.
The main explanatory variable is the leverage ratio. In this research the leverage is measured in interest-bearing debt. This is the debt which influences the free cash flow due to the monthly interest payments. The interest-bearing debt will be scaled with total equity or the total market value.
The control variables are determined with the existing literature. The research on the determents of the compensation package is sometimes contradictive, therefore the control variables cannot be picked from a large pole. Yermack (1995) found out that company size is positively related with compensation. A larger company makes it more difficult for shareholders to monitor the actions, therefore it is expected that the agency costs will be higher and thus the compensation too. Murphy (1985) found out that firm performance is positively related to compensation. However, variable compensation effects profit, which makes the variable endogenous. Therefore, profit will not be inserted in the model. Lewellen et all (1987) finds a positive relationship between the age of the manager and the amount of compensation. They argue that younger managers have longer tenures for the future. Therefore, a younger manager has more to lose and the chance a younger manager engages
more to reduce the agency conflict. Another control variable which can be included is the CEO tenure. Chemmanur et al (2013) found evidence of a negative relation between the CEO tenure and equity-based compensation. The CEO tenure is measured in years from when the manager was appointed, till the corresponding year. The last control variable is the number of employees. This variable can also be seen as a proxy of firm size. As well as the total assets. More employees imply more risk because the monitoring process is larger.
Due to multicollinearity company size and number of employees can’t be in the same model. When both variables where inserted in the model, they became insignificant.
Linear regression model
The following model is now formed with X1’ as endogenous variable.
𝐿𝑁(𝑌) = 𝛽0 + (b1) ∗ 𝑋1′ + b2c ∗ 𝑋2 + 𝑏3𝑐 ∗ 𝑋3 + 𝑏4𝑐 ∗ 𝑋4 + 𝑒
Y = Variable compensation
B1 = Leverage (interest-bearing debt divided by total market value) B2c = CEO tenure
B3c = Number of employees B4c = The age of the CEO in years Z = Financial industry dummy
To test which regression method can be used, the assumptions of OLS need to be tested. The first assumption is that the observations need to be from a random sample. This assumption is fulfilled. The second assumption concerns the exogeneity. The explanatory variable leverage might be endogenous. Endogeneity is caused by a correlation between the error term and a variable. If the leverage ratio is endogenous, the coefficient will be biased. To test if the model is valid, the error term needs to be determined. The error term of the
multicollinearity. In Appendix 1B, the output of the variance inflation factor is shown. The critical value of the variance inflation factor is around four. Therefore, there is no evidence of multicollinearity. The fourth assumption is homoskedasticity. In Appendix 1C the output of the heteroskedasticity test is shown. The null hypothesis of this test validates constant variances (homoskedasticity). The test rejects the null hypothesis which means the fourth assumption is not valid. To solve this problem a robust estimation with robust standard errors will be used. For the robust estimation the variances can be non-zero and outliers will not bias the regression output. The fifth assumption is therefore automatically fulfilled, no outliers. The sixt and last assumption concerns the normal distribution of the error term. Appendix 1D shows the histogram concerns the normal distribution of the error term. Appendix 1D shows the histogram of error term. From this histogram the approximately normal distribution of the error term can be derived.
Instrumental variable regression
With instrumental variable regression, a valid instrument is needed. If the model is truly exogenous, the instrumental variable regression should generate approximately the same coefficients. Ortiz-Molina (2007) and Hsuan-Chu-lin et al (2012) summarize potential instruments. Zhang (2009) used the return on assets as an instrument. The return on assets includes endogeneity as well and can therefore not be used. The instruments with available date are taken. These variables need the requirements of valid instruments, namely the restriction and exclusion criteria. It means that the instrument needs no relationship with the depended variable and needs to be significant related with the explanatory variable. Therefore, the correlation with the variable compensation needs to be close to zero and the correlation with leverage close to one. The potential instruments are divided payments, market to book ratio, firm size, divended payments, liquicity, industry dummies and profit. A correlation matrix is shown at Appendix 4 and 5. The extra financial data reduced the
sample to 677 and further to 619.
However, these instruments cannot be substantiated with economic reasoning. There is not enough evidence that the suggested instruments will not influence the variable
compensation and leverage. In appendix 3 the correlation matrix between the industry dummies and the depended and independent variable is shown. The financial sector is known for their high debt level, due to the large investments they have to make. This variable will be used as instrument in order to mitigate the endogeneity problem. In order to conduct the two stage least square estimation, the following steps are needed.
