FOREIGN TRADE UNIVERSITY

HO CHI MINH CAMPUS

--------***--------

ECONOMETRICS GROUP PROJECT

TOPIC:
AN ANALYSIS OF FATORS AFFECTING
CEO SALARY
Lecturer: Ms. Nguyen Thu Hang – Lecturer of
Economics & International Business Falculties

Group members:
1. Vũ Thị Trúc Quỳnh 1601035127
2. Phan Long 1601035075
3. Nguyễn Thị Mai Thảo 1601035138
4. Nguyễn Lê Bảo Trân 1601035164

5. Lương Nguyễn Hồng

Nhung 1601035127

Ho Chi Minh city, November

2017

1
 Table of contents

ABSTRACT ................................................................................................................ 3

CHAPTER I: INTRODUCTION................................................................................ 3

CHAPTER 2: LITERATURE REVIEW ON THE EFFECT OF VARIABLES ON

CEO SALARY ............................................................................................................ 3

CHAPTER 3: METHODOLOGY ............................................................................. 6

1. Data ..................................................................................................................... 6

2. Model specification ......................................................................................... 8

3. Result and discussion ................................................................................ .10

CHAPTER 4: CONCLUSION ................................................................................ 13

REFERENCES ......................................................................................................... 14

List of Tables
Table 1: Descriptive statistics of all variables in data ................................... 6
Table 2 Regression result using OLS ................................................................ 8
Table 3 Coefficient of variables in model .......................................................... 9

List of Figures
Figure 1: The Heteroscedasticity test with Graphical Method .................... 7

2
 ABSTRACT

This paper investigates Salary of CEO and illustrates a variety of

factors that significantly influence the CEO salary. OLS (ordinary least
squares) is applied in these regressions, which shows the result including
sale, return on equity, change in percentage of ROE, return on firm’s stock,
types of the business. The data shown in this study are provided by Jeffrey M.
Wooldridge , dataset from Introductory econometrics. He collected these
data from a survey of Businessweek.
Keywords: Salary of CEO

CHAPTER I: INTRODUCTION

Executive compensation or executive pay is composed of the financial

compensation and other non-financial awards received by an executive from their
firm for their service to the organization. It is typically a mixture of salary, bonuses,
shares of or call options on the company stock, benefits, and perquisites, ideally
configured to take into account government regulations, tax law, the desires of the
organization and the executive, and rewards for performance. Executive
compensation is a very important thing to consider when evaluating an investment
opportunity. However, CEO who are improperly compensated may not have the
incentive to perform in the best interest of shareholders, which can be costly for
those shareholders.

This study goal is to find out the relation between important factors which
determined salary of CEO and the differences with others CEO in salary.
Therefore, we have generated and run the model to find out which significant
determinants are and the trend of their impact on the salary of CEO.

CHAPTER 2: LITERATURE REVIEW ON THE EFFECT

OF VARIABLES ON CEO SALARY

3
 There is a wide range of researches on the factors affecting CEO salary, among
them there are a lot of researches studying about the relationship between CEO
compensation and firm performance with conflicting results. Ozkan (2007)
suggests that there is a positive relationship between CEO pay and the
performance of a company (at least in the UK). Zhou (2000) doing a research on
Canadian firms finds out that CEO pay is positively related to the firm size and
compensation depends on company performance. On the other hand a study by
Brick et al. (2005) shows that there is a strong negative relation between CEO
compensation and the performance of a company. Another case, Tosi et al. (2000)
reported a weak correlation between CEO pay and performance but a strong
positive correlation with CEO pay and size of the firm.
These following researches will explore in more detail about the factor we
mentioned above:
In 2011, KJ Sigler, comes from University of North Carolina Wilmington,USA,
examined the relationship between CEO pay and company performance. The data
is collected from 280 firms listed on the New York Stock Exchange for a period
from 2006 through 2009. The time frame of the study is a period after the adoption
of the Sarbanes Oxley Act and after the SEC approval of the corporate
governance rules affecting executive pay for New York Stock Exchange
companies. In KJ Sigler’s research, he found that that the size of the firm has the
most impact on the compensation of the firm’s chief executive officer. Tenure is
the next most important variable implying that either the CEO acquires more
knowledge and expertise over time in the position or he or she attains more power
over the compensation committee that decides the level of CEO pay. ROE also
has a positive effect on CEO pay. This is consistent with the premise that CEO
pay is paid in relation to how the company performs. The last factor that he
included in his model is the risk taken by CEO, and the result is noteworthy: the
more risk taken by the CEO than the larger his compensation.

