Executive Compensation, Firm Performance and
Corporate Governance: Evidence from Pakistan
By
Muhammad Fayyaz
CIIT/FA11-PMS-004/ISB
PhD Thesis
In
Management Sciences
COMSATS Institute of Information Technology,
Islamabad Campus, Pakistan
Fall, 2016
ii
COMSATS Institute of Information Technology
Executive Compensation, Firm Performance and
Corporate Governance: Evidence from Pakistan
A Thesis Presented to
COMSATS Institute of Information Technology, Islamabad
In partial fulfillment
of the requirement for the degree of
PhD (Management Sciences)
By
Muhammad Fayyaz
CIIT/FA01-PMS-004/ISB
Fall, 2016
iii
Executive Compensation, Firm Performance and
Corporate Governance: Evidence from Pakistan
A Post Graduate Thesis submitted to the Department of Management Sciences as
partial fulfillment of the requirement for the award of Degree of PhD in Management
Sciences
Name Registration No.
Muhammad Fayyaz CIIT/FA11-PMS-004/ISB
Supervisor
Dr. Syed Zulfiqar Ali Shah
Professor, Department of Management Sciences
COMSATS Institute of Information Technology (CIIT)
Islamabad, Campus.
iv
Certificate of Approval
This is to certify that the research work presented in this thesis entitled “Executive
Compensation, Firm Performance and Corporate Governance: Evidence from Pakistan” was
conducted by Muhammad Fayyaz (PhD Scholar) under the supervision of Dr. Syed Zulfiqar
Ali Shah. No part of this thesis has been submitted anywhere else for any degree. This thesis
is submitted to the Department of Management Science, COMSATS Institute of Information
Technology, Islamabad Campus in the partial fulfillment of the requirement for the degree of
Doctor of Philosophy in the field of Management Sciences.
Student Name: Muhammad Fayyaz Signature: _________________
Examination Committee:
Dr. Eatzaz Ahmad
Professor,
University of Peshawar
----------------------------------------------------
----------------------------------------------------
Dr. Abdul Qayyum
Professor,
Pakistan Institute of Development Economics
----------------------------------------------------
----------------------------------------------------
Dr. Syed Zulfiqar Ali Shah
Supervisor,
Department of Management Sciences
CIIT, Islamabad.
Dr. Muhammad Tahir
HoD,
Department of Management Sciences
CIIT, Islamabad.
Dr. Farzand Ali Jan
Chairman, Management Sciences
CIIT
Dr. Khalid Riaz
Dean, Business Administration
CIIT
v
Author’s Declaration
I Muhammad Fayyaz CIIT/FA11-PMS-004/ISB hereby state that my PhD thesis titled
“Executive Compensation, Firm Performance and Corporate Governance: Evidence from
Pakistan” is my own work and has not been submitted previously by me for taking any
degree from this University i.e. COMSATS Institute of Information Technology or anywhere
else in the country/world.
At any time if my statement is found to be incorrect even after I graduate the University has
the right to withdraw my PhD degree.
Date: __________________
Muhammad Fayyaz
CIIT/FA11-PMS-004/ISB
vi
Plagiarism Undertaking
I solemnly declare that research work presented in the thesis titled “Executive Compensation,
Firm Performance and Corporate Governance: Evidence from Pakistan” is solely my research
work with no significant contribution from any other person. Small contribution/help
wherever taken has been duly acknowledged and that complete thesis has been written by me.
I understand the zero tolerance policy of HEC and COMSATS Institute of Information
Technology towards plagiarism. Therefore, I as an author of the above titled thesis declare
that no portion of my thesis has been plagiarized and any material used as reference is
properly referred/cited.
I understand if I am found guilty of any formal plagiarism in the above titled thesis even after
award of PhD degree, the University reserves the right to withdraw/revoke my PhD degree
and that HEC and the university has the right to publish my name on the HEC/university
website on which names of students are placed who submitted plagiarized thesis.
Date: __________________
Muhammad Fayyaz
CIIT/FA11-PMS-004/ISB
vii
Certificate
It is certified that Mr. Muhammad Fayyaz, CIIT/FA11-PMS-004/ISB has carried out all the
work related to this thesis under my supervision at the Department of Management Sciences,
COMSATS Institute of Information Technology Islamabad and the work fulfills the
requirement for award of PhD degree.
Date: ____________________
Supervisor:
______________________________
Dr. Syed Zulfiqar Ali Shah
Professor
Head of Department:
______________________________
Dr. Muhammad Tahir
Assistant Professor,
Department of Management Sciences
viii
DEDICATION
I dedicate my work to my beloved Parents and Family
ix
ACKNOWLEDGEMENT
First and foremost, I am grateful to my ALLAH Almighty who bestowed upon me the
knowledge, skills, courage and strength to complete this thesis. I pay my sincere gratitude
from the core of my heart to Holy Prophet Hazrat Muhammad (S.A.W) who is the sole reason
for creation of this universe and is the role model for the whole mankind.
I wish to pay my thanks to my supervisor, Dr. Syed Zulfiqar Ali Shah, Professor, Department
of Management Science, CIIT, Islamabad, who imparted his knowledge, broad experience
and positive vision to me. He has always been a source of inspiration for me since I started
my research work. I would never forget his unflinching readiness to help me out whenever I
was in a trouble, no matter how trivial the trouble was.
I am obliged to Dr. Khalid Riaz, Dean Faculty of Social Sciences and Dr. Qaiser Abbas,
former Dean Faculty of Social Sciences for their guidance and help in research work as well
as in administrative matters whenever I seek. Likewise, I am also grateful to Dr. Muhammad
Tahir, Head of the Department of Management Sciences and Dr. Zahid Iqbal, former Head of
the department of Management Sciences for their great help in administrative matters.
I owe my heartily gratitude to my lovely wife, Sadia Fayyaz Sheikh, sweet daughters,
Fareeha and Sabaina, and a charming son, Muhammad Ibraheem, as without their love,
cooperation and continuous encouragement this research work was not possible. They
cooperated in every respect throughout my PhD program right from the beginning till the
end. I would like to pay special thanks to my brothers and sisters particularly my brother
Muhammad Asad for his never ending encouragement and moral support throughout the PhD
program.
Finally, I am pleased to pay my thanks to my colleagues, friends and classmates especially
Mr. Ahsin Waqas, Senior Joint Director SBP, Mr. Shahid Mahmood, Mr. Fahim Javed, Mr.
Muhammad Shahid, Mr. Abdul Qayyum and Mr. Muhammad Waqas Maharvi for their
continuous support.
Muhammad Fayyaz
x
ABSTRACT
Executive Compensation, Firm Performance and
Corporate Governance: Evidence from Pakistan
The study aims at examining how compensation of chief executive officer (CEO) is
influenced by firm performance and corporate governance mechanisms such as
concentrated/family ownership and board structure in Pakistan. It further extends its
scope to analyzing the motivation aspects of CEO compensation towards future firm
performance and earnings management. The study uses a sample of 1508 firm-year
observations from 225 non-financial listed companies in Karachi Stock Exchange
(KSE) for period 2005 to 2012. Robust methodological procedures that control for
heteroskedasticity, serial correlation and endogeneity issues are used.
The study finds that current and previous year firm accounting performance as
measured by return on assets has significant influence on CEO compensation.
However, firm market performance does not have positive influence on CEO
compensation. Ownership concentration appears to have positive influence on CEO
compensation, indicating some sort of collusions between management and largest
shareholder to get personal benefits. CEOs who are also chairman board of directors
receive lower compensation as compared to their counterparts. Board size and board
independence have no convincing influence on CEO compensation in any direction,
indicating irrelevance of these mechanisms in CEO compensation decisions making.
There is weak evidence that CEOs in family firms receive lower pay as compared to
their counterparts. Further, it is found that CEOs are not rewarded for expected
future firm performance. Similarly, CEOs do not appear to induce earnings
management to receive higher compensations.
The study indicates that incentive based CEO compensation does not exist in
Pakistan. In addition, behavior of corporate governance variables in setting CEO
compensation appears to be pretty different from what is expected. Therefore, all
the stakeholders should be concerned and need to notice the decision-making
process within the boards in addition to ensuring the effective internal control
system to align the CEO interests with that of shareholders and restrain the possible
expropriation of minority shareholders’ interests.
xi
Keywords: Executive Compensation, CEO Pay, Corporate Governance, ROA, Stock
Returns, Earnings Management, Fixed Effects, Endogeneity, Non-Financial
Firms.
xii
TABLE OF CONTENTS
CHAPTER 1 ....................................................................................................................... 1
INTRODUCTION ............................................................................................................. 1
1.1 INTRODUCTION ............................................................................................................ 2
1.2 MOTIVATION AND RESEARCH OBJECTIVES .................................................................. 4
1.3 RESEARCH QUESTIONS ................................................................................................ 7
1.4 INSTITUTIONAL CONTEXT OF PAKISTAN ...................................................................... 7
1.4.1 The Code of Corporate Governance ......................................................................... 9
1.4.2 Brief Overview of Corporate Governance Practices in Pakistan ............................ 10
1.4.3 Corporate Governance under Concentrated/Family Ownership in Pakistan .......... 11
1.5 BRIEF SUMMARY OF RESULTS AND STUDY CONTRIBUTION ....................................... 12
1.6 STRUCTURE OF THE THESIS ....................................................................................... 15
CHAPTER 2 ..................................................................................................................... 17
LITERATURE REVIEW ............................................................................................... 17
2.1 THEORETICAL BACKGROUND .................................................................................... 18
2.1.1 Overview of Agency Theory ................................................................................... 18
2.1.2 Optimal Contracting Approach ............................................................................... 22
2.1.3 Managerial Power Approach .................................................................................. 23
2.1.4 CEO Compensation and Corporate Governance..................................................... 25
2.1.5 CEO Compensation and Ownership Structure ........................................................ 26
2.1.6 CEO Compensation and Performance Manipulation .............................................. 28
2.2 EXISTING EMPIRICAL EVIDENCE ................................................................................ 31
2.2.1 Executive Compensation and Firm Performance .................................................... 31
2.2.1.1 US Context ............................................................................................................... 31
2.2.1.2 Australian and European Context ............................................................................. 38
2.2.1.3 Asian Context ........................................................................................................... 43
2.2.1.4 Pakistani Context ...................................................................................................... 51
2.2.2 Effect of Executive Compensation on Firm Future Performance ........................... 52
2.2.3 CEO Compensation and Earnings Management ..................................................... 57
2.2.4 Summary of Empirical Research ............................................................................ 62
CHAPTER 3 ..................................................................................................................... 69
METHODOLOGY .......................................................................................................... 69
3.1 RESEARCH PHILOSOPHY ............................................................................................ 70
3.1.1 Ontological Considerations ..................................................................................... 70
3.1.2 Epistemological Considerations .............................................................................. 71
3.1.3 Methodological Considerations .............................................................................. 72
3.2 HYPOTHESES DEVELOPMENT ..................................................................................... 74
3.2.1 CEO Compensation and Firm Performance ............................................................ 74
3.2.2 CEO Compensation and Corporate Governance..................................................... 75
3.2.2.1 Ownership Concentration, Family Ownership and CEO compensation .................. 75
xiii
3.2.2.2 Board Size and CEO Compensation ......................................................................... 76
3.2.2.3 Board Independence and CEO Compensation ......................................................... 77
3.2.2.4 CEO Duality and CEO Compensation ..................................................................... 78
3.2.3 CEO Compensation and Future Firm Performance ................................................ 79
3.2.4 CEO Compensation and Performance Manipulation .............................................. 79
3.3 MEASUREMENT OF KEY VARIABLES.......................................................................... 80
3.3.1 CEO Compensation ................................................................................................. 80
3.3.2 Firm Performance ................................................................................................... 81
3.3.3 Corporate Governance Variables ............................................................................ 82
3.3.3.1 Family Ownership .................................................................................................... 82
3.3.3.2 Ownership Concentration ......................................................................................... 82
3.3.3.3 Board Size ................................................................................................................ 83
3.3.3.4 Board Independence ................................................................................................. 83
3.3.3.5 CEO Duality ............................................................................................................. 83
3.4 DATA ......................................................................................................................... 83
3.5 MODEL SPECIFICATIONS ............................................................................................ 86
3.5.1 CEO Compensation, Firm Performance and Corporate Governance ..................... 86
3.5.2 CEO Compensation and Past Firm Performance .................................................... 90
3.5.3 Controlling for Endogeneity ................................................................................... 90
3.5.3.1 Regressing on Lagged Values of Explanatory Variables ......................................... 90
3.5.3.2 Dynamic Panel Estimation using Generalized Method of Moments (GMM) .......... 91
3.5.4 CEO Compensation and Future Firm Performance ................................................ 93
3.5.5 CEO Compensation and Performance Manipulation .............................................. 95
3.5.5.1 Detecting Earnings Management .............................................................................. 95
3.5.5.2 Regression Model ..................................................................................................... 96
CHAPTER 4 ..................................................................................................................... 99
CEO COMPENSATION, FIRM PERFORMANCE AND CORPORATE
GOVERNANCE: EMPIRICAL RESULTS AND DISCUSSION .............................. 99
4.1 DESCRIPTIVE STATISTICS ......................................................................................... 100
4.1.1 CEO Compensation ............................................................................................... 100
4.1.2 Firm Performance ................................................................................................. 100
4.1.3 Corporate Governance .......................................................................................... 101
4.1.4 Control Variables .................................................................................................. 104
4.2 CEO COMPENSATION AND REALIZED FIRM PERFORMANCE .................................... 105
4.2.1 Correlation Analysis ............................................................................................. 105
4.2.1.1 Addressing Multicollinearity .................................................................................. 106
4.2.2 Regression Results ................................................................................................ 108
4.2.2.1 Total Compensation ................................................................................................ 108
4.2.2.2 Cash Compensation ................................................................................................ 113
4.2.2.3 Regression on Lagged Explanatory Variables ........................................................ 118
4.2.2.4 Dynamic Panel Model – GMM Estimation ............................................................ 120
4.2.3 Robustness Checks ................................................................................................ 122
4.3 CEO COMPENSATION AND FUTURE FIRM PERFORMANCE ....................................... 125
4.3.1 Correlation Analysis ............................................................................................. 125
xiv
4.3.1.1 Addressing Multicollinearity .................................................................................. 125
4.3.2 Regression Results: Compensation and Future Firm Performance ....................... 127
4.3.3 Robustness Checks ................................................................................................ 128
CHAPTER 5 ................................................................................................................... 130
CEO COMPENSATION AND EARNINGS MANAGEMENT: EMPIRICAL
RESULTS AND DISCUSSION .................................................................................... 130
5.1 CEO COMPENSATION AND EARNINGS MANAGEMENT ............................................. 131
5.1.1 Descriptive Statistics ............................................................................................. 131
5.1.2 Correlation Analysis ............................................................................................. 132
5.1.2.1 Addressing Multicollinearity .................................................................................. 132
5.1.3 Regression Results: Compensation and Earnings Management ........................... 135
5.1.3.1 Income Increasing Discretionary Accruals ............................................................. 137
5.1.3.2 Income Decreasing Discretionary Accruals ........................................................... 138
5.1.4 Robustness Checks ................................................................................................ 138
CHAPTER 6 ................................................................................................................... 142
CONCLUSIONS AND POLICY IMPLICATIONS .................................................. 142
6.1 CONCLUSIONS.......................................................................................................... 143
6.2 POLICY IMPLICATIONS ............................................................................................. 146
6.3 LIMITATIONS AND SCOPE OF FUTURE RESEARCH .................................................... 147
REFERENCES .............................................................................................................. 149
xv
LIST OF FIGURES
Figure 3.1 Research Philosophy………………………………....………... 73
Figure 4.1 Average CEO Compensation over the Sample Period.………... 104
xvi
LIST OF TABLES
Table 3.1 Industry wise Observations ……………………………….……………. 85
Table 3.2 Observations by Pay Components ……………………………………… 85
Table 3.3 Heteroskedasticity and Serial Correlation tests ………………………... 87
Table 3.4 Hausman specification test for random effect model ……………….…. 88
Table 4.1 Descriptive Statistics……………………………………………………. 103
Table 4.2 Correlation Matrix (Variables of Model 1, 2, 3 & 4) ……………………. 107
Table 4.3 Variance Inflation Factor (Variables of Model 1, 2, 3 & 4) …….………. 108
Table 4.4 Total Compensation and Firm Performance (Results)………………...... 116
Table 4.5 Cash Compensation and Firm Performance (Results)………………....... 117
Table 4.6 CEO Compensation and Firm Performance (Regression on lagged
explanatory variables) ………….…………………………………….…. 119
Table 4.7 CEO Compensation and Firm Performance (Dynamic Panel Model-
GMM Estimation) ……………………………………………….………. 124
Table 4.8 Correlation Matrix (Variables of Model 6) ……………………...………. 126
Table 4.9 Variance Inflation Factor (Variables of Model 6) ………………………. 127
Table 4.10 CEO Compensation and Future Firm Performance (Results)…………… 129
Table 5.1 Descriptive Statistics (CEO Compensation and Earnings Management). 133
Table 5.2 Correlation Matrix ………………………………………………….…… 134
Table 5.3 Variance Inflation Factor ….….….….….….….….….….….….….….…. 135
Table 5.4 CEO Compensation and Earnings Management (Results)…………….... 139
Table 5.5 CEO Compensation and Income Increasing Discretionary Accruals….... 140
Table 5.6 CEO Compensation and Income Decreasing Discretionary Accruals…... 141
xvii
LIST OF ABBREVIATIONS
CEO Chief Executive Officer
OCA Optimal Contracting Approach
MPA Managerial Power Approach
BMH Bonus Maximization Hypothesis
SECP Security and Exchange Commission of Pakistan
SBP State Bank of Pakistan
KSE Karachi Stock Exchange
AGM Annual General Meeting
OLS Ordinary Least Square
GMM Generalized Methods of Moment
TMT Top Management Team
ROA Return on Assets
EPS Earnings per Share
SOE State Owned Enterprise
1
Chapter 1
Introduction
2
1.1 Introduction
There has been enormous growth in research on executive compensation over the last
two decades with primary focus on compensation of chief executive officer (CEO). For
instance, Murphy (1999) notes that “CEO pay research has grown even faster than CEO
paychecks”. The research on CEO compensation has become multidisciplinary with greater
part coming from finance and management disciplines. Most of the finance literature on CEO
compensation is grounded in agency theory (Murphy, 1999).