Run the first stage linear regression. Regress with Z as an instrument and X2, X3 and X4 as independent variables and the endogenous covariant as depended variable to obtain X*.
First stage regression
𝑋’ = 𝑗0 + 𝑗1 ∗ 𝑍 + 𝑗2 ∗ 𝑋 + 𝑗3 ∗ 𝑋 + 𝑗4 ∗ 𝑋 + 𝑣
Plug fitted values into the main equation with v as a composite error term which is uncorrelated with X*, X2, X3 and X4.
𝐿𝑁(𝑌) = 𝛽0 + (b1) ∗ X ∗ +b2 ∗ 𝑋2 + 𝑏3𝑐 ∗ 𝑋3 + 𝑏4𝑐 ∗ 𝑋4 + 𝑣
Logistic regressions can cause bias in small samples (Bergtold et al, 2018). Since the Dutch sample only consists of fifty companies, the sample can be considered small. The regression model for the Dutch sample will therefore use the variable to total compensation ratio.
The American dataset is used to check if the relationship between leverage and variable compensation really exists. The dataset consists of American companies from the New York Stock Exchange in a time period of 5 years. With company financial data between 2010 and 2015 and CEO compensation between 2011 and 2016. The financial data are therefore linked to the compensation in the next year. In appendix 2 the descriptive statistics of the data set are shown.
Results
American and Dutch sample
In the table 2 the results of the regression analysis are given. The linear regression model in column one shows the model which is explained in the methodology. The Dutch data are left out for robustness check. The leverage ratio shows a positive and significant (10%) relation with the variable compensation. It suggests that when the leverage ratio increases with 10%, the variable compensation increases with 3,38%. The explanatory variables are taken at time t-1, this is 2015. In 2015 the targets for the variable compensation are determined for 2016. Table 2 provides the following results. An increase in employees with ten percent increases the amount of variable compensation with 0,905% (1%). A larger number of employees is associated with a larger firm. The CEO age coefficient is also significant, which implies there is enough evidence that the age of the CEO influences the variable compensation. An increase of the age of the CEO with one year, increases the amount of variable compensation with 1.33%. For the CEO tenure, the coefficient of the tenure in years is insignificant. In column 2 the dummy variable is included. The coefficient is negative which means if the company is Dutch the variable compensation is on average 70.56% lower. The other coefficients in the model provide evidence for the robustness check. No significant differences are detected when a new variable is included in the model. The robustness check supports the reliability of the existing coefficients.
However, there is still the problem of endogeneity. Variable compensation can be affected by leverage but also the other way around. Managers can affect the amount of leverage by making investment decisions and therefore increase the leverage. When the leverage is determined one year before the variable compensation it can be affected by the compensation in 2016 due to the proposed targets. The variable compensation will be determined in 2016 and depends on the achieved targets for the year.
Table 2
A method to deal with the reverse causality is instrumental variable regression. A valid instrument needs no correlation with the dependent variable and significant correlation with the endogenous variable.
(1) (2) (3) (4)
VARIABLES Var. Comp.
Linear Var. Comp. Linear Leverage ratio First stage Variable Comp. IV financial
Interest bearing debt 0.3379** 0.2917** -0.0370
to market value (0.13265) (0.1298) (0.0434) CEO age 0.0140** 0.0133** 0.00055 0.0139** (0.0059) (0.00572) (0.0014) (0.0053) Number of employees 0.1134*** 0.0905*** 0.0072* 0.0892*** (0.0101) (0.0100) (0.0035) (0.0125) Dutch dummy -0.7056*** -0.0713*** -0.811*** (0.1800) (0.0356) (0.170) Financial Dummy 0.0856*** (0.0262) CEO Tenure -0.0059 -0.0058 -0.0022 -0.0059 (0.0074) (0.007) (0.0016) (0.0077) Constant 6.373*** 6.577*** 0.5757*** 6.853*** (0.339) (0.330) (0.0748) (0.347) Observations 720 720 720 720 R-squared 0.201 0.226 0.039 0.126
instrumental variable regression, the endogenous variable is no longer significant, which can be explained in two ways. The instrument is not valid or the leverage ratio has no effect on the variable compensation. To test the validity of the instrument, the Wu-Hausman test statistic will be used. Appendix 5 shows the Wu-Hausman statistics and the robust score for the financial dummy. The statistics do not reject the null hypothesis which means there is enough evidence to conclude that the variables are exogenous.
