The Executive Compensation Disclosure Rules

Published Date: 02 Nov 2017

"The executive compensation disclosure rules provide for enhanced disclosure of executive and director compensation aiming at dealing with corporate governance matters.  With a goal of transparency, perhaps organizations should strive to provide customised narratives of how they derived "pay for performance" compensation plans and how they ultimately will provide shareholder value."

This study will look at the relationship between CEO compensation and its tenure, firm size as well as company performance.

Module Code: MAN2905M

Module Leader: Dr. Chengang Wang

Word Count: 2051

Student Name: Ruochen JIA

Student No.:10018609

Content

1.0 Introduction…………………………………………………….………….………..........1

2.0 Literature Review ……………………………………………..........................................1

3.0 Econometric Module…......................................................................................................2

4.0 Data...............................................................................................................................…...3 4.1 Sample Selection and Variable Measurements..........................................................3

4.2 Descriptive Statistics....................................................................................................3

5.0 Inference Procedure and Diagnostic Test………………………………….…………...5

5.1 RESET Test………………………………………………………….………………..6

5.2 Write Test……………………………………………………………………………..7

6.0 Empirical Results and Conclusion………………………………………………............8

7.0 The Practical Implication and Limitations...….………………………………………..9

8.0 References……………………………………………………………………………….10

9.0 Appendices...…………………………………………………….………………………11

Appendix A………………………………………………….…………………………..11

Appendix A.1 Regression Results………………………………………………….......11

Appendix A.2 R2 Significance…………………………………………………………..12

Appendix A.3 Coefficient Significance………………………………………………...13

Appendix B RESET Test……………………………………………………………….15

Appendix C Write Test…………………………………………………………………17

Appendix D Variance Inflation Factor...………………………………………………18

Introduction

With the recent corporate failures including financial crisis and Enron scandal, the media and public have to pay more attention to Chief Executive Officer (CEO) performance and executive compensation. This study will research whether there is a relation between CEO compensation and CEO’s tenure, firm size measured by sales, as well as firm performance in line with profits. If there exists a relationship between those variables, this study will then analyse, to what extend and how CEO compensation is correlated to its tenure, firm size and performance.

The subject of this study is relevant to corporate governance. Due to the separation of ownership, managers will focus on their own interests rather than the wealth of shareholders. When the corporate governance is efficient, the directors can monitor CEOs through changing CEO compensation levels. An increase in CEO compensation will require greater accountability and create linkages of compensation to corporate performance.

Literature Review

One of earliest researches conducted by Cisel and Carroll (1980) focused on two explanatory variables (profits and sales) to analyse whether the individual has an influence on CEO compensation. According to their results, both two variables were significant. However, this research has a significant limitation because the authors did not take other variables which could affect the compensation into consideration and only used one year of data.

Another study by Roberts (1959) argued that CEO compensation was closely related to firm size which is measured by sales. A strong relationship between compensation and sales indicated that it is easy for CEO to increase the company size rather than profits. In practice, sales growth is really the main incentive for CEO to boost compensation. Some CEOs would scarify long-term profitability to stimulate the current level of compensation. Therefore, recently IASB [1] drafted to modify the IAS [2] 24 requirement to disclose compensation paid to key management personnel so as to improve transparency and increase quality of financial information. This problem has also brought the discussions of the disadvantages of the separation of ownership when managing the corporation.

Furthermore, Deckop (1998) studied on those factors which may affect CEO compensation. Through analysing 120 firms from 1997 to 1981, he found that increasing the firm size did not develop the compensation of CEOs but profits as a percentage of sales had a positive effect on CEO compensation. Additionally, from the result of Sigler (2011), there was a significant relationship between CEO compensation and firm performance measured by Return on Equity (ROE) for a particular period.

Hill and Phan (1991) looked at 104 firms for the period 1977 to 1988 in order to determine whether there is an effect of a CEO’s tenure at a firm on the relationship between firm size and its compensation. According to their result, it shows that if the tenure increases in time horizon, there would be a stronger relationship between pay and firm size. Furthermore, Cordeiro and Veliyath (2003) added, the tenure of CEO played a significant determinant of CEO compensation. This relationship can be explained by the learning theory proposed by Murphy (1986): CEO's managerial ability is limited in the early period so that CEOs will focus on improving company size or performance in short period. However, as time went by, the level of compensation increases with more experience and ability absorbed from annual operations. What's more, weaker CEOs are less likely to survive. Therefore, the survivals will be paid satisfactorily.

According to previous literature, CEO compensation is still affected by multitude of factors such as personal experience (education and qualification), firm’s stock performance, level of risk undertaken by CEO and etc.