Agency theory, proposed by Jensen and Meckling (1976), analyzes the relationship that
emerges between managers (agents) and owners (principals) in an economic environment
when principals delegate authority to the agents to act in their name, so that their wealth is
maximized by the agents’ decisions (Cuevas-Rodríguez et al., 2012). The agents, presumed to
be self-interested in agency theory, may not act in the best interests of the principals, giving
rise to costs, called agency costs, to the principals. Agency theory focuses on minimizing the
agency costs by aligning the interests of agents with those of principals using efficient
compensation contracts. This is called optimal contracting approach (OCA) (see, Bebchuk
and Fried, 2003). Optimal contracting does not mean “ideal” contracting; instead it means
that firm makes best effort to construct a contract within the contracting restraints to avoid
managers’ opportunistic behavior (Conyon, 2006). Optimal contracting attempts to trade-off
between marginal benefits and marginal costs related to the contract. Evolutions in corporate
governance or regulations may possibly change these costs and benefits and hence the
structure of the contract (Conyon, 2006). Therefore, today’s efficient contract may not be
efficient tomorrow.
In optimal contracting approach, the corporate boards, assuming the power to look after
the firm, involve in arm’s length transaction1 with CEO and design such compensation plans
which provide CEO with efficient incentives to maximize the shareholder value. This
predicts a link between CEO compensation and corporate performance. Better monitoring as
induced by effective board, concentrated/family ownership and board independence
contributes towards the level of CEO compensation and its relationship with firm
performance (see, Devers et al., 2007).
Contrary to optimal contracting approach, the other view to analyze the CEO
compensation is the managerial power approach (MPA) advocated by Bebchuk and Fried
(2003). MPA defies the fundamental assumption of optimal contracting that boards involve in
1 An arm’s length transaction is one in which two parties act independently in their own self-interest with balanced bargaining power and without any pressure or any duress from the counter party.
3
arm’s length transactions with managers over compensation arrangements and that such
transactions help reduce the agency problems by designing pay contracts that align the
interests of managers with those of shareholders (van Essen et al., 2012b). MPA states that
CEOs, being in power in a dispersed shareholding environment, set their own pay. However,
this pay is not unlimited rather subject to some potential “outrage” constraints and costs
stemming from an “outrage” – a negative reaction of relevant outsiders to the proposed
compensation contract. The optimal contracting approach considers executive compensation
as a tool to align the interests of agents and principals while MPA considers executive
compensation as an agency problem itself (see, Bebchuk and Fried, 2004).
Bebchuk and Fried (2003) and Bebchuk and Fried (2004) mention that MPA is not a
replacement for principal-agent model rather simply that pay practices cannot be explained
by principal-agent model alone. MPA predicts that CEOs compensation will be greater and/or
less sensitive to firm performance if CEOs are relatively more powerful than the board of
directors (see, Bebchuk and Fried, 2003). Weak boards in relation to CEO, CEO duality,
higher percentage of outside shareholders and less board independence, all contribute to
relative power of CEO (see, van Essen et al., 2012b).
The empirical evidence does not fully support any of the two approaches (OCA or
MPA). For example, several studies (e.g., Chalmers et al., 2006, Conyon and He, 2011, Frye,
2004, Hermalin and Wallace, 2001, Jensen and Murphy, 1990b, Laan et al., 2010, Mehran,
1995, Murphy, 1999, Ozkan, 2011) document a significant positive link between CEO
compensation and firm performance which is consistent with optimal contracting approach.
However, several other (see, Bebchuk et al., 2002, O'Reilly and Main, 2010, van Essen et al.,
2012b) find insignificant or weak pay-performance link along with excessive CEO
compensation which is consistent with MPA. Therefore, empirical evidence is mixed.
In a recent meta-analytic study assessing MPA against optimal contracting, van Essen et
al. (2012b) suggest that executives can and do influence their compensation and their power
can be controlled by principals. However, they do not endorse MPA as the only explanation
for executive compensation especially when it comes to pay-performance relationship. In
sum, both optimal contracting and managerial power influence the pay setting process.
Concluding their study, van Essen et al. (2012b) state that MPA and optimal contracting “do
not represent competing explanations but describe points on a continuum of types of
contracting arrangements that can be encompassed within agency theory”.
Although CEO compensation has spawned a large body of literature, however, most of
the literature is focused on developed countries like USA, Canada, UK and other European
4
countries (see, Devers et al., 2007, Frydman and Jenter, 2010, Gomez-Mejia and Wiseman,
1997, Tosi et al., 2000, van Essen et al., 2012a, van Essen et al., 2012b). This literature may
not have implications for developing countries as many economists strongly argue that there
are basic structural and institutional differences in markets and organizations of developing
and developed countries (Ghosh, 2006). In developing countries, the institutional
environment makes application of agency contracts difficult and costly (Young et al., 2008),
leading to dominance of concentrated ownership (Dharwadkar et al., 2000). In addition,
emerging markets are also characterized with weak corporate laws such as listing agreements,
codes of corporate governance and bankruptcy laws (Ghosh, 2006). When concentrated
ownership combines with ineffective external governance system, more conflicts emerge
between controlling shareholders and minority shareholders (Young et al., 2008). Besides
that, managerial markets are not well developed in emerging markets (Ghosh, 2006). All
these characteristics of developing economies are materially different from the characteristics
of developed economies (see, Fan et al., 2011), hence, provide an interesting framework to
study the relationship between CEO compensation and firm performance in the light of
theories evolved in developed economies.
1.2 Motivation and Research Objectives
In Pakistan, concentrated and family ownership environment accompanied with weak
governance systems suggests a tendency of expropriation of minority shareholders’ interests
by controlling shareholders through various rent seeking activities. For instance, Abdullah et
al. (2011) show that insider-controlling shareholders in Pakistan play a dominant role and
expropriate the external shareholders’ interests by not paying out dividends especially in the
absence of powerful external shareholders. Similarly, executive compensation is another way
by which interests of minority shareholders can be expropriated. In Pakistan, executives are
generally part of controlling families or have direct association with controlling shareholders
Therefore, CEOs tend to be more powerful than the boards of directors. The boards of
directors act more like rubber stamps and just approve the decisions made somewhere else
(Javid and Iqbal, 2008, Kamran and Shah, 2014). Bebchuk and Fried (2003) argue that
powerful CEOs set their own pay excessively which may not be justified against firm
performance. However, whether CEO compensation in Pakistan is justified against firm
performance is virtually an unexplored issue in Pakistan.
Securities and Exchange Commission of Pakistan (SECP) issues the first Code of
Corporate Governance in March 2002. The purpose of the Code is to improve corporate
5
governance practices and reduce the trust deficit among the business community, owners, and
agents. One of the primary focuses in the Code is improvement of board structure. Board
independence and separation of position of chairman board of directors from CEO are the
main focal points. However, no provision of compensation committee exists until 2012. In
addition, unlike many other Asian countries, executive compensation decisions do not require
approval from the shareholders in Pakistan. Further, the adoption of best practices of
corporate governance such as board independence and separate positions for CEO and
chairman remains a voluntary choice for the companies until 2012. In this context, how board
structure contributes towards CEO compensation decisions draws attention especially when
executive pay packages are sole responsibility of the boards and are not discussed with the
shareholders in annual general meetings.
A well-established fact is that Asian socio-economic and institutional settings (formal
and informal) are different from Western World and studies conducted in Western World
have limited implications for Asian countries (Fan et al., 2011, Ghosh, 2006, Hofstede, 1980,
Sun et al., 2010, van Essen et al., 2012a). However, within Asia, Pakistani context is different
from other Asian countries like China, Japan and Korea where most of the Asian studies on
executive compensation are based. Amongst others, Pakistani context is different in respect
of CEO pay disclosure and concentrated/family control. Disclosure requirement about CEO
compensation is stronger in Pakistan. Companies are required to report all the components of
CEO compensation. This is not the case for most of the other Asian countries (see, e.g., Basu
et al., 2007, Conyon and He, 2011, Kato et al., 2007). Therefore, there is more accurate data
available in Pakistan which can provide better insight into CEO compensation relationships
with firm performance and corporate governance. Similarly, concentrated and family
ownership is more common in Pakistan than in Japan and Korea. Although, in China, firms
have more ownership concentration than in Pakistan however that ownership concentration is
due to high stake of the government. In Pakistan, concentrated ownership is maintained by
non-government shareholders. Concentrated/family ownership can reduce CEO entrenchment
because of better monitoring incentives which in turn lead to lower CEO compensation.
However, very little is known about how concentrated/family ownership affect CEO
compensation in Pakistan.
Despite considerable importance of the subject of CEO compensation, the evidence from
Pakistan is virtually absent. To the best of my knowledge, there are only two published
studies (i.e., Kashif and Mustafa, 2012, Shah et al., 2009) that focuses on CEO compensation
in Pakistani context. Both studies explore the determinants of CEO compensation and finds
6
that firm performance is not an important determinant. However, they are limited in scope.
They do not control for unobserved heterogeneity in the firms and used only pooled
regression model. Introduction of firm fixed effects and time fixed effects to control for
unobserved heterogeneities may change the entire results. Further, they ignore the
endogeneity problem which has been vastly documented in the corporate governance field
(see, e.g., Conyon and He, 2012, Devers et al., 2007). In addition, they consider only current
firm performance, however, several studies have documented the relationship of CEO
compensation with past firm performance (see, Devers et al., 2007, Frydman and Jenter,
2010, Sun et al., 2010).
Besides that, in Pakistan, Shah et al. (2009) report that about 99% of variation (as
measured by R-square) in CEO compensation is explained by the variables they introduce in
the pooled regression. This casts some doubts on their findings especially when they have not
used fixed effects to control for unobserved heterogeneity. On a soft side, previous studies
(Kashif and Mustafa, 2012, Shah et al., 2009) in Pakistan use relatively short data period and
small sample size. The current study attempts to account for all the shortcomings of previous
which makes this study different from them.
Several other studies (e.g., Awan, 2012, Ibrahim et al., 2010, Javed and Iqbal, 2006,
Javid and Iqbal, 2008, Khatab et al., 2011, Nishat and Shaheen, 2006-2007, Yasser et al.,
2011) examine the impact of corporate governance on firm performance in Pakistan,
however, none of these examines the relationship of executive compensation with firm
performance and corporate governance. This reflects lack of research on the subject area,
thus, providing great incentive to study CEO compensation in Pakistan.
Given the above discussion, the study has the following objectives:
1. To examine the relationship between CEO compensation and firm performance in
Pakistan.
2. To explore how concentrated/family ownership and board structure influence the
CEO pay setting process.
3. To investigate the relationship between CEO compensation and manipulation of firm
accounting performance.
4. To overcome the methodological difficulties of the existing studies on CEO
compensation in Pakistan and perform rigorous analysis using more robust
estimation techniques including Generalized Methods of Moments (GMM).
7
1.3 Research Questions
The literature on the executive pay-performance relationship has two prominent
dimensions. One considers the pay as reward for the realized (current and past) firm
performance or as a mean of ex-post settling up (e.g., Fama, 1980) and test the influence of
firm performance on pay while, other conceptualizes the pay as a motivational tool and test
the effect of pay on firm performance (Devers et al., 2007). These two lines of research are
fairly grounded in agency theory. And agency theory posits that CEO compensation is related
to firm performance (past, current and future). Further, the recommended best practices in the
Code of corporate governance in Pakistan are generally justified using arguments based in
agency theory such as separation of CEO position from chairman board of directors and
provision for independent directors in the boards.
Therefore, given the research objectives and framework of agency theory, this study
investigates the following research questions in Pakistani context which is characterized by
concentrated/family ownership structure and weak legal environment.
1. Does realized (current and past) firm performance contribute towards CEO pay
setting process in Pakistan?
2. Does CEO compensation have any relationship with future firm performance in
Pakistan?
3. Does concentrated/family ownership structure influence CEO compensation in
Pakistan?
4. Does the board structure have any effects on CEO compensation decisions in
Pakistan?
5. Does CEO compensation affect earnings management decisions in Pakistan?
1.4 Institutional Context of Pakistan
Pakistani context is complex and unique in respect of its Indian sub-content origin with
traits of a collectivist society, with higher preference for uncertainty avoidance, and relatively
higher power distance (Hofstede, 1980). Further, Pakistani context is characterized by the
influence of British legacy, US influence, and its portrayal as an Islamic republic (see, e.g.,
Khilji, 2003) in the face of heightened political instability, prolonged and fragile law and
order situation, and corrupt social and economic environment. Pakistani context is considered
to be highly vulnerable to opportunistic behavior due to weak governance and political
systems (Mujtaba and Afza, 2011). This opportunistic behavior in general may have effects
on CEO compensation contracts. For example, opportunistic and powerful CEOs as
8
compared to the boards can set their own pay without the fear of check and balance, and that
is not justified against firm performance.
After the Asian financial crisis 1998 and corporate failure like Enron, effective
institutional framework becomes an important consideration (Javid and Iqbal, 2010). The
establishment of the Security and Exchange Commission of Pakistan (SECP) represents an
important milestone in the development of the regulatory framework of the capital markets in
Pakistan. The commission is established in under SECP Act, 1997 and becomes operational
in January, 1999. The continuing challenges of the Commission has been 1) to develop
regulatory principles to modernize the corporate sector of Pakistan to meet the international
standards and 2) to develop an efficient and dynamic regulatory body to foster principles of
good governance and protect investors through responsive policy measures and enforcement
practices (Javid and Iqbal, 2010).
To encourage corporate governance in Pakistan, Pakistan Institute of Corporate
governance (PICG), a not-for-profit organization, is established under Section 42 of
Company ordinance, 1984. SECP, State Bank of Pakistan (SBP), stock exchanges of Pakistan
and certain other banking/insurance institutions are founding members of PICG under public
private partnership. In 2006, PICG in collaboration with International Financial Corporation
(IFC) and SBP conducts a conference on reforms in Pakistan. The primary objective of the
conference is to develop understanding of the need for good governance among Pakistan’s
banking and financial sectors.
Stock markets play a vital role in channeling the investments for corporations in
developed world as well as in developing countries. Three stock exchanges, Karachi Stock
Exchange (KSE), Lahore Stock Exchange (LSE) and Islamabad Stock Exchange (ISE) were
working in Pakistan. However, recently in 2016, these are merged to form one stock
exchange, Pakistan Stock Exchange (PSX). The stock exchange is linked to Central
Depository System (CDS). Central Depository Company of Pakistan Limited (CDC) is
established in 1993 which became operational in 1997 to manage and operate the Central
Depository System (CDS). CDS is an electronic book entry system to record and transfer
securities without changing hands physically. Almost all the settlements of the stocks are
done through CDS. Thus, CDC is backbone of smooth and efficient settlement in Pakistani
capital markets.
9
1.4.1 The Code of Corporate Governance
SECP issues the first corporate governance Code in March 20022 to improve governance
practices and transparency and to protect investors’ interest by upgrading disclosure
requirements. Issuance of the Code is result of the joint efforts by SECP and Chartered of
Pakistan in collaboration with Institute of Cost and Management Accountants of Pakistan
(ICMAP). The Code includes more than 40 clauses in line with international good practices
of corporate governance. All the listed firms are required to publish and circulate a statement
of compliance with the best practices of corporate governance along with annual reports.
However as noted by Ibrahim (2006) that “Pakistan’s common law background, the country’s
legal system resembles the Anglo-American model; however, Pakistan’s ownership structure
is the opposite of the Anglo-American structure of dispersed ownership. The issuance of the
Code ignores this distinction and benefits from the U.K. and South African reform
initiatives” (p.328).