Another innovative work which focuses on the monopoly industries is that of

4
Yun Dai (2015), which concentrated on the compensation of CEOs in China’s
monopoly enterprises. Finding out the factors affecting executive compensation is
of great importance to establish effective and equitable compensation mechanism,
Yun Dai conducted a research to study the main affecting factors of the executive
compensation of state monopolies, with 1280 firm-year observations of 183
monopoly companies. Yun Dai got 3 conclusions, which is executive
compensation is positively correlated to firm performance, firm size, CEO duality.
Secondly, executive compensation is negatively correlated to state share
proportion and independent director proportion. Finally, company location, industry
both have significant influence on executive compensation.

Berkema and Mejia (1998) have also indicated a few factors which influence the
pay scale of a CEO (their study based on US firms). They are of the opinion that
market forces play an important role in determining the pay of the CEO. The
ownership structure of the firm influences greatly. According to them the firms with
large blockholders influence the pay given to the CEO. Another important factor
that is included in their findings is the recent development of remuneration
committees. These committees are responsible for developing a proposal of pay
for the CEO. This proposal is then passed by the Board of directors (Berkema et al.
1998). The members of the remuneration committee are usually outside the
company, in this way there is a separation of control and management (which is a
good thing according to the agency theory).

One more factor that many people decided to conduct a research on it is the
influence of Board of Directors. According to Boyd (1994), the Board members are
an important “internal control mechanism” for setting CEO compensation. Guest
(2009) researched on 1880 public firms in UK from a period of 1983-2002 and
found that as the size of the Board increases, the pay of the chief executive officer
increases. Similarly Core et al. (1999) did a study on 205 publicly traded US firms
and found larger Boards giving out more compensation to CEOs.

5
 CHAPTER 3: METHODOLOGY

1. Data

The source of data for researching purposes was given in Wooldrigde, J (2003):
Introductory econometrics: A modern approach, South-Western College
Publishing.
The data contains the number of observations equal 209 CEOs (Chief Executive
Officers). These variables in our survey includes Salary and Sales in 1990, which
are measured by thousand dollars and million dollars, perspective. The additional
observations for these research objects consist of return on equity, return on stock
of a company and type of these company (industry, finance, transport or utility,
consumer product firm).

Table 1: Descriptive statistics of all variables in data

Variable Observation Mean Std. Dev Min Max

salary 209 1281.12 1372.345 223 14822
sales 209 6923.793 10633.27 175.2 97649.9
roe 209 17.18421 8.518509 .5 56.3
ros 209 61.80383 68.17705 -58 418
indus 209 .3205742 .4678178 0 1
finance 209 .2200957 .4153057 0 1
consprod 209 .2870813 .4534861 0 1
utility 209 .1722488 .3785031 0 1
lsalary 209 6.950386 .5663741 5.407172 9.603868
lsales 209 8.292265 1.013161 5.165928 11.48914

Source: Completed by Stata

Table 1 described partly about the expected value (Mean), Standard error of the
variables we are analyzing. In some variables, such as Indus, Finance, Consprod,
Utility, mean value of them are getting closer to the minimum and maximum value
and in the range of [0,1], this indicates that the distribution is extremely on one side
of these variables. These additional variables indicate that these ceo work for each
job in different occupations. For example: A ceo working for a finance firm will have
6
 1 point in the finance column and the others (indus, consprod, utility) will be 0.
Increasing dramatically the number of observations will help to decrease the
chance of multilinearity in our model and make our model become more reliable
(BLUEs)

Figure 1: The Heteroscedasticity test with Graphical Method

3
2
Residuals

1
0
-1

6 6.5 7 7.5 8
Fitted values

Source: Completed by Stata

According to the graph above, notwithstanding the number of variables in our

model was enormous, heteroscedasticity still happended. In order to solve this
problem, we have ran the datasets in Stata with robust to reduce the
heteroscedasticity’s affect to the result of our research.

7
2. Model specification:

We illustrate the problem studied by using the regression model to explain the
impact of these factors on dependent variable, lsalary results of a CEO, through
some independent variables namely lsales, roe, ros, indus, finance, consprod,
utility. We can illustrate the regression model:

lsalary =0 + 1 lsales + 2 roe + 3 finance + 4 consprod + 5 indus + 6 ros +u

(Repace ros by negros , which equals to 1 if a firm’s ros is less than 0, and 0
otherwise)

According to previous research projects in conjunction with actual observations,

we can make assumptions about the impact of these factors on lsalary. The
regression results using Ordinary Least Squares method are represented in the
Table 2.