Econometric Module

This study will look at how CEO compensation changes in percentage due to the change in tenure years, sales percentage as well as the profit. For this purpose, a semi-log module is used to illustrate the relationship. It is also useful to increase specification with the semi-log module.

where compensation means CEO’s annual compensation in year 1999; tenure is the year CEO stays in its position; sales mean the sale revenues in the year of 1999; profits stand for the company profits measured by profit ratios. Two measures of profits are used in this study. Therefore, two estimations of specification are generated in line with profits. The dependent variable in each specification is the nature logarithm of CEO compensation. In the first specification, the profit is measured by the ratio of profit per sale, which shows how much profit a firm generated per dollar of sales. The other is ROA which stands for return on assets and indicates how profitable a company is relative to its total assets.

Data

Sample Selection and Variable Measurements

The data on CEO compensation and its tenure was collected from Forbes' 1999 list of corporate America's most powerful CEOs. The sample consists of 360 US companies and it was obtained from an initial sample of 480 after taking account of full years of tenure which is more than 6 months. The compensation of CEO is one year pay, including base salary, bonus, stock options and etc. To measure the size related to individual companies, this study makes use of sales generated in the financial year 1999. This is reasonable because it is easy to collect sales turnover from Fortune Magazine's 1999 Fortune 500 list. Additionally, compared with the measurement of employee numbers and market shares, this measure is useful for the purpose of comparison since both of the other two measures are highly limited in the same industry or market. The measurement of sales is also used by Mahmoud and Frank (2008). With regards to the performance of companies, this study applies profitability ratio - profit per sale as the measure as well as return on assets as an alternative. The reason to choose the two measurements is that under accounting assumption, the objective of organisations is to maximise the profit. Therefore, the directors of the organisations will reward the CEO according to the company’s overall profit (McLaney and Atrill, 2010).

Descriptive Statistics

The descriptive statistics including the average of variables and relative basic summary are shown in Figure1. From statistical correlations and correlation matrix presented in Figure 2 and Figure 3 respectively, the figures illustrate that the correlation between nature logarithm of compensation (lnCompensation) and tenure is 0.063. It means those two variables are in a positive linear relationship but the relationship is very week. The case of tenure and profit is similar since their correlation is 0.030. The correlation between lnCompensation and lnsales is 0.318. It is the highest positive linear relationship in this model. Additionally, the correlation between lnCompensation and profit, tenure and lnsales, profit and lnsales is -0.104, -0.069 and -0.023 respectively, which means there is a decreasing linear relationship between those variables.

Mean

Std. Deviation

N

lnCompensation

8.2055

1.04579

360

tenure

8.6611

8.53579

360

lnsales

8.9650

.74998

360

profits

6.9571

5.65153

360

(Figure 1: Descriptive Statistics)

lnCompensation

tenure

lnsales

profits

Pearson Correlation

lnCompensation

1.000

.063

.318

-.104

Tenure

.063

1.000

-.069

.030

Lnsales

.318

-.069

1.000

-.023

Profits

-.104

.030

-.023

1.000

Sig. (1-tailed)

lnCompensation

.

.117

.000

.024

Tenure

.117

.

.094

.287

Lnsales

.000

.094

.

.332

Profits

.024

.287

.332

.

N

lnCompensation

360

360

360

360

Tenure

360

360

360

360

lnsales

360

360

360

360

Profits

360

360

360

360

(Figure 2: Correlations)

(Figure 3 Correlation Matrix)

Inference procedure and diagnostic test

According to the literature review, the relationship between CEO compensation and other variables is linear. Therefore, this section will focus on testing whether the modules are compliant with assumptions of classical linear regression model. The results are summarised in the Table 1. Detailed diagnostic tests are listed in Appendice.

Assumption

Test

Result

1. Linear regression model

The model is linear in the parameters

Linear

2. X values are fixed in repeated sampling

Zero covariance between ui and each X variable

Xs are independent of the error term

3. Specification

Appendix B- RESET Test

Specific at 1% significant level

4. Homoscedastic

Appendix C- White Test

Hemoscedastic at 2.5% chi-square level

5. No collinearity

Appendix D- VIF

Values are below 5

6. No autocorrelation

The module is cross-sectional

Satisfied

7. The term u is normally distributed

Assumed to be acceptable

Satisfied

No autocorrelation

8. The mean value of disturbance is 0

Assumed to be acceptable

Satisfied

9. Variability in x variables

3 of x variables

There are sufficient variation in x variables

10. No. of observations is larger than the No. of parameters

360> 3+1

No. of observations is larger than parameters

(Table 1, Assumptions of the Classical Model, utilised by Gujarati and Porter,2009)

RESET test

The result of RESET test is that both of the two specifications are highly specified at 5% significant level, which means there are likely no specification errors in these two modules. That is to say, the modules not only include relevant variables but also free from unnecessary variables as well as adopt the correct functional form. Moreover, the first specification is preferred (The regression results are shown in Table 2), since the R square of first specification module whose profit is measured by profit per sales is higher than that of second specification with ROA as the profit measure. Additionally, the change of CEO compensation is better illustrated by the profit measured by the profit per sale.