Primarily, the Code is aimed at establishing systems whereby companies are directed and
controlled by their boards of directors in compliance with the best practices around the world
to protect diversified shareholders’ interests. The Code proposes restructuring the
composition of the board of directors in order to introduce broad based representation by
minority shareholders and by executive and nonexecutive directors. Transparency and
openness is emphasized in decisions making and corporate affairs, and directors are urged to
perform their fiduciary duties in the largest interests of all the shareholders in diligent,
informed, transparent and timely fashion. The salient features of the Code include, emphasis
on separation of CEO position from chairman board of directors and appointment of
independent directors, and setting up of audit committees and internal audit functions by all
listed firms.
In August 2002 SECP, in collaboration with UNDP and Economic Affairs Division of
Government of Pakistan, launches a project for implementation of the best corporate
governance practices prescribed in the Code of corporate governance and strong regulatory
framework for the corporate sector in Pakistan. In 2007, SECP, IFC and PICG conduct a
Survey on “Code of Corporate Governance of Pakistan”. The survey is targeted to the
financial institutions, listed and large non-listed firms. The survey comes up with key finding
among others that there is a strong need for creating awareness amongst the directors of the
companies about the benefits of the Code so that they could go further than the tick-box
2 The Code of corporate governance is revised by SECP in March, 2012 applicable from year 2013 with some tightened corporate governance requirements.
10
approach to implement the Code (Javid and Iqbal, 2010). SECP develops a board
development series (BDS) with the help of IFC. PICG has conducted many workshops for the
purpose of understanding corporate governance and responsibilities of boards of directors.
SECP has been making efforts to enforce corporate governance regulations. SECP has
received technical help from Asian Development Bank (ADB) to improve corporate
governance enforcement program and from World Bank to improve awareness and training.
Other elements of enforcement system such as ICAP and stock exchanges are not that
convincing (see, Javid and Iqbal, 2010).
1.4.2 Brief Overview of Corporate Governance Practices in Pakistan
In Pakistan, the basic shareholders’ rights are protected at least by laws in the book. The
registration is secured through CDC. Shareholders are able to get variety of information
directly from the company. They have right to participate in Annual General Meetings
(AGM). Shareholders have cumulative voting rights to elect the directors who can be
removed as well through shareholder resolution. Shareholders’ approval is required for
alteration on the company articles, change in authorized capital, sale of company’s assets and
mergers and acquisitions. While compliance has improved through effective enforcement
however some companies do not seem to be interested in holding annual general meetings
(AGM) (Javid and Iqbal, 2010). The law does not support electronic voting or voting by post.
Influence of minority shareholders is very limited due to concentrated control which
effectively reduces the protection of minority shareholders’ interests from abuse.
The board meetings and directors’ accountability to non-family directors are virtually
absent in many family dominated firms. Some companies also do not pay dividend on time,
and take longer than 5 days to re-register shares in the name of depository (Abdullah et al.,
2011, Javid and Iqbal, 2010). As appointment of independent director is encouraged but not
mandatory in the Code till 2012, therefore it remains a voluntary choice. Also the assessment
of independence directors is based on very scanty criteria till 20123. In addition, since
separating the CEO position from chairman is encouraged in the Code but not mandatory till
2012 therefore many companies do not separate offices of the chairman board and CEO
which is contradicting with best corporate governance practices4. Since compliance with the
Code is weakly defined therefore listed companies are prone to submit so called verified
3 In the revised Code issued in 2012, it is a mandatory requirement that board should include at least one independent director. The assessment criteria for independent director have also been improved. 4 This requirement has been improved with mandatory separation of position of chairman from CEO in the new Code issued in March, 2012.
11
statement of compliance verified by auditors who may sign the statement without ensuring
the actual position. Further, the boards are not evaluated for their performance.
1.4.3 Corporate Governance under Concentrated/Family Ownership in
Pakistan Similar to other emerging markets, corporate governance and ownership structure in
Pakistan is tempting. Family and concentrated ownerships are common in Pakistani firms.
Family owned firms are typically managed by owners themselves, presumably to avoid
agency problems such as misappropriation or power abuse which may arise if executives are
hired from outside family (Kamran and Shah, 2014). In business groups, corporate boards are
dominated by executive and non-executive members from controlling family or by proxy
directors employed to act on behalf of the family. Majority control is often maintained
through interlocking directorship, complex pyramid structures, cross shareholdings, voting
agreements and/or dual class voting shares. This way, the ultimate owners keep (voting)
control while having a smaller fraction of ownership (cash flow rights) (Javid and Iqbal,
2008). Consequently, the dominant shareholders make the decisions without bearing the full
cost.
In instances where firms are multinational or state owned, typically, there is a direct
relationship between management and state/foreign owner, bypassing boards in making
important corporate decisions5 (Ameer, 2013, Kamran and Shah, 2014, World Bank, 2005).
In such a corporate environment more shareholder to shareholder conflicts arise than the
conflicts between shareholders and managers. Controlling shareholders or families promote
their own objectives and expropriate minority shareholders through tunneling or other rent
seeking activities. This environment is different from the environment in developed countries
like US and UK where ownership is diffusely held and where agency problems generally
arise between managers and shareholders (see, Fan et al., 2011). Although agency theory
predicts that concentrated and family ownership provide strong incentives for monitoring,
leading to reduced agency problems. However, the situation in Pakistan seems different due
to higher opportunities of expropriation.
Given the Pakistani context, it is worth exploring that how CEO compensation contracts
are related to firm performance and whether performance manipulation is related to CEO pay
in Pakistan. Understanding top executives’ compensation in Pakistan would benefit
stakeholders in other Asian markets as they can update their executive pay setting processes
5 For detail discussion on corporate governance in Pakistan see Cheema (2003), Cheema et al. (2003),Ibrahim (2006) and Burki (2012)
12
and take measures to solve the manager-shareholder and shareholder-shareholder agency
problems.
1.5 Brief Summary of Results and Study Contribution
Using a dataset consisting of 1508 firm-year observations from 225 listed firms in KSE
for period 2005 to 2012 and modern methodological procedures, this study finds that current
and previous year firms’ accounting performance have significant influence on CEO
compensation in Pakistan. However, market performance does not significantly affect CEO
compensation positively. Thus, CEOs in Pakistan are rewarded against realize firm
accounting performance only. As of motivation aspect of CEO compensation, higher CEO
compensation does not appear to induce better future firm performance. Similarly, no
evidence is found for compensation induced earnings management by CEOs.
As of corporate governance variables, ownership concentration appears to have a
positive relationship with CEO compensation, indicating that higher ownership concentration
leads to higher CEO pay. This is an indication of some sort of collusions between
management and largest shareholder to get personal benefits. Contrary to managerial power
hypothesis, CEOs who also hold the position of chairman board of directors (CEO duality)
agree to receive lower compensation as compared to their counterparts. This is contrary to
existing evidence in US, Europe and other Asian countries. Further analysis shows that this
negative association of CEO duality with CEO compensation is driven by family firms in
Pakistan which means that most of the CEOs in family firms who are also chairman board of
directors receive lower compensation as compared to CEOs holding only one position. The
findings also suggest that CEOs in family firms are paid lower compensation as compared to
CEOs in non-family firms. However, this finding is not consistently found in all model
specifications.
Board size and board independence, have no convincing influence on CEO
compensation in any direction, indicating irrelevance of these variables in CEO compensation
decisions. This seems consistent with Pakistani context as there is common perception that
boards are just rubber stamps in the hands of large shareholders. Thus, board size or number
non-executive do not matter in corporate decision making process.
An important finding of the study is that dynamic panel model shows that CEO pay is
persistent and takes time to adjust to long-run equilibrium. This indicates that boards of
directors in Pakistan adjust CEO compensation to target levels as they learn more about CEO
13
actions over time. Firm size appears to be a very important factor that contributes towards
higher CEO compensation.
Given the results, the study contributes in a number of ways. First, Fan et al. (2011)
notes that “Until now, we still do not know much about how managers of emerging market
firms are paid and promoted and factors that influence these decisions” (p. 211). Thus, by
analyzing CEO compensation in an emerging market of Pakistan, the study provides
important contribution in international literature on executive compensation. The findings of
this study can be used to shed light on host of other Asian markets with similar institutional
settings.
Second, the study provides first hand evidence in Pakistan that firm accounting
performance does contribute towards CEO compensation decisions. This is contrary to
existing evidence in Pakistan. Thus, this study challenges the existing studies, Shah et al.
(2009) and Kashif and Mustafa (2012), which find no contribution of firm accounting
performance in CEO compensation decisions. In addition, since Pakistani firms are
dominated by family ownership therefore positive contribution of firm accounting
performance towards CEO compensation seems to be contradicting to the view that family
firms do not have pay-performance relationship. This view seems true in case of some Asian
countries like Korea (Kato et al., 2007) where pay-performance relationship is not found for
family firms. Thus, by finding a positive pay-performance relationship for accounting
performance, this study adds to the debate on pay-performance relationship in family
dominated ownership environment in Asian context.
Third, the study highlights that corporate governance variables do not influence or
influence CEO compensation opposite to the expected direction. For example, board structure
variables (board size and board independence) have no influence while CEO duality has
unexpected negative association with CEO compensation. Similarly, ownership concentration
has positive association with CEO compensation which is opposite to the view of agency
theory. All these findings have important policy implications for practitioners and regulators
in Pakistan. For example, according to agency theory on which Code of Corporate
Governance based, combining CEO position with chairman board of directors is considered
to be an unhealthy practice which should lead to misalignment of mutual shareholders-
management interests. However, negative association of CEO duality suggest it is beneficial
for the firm as it leads to lower CEO compensation and better alignment of mutual interests.
Thus, this cast some important policy implications. Similar is the case with ownership
concentration. The positive association between ownership concentration and CEO
14
compensation is indicating some sort of hidden arrangements between largest shareholders
and CEOs which needs attention of policy makers and regulators.
Fourth, an important finding of the study is that CEO duality is negatively associated
with CEO compensation. This finding challenges the managerial power hypothesis. Also, it
contradicts with the existing empirical evidence. The study shows that this finding is driven
by family firms in Pakistan. Thus, this study highlights that behavior of CEO duality in
family firms is different as compared to non-family or diffusely held firms. This is an
important contribution of the study which warrants more research on the behavior of CEO
duality in area of management and finance.
Fifth, an important feature of studying influence of firm performance and corporate
governance practices on CEO compensation in Pakistani context is that the corporate
governance regulations in Pakistan have been paid a great attention during the last decade or
so with strong emphasis on board independence and separation of CEO position from
chairman board of directors. This study conducts the first in-depth examination of the
relationship between CEO pay practices and board structure variables with rich dataset and
using advance methodological procedures since the issuance of Code of Corporate
Governance (2002) in Pakistan. It is particularly important when the Code confers all the
responsibility to set executive compensation to the boards and itself provides no guidelines
and when remuneration committee is not required until the reissuance of the Code in 2012.
This study is expected to shed light on the effectiveness of the recommendations provided in
the Code for improvement of corporate governance practices. In addition, it would guide the
remuneration committees in firms which are mandatory from year 2013 on what different
factors can contribute towards devising executive compensation contracts. Moreover, in
future, this study would server as anchor for comparing the effectiveness of the recommended
corporate governance practices related to board structure and remuneration committee in the
Code of Corporate Governance (2002) and revised Code of Corporate Governance (2012).
Sixth, the study provides first hand evidence that CEO compensation in Pakistan neither
influences future firm performance nor it induces earnings management behavior. This
provides important directions for future research on the causes of these insignificant
relationships and influence of compensation on other actions of CEOs, for example, on
dividend policy, capital structure decisions and asset management decisions etc.
Seventh, the study is one amongst few international studies and first in Pakistan that
investigate dynamic nature of CEO pay using dynamic panel model. Thus, unlike most of the
15
existing studies on CEO compensation, this study accounts for adjustment process of CEO
compensation as indicated by Conyon and He (2012).
Finally, the study also has important contributions on methodological front. For example,
study uses fixed effect models which control for unobserved heterogeneity across the cross-
sectional groups. To control for heteroskedasticity and serial correlation in error terms, robust
standard errors clustered at firm level are also used. Certain measures are also taken to
control the endogeneity problems. Importantly a dynamic panel model is also estimated using
three well-known generalized methods of moment (GMM) techniques. In sum, rich
methodological procedures are adopted to produce robust results. Thus, the results provided
in this study are robust and not likely to be plagued with methodological difficulties like
existing studies (Kashif and Mustafa, 2012, Shah et al., 2009) in Pakistani context.
1.6 Structure of the Thesis
This chapter of the thesis provides background of the subject area, motivation and
objectives of the research followed by specific research questions, institutional context of
Pakistan and brief summary of results and study contributions. The remainder of thesis is
organized as follows. Chapter 2 presents the literature review which is divided into two
sections. While, first section provides overview of theoretical frameworks such as agency
theory, optimal contracting and managerial power hypothesis that are mostly used in the
literature to explain the relationship of CEO compensation with firm performance and
corporate governance practices, the second section compiles the empirical evidence related to
these relationships. The empirical evidence is further divided on the bases of the contexts
under examination, for example, US context, European context and Asian context. Chapter 3
describes the research methodology adopted in this thesis. The chapter starts with research
philosophy followed by development of hypotheses, measurement and operationalization of
key variables such as CEO compensation, firm performance and corporate governance
variables. Then, econometric models used in this research are discussed. The tests for choice
of appropriate models such as fixed effect or random effect models are also discussed in
detail. Chapter 4 and 5 describes the empirical results. Chapter 4 presents results related to 1)
the influence of firm performance and corporate governance practices on CEO compensation
and 2) the effect of CEO compensation on future firm performance, while Chapter 5 presents
the empirical results related to the influence of CEO compensation of earnings management
decisions. Each chapter starts with discussion of descriptive statistics followed by correlation
matrices, variance inflation factors, regression results and robustness checks. Finally,
16
conclusions are presented in Chapter 6 followed by policy implications and limitations and
scope of future research at the end.
17
Chapter 2
Literature Review
18
The research on CEO compensation is voluminous and multidisciplinary. Financial
economists, sociologists and psychologists are putting their efforts to understand the
determinants of CEO compensation. Accordingly, much theoretical and empirical work has
been done. A number of theories have been applied to explain the CEO compensation and its
relationships with firm performance and corporate governance and socio-economic variables.
Countless empirical studies can also be found on the subject area which examines the
implications of theoretical explanations. As focus of this study is on the association of CEO
compensation with firm performance and manipulation of firm performance, therefore only
related work is reviewed. The review is divided into two sections. First section aims to
discuss dominating theoretical work in analyzing CEO compensation. Second section
compiles the related empirical work.
2.1 Theoretical Background
Several theories can and have been used to understand CEO compensation. However, the
most dominant approach used as a theoretical background is optimal contracting approach
derived from agency theory. In fact, Bebchuk and Fried (2004) label this approach as
“official story” which contends that executive compensation is a tool to mitigate agency
problems and to align executives’ interests with that of shareholders. The other dominant
approach is managerial power approach which argues that executive compensation is an
agency problem itself because executives are relatively more powerful than the boards
therefore they can significantly influence their own pay. This section first provides overview
of agency theory then review optimal contracting and managerial power approaches and
documents their implications about the relationship between CEO compensation and firm
performance. After that theory about how corporate governance and ownership concentration
may influence CEO compensation is discussed. Finally, theoretical work regarding CEO
compensation and manipulation of firm performance is covered.
2.1.1 Overview of Agency Theory
Agency relationship exists when actions of one person affect his/her own interests as
well as interests of another person or a group of persons in an implicit or explicit contractual
relationship. Jensen and Meckling (1976) define agency relationship as “a contract under
which one or more persons (the principal(s)) engage another person (the agent) to perform
some service on their behalf which involves delegating some decision making authority to the
agent”. However, the agent, assumed to have his own interests, may not always act in best
19
interests of the principals. This gives rise to conflicts between agents and principals, called
agency problems. Agency theory resolves the agency problems by aligning the interests of
agents with that of principals using efficient (optimal) compensation contracts. Primarily, the
theory explains that how contracting parties design contracts to minimize the costs related to
agency problems. The review of literature suggests that agency theory has been the most
widely used theoretical framework to analyze the executive compensation (see, Bebchuk and
Fried, 2004, Bratton, 2005, Devers et al., 2007, Eisenhardt, 1989, Frydman and Jenter, 2010,
Gomez-Mejia and Wiseman, 1997, Murphy, 1999, van Essen et al., 2012b). Therefore, it is
important to understand the agency theory and how this theory addresses the agency
problems.