Table 3
(1) (2) (3) (4) (5)
VARIABLES Variable comp. Interest bearing debt equity First stage
Variable comp. Interest bearing debt Total First stage
Variable comp.
Interest bearing debt /equity
0.119 (0.134) Interest bearing debt
/market value 0.430*** 0.464 (0.134) (0.464) CEO age 0.00762* 0.01938 0.00765 0.00546 0.00743 (0.00423) (0.01480) (0.00580) (0.005478) (0.00509) CEO tenure -0.0101** -0.02566* -0.0102** -0.00755 -0.00977* (0.00444) (0.01438) (0.00518) (0.00499) (0.00529) Number of employees 0.0136* -0.01295 0.001108** 0.0136 0.0136* (0.00807) (0.04069) (0.01064) (0.00762) (0.00762)
Marginal tax rate -0.42020** -0.1079988**
(0.18688) (0.0553796)
Constant -0.0708 -0.0198 -0.0198 0.1130712 -0.0749
(0.224) (0.250) (0.250) (0.31708) (0.198)
Observations 50 50 50 50 50
6.2 Dutch sample
Table 3 provide the results from the Dutch sample. The linear regression model in column 1 provides significant results for the leverage ratio, CEO age, CEO tenure and the number of employees. If the leverage ratio increases with ten percent the variable compensation increases with 4.30% (1%). Just like the US, a positive effect. The coefficient of the CEO age shows a positive effect. An increase of the CEO’s age by one year, increases the variable compensation with 0.762%. The number of employees indicates a positive effect, if the number of employees increases with one the variable compensation will increase with 1.36% (5%). The CEO tenure shows a negative effect. This means that if the CEO stays one year longer by the company, the variable compensation decreases with 1.01% (5%) on average.
Column 2 shows the results of the instrumental variable regression. The instrument provided by Chemmanur et al (2013) is the marginal tax rate. No literature finds a significant effect between the marginal tax rate and the variable compensation whereas Leary & Roberts (2010) find a positive effect between the marginal tax rate and leverage. Appendix 10 shows the endogeneity test statistics which does not rejects the null hypothesis of exogenous variables. Column 2 shows the results of the first stage regression with the leverage ratio defined as interest bearing debt divided by total equity as depended variable. The results of the second stage regression in column 3 do not provide significant results for the leverage ratio. In column 5 and 6 the results of the regression model with leverage ratio defined as interest bearing debt divided by the total market value are shown. In appendix 11 the endogeneity test statistics is given. The test statistics does not reject the null hypothesis of exogenous variables. The leverage is defined different for robustness check but is still insignificant.
American sample
To further investigate our hypothesis and our instrument, the results of the American data between 2010 and 2015 are given in table 4. In column 1 the results of the linear regression
regressions with the leverage ratio as depended variable. In column 3 the results of the second stage regression are given, with the financial dummy as an instrument. Appendix 12 does not reject the null hypothesis of exogeneity. The leverage ratio is still insignificant. The coefficient of the number of employees indicates a positive effect. An increase in number of employees with one corresponds with an increase in variable compensation of 25.0%. If the age of the CEO increases with one year, the variable compensation increases with 0.53%. The last control variable is the CEO tenure. An increase of the time period with one year increases the variable compensation with 1.684%.
In column 4 and 5 the leverage ratio is defined differently for robustness check. The test statistic in appendix 13 does not rejects the null hypothesis of exogenous variables (10%). The variables are therefore exogenous. The robustness check makes the results in column 2 and 3 more reliable.