Table 2: Regression results using OLS

(1) (2) (3)
VARIABLES Model 1 Model 2 Model 3

lsales 0.256*** 0.272*** 0.257***

(0.0342) (0.0320) (0.0345)
roe 0.0112** 0.00916**
(0.00435) (0.00430)
ros -5.35e-05
(0.000543)
indus 0.282*** 0.333***
(0.0998) (0.0995)
finance 0.441*** 0.472***
(0.104) (0.103)
consprod 0.466*** 0.495***
(0.111) (0.108)
negros -0.283***
(0.105)
Constant 4.317*** 4.218*** 4.822***
(0.306) (0.279) (0.288)

Observations 209 209 209

R-squared 0.357 0.379 0.211
8
 (Model 2)
Observation 209
R-squared 0.3790
F-statistic 20.55
Prob>F 0.0000
RSS 41.432342
Standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
roe= return on equity; ros: return on stock; indus: industrial firm; finance: finace
firm; consprod: consumer product firm

Strictly following the theory from literature and various research, we obtain the
final model by starting with the different groups of independent variables and
adding the important ones to not only avoid omitting important explanatory
variables, which cause the assumptions to be violated; but also acquire the most
accurate one. As based on the literature and economic theory, we are able to
recognize the impact of these factors on the salary of CEO. Firstly, looking at the p-
value of F-statistics, we can see that overall model is statistically significant at the
level of 1%. Turning to the regressors, most of the coefficents are significant at the
1% except the effect of roe β2, which is significant at the level of 5%.

Table 3: Coefficient of variables in model

e(V) lsales roe negros indus finance consprod cons

lsales 1.0000
roe 0.1218 1.0000
negros -0.1694 0.1728 1.0000
indus -0.1000 -0.2962 -0.1884 1.0000
finance -0.0854 -0.1513 -0.1115 0.6221 1.0000
consprod -0.1013 -0.4615 -0.1065 0.6760 0.5908 1.0000
cons -0.9464 -0.2903 0.1162 -0.0583 -0.0917 -0.0131 1.0000

From table 3, we have recognized the coefficient matrix of variables. The

coefficients of roe, negros is 0.1728, which are far away from perfect correlation or
multicollinearity. Hence, the requirement for no exact collinearity is considered to
be fulfilled by the model suggested.

9
 There are questions about the story: Will those factors really affect a CEO's
salary? And there are a lot of hypotheses out there. These are some research
which are related to this question.

In a report of research, Lewellen and Huntsman, found that the accounting

profits and sales revenue both affected executive compensation.

(Lewellen, W.G. and Huntsman, B. (1970) Managerial Pay and Corporate Performance.
The American Economic Review, 60, 710-720.)

In accordance with Yun Dai’s research (2014), Company location, industry both
have significant influence on executive compensation. Yun Dai has stated that the
industry of a CEO leading will affect his/her salary or we stated the effects of

on a salary of a CEO.
(Dai, Y. (2014) Research on Influencing Factors of Executive Compensation in China’s
Monopoly Industries. Open Journal of Business and Management, 2, 210-218. Ht)

As we all know, The net asset return rate (ROE), reflecting the profitability of a
corporation, is adopted to measure firm performance. As a research of KJ Sigler,
which was published in April 30, 2011, stated the effect of ROE (return on equity)
on a CEO salary. So we can restate the null hypothesis:

(KJ Sigler research - University of North Carolina Wilmington, Cameron School of

Business, Wilmington, NC 28403, USA. Correspondence: Kevin J Sigler,
siglerk@uncw.edu Accepted: March 17, 2011; Published: April 30, 2011)
We will examine these 3 hypothesises independently.