The agency literature has developed in two dimensions: a mathematically complex
agent-principal literature and a more empirically-oriented Positive Agency Theory (Shapiro,
2005). Both literatures use the agency cost minimizing tautology under same set of
assumptions and address the issue of contracting between self-interested utility maximizing
agents and principals (Cuevas-Rodríguez et al., 2012, Eisenhardt, 1989, Jensen, 1983) but, by
modeling the effects of different factors on contracts between parties. Positive literature
focuses on identifying situations in which agent is likely to have conflicting goals with that of
principal and explaining governance mechanism that curtails agent’s self-interested behavior
(Eisenhardt, 1989). Furthermore, positive agency theory deals almost exclusively with special
case of agency relationship between managers and owners of large public corporations
(Cuevas-Rodríguez et al., 2012). Conversely, agent-principal literature is concerned with
general theory of agency relationship which can be applied to any type of agency relationship
such as lawyer-client, doctor-patient and buyer-supplier (Eisenhardt, 1989). The agent-
principal research involves logical deduction and mathematical proof under carefully
specified assumptions. In addition, agent-principal literature has a broader focus and greater
interest in general theoretical implications. Although the two streams of agency literature
differ epistemologically, they are complementary: positive agency theory identifies different
contract alternatives and agent-principal theory identifies the most efficient contract under
varying level of outcome uncertainty, information, risk aversion and other factors (Cuevas-
Rodríguez et al., 2012, Eisenhardt, 1989).
Basically agency theory deals with resolving two problems that exist in agency
relationship (Eisenhardt, 1989). First is the agency problem that emerges from either goal
misalignment or information asymmetry between agent and principal. Second is the problem
of risk sharing that emerges from difference in attitudes of agents and principal towards risk.
20
Agency theory focuses on finding the most efficient contract governing the agent-principal
relationship given certain assumptions. Gomez-Mejia and Wiseman (1997) identify three
main assumptions of agency theory. First, agent is self-interested as he is also a utility
maximizer. Second, agent is risk averse because he cannot diversify his employment across
the firms. Third, there is misalignment of interests between agent and principal. Given these
assumptions two cases can be identified. First is the complete information case. In this case
principal has complete information about the agent’s behavior and actions. Because principal
is well aware of how results would be achieved therefor transferring risk to a risk-averse
agent by providing him/her outcome based incentives would be unnecessary (Eisenhardt,
1989). The second case is incomplete information case. In this case, principal is not fully
aware of agent’s behavioral tendencies and actions, implying that principal would not
completely know when agent departs from principal’s interests. Technically, there is a
problem of information asymmetry between agent and principal (Jensen and Meckling,
1976). Information asymmetry can lead to two types of agency problems. One is moral
hazard and other is adverse selection.
Moral hazard may be referred to as “actions of economic agents in maximizing their own
utility to the detriment of others, in situations where they do not bear the full consequences
or, equivalently, do not enjoy the full benefits of their actions due to uncertainty and
incomplete or restricted contracts which prevent the assignment of full damages (benefits) to
the agent responsible” (Kotowitz, 1987). A moral hazard problem arises when the agent’s
actions are unobservable by the principal. This can possibly be because of two reasons. First,
due to positive cost associated with getting information about agent’s actions; second, the
principal is unable to completely deduce agent’s actions from the outcome because outcome
may not be resultant of agent’s actions. The cost of getting information (monitoring cost) and
uncertainty about the contribution of agent’s actions in the outcome make it difficult for
principal to design an efficient contract that make agent compensation subject to actions of
the agent or the outcome of agent’s unobserved actions (Eisenhardt, 1989). Put it differently,
principal is unable to contractually make the agent bear complete fallouts (positive or
negative) of his/her actions. Consequently, agent may choose to act in his/her own interests at
the expense of interests of principal.
Adverse selection occurs when agent possesses some information that may prove fruitful
in agent’s decision making and he/she does not disclose this information when making
compensation contract with principal. Therefore, the principal would not know if agent has
taken the most appropriate decision on the basis of hidden information. Agent’s access to
21
privileged information about company’s environment and agent’s knowledge and expertise in
the area are the potential source of hidden information. Making an attempt to have this hidden
information is costly to the principal, making it difficult to reach an efficient contract.
Principal makes a trade-off between costs and benefits of acquiring this information when he
designs a compensation contract. The difference between adverse selection and moral hazard
is that in an adverse selection, agents have more information than principal which they can
use to benefit themselves during the contract, and in moral hazard agents develop actions that
are unobservable or immeasurable to the principal and can benefit agents while incurring
losses to principals.
To solve the agency problems, the principal has two choices (Eisenhardt, 1989). First,
obtain more information about agent’s behavior and actions (effort) with better monitoring
and curtail self-interested behavior. Second, provide agent with incentive in a way that
agent’s interests become aligned with interests of principals. As it is assumed that agent is
risk-averse and self-interested, therefore presumably, agent will adhere to the incentives and
make more effort, leading to an outcome that is more desirable by the principal. By providing
incentives, the risk of divergence from principal’s interests would also be shifted back to the
agent (Eisenhardt, 1989).
Agency theory focuses on the costs related to the agency problems and use cost
minimizing tautology to reach a pay contract that align the agent’s interests with that of
principal and adequately transfer the risk to the agent (Cuevas-Rodríguez et al., 2012). These
costs are called agency costs. Jensen and Meckling (1976) identify three types of agency
costs: monitoring cost, bonding cost and residual losses. Monitoring costs can include cost of
efforts that principal make to monitor agent’s actions and cost of incentives that are provided
to the agent. Bonding costs can be referred to as pecuniary and non-pecuniary expenses
incurred by principal or agent to guarantee that agent will act in the best interest of the
principal and if he/she does not in principal’s interests, the principal will be compensated for
agent’s actions. The bonding costs may take the form as contractual guarantees to do or not to
do certain actions such as to have the financial accounts audited by a public account (Jensen
and Meckling, 1976). Finally, residual losses are dollar equivalent of the reduction in
principal’s wealth due to deviation of agents’ decisions from decisions which would
maximize principals’ wealth given the optimal monitoring and bonding costs (Jensen and
Meckling, 1976).
Jensen (1983), subsequently, presents a broader definition of agency costs as “the sum
of the costs of structuring, bonding and monitoring contracts between agents… [which] …
22
also include the costs stemming from the fact that it doesn’t pay to enforce all contracts
perfectly” (p: 331). Positive agency theory suggests that these agency costs can be mitigated
by making an appropriate incentive package for the agents and by incurring monitory costs
(Pepper and Gore, 2015). Eisenhardt (1989) formalizes monitoring and incentivizing agents
for interest alignment in two propositions: first, regarding incentives: “when the contract
between the principal and agent is outcome based, the agent is more likely to behave in the
interests of the principal” (p: 60), second, regarding monitoring, “when the principal has
information to verify agent behavior, the agent is more likely to behave in the interests of the
principal” (p: 60).
2.1.2 Optimal Contracting Approach
In large corporations where ownership is separated from control, shareholders
(principals) are assumed to design efficient compensation contracts that attract, retain and
motivate CEOs (agents) to perform in the best interest of the shareholder i.e. to maximize
shareholders’ wealth (Conyon, 2006). This is labeled as “optimal contracting approach”
(Bebchuk and Fried, 2003). Under optimal contracting approach, compensation contracts are
viewed as an instruments or tools to solve the agency problems. Bebchuk and Fried (2003)
observe that “Optimal compensation contracts could result either from effective arm’s length
bargaining between the board and the executives or from market constraints that induce these
parties to adopt such contracts even in the absence of arm’s length bargaining”. A
compensation contract is efficient one if it (1) is able to retain talented executives, (2)
properly incentivize executives to take decisions in shareholders’ interests and (3) reduces the
overall agency costs (Bebchuk and Fried, 2004, Jensen and Murphy, 1990b). Core et al.
(2003) state this in a broader way that an efficient (or an optimal) contract is one “that
maximizes the net expected economic value to shareholders after transaction costs (such as
contracting costs) and payments to employees”.
Bebchuk et al. (2002) argue that in order to retain the CEO, his/her compensation should
exceed his/her opportunity cost, labeled as “reservation value”. A CEO’s reservation value is
partly a function of willingness to take risk. If a firm wishes a risk-averse CEO to accept
risky components (that are linked to firm performance) of a compensation contract, it would
have to offer larger total compensation on an expected value basis to offset CEO risk-bearing
(Bebchuk and Fried, 2004, Bebchuk et al., 2002). Although CEO’s reservation value places a
lower limit on CEO pay however, a firm may be willing to set CEO compensation level well
beyond CEO reservation value to provide incentives to make him/her behave in a way that
23
maximize shareholders’ wealth (Bebchuk et al., 2002). A firm would give value to CEO until
the marginal benefits are greater the marginal costs of doing so. Therefore, boards
considering the compensation plans to maximize shareholder value would focus on
alternative choices both in term of plans’ costs and incentive benefits to the shareholders
(Bebchuk and Fried, 2004, Gomez-Mejia et al., 2010).
There are several implications of an optimal contract. First, it reduces executive
opportunism, second, it links executive compensation with firm performance, third, it
motivates CEO to work against his “assumed” risk-averse behavior and last but not the least
it motivates CEO for future performance. Optimal contract does not imply an “ideal” contract
that totally eliminate agency problems, only that board of directors designs the best contract
that is possible under contracting constraints to avoid opportunistic and malfeasant behavior
of the executives. Accordingly, in optimal contacting, the board does not essentially eliminate
the agency costs but make a trade-off between (marginal) costs and (marginal) benefits of
implementing the contract (Conyon, 2006).
2.1.3 Managerial Power Approach
Contrary to optimal contracting approach, Bebchuk and Fried (2003) advocate another
view to analyze the CEO compensation, labeled as managerial power approach (MPA).
Bebchuk and Fried (2003) argue that CEOs have substantial influence over their own pay.
The board of directors and compensation committee, being influenced by CEO, compromise
their fiduciary duties and settled upon CEO compensation that is not in the interest of
shareholders. Basically, Bebchuk and Fried (2003) challenge the fundamental assumption of
optimal contracting approach that boards involve in arm’s length transaction with managers
over compensation arrangements. Bebchuk and Fried (2003) contend that “Just as there is no
reason to presume that managers automatically seek to maximize shareholder value, there is
no reason to expect a priori that directors will either. Indeed, directors’ behavior is also
subject to an agency problem, which in turn undermines their ability to address effectively
the agency problems in the relationship between managers and shareholders” (p: 73). Thus,
directors may not be true representatives of shareholders.
Bebchuk and Fried (2003) provide several reasons as to why directors would
compromise their fiduciary duties to shareholders and would not effectively challenge the
CEO compensation. First, CEO can affect directors’ compensation and perks; going against
CEO compensation may fire back. Second, CEO has significant influence over directors’
nomination process. As directors wish to be reappointed to get attractive compensation and
24
maintain the prestige; opposing CEO on the matters dear to him (such as compensation)
would not be a good choice for directors. Third, directors would not wish to lower their
chances of being invited to join other companies by having a reputation of quibbling up with
CEO over compensation. Last, directors usually have little access to independent information
regarding compensation practices to challenge the CEO compensation. Overall, there are
plentiful reasons to why directors have incentives to go along the desires of the CEO (see,
Weisbach, 2007). Bebchuk and Fried (2003) further contend that market forces are not strong
enough and prepared to assure the arm’s length contracting.
Although CEOs can influence the directors to settle upon the compensation that is
beyond the level of what optimal contracting suggests. However, this excessive compensation
is not unlimited rather subject to some potential “outrage” constraints and costs stemming
from an “outrage” – a negative reaction of relevant outsiders to the proposed compensation
contract. Outrage might cost CEO and directors a reputational loss or embarrassment, or it
might lead to unwillingness of shareholders to support executives in takeover bids or proxy
contests (Bebchuk and Fried, 2003). The gravity of these costs depends upon the gravity of
outrage. Therefore, facing potential outrage costs and constraints, how far directors and
compensation committee can go in favoring CEO will depend not only on how much
expected compensation plan will actually benefit CEO but also on how this plan will be
perceived by relevant outsiders (Bebchuk and Fried, 2003). The greater the outrage a
compensation contract is expected to cause, greater the reluctance will be to approve the
contract by directors and greater the hesitation will be to propose it by the executives
(Bebchuk and Fried, 2003). All this leads to consider that executive compensation is an
agency problem itself rather than a tool to solve agency problems between shareholders and
executives (see, Bebchuk and Fried, 2004). The major implication of MPA is that CEOs’
compensation would be greater and/or less sensitive to firm performance if CEOs are
relatively more powerful than the board of directors (see, Bebchuk and Fried, 2003). Weak
boards in relation to CEO, CEO duality, higher percentage of outside shareholders i.e., less
concentrated ownership, less board independence and possibly less institutional holding, all
contribute to relative power of CEO (see, van Essen et al., 2012b), leading higher
compensation with less sensitivity to firm performance.
Although MPA is a direct challenge to optimal contracting approach and view CEO
compensation with different lens, however, Bebchuk and Fried (2003) and Bebchuk and
Fried (2004) mention it clearly that MPA is not a replacement for optimal contracting rather
simply that pay practices cannot be explained by optimal contracting approach alone. In a
25
recent meta-analytic study assessing MPA against optimal contracting, van Essen et al.
(2012b) suggest that executives can and do influence their compensation and their power can
be controlled by principals. However, they document that MPA is unable to provide single
combined explanation for all the outcomes of executive compensation, especially for
arrangements that related compensation to performance. In sum, both optimal contracting and
managerial power influence the pay setting process. Concluding their study, van Essen et al.
(2012b) document that MPA and optimal contracting “do not represent competing
explanations but describe points on a continuum of types of contracting arrangements that
can be encompassed within agency theory”.
2.1.4 CEO Compensation and Corporate Governance
Agency theory focuses on two mechanisms to solve agency problems: first, an efficient
compensation contract, second, better monitoring. Both mechanisms go side by side in
solving agency problems. A weak monitoring system would exacerbate the agency problems
and increases the agency costs to the firm. In order to mitigate the agency costs, firms need to
devise internal system to effectively separate control decisions (ratification and monitoring
decisions) from management decisions (initiation and implementation decisions) (Fama and
Jensen, 1983). Such a system would lead to better monitoring and control, and would curb
managerial opportunism, leading to reduced agency costs.
Corporate governance assumes this role of monitoring and curbing managerial
opportunism so that shareholder’s interests are protected (Fama, 1980, Fama and Jensen,
1983, Shleifer and Vishny, 1986). Daily et al. (2003) define corporate governance as “the
determination of the broad uses to which organizational resources will be deployed and the
resolution of conflicts among the myriad participants in organizations” (p.371). Therefore,
corporate governance is meant for designing effective control mechanisms (internal or
external) that enhance firm performance, mitigate agency problems, align the interests of
agents and principals and protect shareholders’ wealth.
A number of governance mechanisms are used to accomplish this which can be
categorized into two broad categories: internal governance mechanisms and external
governance mechanisms (Gillan, 2006). These mechanisms include effective boards, board
independence, well designed compensation plans to motivate agent to work in shareholders’
interest, concentrated ownership to monitor and discipline executives, and an effective
outside controls by market of corporate governance where internal governance mechanisms
are inefficient to discipline executive (Clarke, 2007, Walsh and Seward, 1990). The major
26
participants of internal corporate governance mechanism include boards, shareholder,
debtholders, executives, employees, suppliers and customers (Gillan, 2006). All these make
nexus of contracts view of an organization as formulated by Jensen and Meckling (1976).
The participants of external governance mechanism include law/regulations, markets,
politics, culture and communities (Gillan, 2006). External factors have significant importance
in designing internal governance systems as they provide an environment in which firm is
working (Gillan, 2006).
Agency theory predicts that more information about executives’ actions through better
monitoring system would result in less performance based remuneration (Eisenhardt, 1989).
This would lead to less total executive compensation. As an efficient corporate governance
structure provides close monitoring and oversight of management, therefore in the presence
of strong corporate governance, the total executive compensation would be low. On the other
hand, weak corporate governance may make executives more powerful relative to board of
directors, leading to more influence on pay-setting process for higher compensation.
2.1.5 CEO Compensation and Ownership Structure
Agent-principal paradigm predicts that in a dispersed ownership structure, individual
owners have weak incentives to exercise the power over major corporate decisions and to
invest in monitoring of management (Fama and Jensen, 1983, Jensen and Meckling, 1976).
This problem of management’s free-ride may be reduced through concentrated ownership.
Concentrated ownership has significant impact on the pattern of executive pay and CEO
equity incentives (Core et al., 1999). In a concentrated ownership structure, large
shareholders have strong incentives to oversee agents’ activities (Jensen and Warner, 1988).
Therefore, concentrated ownership generally suggests that shareholders are able to better
protect their interests in their companies, leading to mitigating potential CEO entrenchment
and relatively strong association between CEO pay and firm performance.
A family ownership is a special form of concentrated ownership. The unique nature of
family firms affects the top executive compensation contracts and the impact of these
contracts on firm performance and managerial actions. Several features of family ownership
reduce agency problems which consequently influence executive compensation. For example,
concentrated family ownership provides strong monitoring incentives, mangers from family
as being insiders can improve monitoring by providing information to other family members
outside (see, Harris and Raviv, 2008, Villalonga and Amit, 2010), strong commitment and
27
firm specific knowledge of family members make them better monitor (Bertrand and Schoar,
2006).
Although concentrated ownership provides special monitoring incentives and seemingly
enforce the relationship between executive compensation and firm performance however, it
may introduce the conflicts between controlling shareholders and minority shareholders
(principal-principal agency problems), leading to expropriation of minority interests by
controlling shareholders. This expropriation is very likely in emerging markets where formal
institutions are weak to support mutually beneficial impersonal exchange between economic
players (Young et al., 2008).