Table 4
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 (1) (2) (3) (4) (5) VARIABLES Variable compensation Linear Total Interest bearing debt to total value First stage Variable comp. Second stage Interest bearing debt to equity First stage Variable compensati on Leverage ratio -0.00613*** (0.0016) 0.17172 (0.1919) 0.01028 (0.01155) Number of employees 0.24915*** 0.01048*** 0.25040*** 0.48444** 0.2472*** (0.00803) (0.0029) (0.00828) (0.1577) (0.0098) CEO age 0.00552** -0.00113* 0.00531** 0.02926 0.00481* (0.00265) (0.00065) (0.00264) (0.02366) (0.00272) Financial dummy 0.25028*** 4.17692*** (0.01030) (0.35301) CEO Tenure 0.01693*** -0.00084 0.01684*** 0.14852 0.01517*** (0.00254) (0.00072) (0.00252) (0.09922) (0.00330) Constant 5.2316*** 0.48900*** 5.1297*** -5.649259** 5.27180*** (0.1485) (0.03604) (0.16977) (2.65889) (0.169439) Observations 5,441 5,441 5,441 5,441 5,441 R-squared 0.1758 0.0456 0.1710 0.0157 0.1572
Conclusion
The agency costs theory expected a reduction in variable compensation when the leverage ratio increases. The assumption for this hypothesis was that variable compensation and leverage could be seen as substitutes. An increase in leverage would therefore reduce the agency costs and a reduction in variable compensation would follow. Jensen and Meckling (1976) confirmed that an increase in debt reduces the incentive for managers to act in their own interest. The increase in debt increases the monthly interest payments and therefore reduces the free cash flow which increases discipline by the managers. In addition, Zhang (2011) found that an increase in leverage reduces the compensation due to the fact that they both are used to reduce the agency costs. This hypothesis is not confirmed.
The managerial power theory suggested a negative relationship. Berger et al (1997) found evidence that CEO’s with more entrenchment prefer debt over equity. Therefore, an increase in debt could be associated with less managerial entrenchment and less variable compensation. This hypothesis is also not confirmed.
Chemmanur et al. (2013) found a positive relationship between debt and variable compensation in the US. They argued that an increase in debt increases the indirect bankruptcy costs of human capital. The instrumental variable regression model did not confirm this finding. The results of the linear regression model suggested a positive effect for both countries. Due to the endogeneity problem these findings are not reliable. Further research is needed to mitigate the endogeneity problem.
Hofstede provided a framework which indicated a difference in masculinity between the US and the Netherlands. For further research it might be interesting to investigate if this cultural difference influence the compensation package of a company.
Discussion
Theory predicts an increase of discipline for managers due to the reduction in cash flow when the leverage increases. The increase in leverage will induce the agency costs. Variable compensation decreases agency costs as well due to the alignment between shareholders and managers. Zhang (2009) argued therefore that leverage and agency costs are substitutes. This assumption is critical for a potential negative relationship.
The CEO compensation of the Dutch companies was hand searched. WRDS restricted the managerial compensation data for Dutch companies which made it difficult to implement a larger time period. Therefore, only the American data could be used for a larger time period. The small sample made it difficult to use the financial industry dummy as an instrument. A larger sample is needed.
Another reason for the insignificant results can be the instrument that is used. The financial dummy may not take away the complete endogeneity. The remaining endogeneity could cause biases in the results.