Model Summary

Model

R

R Square

Adjusted R Square

Std. Error of the Estimate

1a: profits- profit per sale

.344a

.118

.111

.98616

2a: profits- ROA

.338a

.114

.107

.9884

1a. Predictors: (Constant), lnsales, tenure profits - profit per sale

1b. Dependent Variable: lnCompensation

2a. Predictors: (Constant), lnsales, tenure profits –ROA

2b. Dependent Variable: lnCompensation

(Table 2)

Write test

The heteroscedastic problem is significant in this module. The heteroscedasticity happens at the 2.5% chi-square level. Gujarati and Porter (2009) argued that heteroscedasticity may be caused by the presence of outliers. According to Figure 4, exclusion of the outliners would substantially fix this kind of problem. Another remedy is to redefine the variables. In this case, what this study can do is to add some new variables according to the previous researches. Alternatively, the double-log form has inherently less variation than the linear form so it is less likely to encounter heteroskedasticity. Therefore, theoretically the underlying module can be switched from semi-log form to double-log functional form. However, due to the high specification, it must be highly careful to refine the variables and functional form otherwise corresponding changes in specification would dramatically damage the underling model. According to Studenmund (2006), another popular remedy is heteroskedasticity-corrected standard error. It focuses on improving the estimation of the SE(βPRE) to avoid the consequences of heteroscedastic problem

(Figure 4)

Empirical Results and Conclusion

The regression results are shown in table 3 and the regression module will be as followed.

Dependent variable: lnCompensation

Specification 1 Specification 2

Intercept

4.218

(0.636)***

4.152

(0.636)***

tenure

0.011

(0.006)*

0.011

(0.006)*

lnsales

0.449

(0.070)***

0.451

(0.070)***

profits

-0.018

(0.009)*

-0.015

(0.010)*

R-squared

0.118

0.114

Adjusted R-squared

0.111

0.107

VIF:

tenure

1.006

1.006

lnsales

1.005

1.005

profits

1.001

1.001

Number of observations is 360

Standard errors are in parentheses

***significant at 1%, ** significant at 5%, * significant at 1%

(Note: detailed test of R2 significance and coefficient significance is presented in Appendix A)

(Table 3: summary of regression results)

To reflect the CEO compensation that changes in percentage, this study applies semi-log module. The CEO compensation has been logged to represent a relative change. The model does not have a high explanatory power since the adjusted R2 value is only 0.111. However, as hypothesis testing (t-test) suggests, all the explanatory variables are relevant at 10% significant level; moreover, one of the major determinants of CEO compensation is firm size, measured by the company salsas turnover (Figure 5). Evidences from coefficient 0.449 and t-value 0.636 indicate that sales of a firm have a significant positive influence on CEO compensation. The elasticity coefficient of 0.449 of proposes that a 1 percent bigger firm would pay its CEOs 0.449 percent greater compensation. Simon (1957) also found the positive relation between firm size and compensation because the larger a firm is the high demand for skilled CEO as well as greater ability to pay for its compensation. As regards to tenure, holding lnsales and profits constant, one more year a CEO staying in his/her place will lead to 0.011% increase in the compensation (shown in figure 6). Although, the increment is small, it significantly affects the level of CEO compensation at the given significant level. With time elapsing, the abilities of the CEO are supposed to improve and so does its influence over the board of directors. Therefore, the longer time a CEO stayed in a company, the greater was its compensation. However, the study observed profit in a company has a negative relationship between CEO compensation (Figure 7). In addition, the CEO compensation was not highly sensitive to company performance. Although the negative relationship is not that significant, it reflects the contemporary issues in accounting. In the past, it is not surprising that directors and executives still received massive compensation even when the company performance was deficient. However, in recent years, the corporate governance system has changed to avoid financial disasters. Throughout the reforms of corporate governance, remuneration committee consisting of non-executive directors is introduced to determine appropriate levels of remuneration of executive directors, aiming at preventing from the executive influence to increase the transparency of paying system and improving accountability to shareholders (Cadbury Report 1992).

(lncompensation and sales) (lncompensation and tenure ) (lncompensation and profits)

Figure5 Figure 6 Figure 7

The Practical Implication and Limitation

The study conducted a research on the relationship between CEO compensation and tenure, company size as well as company performance. It is practically useful for having a new look at contemporary issues in accounting, especially the corporate governance system and agency problems. The result shows that boards of directors should try to optimise CEO compensation level on the basis of firm size; the negative relationship between CEO compensation and company profits measured by profit per sale shown from this study illustrates the problems caused by the separation of ownership. The owner of companies should take actions to monitor the CEO performance to avoid the situation where CEO still received massive compensation even when the company performance was deficient. However, this study also experiences some limitations. The result came from only one year data which was collected from the region of America without covering the global environment. Additionally, CEO compensation concludes only short-term components but does not cover the long-term reward. Despite high specification of this model, it suffers from the occurrence of heteroskedasticity in the linear modul. In the future research, changing functional form or introduction of other variables such as firm’s stock performance, level of risk undertaken by CEO may substantially remedy heteroscedastic problem.