Expropriation can be achieved through many ways – legal or illegal, or the ways that are
in grey areas (La Porta et al., 2000). For example, it can be achieved 1) through appointing
unsuitable friends, family members and cronies in important positions (Faccio et al., 2001,
Yeung, 2006), 2) devising such strategies that promote family or political interests against
firm performance (Young et al., 2008), 3) engaging in transactions with associated firms or
owned by controlling shareholders at prices unfavorable to the firm. In addition, controlling
shareholders may collude with management, called tunneling, for rent extraction. For that
matter they are willing to pay executives excessively but in return they demand convenient
way to expropriate minority interests (Su et al., 2010). Since tunneling has negative effects on
firm performance (Cheung et al., 2006, Jiang et al., 2010a), a strong pay-performance
relationship would induce executives to increase firm performance and reduce the incentives
to collude with controlling shareholders. Therefore, controlling shareholders involved in
tunneling activities may compromise the pay-performance link. Further, in the presence of
tunneling activities, controlling shareholders are somewhat responsible for firm performance.
Therefore, performance measures are likely to become less informative of executive effort
and less useful in executive pay setting process (Wang and Xiao, 2011). Thus, when
management and controlling shareholders collude, they attempt to maximize their private
benefits at the expense of minority shareholders. In sum, concentrated ownership may lead to
expropriation of minority interests and excessive pay to executives presumably not linked to
firm performance (Su et al., 2010, Wang and Xiao, 2011).
Similarly, although family ownership help reduces the agent-principal conflicts, it also
introduces new conflicts between family and non-family shareholders due to existence of
private benefits of control. These private benefits include utility derived from preserving the
family legacy for future generations or ensuring the well-being of other family members
(Becker, 1981, Bertrand and Schoar, 2006, Burkart et al., 2003).
28
2.1.6 CEO Compensation and Performance Manipulation
Agency theory assumes that CEOs are self-interested and may behave opportunistically
at the expense of shareholders’ interests. Accordingly, corporate boards are presumed to
restrain CEO opportunism and align the CEO actions with shareholders’ interests by
designing efficient compensation contracts that typically link CEO compensation with firm
performance. However efficient compensation contracts may encourage managers to
manipulate performance because their compensation is tied to some pre-specified
performance criteria (Sun, 2012). These performance criteria generally include earnings
targets, share price performance and some specific financial ratios that signal the financial
well-being of the company (Ibrahim and Lloyd, 2011, Puffer and Weintrop, 1991).
Watts and Zimmerman (1978) develop a hypothesis that managers tend to manipulate
earnings through the choice of accounting policies because their compensation is tied to firm
performance. Similar to agency theory, Watts and Zimmerman (1978) assume that there is
misalignment of interests between shareholders and managers and, managers are self-
interested and tend to maximize their own utility. They further assume that management's
utility is a positive function of the expected compensation in future periods (or wealth) and a
negative function of the dispersion of future compensation (or wealth). Watts and
Zimmerman (1978) focus on performance related components of managerial pay i.e.
incentive cash bonuses and stocks or stock options. They contend that through the choice of
accounting methods, managers can increase incentive cash bonuses or value of the firm share,
leading to increase in managers’ wealth. Thus, managers are prone to pick those accounting
policies which transfer reported earnings from future to current period, or vice versa, under
specific situations (Watts and Zimmerman, 1986). Watts and Zimmerman (1978) identify
various factors by which the choice of accounting policies can affect both components of
managerial compensation directly or indirectly. These factors include 1) taxes, 2) political
costs, 3) regulatory procedures if the firm is regulated, 4) information production costs and 5)
managerial compensation plans. The first four factors increase cash flows which in turn have
positive effect on share price, leading to increased managerial wealth. The last factor can
increase managerial wealth through the alteration in the terms and conditions of incentive pay
plans (Watts and Zimmerman, 1978).
Management can increase (decrease) current reported tax by adopting certain accounting
methods which in turn decrease (increase) the reported cash flows. To cope with political
interventions, firms employ different methods such as government lobbying, social
29
responsibility movement and adopting certain accounting procedures to avoid high reported
earnings. Management reduces chances of adverse political actions by avoiding public
attention due to high profits that may be perceived as monopoly rents, hence, reduces
political costs and increases cash flows indirectly (Watts and Zimmerman, 1978).
Management can take benefit from favorable regulation that can increase current earnings
and cash flows. A new accounting standard may provide the managers an excuse to raise the
prices in a regulated firm due to standard’s tendency to reduce firm’s reported income (Watts
and Zimmerman, 1978). Changes in accounting method can increase information production
costs that may include book keeping costs, salaries to accountants and any additional training
costs. Managers would favor those methods that incur minimum information production
costs.
One of the major components of management compensation is incentive (bonus) plans.
These plans are typically linked to accounting income. Any change in accounting policy or
method that increases the firm’s reported income would lead to higher incentive earnings for
managers. However, this would probably reduce company cash flows, leading to decline in
share price. Managers would favor change in accounting method as long as the present value
of incentive income (after tax basis) is greater than the reduction in value of manager’s
portfolio (Watts and Zimmerman, 1978, Watts and Zimmerman, 1986). Existing literature
finds evidence that executive compensation is related to earning manipulation driven by a
managers’ desire to meet the performance targets, leading to higher compensation (see,
Devers et al., 2007, Frydman and Jenter, 2010, Ibrahim and Lloyd, 2011).
Recently, Sun (2014) theoretically examines the implications of earnings management
for executive compensation in the context of corporate governance reforms and where
managers may take costly decisions to manipulate earnings. She presents that increased
chances to manipulate earnings calls for equilibrium pay-performance sensitivities to be
stronger. In her optimal contracting model, shareholders are faced with the information
asymmetry about the accuracy of financial information presented by the management. An
opportunity to manipulate earnings serves as protection against lower compensation pay offs
to executives and thereby reducing the incentives to exert costly effort. Therefore, when
earnings manipulation is possible, high-powered pay contracts are required to incentivize
executive efforts. Her model also suggests that tightened corporate governance reforms may
lead to lower pay-performance sensitivity because they generate the incentives for executive
efforts and increase the cost of performance manipulation.
30
Theoretical arguments presented above suggest that executives have different incentives
to manipulate firm accounting performance depending upon the circumstances and
motivations. However, given the assumption that executives are self-interested, one of the
motivational factors of manipulating performance is to increase executives’ own
compensation and wealth. Further, such choice and opportunity of manipulating performance
depends on prevailing corporate governance policies and practices. Given the Pakistani
corporate governance context, this study empirically examines the proposition that CEOs
tend to manage earnings to get higher compensation while controlling for corporate
governance practices. The existing empirical evidence to such propositions is discussed in the
end of next section.
31
2.2 Existing Empirical Evidence
This section aims at discussing the empirical evidence on the relationship of executive
compensation with firm performance and manipulation of firm performance. It contains two
subsections. The first subsection covers the relationship between executive compensation and
firm performance. Because several studies have pointed out institutional differences across
the world and institutional context is likely to have significant influence on the relationship
between executive compensation and firm performance (see, Fan et al., 2011, Ghosh, 2006,
Sun et al., 2010, van Essen et al., 2012a), therefore within this first subsection empirical
studies are grouped according to the contextual settings in which studies are performed, for
example, US context, European context etc. The first subsection ends with review of
empirical studies that examine the relationship between executive compensation and future
firm performance. The second subsection reviews the empirical studies that examine the
relationship between executive compensation and performance manipulation. In the last, a
brief summary of empirical work is presented.
2.2.1 Executive Compensation and Firm Performance
Optimal contracting approach, derived from Agency theory, views the CEO
compensation arrangements as one of the tools to reduce the conflicts of interest between
agents and principals. Under this approach corporate boards are presumably responsible to
devise such compensation arrangements that provide agents with efficient and reasonable
inducements to maximize shareholder value (Bebchuk and Fried, 2003). This leads to the
argument of aligning the CEO compensation with firm performance. However, findings of
the empirical literature are mixed regarding the relationship of CEO compensation with firm
performance (see, Bebchuk and Fried, 2004, Devers et al., 2007, Sun et al., 2010)
2.2.1.1 US Context
In a seminal paper, Jensen and Murphy (1990b) examine the relationship between CEO
compensation and shareholders’ wealth – the ultimate goal of shareholders. They use data for
2,213 CEOs listed in the Executive Compensation Surveys published in Forbes from 1974 to
1986. Their final sample after eliminating observations with missing entries includes 7750
first differences in yearly compensation for 1688 executives from 1049 firms. Using least
square regressions, they document that for every $1,000 increase in shareholder wealth there
is about $3.25 increase in CEO pay including compensation, stockholdings and dismissal.
They show that, out of $3.25 change in CEO compensation, seventy-five cents correspond to
32
wealth consequences associated with outstanding stock options, salary revisions and
performance-related dismissals while, $2.50 correspond to firms’ common stock held by
CEOs including common stock and exercisable stock options held by their family members
or associated trusts. Given that higher pay for performance sensitivity should represent
greater interest alignment, they conclude that although their results are statistically significant
but sensitivity of CEO compensation to firm performance is very low which seems
inconsistent with agency theory. The potential reasons for this inconsistency they quote are:
boards may not consider CEO as an important input in production process as CEO actions
may be monitored and evaluated easily; there may be political forces internal and external to
the organization that influence the contracting process between shareholders and
management; public condemnation of higher compensation may truncate the up side of the
earnings distribution of executives.
Although findings of Jensen and Murphy (1990b) seem inconsistent with agency theory,
however, it is observed that pay-performance sensitivity varies considerably with the time
period examined, sample used and variables included (Devers et al., 2007). For instance,
using sample period from 1980 to 1994 and incorporating more comprehensive data on CEO
stock and option holdings for 478 companies from Forbes, Hall and Liebman (1998)
document that pay-performance sensitivity is nearly four times higher than it is in Jensen and
Murphy (1990b). They argue that CEO are not paid like bureaucrats as it was stated by
Jensen and Murphy (1990a). In fact, by using different measures of the pay-performance
relationship they find a strong association between wealth of CEO and wealth of the firm
CEO manage. Hall and Liebman (1998) incorporate broad measures of compensation that
include change in value of stock and stock options and employ ordinary least square
regression, median regression and robust regression models to analyze pay-performance
relationship. They report that both level of CEO compensation and pay-performance
sensitivity have risen considerably over their study period. The substantial increase in pay-
performance sensitivity is because of the fact that most of the pay increase is attributed to
increase in stock options. Indeed, they find that, for a given change in firm value, the
contribution of stock and stock option revaluation toward change in CEO wealth is more than
50 times greater than the contribution of salary and bonus. Overall the findings of Hall and
Liebman (1998) challenge the claim of Jensen and Murphy (1990b) that there is weak
relationship between CEO pay and firm performance.
In the same vein, Frydman and Saks (2010) also reject the position of Jensen and
Murphy (1990a) that CEOs are paid like bureaucrats. They examine the level and structure of
33
executive compensation of 101 firms for period 1936 to 2005. They show that compensation
increases with slow rate from mid-1940s to mid-1970s but after mid-1970s it rises at an
increasing rate till the end of sample period. Important finding, they show, is that pay-
performance sensitivities are remarkable during all the sample period except 1940s. They
pose this finding as indication of utilization of pay-performance relationship to align the
executive incentives with shareholders’ interests for most of the sample period.
As research on executive compensation flourishes in 1990s, scholars start examining the
pay-performance relationship by controlling for the influence of other variables. For example,
Aggarwal and Samwick (1999b) include firm risk as measured by volatility of stock returns
in their analysis of pay-performance sensitivities. They argue for an economic trade-off
between the amount of unobservable effort that an agent makes and minimizing the amount
of risk an agent is supposed to bear. Basically, they test the prediction of principal-agent
model that pay-performance sensitivity would decrease as variance of firm performance
increases. In other words, executives in firms with more volatile stock returns are less likely
to have performance based compensation. Using the data on compensation of top executives
at 1,500 of the largest US publicly traded firms for period 1993 to 1996 from ExecuComp
database and controlling for stock return volatility, they find that median CEO pay–
performance sensitivity is $14.52 for every $1,000 change in shareholder value which is
greater than the sensitivity documented by Jensen and Murphy (1990b) i.e., $3.25 per $1,000.
However, they find that median pay–performance sensitivity for executives other that CEO is
$3.30 per $1,000. Further they find that pay-performance sensitivity is greater for the
executives at the firms having less volatility of stock returns as compared to the executives at
firms having higher volatility of stock returns. They conclude that without accounting for the
volatility, the examinations of pay–performance sensitivity may be biased toward zero.
Similarly, Guillet et al. (2012) examine the influence of executive characteristics and
firm performance on executive compensation. They use 16-year data from 1992 to 2007
consisting of 2200 firm-year observations from restaurant industry. Both accounting
performance and stock performance are considered along with executive characteristics such
as gender, age and tenure. Using generalized least square (GLS) model, they find that equity
based compensation varies with the type of executive and it is tied not only to firm
performance but also to executive characteristics such as executive age and tenure. Firm size
also appears to have significant positive influence on executive compensation.
Brick et al. (2006) also model directors and CEO compensation using governance
variables, CEO characteristics and firm characteristics. However, they test slightly different
34
hypothesis that excessive compensation of both CEO and directors is related to firm
underperformance. They use fixed effect to control for unobserved heterogeneity. Using
ExecuComp and COMPUSTAT data sets and hand collected data for 1163 to 1441 firms for
the period from 1992 to 2001, they find that the two compensations (CEO’s and directors’)
are negatively related to firm performance. They further find that there is a strong
relationship between CEO compensation and directors’ compensation. They argue that a
positive relationship between CEO compensation and directors’ compensation coupled up
with a negative relationship of both the compensations with firm performance is consistent
with cronyism hypothesis in which protection of shareholders’ interests in undermined.
Moreover, they show that excessive pay has an effect on firm performance that is
independent of poor governance variables.
A position in management hierarchy is another factor which may lead to different pay-
performance sensitivities for executives. Building on this, Aggarwal and Samwick (2003)
show that pay-performance sensitivity increases with the level of responsibility in an
organization hierarchy. They use data for the period 1993 to 1997 from ExecuComp for
13109 executives with 33607 executive-year observations. By employing regression with
executive fixed effects and year fixed effects, and controlling for level of compensation and
variables associated with firm’s contracting environment, Aggarwal and Samwick (2003)
show that the median pay-performance sensitivity of CEO is $5.65 per $1,000 increase in
shareholder value greater than that of executives with divisional responsibility. They further
show that aggregate median pay-performance sensitivity for top management team (TMT) is
$32.32 per $1,000 and CEO accounted for 42% to 58% of these aggregate incentives.
Concentrated ownership and governance factors are important factor that can redefine
pay-performance relationship. Amoako-Adu et al. (2011) examine executive compensation in
two types of concentrated ownerships using 3620 executive-firm-year observations and 700
firm-year observations from all listed companies in Toronto Stock Exchange for the period
1998 to 2006. They show that dual class firms as compared to single class firms with
concentrated control pay more to executives who belong to the controlling family. This
excessive pay appears in the form of either bonuses or stock options. This finding is claimed
to be consistent with optimal contracting approach in that higher compensation prevents
executives in dual class firms from taking benefit of greater voting power.
Core et al. (1999) show that board structure also affects CEO compensation along with
ownership structure. They examine 495 observations over a three-year period (1982 to 1984)
for 205 publicly traded U.S. firms and document that CEO compensation is lower when
35
percentage of inside directors is higher in the boards. Further, large board size, greater
number of outside directors appointed by the CEO, greater percentage of gray outside
directors, greater percentage of outside directors over ag 69, greater percentage of directors
serving in three or more other boards and CEO duality are associated with higher CEO
compensation. With respect to ownership structure, they find that CEO compensation is
lower when CEO ownership is higher, external blockholders or non-CEO internal board
member owns more than 5% shares.
Conceptualizing as if CEO compensation may be an outcome of abnormal performance,
Boschen et al. (2003) study the effects of performance surprises on CEO compensation. They
use vector autoregression (VAR), a time series model, to examine the effects of accounting
and stock performance surprises on CEO compensation. Their dataset consists of 1,110 firm-
year observations of 30 firms for period 1959 to 1995. They show that in the long run, stock
performance and accounting performance have different impact on CEO compensation. Both
positive accounting and stock performance surprise lead to higher CEO compensation
initially. However, in the long run, responses of performance surprises are different. In case
of positive accounting performance surprise, the higher initial compensation is offset by
lower compensation in later years, leading to no effect on CEO compensation in the long run.
Contrary to this, CEO continues to receive higher compensation (although diminishing) for
many years in response to a positive stock performance surprise.