Appendix
1A 1B Residu~s Dutch dummy Leverage ratio Return on assets Number of employees Variable compensation Residuals 1 Dutch Dummy 0 1 Leverage ratio 0 -0.1338 1 Return on assets 0 -0.0041 -0.2283 1 Number of employees 0 -0.4794 0.1459 0.1117 1 Variable compensation 0.8798 -0.3512 0.1463 0.0533 0.4328 1Variable VIF 1/VIF
Number of employees 1.34 0.743585 Dutch dummy 1.31 0.764648 Leverage ratio 1.09 0.914659 Return on assets 1.08 0.92465 CEO age 1.01 0.994153 Mean VIF 1.17
1C
Breusch-Pagan / Cook-Weisberg test for heteroskedasticity
Ho: Constant variance
Variables: fitted values of (ln)Variable
chi2(1) = 12.82 Prob > chi2 = 0.0003 1D 0 .1 .2 .3 .4 .5 D e n si ty -4 -2 0 2 4 Residuals
Appendix 2
Descriptive statistics American data
(1) (2) (3) (4) (5)
VARIABLES N mean sd min max
Total assets 5,441 31,352 155,705 26.58 2.573e+06
Ln (total assets) 5,441 8.566 1.627 3.280 14.76
Leverage ratio 5,441 0.619 0.237 0.0622 3.793
Book value per share 5,441 20.38 22.90 -63.42 343.5
Cash and short-term investments 5,441 4,085 30,449 0 728,111
Common shares outstanding 5,441 298.9 780.9 4.049 10,778
Total long-term debt 5,441 5,165 19,419 0 324,782
Earnings before interest and tax 5,441 1,431 4,356 -25,913 56,079
Return on assets 5,441 0.0835 0.122 -3.144 0.643
Number of employees 5,441 28,639 94,635 7 2.200e+06
Ln (employees) 5,441 8.822 1.848 1.946 14.60
Earnings per share 5,441 2.264 4.499 -136.9 65.83
Total liabilities 5,441 24,884 138,996 10.72 2.341e+06
Net income 5,441 773.1 2,602 -23,119 44,880
Operating activities net cash flow 5,441 1,583 5,253 -13,858 107,953 Cash flow divided by total assets 5,441 0.0892 0.0711 -0.828 0.500 Total stockholders’ equity 5,441 6,446 19,673 -11,476 256,205
Deferred taxes cash flow 5,441 17.82 669.2 -35,561 8,003
Total market value 5,441 13,753 33,464 4.058 438,702
Ln (market value) 5,441 8.296 1.530 1.401 12.99
Total book value 5,441 6,106 18,513 -13,244 233,932
Market to book ratio 5,441 3.859 47.02 -1,107 1,542
GIC sector codes 5,441 31.31 15.20 10 60
Base salary 5,441 860.1 393.2 0 4,550
Cash bonus 5,441 243.8 1,289 0 32,000
Value of stock awards 5,441 3,088 3,451 0 36,847
Value of option awards 5,441 1,011 2,334 0 48,316
Variable compensation 5,441 4,343 4,905 0.600 63,947
Ln (Variable compensation) 5,441 7.872 1.105 -0.511 11.07
All other compensation 5,441 240.3 1,181 -758.7 46,348
Total compensation as reported 5,441 7,243 6,522 60.60 69,901 Total compensation stock value 5,441 7,778 11,754 -1,642 258,918
Shares owned by CEO 5,441 1,101 3,251 0 68,417
% of shares owned by CEO 5,441 1.095 3.132 0 57.93
CEO age 5,441 54.70 6.768 31 84
Total pension payments made 5,441 24.59 558.9 0 30,992
Interest bearing debt to equity 5,441 2.193 15.86 -257.5 678.5
Appendix 3
Appendix 4
Utility Informational Energy Telecom Real estate
Healthcare Industrial Financial Variable comp. Leverage Ratio Utility 1 Informational 0.0726 1 Energy 0.0977 -0.0921 1 Telecom 0.0264 -0.0249 0.0334 1 Real estate 0.0762 -0.0718 0.0966 -0.0261 1 Healthcare 0.0886 -0.0835 0.1123 -0.0303 0.0876 1 Industrial 0.1387 -0.1307 0.1759 -0.0474 0.1371 -0.1594 1 Financial 0.0947 -0.0893 0.1202 -0.0324 0.0937 -0.1089 -0.1583 1 Variable comp. 0.0887 0.1011 0.0965 0.049 0.1071 0.0823 -0.0361 -0.0666 1 Leverage Ratio 0.0306 -0.0079 -0.073 0.0624 0.0975 -0.0959 -0.0301 0.1008 0.0998 1 Variables Net turnover Gross margin Market value Market to book ratio
Dividends Net cash flow EBIT Variable compensation Leverage ratio Net turnover 1 Gross margin -0.0011 1 Market value -0.1116 0.0731 1
Market to book ratio 0.0056 0.0157 0.022 1
Dividends -0.1246 0.0615 0.9447 0.023 1
Net cash flow -0.1098 0.0501 0.8191 0.0113 0.8349 1
EBIT -0.0561 0.3053 0.6784 0.009 0.6334 0.7949 1
Appendix 5
Number of observation: 619
Appendix 6