3. Results and discussions:

After running the regression model on Stata 12, we had the following result:

10
 lsalary = 4.218 + .272 lsales + .00916 roe + .333 indus + .472 finance + .495
consprod+u

Looking at the Model 1, we realize that ros abs tstatistic is lower than the critical
value (1.6), which means that the coefficient of the ros is insignificant, even at the
level of 10%. As a result, we exclude it from the model and generate a new dummy
variable, called negros, which is equal to 1 if the ros is less than 0, and 0
otherwise, and run the second model (Model 2). The result received is that with
the absolute t-statistic of 2.69, the coefficient of the negros is significant at the
significant level of 1%. The sign of β6 is negative to indicate the relationship
between CEO salaries and ros. If the CEO’s firm has a negative return on its stock,
the CEO salary is predicted to be about 28.3% lower, for given levels of sales and
roe. Ros measures the amount of value an investor earns from a security over a
specific period, typically one year, when all distributions are reinvested. It was not
directly related to the job of the CEO as well as the operating of the firm. The
investors, who determine the salary of the CEO, will react to negative ros - the fall
in the price of stocks- by narrowing the compensation paid to ceo because this
situation can diminish their benefits.
The lsales which has the t-statistics = 8.5 to be very significant at the level
of 1%. So we can reject the first hypothesis at the significant level at 1%, and
conclude that the sales of the firm have a positive effect on the CEO’s salary. If the
firm’s sale increase by 1%, the CEO’s salary increase by 0.272%. This result is just
like what we have already expected as for many companies, higher sales will
ensure higher net income, which results in the higher salary for all of the
employees, including the CEO, assumed that the firm do not change their
operational strategy.

The second hypothesis to test is whether the CEO’s salary depends on their
type of business, including industrial, finance, consumer product and utility and the
difference in the salary of the CEO among those business. We compute the
restricted least square estimation and received the F-statistic which is equal to
15.3- much greater than the critical value. As a result, we can indicate that the

11
effect of the business on compensation for the executive is substantial.
Furthermore, as can be seen from the Model 2, the income of the CEO in the
industrial firm, finance firm and consumer product firm is 33,3%;47,2%;49,5%
higher than that of the utility firm respectively.
With the t-statistic equals 2.13, the coefficient of the ROE is significant at the
significant level of 5%. If the firm’s roe increase by 1%, then the salary of that firm
increase by 0.00916%. ROE measures a corporation’s profitability by revealing
how much profit a company generates with the money shareholders have invested.
The increase in roe means that companies use its capital more efficiently, which
may be partly due to the good governance of the company leadership, especially
the CEO. Consequently, the CEO's income become higher, despite how negligible
it is.
Overall, by looking at the R2, we can claim that nearly 37.9% total variation
of the CEO salary can be explained by the total variation of 6 independent
variables, including the lsales, roe, negros, indus, finance and consprod.

12
 CHAPTER 4: CONCLUSION

In this report, we study about the determinants of the CEO salary. Looking at
the model assumption, we can make a conclusion that the variance of CEO salary
was determined by these factors: the variance of sales, roe, the type of the
business and whether the ros is negative or not. The coefficients of all the variables
are statistically significant. The lsales, roe have positive effect on the variance of
the CEO salary, while the negros adversely affect the variance compensation for
the CEO. Furthermore, taking utility firm as a base group, the CEO in industrial
firm, consumer product firm and finance firm enjoy a much better salary.
The result shows us how some factors of the company affect the variation of
the CEO salary. Therefore, this empirical study can provide analysis and insights to
help firms effectively control and adjust to pay for their CEO effectively and make
comparison about the difference in payment for CEO in different industries

13
 REFERENCES

1. Ozkan (2007), CEO Compensation and Firm Performance:

An Empirical Investigation of UK Panel Data.
2. Zhou, X., 2000. CEO Pay, Firm Size, and Corporate
Performance: Evidence from Canada. The Canadian Journal of
Economics / Revue canadienne d'Economique, 33(1), 213-251.
3. Brick et al. (2005), CEO compensation, director
compensation, and firm performance: Evidence of cronyism?
4. Tosi et al. (2000), How Much Does Performance Matter? A
Meta-Analysis of CEO Pay Studies
5. KJ Sigler research - University of North Carolina
Wilmington, Cameron School of Business, Wilmington, NC 28403,
USA. Correspondence: Kevin J Sigler, siglerk@uncw.edu Accepted:
March 17, 2011; Published: April 30, 2011.
6. Dai, Y. (2014) Research on Influencing Factors of Executive
Compensation in China’s Monopoly Industries. Open Journal of
Business and Management, 2, 210-218. Ht.
7. Berkema and Mejia (1998), Managerial compensation and
firm performance: A general research framework.
8. Boyd (1994), Board control and CEO compensation.
9. Guest (2009), Board structure and executive pay: evidence
from the UK.
10. Core et al. (1999), Corporate governance, chief executive
officer compensation, and firm performance.

14