While Boschen et al. (2003) document different long term effect of accounting and stock
performance on CEO compensation, Leone et al. (2006) document an asymmetric
relationship between cash compensation and stock returns. They use data from ExecuComp
for 2,751 CEOs (9,858 year-observations) for period 1993-2003. They argue that as the board
of directors include most unrealized losses and exclude most unrealized gains in CEO cash
compensation to reduce the cost of ex-post settling up for the shareholders, and since stock
returns include both unrealized losses and unrealized gains therefore, cash compensation
would be less sensitive to positive stock movements due to unrealized gains than to negative
stock movements due to unrealized losses. However, they argue that the response of other
types of compensation to stock returns would be different. For example, the response of
equity-based compensation to stock returns would be symmetrical because of vesting
constraints. If unrealized gains disappear before the equity claim vests, this information
would be impounded in claim’s price, therefore there would be no ex post settling up problem
for the stockholders, hence symmetrical response to positive and negative movements of
stock prices. Consistent with their arguments, they find that the sensitivity of cash
36
compensation to negative stock returns is twice more than the sensitivity to positive stock
returns and sensitivity of equity-based compensation is symmetrical to positive and negative
stock returns.
Given ex post setting up explanation to asymmetric pay-performance sensitivity of cash
compensation to stock returns by Leone et al. (2006), Dechow (2006) argues that importance
of ex post settling up explanation is of second-order, hence, she provides alternative
explanations. Compensation includes several components which are adjustable according to
the alignment of agents’ interests with that of principals. Consistent with optimal contracting
approach, she takes the position that bonus contracts (part of cash compensation) are typically
designed to offer more downside risk and limited upside risk, while equity compensation is
typically designed to offer more upside risk and limited downside risk. Therefore, pay-
performance sensitivity may be asymmetric to firm performance for different components of
compensation. However, overall, total compensation is sensitive to both upward and
downward stock price movements. She also provides explanation consistent with managerial
power approach. Excessive cash compensation is very visible thing and difficult to
camouflage and it may ignite significant outrage especially when firm is not performing well.
In order avoid this problems board being influenced by CEO under managerial power
approach can offer substitute forms of compensations such as stock options and perks. The
board of directors, possibly due to outrage concerns, wants to be able to assert that bonus
payment to CEO is based on performance and is tax deductible, leading to potential
asymmetric pay-performance sensitivities.
Shaw and Zhang (2010) defy the findings of Leone et al. (2006) that CEO cash
compensation is more sensitive to lower stock performance than to positive stock
performance because of ex post settling up. They argue that Leone et al. (2006) results are
due to misclassification of firms into good and bad performers. By classifying firms into
good and bad performers in a more sophisticated way, named as “three-way performance
partition”, and using a sample of14,632 firm-years over 1993–2005, Shaw and Zhang (2010)
find no evidence that cash compensation is more sensitive to lower stock returns than to
higher stock returns. Further, they find that sensitivity of cash compensation with firm
performance is lower for poor earnings performance than for better earnings performance.
Thus, they find no support for ex post settling up for poorly performing firms. This is even
true for worst performing firms with efficient corporate governance. They report similar
results when firms are partitioned on the basis of earnings-based measures. Overall, they
conclude that CEO cash compensation is not punished for poor firm performance.
37
A level of competition in the industry may have impact on the CEO compensation.
Aggarwal and Samwick (1999a) argue that in a more competitive industry, executive are
incentivized more on the basis of minimizing the value of other firms in the industry as
compared to maximizing the value of their own firm. They predict that the level of
competition in the industry decreases the ratio of pay-performance sensitivity of own-firm to
pay-performance sensitivity of rival firm. To test the predictions of their model, they employ
median regression with executive fixed effects using compensation data on 1,519 CEOs and
6,305 other executives from ExecuComp database for period 1992-1995. They find positive
pay-performance sensitivities for both own and rival firm. However, the ratio of own-firm
pay-performance sensitivity to rival-firm pay-performance sensitivity is lower in a more
competitive industry.
While Aggarwal and Samwick (1999a) examine compensation relative to other firms in
the industry, Bebchuk et al. (2011) explore the CEO compensation relative to other top five
executives in the firm. Specifically, they examine the relationship between CEO’s share in
compensation of top-five executive, namely CEO pay slice (CPS), and performance (value)
and behavior of publically traded firms. The dataset they used consists of 12,011 firm-year
observations representing 2,015 firms and 3256 CEOs for period 1993 to 2004. After
controlling for firm, corporate governance and CEO characteristics in regressions, they find
that as CPS increases the value of firm as measured by industry-adjusted Tobin’s Q
decreases. They further find that accounting profitability is negatively related to CPS.
Increase in CPS is negatively related to stock market returns accompanying the filing of
proxy statements. The CEOs with high CPS are more likely to be granted opportunistically
timed options. Finally, they conclude that higher CPS exacerbates the agency problems.
Again building on agency theory, Sun et al. (2013) relate CEO compensation to firm
efficiency as proxy for firm performance. First, they estimate cost efficiency (CE) and
revenue efficiency (RE) by applying data envelopment analysis (DEA). In the second step,
these efficiencies are used as explanatory variables in a regression model with CEO
compensation as dependent variable. The dataset they used consists of 139 year-observations
from 31 unique firms during 2000 to 2006. The result show that total CEO compensation as
measured by sum of cash compensation (base salary and bonus) and incentive based
compensation (stock grants, stock options and long term cash incentive payments) is
positively related to CE as well as to RE. Further analysis shows that CE is positively related
to incentive based compensation while RE is positively related to cash compensation.
38
Similarly, Nyberg et al. (2010) develop their arguments on agency theory but considers
firm’s CEO as significant co-investor instead of an employee only. They gauge the financial
alignment between shareholders and CEOs by assessing the relationship between CEO return
and shareholder return. After analyzing a sample of 9,469 CEO-year observations from 2,166
companies in US, they report that intensity of interest alignment between CEO and
shareholder is larger than reported earlier. They also show the moderating effect of relative
share of total shareholder return and CEO’s wealth on this relationship.
In a recent meta-analytic study, van Essen et al. (2012b) focus on finding evidence on the
relationship between CEO compensation and proxies of managerial power and pay-
performance sensitivities of CEO. In fact, they assess which approach is dominant in
explaining the CEO compensation: optimal contracting approach or managerial power
approach. The results based on 219 US-based studies indicate that managerial power theory is
good at predicting the level of cash and total compensation but not very successful in
predicting pay-performance sensitivities. They use four proxies for relative power of
management: board size, CEO tenure, CEO duality and fraction of independent directors in
the board. In most of the cases where CEOs are likely to have more relative power than the
boards to influence the pay-setting process, CEOs cash and total compensation are higher. In
contrast, where boards are more powerful relative to CEOs, CEO cash and total
compensation are lower. Furthermore, directors with more relative power seem to be able to
create a stronger relationship between CEO compensation and firm performance even against
the powerful CEOs they restraint. Although managerial power approach can explain level of
compensation however, this approach cannot be taken as a single unifying approach to
explain CEO compensation as it is not very successful in explaining the pay-performance
relationship. They finally conclude that managerial power has important influences on
compensation setting process but optimal contracting also exists at the same time. Therefore,
both optimal contracting and managerial power approaches do not seem to compete with each
other in describing CEO compensation. They rather describe points on a continuum of types
of contracting arrangements that can be encompassed within agency theory.
2.2.1.2 Australian and European Context
Deviation from efficient compensation structure may lead to demotivated CEOs which in
turn, can lead to firm underperformance. Matolcsy and Wright (2011) test this hypothesis
using data from Australian market for period 1999 to 2005. They select 3503 year-
observations for Top 500 firms from Connect 4. Using both accounting and stock based
39
performance criteria they find consistent results with their hypotheses. The firms having CEO
compensation structure inconsistent with the firm characteristics underperform as compared
to the firms having CEO compensation consistent with the firm characteristics. The firm
characteristics they used include firm size (market value of equity), market-to-book ratio,
number of subsidiaries, number of foreign subsidiaries, standard deviation of ROA over prior
3 years, operating cash flows to total assets ratio, prior year stock return (dividend adjusted),
ROA, CEO share ownership, debt-to-equity ratio, total equity owned by all blockholders with
more than 5 per cent of the shares outstanding.
In a study on a sample of 1152 managers from 160 Danish firms for period 1990 to 1994,
of which 120 are CEOs, 302 are upper-level managers and rest of 730 are lower-level
managers, Eriksson and Lausten (2000) find that the compensation of CEO and other
mangers is positively related to return on firm’s own capital. However, CEO compensation is
not driven by the changes in firm accounting performance as operationalized by different
dummy variables. Further, they show that, contrary to existing evidence, pay-performance
sensitivity is slightly different among various levels of managers in an organizational
hierarchy. Moreover, they find that CEO tenure and time left to retirement do not affect CEO
compensation.
In a survey study held on privately owned firms, Banghoj et al. (2010) conjecture that
pay-performance relationship in Denmark is weak. This is because of a couple of convincing
reasons. First, corporate governance practices are relationship oriented and there is less
variability in wage levels in Demark. Second, about 99 percent of the companies in Denmark
are privately owned. In privately owned firms, owners have better incentives for monitoring
executive, leading to less moral hazard problems and consequently, less performance based
compensation. Third, size of privately owned firms is generally small as compared to the size
of the public firms therefore it is not likely that privately held firms choose performance
based compensations. Finally, noisier nature of earnings as a measure of performance in
privately held firms may distort the pay-performance relationship. Banghoj et al. (2010) also
test the influence of corporate governance characteristics, executive characteristics and
quality of bonus on compensation. By employing two-stage and three-stage least square
regression, as expected, they find weak relationship between executive compensation and
firm performance in privately held firms of Denmark. Only ownership concentration and
board size appear to influence executive compensation among different corporate governance
variables. This finding is in contrast to several other studies (see, Devers et al., 2007, van
Essen et al., 2012b) in which many of corporate governance characteristics influence
40
executive compensation. One possible explanation they provide for this discrepancy is
difference in institutional context. In executive characteristics, executive position, educational
background and executive skills appear to have influence on executive compensation.
However, contrary to the expectations, they do not find strong pay-performance relationship
in firms with better designed bonus contracts.
Brunello et al. (2001) hypothesize that given the characteristics of Italian institutional
context such as family ownership and pyramidal groups, developing capital markets and
absence of relative hostile takeovers, the ratio of incentive pay with respect to total pay is low
and there is weak relationship between incentive pay and firm performance. To test their
hypotheses, they employ log-linear specifications using a survey data consisting of 2996
observations from 107 firms exclusive of state owned firms. They find that executive
compensation is positively related to firm performance as measured by real accounting
earnings per head. However, pay-performance semi-elasticity is low. They estimate that an
increase of 1billion lire in real accounting earnings lead to an increase of only 31 thousand
lire for middle and upper level managers which is 25 thousand more than for lower level
managers. The firms that are listed in stock exchange, belong to multinational group or
owned by foreign capital appear to have higher pay-performance sensitivity. As firms
associated with multinational groups and/or owned by foreign capital are less likely to be
influenced by unique Italian context therefore it is implied that characteristics of specific
Italian economic environment lead to weaker pay-performance relationship.
In another study in Italian context, Barontini and Bozzi (2011) examine the relationship
of board compensation with ownership structure and future firm performance. The sample
examined consists of 1722 year observations related to 215 firms listed in Milan Stock
Exchange. Using OLS regression, they find that many corporate governance variables
influence the board compensation however excess board compensation is never found to be
associated with future firm performance. Smaller board size and greater representation of
family members on the board are associated with higher board compensation and higher
board compensation appears to reduce future firm performance. Overall, Barontini and Bozzi
(2011) conclude that their evidence favors the rent extraction view of pay setting process
against the social network view which they quote as “high board compensation … could be
related to the need to hire directors with higher professional standing and also to the desire to
create a network with other companies through the enlargement of the board”
Scandinavian countries are characterized by extensive public sector, high taxes and
economic transparency. Randøy and Nielsen (2002) examine the relationship between CEO
41
compensation, corporate governance practices and firm performance in these countries. They
use a sample of 224 companies for the year 1998, of which 120 belong to Norway and 104
belong to Sweden. For financial variables they use 3-year (1996-1998) averages. Using a
cross section OLS regression, they find that CEO compensation and CEO tenure are not
significantly related to firm performance as measured by return on equity, total stock return
and market to book ratio. However, they find CEO compensation to be positively related to
board size, firm size (market capitalization) and foreign board membership. In addition, they
find that CEO compensation is negatively associated with CEO ownership as measured by
total percentage of equity held by CEO.
Similarly, in Portuguese context, Fernandes (2008) focus on the relationship between top
management compensation, firm performance and board structure. They document
significant influence of non-executive directors on compensation level. The top executive
compensation is higher in firms that have larger number of non-executive directors in the
board. They document an interesting result that pay for performance sensitivity is stronger for
the firms in which non-executive directors do not exist. Therefore, fewer agency problems
exist in firms with zero non-executive directors. This questions the effectiveness of
independent board members in aligning the principal-agent interests, especially when major
index firms mostly rely on non-executive directors in Portuguese Market.
In The Netherlands context, Laan et al. (2010) find that there is a positive relationship
between pay and performance after controlling for firm size as measured by number of
employees. They analyze hand collected data on 523 executives in 107 Dutch listed
companies for period 2002 to 2006. They use different components of pay as dependent
variables in panel regression. The performance variables they used include sales, profit,
relative total shareholder return, and earnings per share (EPS). Nevertheless, they conclude
their study by positing two broad questions. First, what drives the pay arrangements? With
regard to this they suggest that the context, especially corporate governance practices,
country settings (legal systems, regulations and the like) and social network are important
factors. Second, how pay arrangements affect executive behavior and firm performance?
In a cross section study on 414 large UK firms from FTSE all share index, Ozkan (2007)
examines the effect of corporate governance practices on the level of CEO compensation.
After controlling for firm characteristics in OLS and Tobit regressions, she shows that the
level CEO compensation is higher in firms that have larger board size and have higher
proportion of non-executive directors. This suggests inefficient monitoring by non-executive
directors. Further, she finds that higher block-holder and institutional ownership reduces the
42
level of cash and total CEO compensation, indicating an active role of block holders and
institutional shareholders in monitoring CEOs. However, in case of equity based
compensation, only block holders seem to have negative impact. Additionally, she documents
that higher directors’ ownership is associated with lower CEO compensation (both cash and
equity-based), while higher CEO ownership is associated with lower equity-based
compensation only.
In another study in UK context, Liu and Stark (2009) empirically examine the cash
compensation of boards of directors. They focus on 169 UK non-financial listed firms that
existed during the period from 1971 to 1998. The performance measures they use include
pre-tax accounting earnings and stock market returns for both own-firm and the industry. By
employing time series model and panel regression, they find that own-firm accounting
earnings have positive effects on cash compensation of board of directors, while value
weighted industry accounting earnings have negative effects on cash compensation of the
board of directors. However, there is weak evidence that stock returns (of both own-firm and
peer group) have impact on cash compensation.
In a more recent study in UK context, Ozkan (2011) examine link between CEO
compensation and firm performance while controlling for corporate governance variables,
CEO characteristics and firm characteristics. She uses hand-collected data on 390 non-
financial companies from FTSE all share index for six years from 1999 to 2005. She focuses
on cash compensation that includes (salary and bonus), equity-based compensation that
includes long run incentive plans and options, and CEO wealth measured by CEO
shareholdings, stock awards and stock options. The set of corporate governance variables
includes institutional ownership, blockholder ownership, number of block holders, executive
directors’ ownership, non-executive directors’ ownership, board size and proportion of non-
executive directors. Firm characteristics include market capitalization, firm sales and Tobin’s
Q, while CEO characteristics include CEO age, CEO tenure. The performance measure she
uses shareholder returns and shareholder wealth. Controlling for year-fixed effects and
industry fixed-effects, she finds that CEO pay-performance elasticity is 0.095 (0.075) for
total direct compensation (cash compensation) which seem to be lower than US CEOs
documented in previous studies. The stock-based sensitivity and median share holdings of
UK CEOs are also lower than that of US CEOs. She further finds that higher institutional
ownership increases the CEO pay-performance sensitivity of option grants. Moreover, pay-
performance sensitivity of option grants appears to be negatively related to CEO tenure
which indicates an entrenchment effect of CEO tenure.
43
Similar to Pakistan, concentrated family ownership is common in Continental Europe.
Building on this, Croci et al. (2012) examine the influence of family and institutional
investors on CEO compensation in 14 Continental European countries. A sample of 3731
year observations from 754 listed firms is used and sample period is from 2001 to 2008.
Regression analysis that controls for specific CEO and firm characteristics reveals that family
control does curb the level of CEO cash and total compensation. It also limits the ratio of
CEO stock-based compensation. The results suggest that CEO compensation is not a choice
for controlling families to expropriate the minority interests. Concerning institutional
investors, the analysis indicates that institutional ownership is positively correlated with CEO
total and cash compensation and this relationship is more pronounced in family firms.
Foreign institutional investors also appear to have positive influence on CEO compensation.
Finally, fraction of stock-based compensation both in family and non-family firms appears to
be positively influenced by institutional holdings.