Tests of endogeneity Ho: variables are exogenous
Durbin (score) chi2 (1) = 0.880213 (p =0.3481) Wu-Hausman F(1,44) = 0.872722 (p =0.3505) Robust score (chi2)1 = 1.05209 (p =0.3050) Robust regression F(1,713) = 1.07003 (p =0.3013) Fixed Assets/Total assets Current ratio Leverage ratio Variable compensation Return on assets Percentage of total shares owned
Fixed Assets/Total assets 1
Current ratio -0.4267 1
Leverage ratio 0.0617 -0.4173 1
Variable compensation 0.1044 -0.2065 0.1106 1
Return on assets -0.2397 0.1206 -0.2252 0.0679 1
Appendix 7
Descriptive statistics American and Dutch data
(1) (2) (3) (4) (5)
VARIABLES N mean sd min max
CEO age 872 55.93 6.741 34 83
Total variable compensation 872 4,961 4,904 25 63,947
(Ln)variable compensation 872 8.083 1.023 3.219 11.07
Net sales in millions 822 9,556 21,499 0 236,810
Employees in thousands 872 22.60 47.74 0 444
Total assets in millions 872 33,207 154,138 0.207 2.352e+06
Cash and short-term investments in millions 872 4,032 29,524 0 573,080 Earnings before interest and taxes 872 1,177 4,132 -25,913 50,744 Liabilities and stockholders’ equity in millions 872 32,877 153,898 0.135 2.352e+06
Total liabilities in millions 872 26,497 136,029 0.135 2.104e+06
Stockholders equity in millions 872 6,518 20,708 -13,244 256,205
Leverage ratio 872 0.643 0.241 0.0594 3.251
Dutch dummy 872 0.0573 0.233 0 1
Debt to equity ratio 872 3.388 25.54 -80.03 712.9
CEO tenure in years 826 6.770 6.465 0 44.003
Appendix 8 Correlation Matrix CEO age Total variable comp. (Ln)Variable comp Net sales Number of employees Total assets Total liabilities Leverage ratio EBIT CEO age 1
Total variable comp. 0.0777 1
(Ln)Variable comp. 0.0321 0.7987 1 Net sales 0.0667 0.4685 0.4126 1 Number of employees 0.0344 0.4386 0.3877 0.6526 1 Total assets 0.058 0.3035 0.2378 0.4685 0.4046 1 Total liabilities 0.0561 0.2845 0.2209 0.4327 0.3876 0.9978 1 Leverage ratio 0.0004 0.0854 0.1075 0.1108 0.133 0.1374 0.1475 1 EBIT 0.0647 0.3793 0.3063 0.6215 0.5849 0.7746 0.7638 0.1223 1 Number of observations: 872 Appendix 9
Descriptive statistics Dutch data
(1) (2) (3) (4) (5)
VARIABLES N mean sd min max
CEO age 50 55.52 5.853 38 72
CEO Tenure 50 7.135 5.143 0.250 25
Variable comp. in millions 50 1,592 1,741 37 7,315
Marginal tax rate 50 0.129 0.336 -1.846 0.562
Number of employees in thousands 50 14.02 28.81 0.00200 169
Total compensation in million 50 2.487 2.229 0.408 9.480
Leverage ratio 50 0.525 0.164 0.0616 0.826
Shareholder funds in millions 50 5,714 21,652 0.0719 152,789 Total liabilities in millions 50 7,624 25,872 0.135 179,371
Appendix 10
Tests of endogeneity Ho: variables are exogenous
Robust score chi2(1) = 0.007806 (p=0.9296) Robust regression
F(1,5435)
= 0.006427 (p=0.9365)
Appendix 11
Tests of endogeneity Ho: variables are exogenous
Robust score chi2(1) = 0.007227 (p=0.9323) Robust regression
F(1,4998)
= 0.005947 (p=0.9389)
Appendix 12
Tests of endogeneity Ho: variables are exogenous
Robust score chi2(1) = 0.535425 (p=0.4643) Robust regression
F(1,5435)
Appendix 13
Tests of endogeneity Ho: variables are exogenous
Robust score chi2(1) = 0.682955 (p=0.4086) Robust regression
F(1,4998)
References
Bebchuk, L. A., Fried, J. M., & Walker, D. I. (2002). Managerial power and rent extraction in the design of executive compensation (No. w9068). National bureau of economic research.
Berger, P. G., Ofek, E., & Yermack, D. L. (1997). Managerial entrenchment and capital Structure decisions. The journal of finance, 52(4), 1411-1438.
Bergtold, J. S., Yeager, E. A., & Featherstone, A. M. (2018). Inferences from logistic regression models in the presence of small samples, rare events, nonlinearity, and multicollinearity with observational data. Journal of Applied Statistics, 45(3), 528-546.
Berk, J., & DeMarzo, P. (2014). Corporate Finance. Pearson.