The pay-performance relationship may change when endogeneity problem is addressed
in the regression models. Taking the lead from this argument, Ntim et al. (2013) show that
pay-performance relation becomes stronger when the research design is controlled for
possible endogeneity in South African context. They use data for 169 companies listed in
Johannesburg Securities Exchange (JSE) for the period 2002 to 2007. Using single equation
model, they find that pay-performance relationship is relatively weak. However, use of three-
stage-least-square model to account for endogeneity in corporate governance variables
significantly improves the estimates of pay-performance relationship. The results confirm the
simultaneous interdependencies among executive compensation, corporate governance
variables and firm performance. Further, it is shown that institutional ownership and presence
of debt lead to lower compensation, consistent with agency theory in that both creditors and
institutional investors engage in executive monitoring. In addition, board size appears to be
positively correlated with executive compensation, consistent with the argument that larger
boards are sign of weak corporate governance.
2.2.1.3 Asian Context
The research on CEO compensation in Asian context is relatively limited. However, like
in the Western World, CEO pay-performance relationship has been the most studied topic in
Asia. Various empirical studies find a link between top executive pay and firm performance.
For example, Firth et al. (2006) show that CEO pay is positively related to firm accounting
performance measures (return on sales) as well as to stock market performance measures
44
(stock returns). However, statistical significance of the results is conditioned to certain
ownership structures. This is consistent with their argument that different kinds of controlling
shareholder induce distinct types of incentive pay. They use a sample of 1647-year
observation from 549 non-financial listed firms in China for period 1998 to 2000. In addition
to positive association between CEO pay and firm performance, Firth et al. (2006) find that
firms that have ownership dominated by State agency do not exercise performance based
compensation plans. In contrast, firms that have ownership dominated by private
blockholders and SOEs do exercise compensation plans that are based on stockholders’
wealth and firm profitability respectively. They further show that foreign ownership also
induces the use of performance (stock returns) based compensation. They argue that these
differences in pay plan adoption between State agency ownership and SOEs and private
blockholders can be explained by the differences in cash flow rights. Although they find that
pay-performance sensitivity is statistically significant for certain ownership structures,
however the magnitude is very small - close to one for every 1,000 RMB change in
shareholder wealth.
In a subsequent study, Firth et al. (2007) examine the impact of firm performance and
evolving corporate governance systems on CEO compensation in China using a similar
dataset as used by Firth et al. (2006). They find a positive link between CEO compensation
and firm accounting performance as measured by return on assets. However, stock returns do
not appear to have significant influence on CEO compensation. Thus, more importance is
given to accounting performance than market performance in setting CEO compensation.
Further findings suggest that the state ownership tends to decrease the pay levels whereas
foreign holdings tend to increase the pay levels. However foreign holdings have positive
effect on the pay-performance sensitivity. Regarding governance mechanism, they find that
firms with joint position CEO/chairman tend not to support performance based compensation.
Firms with greater number of directors on the board also tend to pay less to their CEOs.
Lastly, presence of many non-executive directors on the board has positive effect on pay-
performance sensitivity.
Mengistae and Xu (2004) examine the implications of agency theory mainly by
examining effect of variance of firm performance on pay-performance sensitivity. They use
10-year data in 1980s on around 400 state owned enterprises from a survey conducted by
Chinese Academy of Social Sciences. Consistent with agency theory arguments, they find
that sensitivity of CEO pay to firm performance is negatively related to variance of the firm
performance. Further, CEO compensation is positively related to firm performance as
45
measured by marginal return to executive action. The pay elasticity to sales is somewhat
lower than that is documented in studies conducted on Western data. However, estimated
semi-elasticity of compensation to profitability is comparable with that estimated for
regulated industry in US. The CEO semi-elasticity to profitability is found to be between 0.08
to 0.77 in China compared to between 0.2 to 0.7 in US found by an earlier study. Moreover,
they find that pay-performance sensitivity is greater when there is more competition in
product market in which the company operates.
Using a relatively rich dataset than previous studies in China consisting of 942 firms for
period 1998-2002, Kato and Long (2006) find a positive elasticity of cash compensation
(salary and bonus) of top executives to shareholder value in China. The magnitude of
estimated elasticity of cash compensation appears to be 3.6 percent for every 10 percent
change in shareholder value, whereas, the magnitude of estimated sensitivity appears to be
0.053RMB for every 1,000RMP change in shareholder value. In addition, they find that
executive compensation is positively associated with sales growth. The results further suggest
that although executives in China are neither rewarded for increased profits and nor punished
for lower profits, they are penalized for losses. They also document the effect of ownership
structure on pay-performance relationship. Specifically, they show that State owned firms
have weak pay-performance relationship, whereas, this relationship is strong for private and
listed firms, suggesting that State ownership is less effective in resolving agency problems.
Again consistent with agency theory predictions, Conyon and He (2011) find that
executive compensation is significantly positively associated with firm performance. They
use fixed effects models and a relatively large sample of 1342 Chinese listed firms with 5928
firm-year observations during 2001 to 2005. They further find that the positive pay-
performance relationship is stronger in firms with more independent directors on the board.
In addition, more State ownership and concentrated ownership lead to lower CEO incentives
and executive compensation. Firm size, firm growth opportunities and presence of
compensation committee have positive effect on executive compensation, while firm risk has
negative effect. Lower firm performance leads to higher CEO turnover. Importantly, firms
with higher percentage of independent directors on the board and lower State control are
more likely to penalize CEOs for poor firm performance resulting in CEO replacement. They
also compare compensation of US and Chinese executives and find that the compensation as
measured by sum of salary and bonus in US is nearly seventeen times greater than the
compensation in China. And these US-China differences in executive compensation persists
even after controlling for general economic and governance variables, suggesting that
46
unobserved factors such as institutions and private benefits of control in China and social
norms might play their role.
In another study, Conyon and He (2012) use a more comprehensive dataset to address
the relationship between CEO compensation and firm performance in China. They use the
dataset that includes 2104 unique companies listed on the two Chinese stock exchanges for
period 2000 to 2010. However, the main regression is estimated using sample of 8600 firm-
year observations from 2024 unique firms for period 2005 to 2010. They relax the
assumption of traditional agency models that CEO pay is in equilibrium. They gain insight
from wage dynamic theory and append it with traditional agency theory. They view executive
pay decision as a dynamic process in which incomplete information and learning are equally
important as interest alignment. For this they use dynamic panel model estimated using
generalized method of moments (GMM). They show that CEO compensation is positively
associated with firm performance (both accounting and stock). However, this association is
stronger for accounting performance as measured by return on assets. They further show that
CEO compensation is highly persistent. Last year CEO compensation has significant positive
influence on current level of CEO compensation, indicating that compensation is set
dynamically and adjusted towards target equilibrium level. In addition, they show that
ownership structure and board have strong influence on CEO equity grants and equity
ownership. Moreover, they find that economic determinants of CEO compensation such as
firm performance, firm size and CEOs’ human capital are more important than ownership and
corporate governance. Especially, effects of corporate governance variable are significantly
weaker when the regression models are controlled for pay persistence and unobserved firm
level heterogeneity.
In China again, Su et al. (2010) focus on the relationship between executive
compensation and ownership concentration. They study state owned enterprises (SOE) and
non-state owned enterprises (non-SOE) separately. Using data on 967 firms (307 SOEs and
660 non-SOEs) for the year 2005, they find no significant evidence of the relationship
between ownership concentration and executive compensation in SOEs however results
suggest a U-shaped association in non-SOEs. In addition, results of all the models suggest a
significant positive association between executive compensation and return on assets.
Presumably, concentrated ownership and blockholders have better incentives to monitor
managers through incentive based pay contracts which in turn lead to stronger pay-
performance relationship. However, Wang and Xiao (2011) argue that agency problems
between minority shareholders and controlling shareholders can impede the adoption of
47
incentive based compensation in China. This is possibly due to tunneling effect refers to as
extracting private benefits from controlling shareholders’ position through making self-
dealing transactions such as transferring assets to member companies under controlling
shareholders’ control at favorable prices. Using 6670 year-observations from Chinese listed
companies during 1999 to 2005, Wang and Xiao (2011) show that tunneling effect does
influence the pay-performance relationship. Firms with controlling shareholders tunneling the
economic resources for their private benefits have lower pay-performance sensitivity of
executive compensation. Large amount of tunneling transactions disturb the firm
performance and reduce the demand of controlling shareholders to adopt performance based
compensation contracts. Typically, these transactions are not under managerial control,
making performance measures less informative of executive effort and less suitable for
setting executive pay. Wang and Xiao (2011) assert that tunneling effect can explain the
weak but statistically significant positive relationship between executive compensation and
firm performance in China and suggest that interest alignment between minority and majority
shareholding is an important determinant of adopting performance based compensation
schemes.
While other studies in China study the implications of agency theory, Chen et al. (2011)
test the implications of managerial power theory and tournament theory of executive
compensation in Chinese market. Concerning managerial power theory, they study influence
of three types of managerial power on executive compensation: structural power, prestige
power and political power. The data includes all the firms that disclose compensation for
three highest paid executives in eleven-year span from 1999-2009 (432 firms in 1999 and
1458 firms in 2009). They find that structural power as measured by executive share
ownership and prestige power as measured by executive education are strongly positively
related to executive compensation. In case of relationship between political power, as
measured by executive/party secretary duality, and executive compensation, the sign is
positive but of the significance is weak. Further, they find that position in top organization
hierarchy, as indicated by executive pay level for each of the three highest paid managers, has
positive influence on executive compensation. However, this influence is diminishing
(convex) and when government ownership interacts with executive position in top
organization hierarchy, the influence becomes negative. In addition, executive pay is not
influenced by the number of persons competing for the executive position. Moreover, firm
performance as measured by EPS is positively related to contestants’ compensation gaps and
this relationship becomes negative when government ownership interacts with these gaps.
48
Similar to studies in Chinese context, Kato and Kubo (2006) study the CEO pay-
performance relations in 51 Japanese firms using 10-year panel data on CEOs’ salary and
bonuses from 1986 to 1995. The data is from a private survey of CEO compensation
conducted annually by the consulting firm. Using panel regression, they find that CEO cash
compensation is sensitive to firm performance (especially firm’s accounting performance).
They report that “semi-elasticity” of CEO cash compensation with respect to return on assets
varies from 1.3 to 1.4. Consistent with the argument that Japanese corporate governance
tends to restrain the role of shareholder and emphasizes the role of banks and employees, they
find that stock market tends to have a limited role in CEO compensation decisions. Contrast
to the literature on employee compensation in Japan arguing that bonus is a disguised base
wage; they find that bonus system makes CEO compensation more sensitive to firm
performance.
In a comparative study between US and Japan, using a sample of 613 year-observations
from Japan and 2399 year-observations from US during 1992 to 1996, Mitsudome et al.
(2008) examine the significance of the relationship between CEO wealth and firm value. The
CEOs in both countries appear to be rewarded for contemporaneous firm performance
(operating income and stock returns). However, it is stated that this relationship is more
persistent for Japanese firms than for US firms, suggesting that Japanese managers are
rewarded for both short-term and long term performance. Further, there is no link found
between CEO compensation and sales growth for both countries’ firms. This result seems
inconsistent with the general view in Japan that firms emphasize more on maximizing market
shares than on short-term profitability.
The ‘keiretsu” culture is common in Japan. Basu et al. (2007) use this cultural dimension
and focus on top executive’s pay and firm performance within the corporate governance
environment of Japan. They use data of 300 executives from 174 large Japanese firms during
1992-1996. Using different model specifications and controlling for standard economic
variables, they find that weak corporate governance practices lead to higher executive pay
and vice versa. Further they find that CEO pay is lower in firms associated with keiretsu or
have at least one outside director. Similarly, lower CEO pay is also associated with lower
board size. However, higher stock ownership by the board and family-influence appear to be
related to higher executive compensation, suggesting an entrenchment effect of management.
Finally, they argue that if set of monitoring and ownership variables they used is a good
proxy for decreased agent-principal conflicts, then excessive compensation associated with
poor governance should lead to poor future firm performance. Consistent with their
49
hypothesis, they find that compensation explained by ownership and monitoring variable is
negatively associated with future accounting performances. Therefore, ownership and
monitoring echo the effectiveness of governance structure instead of proxies for neglected
determinants of executive equilibrium compensation. Overall results suggest that weak
corporate governance in Japan leads to increased agency problems which subsequently lead
higher executive pay and finally poor future firm performance. However, keiretsu and outside
directors help reduce agency problems in Japan.
Similar to keiretsu in Japan, a popular institutional setting common in Korea is
“Chaebol” – a conglomerate business structure often run by a family. The presence of
Chaebols make Korean market an interesting environment to study the CEO compensation,
Kato et al. (2007) do this. Using a panel dataset covering 246 firms included in KOSPI200
index for at least two consecutive years between 1998 and 2001, they show that cash
compensation of Korean executives is significantly related to firm stock performance. The
magnitude of this relationship is quite comparable to what has been documented for Japan
and US. Alternative measures such as accounting performance and sales seem to have limited
influence on Korean executive compensation. An important result uncovered is that pay-
performance sensitivity does not exist for firms that are associated with Chaebol and that all
the pay-performance sensitivity is driven by non-Chaebol firms. This suggests that Chaebol
firms in Korea lack in well-functioning internal corporate governance.
In Indian context, Ghosh (2006) investigates determinants of CEO and directors’
compensation using 462 firms from the Indian manufacturing sector for period 1997 to 2002.
Specifically, he examines the influence of firm performance, corporate governance practices
and firm diversification on board and CEO compensation. He finds that both last-year and
current firm performance as measured by ROA and diversification largely influence the board
compensation. However, CEO compensation is influenced by current firm performance only.
Number of outside directors on the board appears to have inverse relationship with fixed as
well as total directors’ compensation. CEO duality increases the board compensation. The
presence of founder CEO and more than one CEO also tend to increase the board
compensation. CEO compensation appears to be more influenced by firm size rather than
firm performance. Regarding influence of CEO characteristics on CEO compensation, only
CEO tenure appears to have positive influence on CEO compensation. This is seemingly
contradicting with existing studies which find positive influence of CEO characteristics such
as age, education and experience on CEO compensation.
50
In another study in Indian context, using cross section data on 324 Indian manufacturing
companies for the year 2005, Ghosh (2010) examine the association between CEO
compensation and firm performance. While controlling from industry effects he finds that
firm performance has significant influence on CEO compensation however, the magnitude is
apparently small. He also reports that big size firms pay more to their CEOs. Further analysis
shows that group associated firms, privately owned firms and foreign private firms have
higher CEO compensation compared to state owned firms.
In three South American emerging markets: Brazil, Argentina and Chile, Gallego and
Larrain (2012) examine the relationship between CEO compensation and types of
concentrated ownership. Using data of about 1700 executives from 720 firms, they show that
large shareholders do not represent a uniform class of “hands-on” principals when it comes to
compensating the executives. They find that ownership concentrated in families is related to
higher CEO compensation but not to the compensation of other executives lower in the
organizational hierarchy. There is a compensation premium of 30 log points for professional
CEOs employed in family firms. Contrary to this, CEOs in foreign concentrated ownership,
shareholder coalitions and the state concentrated ownership do not get any premium. It is
hard to distinguish the contribution of managerial skills, career risk and private benefits or
rent extraction towards family premium. All these might contribute simultaneously and their
relative importance and contribution may vary with time and firm. However, most of the
family premium comes from firms where founders are absent but their children are involved.
This distinguishes the family ownership from other concentrated ownerships.
Recently in a meta-analytic research paper based on 332 studies on compensation in 29
countries, van Essen et al. (2012a) document that the relationship between compensation and
firm performance is moderately positive but statistically strongly significant. This suggests a
modest support for optimal contracting theory. However, there is a considerable cross-
country variability in pay-performance relationship. They find that this variance is due to the
differences in formal and informal institutional settings that protect investors from executive
underperformance and overcompensation. Both formal (rule of law and investor protection)
and informal (ownership concentration and code of good corporate governance) institutions
have significantly positive moderating effect on pay-performance relationship. Further they
find that formal and informant institutions complement each other in influencing the pay-
performance sensitivity. Pay-performance sensitivity appears to be stronger when
concentrated ownership accompanies access to well-effective law courts and when
shareholder protection laws accompany informal norms of good corporate governance.
51
Finally, they suggest supplementing the institutional-based view of compensation with
agency theory in order to account for conditional influences of formal and informal
institutions on national pay practices.
2.2.1.4 Pakistani Context
In Pakistan, Shah et al. (2009) explore the determinants of CEO compensation and use
data on 114 KSE listed companies for period 2002 to 2006. Using pooled regression without
controlling for firm fixed effects and time fixed effects, they find that larger firms pay more
to their CEOs. Larger board and ownership concentration also lead to higher CEO
compensation. However, board independence lead to lower CEO compensation, indicating
that independent board members do have some monitoring role and influence the CEO
compensation decisions. Besides these finding, Shah et al. (2009) report that about 99% of
variation (as measured by R-square) in CEO compensation is explained by the variables they
introduced in the pooled regression. This casts some serious doubts on their findings
especially when they have not used fixed effects to control for unobserved heterogeneity.