Berk, J. B., Stanton, R., & Zechner, J. (2010). Human capital, bankruptcy, and capital structure. The Journal of Finance, 65(3), 891-92
Brockman, P., Martin, X., & Unlu, E. (2010). Executive compensation and the maturity structure of corporate debt. The Journal of Finance, 65(3), 1123-1161.
Bryan, S., Hwang, L., & Lilien, S. (2000). CEO stock-based compensation: An empirical analysis of incentive-intensity, relative mix, and economic determinants. The Journal of Business, 73(4), 661-693.
Calcagno, R., & Renneboog, L. (2007). The incentive to give incentives: on the relative seniority of debt claims and managerial compensation. Journal of Banking & Finance, 31(6), 1795-1815.
employee pay: An empirical analysis. Journal of Financial Economics, 110(2), 478-502.
Core, J. E., Holthausen, R. W., & Larcker, D. F. (1999). Corporate governance, chief executive officer compensation, and firm performance1. Journal of financial economics, 51(3), 371-406.
Dittmann, I., & Maug, E. (2007). Lower salaries and no options? On the optimal structure of executive pay. The Journal of Finance, 62(1), 303-343.
Dong, Z., Wang, C., & Xie, F. (2010). Do executive stock options induce excessive risk taking? Journal of Banking & Finance, 34(10), 2518-2529.
Eisdorfer, A., Giaccotto, C., & White, R. (2013). Capital structure, executive compensation, and investment efficiency. Journal of Banking & Finance, 37(2), 549-562.
Ernst and Young (2017). Executive Remuneration in the Netherlands 2017. Retrieved from: http://www.ey.com/Publication/vwLUAssets/EY-executive-remuneration-in-the-netherlands-2017/$FILE/EY-executive-remuneration-in-the-netherlands-2017.pdf
Fahlenbrach, R., & Stulz, R. M. (2011). Bank CEO incentives and the credit crisis. Journal of financial economics, 99(1), 11-26.
Fama, E. F., & Jensen, M. C. (1983). Separation of ownership and control. The journal of law and Economics, 26(2), 301-325.
Grossman, S. J., & Hart, O. D. (1982). Corporate financial structure and managerial incentives. In The economics of information and uncertainty (pp. 107-140). University of Chicago Press.
Hofstede, G. (2017). Hofstede Insights. Retrieved from:
https://www.hofstede-insights.com/country-comparison/the-netherlands,the-usa/
Humphery-Jenner, M. (2012). The impact of the EU takeover directive on takeover performance and empire building. Journal of Corporate Finance, 18(2), 254-272.
Jensen, M. C., & Meckling, W. H. (1976). Theory of the firm: Managerial behavior, agency costs and ownership structure. Journal of financial economics, 3(4), 305-360.
John, T. A., & John, K. (1993). Top-management compensation and capital structure. The Journal of Finance, 48(3), 949-974.
Leary, M. T., & Roberts, M. R. (2010). The pecking order, debt capacity, and information asymmetry. Journal of financial economics, 95(3), 332-355.
Lefebvre, M., & Vieider, F. M. (2014). Risk taking of executives under different
incentive contracts: Experimental evidence. Journal of Economic Behavior & Organization, 97, 27-36.
Lewellen, W., Loderer, C., & Martin, K. (1987). Executive compensation and executive
incentive problems: An empirical analysis. Journal of accounting and economics, 9(3), 287-310.
Lin, H. C., Chou, T. K., & Wang, W. G. (2012). Capital Structure and executive
compensation contract design: A theoretical and empirical analysis. Journal of Banking & Finance, 36(1), 209-224.
Modigliani, F., & Miller, M. H. (1958). The cost of capital, corporation finance and the theory of investment. The American economic review, 48(3), 261-297.
Ortiz-Molina, H. (2007). Executive compensation and capital structure: The effects of
convertible debt and straight debt on CEO pay. Journal of Accounting and Economics, 43(1), 69-93.
Uhde, A. (2016). Risk-taking incentives through excess variable compensation: Evidence from European banks. The Quarterly Review of Economics and Finance, 60, 12-28.
Yermack, D. (1995). Do corporations award CEO stock options effectively? Journal of financial economics, 39(2-3), 237-269.
Zhang, Y. (2009). Are debt and incentive compensation substitutes in controlling the free cash flow agency problem? Financial Management, 38(3), 507-541.