Moreover, their research design account for only current CEO compensation and current firm
performance. No lagged values of firm performance are introduced in their empirical model
thus ignoring a very important factor. If compensation paid to CEOs is a reward for their
performance (ex-post) then past firm performance would be an important determinant of
current compensation. Several studies find positive link between CEO compensation and past
firm performance (see, Devers et al., 2007, Frydman and Jenter, 2010, Sun et al., 2010).
Like Shah et al. (2009), Kashif and Mustafa (2012) also study the determinants of CEO
compensation in Pakistan. They focus on only 66 family firms listed in KSE for period from
2010 and 2012. Using pooled model, they find that CEO compensation is conditioned to firm
size only. All other factors such as firm performance, board size, number of independent
directors, CEO duality and family CEO are reported to have no impact on CEO
compensation. However, Kashif and Mustafa (2012) is also plagued with similar difficulties
as Shah et al. (2009). For example, they do not control for unobserved heterogeneity and
endogeneity problem. In addition, lagged firm performance is missing in their analysis.
Besides Shah et al. (2009) and Kashif and Mustafa (2012), a number of studies explore
the impact of corporate governance and family ownership on firm performance in Pakistani
context. For example, using survey data for 37 corporate governance factors from 226 firms
of KSE, Nishat and Shaheen (2006-2007) make a broad summary measure (Gov-Score) of
corporate governance. Their performance measures include profit margin, return on equity,
52
sales growth, Tobin’s Q and dividend yield. All of firm performance measures except
Tobin’s Q appear to be positively related to Gov-Score. Javed and Iqbal (2006) make an
index for corporate governance using 22 corporate governance variables from sample of 50
firms and find that corporate governance is positively linked to firm performance. The
corporate governance index they formulate is based on subjective weight assignment to 22
corporate governance variables which is debatable. Javid and Iqbal (2008) investigate the
factors which influence ownership concentration, and the link between ownership
concentration and firm performance using representative sample of 60 firms for the period
2003 to 2008. They find that Pakistani firms have more concentrated ownership which is
interpreted as response of weak legal environment. Further, concentrated ownership, family
ownership and foreign ownership have positive association with firm performance. They
argue that the identity of ownership matters more than the concentration of ownership.
Ibrahim et al. (2010) examine influence of board size, board independence and
ownership concentration of firm performance in chemical and pharmaceutical industries.
They find that return on equity (ROE) is positively influenced by corporate governance
variables. Yasser et al. (2011) test the relationship between four corporate governance factors
(board size, board composition, CEO/chairman duality and audit committee) and two firm
performance measures (profit margin, PM, and return on equity, ROE), for a sample of 30
Pakistani listed firms between 2008 and 2009. Their major finding includes positive
relationship of PM and ROE with three corporate governance factors (board size, board
composition and audit committee). However, the study does not find impact of CEO duality
on performance. In a recent study Jabeen et al. (2012) empirically examine the impact of
ownership structure on firm’s performance while controlling for contextual variables such as
firm size, firm age, risk, leverage using data from 62 non-financial firms for period 2006 to
2009. They find that family firms in Pakistan under perform as compared to non-family
firms. However, founder family and descendant family firms show no significant difference
in the performance.
2.2.2 Effect of Executive Compensation on Firm Future Performance
Besides considering pay as reward for performance, many studies investigate the
motivational role of pay including its components. Specifically, these studies examine how
CEO compensation is related to future firm performance however fewer studies examine this
relationship. For example, Hayes and Schaefer (2000), using data from Forbes Executive
Compensation Surveys from 1974 through 1995, find positive relation between cash pay and
53
future performance. They argue that, under optimal contracting approach, if both observable
and unobservable performance measures are incorporated in CEO compensation and
unobservable performance measures are correlated with observable future performance
measures, then current compensation that is not explained by current observable performance
measures would be correlated to future observable performance measures. This leads to
hypothesize that current compensation can predict future performance. They further
hypothesize that when less preference is given to observable performance measures as basis
for compensation contracts, the current compensation becomes strong predictor of future
performance.
Regarding first hypothesis, Hayes and Schaefer (2000) find strong evidence suggesting
that future performance can be predicted using unexplained variation in current
compensation. This finding is robust to alternative model specifications. For the second
hypothesis, they find that higher variation in publically observable performance measure
leads to stronger relationship between current compensation and future firm performance
because higher variance in observable performance measure makes it noisier and less
preferable to use in devising compensation contracts. This finding is true for both market and
accounting measures of performance. Overall, the evidence is consistent with the argument
that boards of directors use unobservable information (to outsiders) as function of implicit
incentive pay contracts which makes current compensation a reasonable predictor of future
firm performance.
In the same vein, Kubo (2005) examines how pay-performance link motivates executives
to performance in future. Specifically, he tests the hypothesis that firms with higher pay-
performance sensitivity of directors outperform those with lower pay-performance sensitivity.
He uses sample of 210 firms listed in Japan for period 1993 to 1995. Basic characteristics of
directors’ pay show that, for most of the firms, in response to 1 percent increase in firm
performance the directors’ pay increases from zero to 0.33 percent and there are several firms
which have negative sign for pay-performance sensitivity. This suggests that overall pay-
performance relationship is weak in Japan. To test the implications of pay-performance
sensitivity for future firm performance, Kubo (2005) regresses the future performance on the
direct measure of pay-performance sensitivity. The results show that there is no support for
the hypothesis of positive relationship between pay-performance sensitivity and firm
performance. The results also confirm that there is no relationship between change in pay
policy and firm performance. Therefore, performance improvements induced by highly
geared incentive pay plans offered by firms are doubtful in Japan. Moreover, pay-
54
performance sensitivity does not seem to motivate directors to perform in shareholder
interest.
Similarly, in a proclaimed to be the first study in UK context, Balafas and Florackis
(2014) examine whether excessive pay to CEOs lead to superior future stock and accounting
performance. They utilize the argument of Hayes and Schaefer (2000) that boards of directors
use unobservable (to outsiders) performance measures to reward CEOs for actions that may
increase the firm value (observable performance) in the long run. Importantly, consequences
of such actions may not reflect in the current observable performance. This implies that
efficient pay contracts, especially that include incentive-based compensation, may encourage
executives to exert significant effort to enhance the growth prospective of their firms which
in turn reduce the agency problems and creates the long-run value to the shareholders. These
arguments give rise to the potential relationship between current compensation and future
firm performance – ex-post effects of CEO pay on firm value. Using a sample of 1787 listed
firms on London Stock Exchange for the period 1998 to 2010, Balafas and Florackis (2014)
show that excessive CEO incentive compensation is inversely related to short-term future
stock returns. Interestingly, firms with CEOs getting incentive pay at the bottom of the pay-
distribution produce significantly positive abnormal returns and completely outperform their
counterparts that pay their CEOs at top of the incentive-pay distribution. Additional tests
show that the outperformance of low-incentive-compensation firms can be explained by their
excessive exposure to idiosyncratic risk. A panel regression analysis confirms the negative
relationship between incentive pay and future performance including operating performance.
Adithipyangkul et al. (2011) address the motivational aspect of perks (non-cash
compensation) in China. Specifically, they examine relationship of perks with current and
future firm performance. Using manually collected data (3706 firm-year observations) for
period 1999 to 2004, they document that, as expected, perks are positively related to both
current and future firm performance as measured by return on assets, indicating that perks are
used to reward the performance as well as to motivate the executives for future profitability.
In addition, analysis on stratified samples confirms that perks may motivate executives for
firm performance, even after controlling from firm characteristics such as leverage, growth
opportunities and firm size.
Using Granger causality tests, Buck et al. (2008) analyze the role of compensation as
reward and as motivation in China. They argue that although analyzing the sensitivity of
CEO compensation to stock price performance is a dominant approach to explore CEO pay-
performance relationship however, researchers normally ignore the motivational component
55
and possibility of two-way causation in pay-performance relationship despite the fact that
commonly cited rationale for pay-performance relationship is motivation. Keeping in view
the Chinese institutional settings and compensation practices (absence of stock options based
compensation), they formulate two typical hypotheses related to two-way causation. First,
“there will be a positive relation between executive pay and past and current firm
performance, running from performance to pay” and second, “firm performance will be
positively related to levels of executive rewards, running with pay leading performance”.
They use four-year data on executive compensation from 2000 to 2003 for 601 firms out of
934 listed firms in 2000. Controlling for firm size and board size, they document that
executive pay-performance sensitivity is quite comparable to the sensitivity reported in
studies on Western markets. However, importantly, estimates depict that causality also run
from pay to performance, meaning that executive compensation and firm performance have
mutual effects on each other through both reward and motivation.
Option grants are advocated to be an instrument to reduce moral hazard problems and
align the mutual interests of agents and principals however, according to rent extraction view,
option grants can be used to compensate executives excessively if executives are in control of
the pay setting process (see, Bebchuk and Fried, 2003).
Consistent with former, Hanlon et al. (2003) show that stock option grants to top
management team (TMT) positively influence the future firm performance as measured by
operating income. Using 2,627 firm-year observation for the period 1998 to 2000 and
employee stock option (ESO) grants from 1992 to 2000, they find that one dollar value
(Black Scholes) of an option grant to TMT is related to approximately $3.71 operating
income over the next five years. However, tests of specification indicate that this relationship
between future operating earnings and option grant values is concave (increase with
decreasing rate). They further analyze option grant values using two- stage regression. In the
first stage, they regress option grant values on economic determinants such as growth
opportunities and cash constraints, and proxies for governance quality such as CEO duality,
interlock directorship and board structure. They find that ESO grant values are strongly
related to economic determinants however they do not find expected association between
ESO grant values and proxies for governance quality as predicted by rent extraction view. In
the second stage, future earnings are regressed on the predicted and residual ESO grant
values estimated in the first stage. The results show that earnings payoffs to ESO grant values
due to economic determinants are strongly positive, consistent with incentive alignment view.
56
The payoffs to grant values attributed to governance variables are also positive, inconsistent
with rent extraction view.
Contributing to the debate between self-serving view and interest alignment view of
stock option grants, Kato et al. (2005) show that adoption of stock options, following a
regulatory change in 1997, has significant impact on operating performance in Japan. This is
consistent with the view that Japanese firms use stock options primarily to better align the
agents’ incentives with that of principals. Further, using event study methodology on 562
announcements of adoption of stock option plan by 344 firms, Kato et al. (2005) document
that adopting firms experience about 2% 5-day cumulative abnormal returns around the plan
adoption announcement. In addition, they find that abnormal returns are more positive for the
firms that grant larger portion of shares to board members and abnormal returns are less
positive for the firms that grant larger portion of shares to the lower level employees.
Moreover, firms with higher growth opportunities are more likely to adopt stock option plan.
High leverage and ownership of other corporations in the firm lead to less probability of
adopting stock option plans, indicating less use of incentive pay in the presence of other
monitoring mechanisms.
Relatedly, Fich and Shivdasani (2005) examine the effects of provision of stock option
plan on firm value. Using a sample of 2088 observations from 774 firms for the period 1997
to 1999 and fixed effect models, they show that presence of stock option plans for outside
directors have positive influence on market to book ratios. On average, the positive change in
magnitude of market to book ratio is 0.14 in response to presence of stock option plan for
outside directors. The relationship is consistently positive when various measures of
accounting performance are used instead of market to book ratio. Further analysis based on
an event study indicates favorable investor reaction to the announcement of stock option plan
for outside directors. Analysts also seem to revise their earnings forecasts following the
announcement of the option plan. Appointments of outside directors in firms with option plan
generate cumulative abnormal returns (CARs) near to zero but these CARs are significantly
negative when outside directors are appointed in firm without option plans. Thus, market
responds differently to appointment of outside directors when stock option plans are available
for directors. This leads to conclude that directors’ stock-option plans help align the
shareholder-director incentives, which in turn improves the firm value.
Malmendier and Tate (2009) examine firm performance subsequent to CEO awards
conferred by national media groups. Using hand collected list of award winning CEOs during
1975 to 2002, they document that both stock and operating performances decline following
57
CEO awards. Parallel to that, award winning CEOs extract greater pay, mainly in the form of
shares and share options and this increase in pay following award is not shared by other top
executives in the firm. Further, winning CEOs appear to get involved in those activities (e.g.,
sitting on outside boards, writing books, playing golf) which provide more personal benefits
than benefiting to the firm. In addition, winning CEOs are more likely to involve in earnings
management. Moreover, these value destruction effects of CEO award are more prominent in
firms with poor shareholder rights and poor corporate governance practices.
2.2.3 CEO Compensation and Earnings Management
As discussed earlier in related theoretical section, the literature on performance
manipulation is based on the bonus maximization hypothesis (BMH) initially introduced by
Watts and Zimmerman (1986). BMH posits that managers in firms with cash bonus plans are
more prone to pick that accounting method which transfers reported earnings from future to
current periods, or vice versa, under specific situations. In a well cited paper, Healy (1985)
shows a strong link between accounting accruals and executives’ income-reporting incentives
under bonus contracts. He hypothesizes that executives are more likely to manipulate
earnings downward when their bonus plans are binding to some lower or upper threshold
level of earnings and in contrast, executives are more likely to manipulate earnings upward
when such bindings do not exist. In simple words, when income is so low that the choice of
accounting procedure would not lead to required level of income to earn bonus, executives
have incentives to further decrease the current income by adopting a ‘taking a bath’ strategy -
choosing accelerated write-offs or deferring sales. This decrease in current income would be
offset in the next period earnings. Similarly, if bonus touches its maximum that further
increase in income would not affect the bonus, executive have incentives to limit the earnings
at upper threshold.
Using a sample of 1527 year-observation for 94 companies from Fortune 250 over the
period 1930 to 1980, Healy (1985) shows that managers manipulate earnings upward when
earnings lag behind the minimum threshold level of bonus pay and in case, if managers’
bonus pay touches its maximum they manipulate earnings downward. Also, managers tend to
manipulate earnings downward when income is not likely to cross the minimum threshold
level. Further, he documents that adoption or change in bonus contract lead to greater chances
of voluntary changes in accounting procedures. But, executives do not seem to alter
accounting methods to reduce income in the presence of binding bonus plan lower or upper
bounds. Overall, the results are consistent with their hypotheses.
58
Extending work of Healy (1985), Holthausen et al. (1995) investigate the confidential
data (443 year-observations) on CEO-specific short-term annual bonus plans for period 1980
to 1990. Consistent with Healy (1985), they find that in order to increase the present value of
bonus compensation contracts, managers manipulate earnings downward when bonus
payments touch their maximum. However, contrary to Healy (1985), they find no evidence of
downwards earnings manipulation by executives when earnings are below the lower limit
necessary to receive any bonus. These results are obtained using model developed by Jones
(1991) to estimate the discretionary accruals. They point out that Healy’s findings at lower
bound might be affected by his methodology.
Again extending work of Healy (1985), Gaver et al. (1995) examine bonus data of 102
firms in US for period 1980 to1990. However contrary to Healy (1985) they find that
executives choose income-increasing accruals when minimum targets of income before
discretionary accruals at which bonus would be paid are not met. Other results appear to be
consistent with Healy’s results. They argue that their results are more consistent with the
income-smoothening hypothesis than the bonus maximization hypothesis6.
Once again, Guidry et al. (1999) test the bonus maximization hypothesis through accrual
management however they focus on business unit level data of large conglomerates in US
arguing that business unit level data enhance the power of the test in two ways: 1) reduction
in potential aggregation problem that exists in firm level data, 2) elimination of potentially
confusing effects of equity-based compensation and long-term performance. Using 179 year-
observations over period 1993 to 1995, they show that business unit managers do manipulate
earnings to increase their bonus compensation. The results are consistent when different
proxies for discretionary accruals are used. However, the results are more pronounced for
inventory reserve account possibly. This is possibly because greater information asymmetry
that may exist between upper management and business unit managers in case of inventory
valuation.
Change in accounting policy can affect the accrual income and so the executive bonuses.
Healy et al. (1987) examine the link between change in accounting policy and CEO salary
and bonus compensation. They select two changes in accounting policies. One is change of
inventory method i.e. from FIFO to LIFO and the other is change in depreciation method i.e.
from accelerated to straight-line. They first examine the adjustment of earnings based
6 When income before discretionary accruals remains below the lower bound, the bonus-maximization hypothesis predicts negative (income decreasing) discretionary accruals while the income-smoothening hypothesis suggests positive (Income increasing) discretionary accruals.
59
compensation subsequent to change in accounting policy and then adjustment of
compensation model parameters to cancel out the earnings effect of change in accounting
method. Using data on 52 firms that shift from FIFO to LIFO and 38 firms that shift from
accelerated to straight-line depreciation method over period 1967 to 1976, overall, they find
that potential decrease in CEO salary and bonus subsequent to shift to LIFO is 2.3 percent per
year. CEOs do not appear to resist the switch to LIFO despite adjustment to their
compensation does not occur and they bear loss in their remuneration. In the same way,
potential increase in CEO salary and bonus subsequent to shift to straight-line depreciation is
nearly 1.5 percent per year. Therefore, CEO benefit appears to be small even if adjustment to
their remuneration subsequent to accounting policy change does not occur.
As research on earnings management flourished, researchers start examining the
relationship between executive compensation and earnings management using discretionary
accruals. For example, Balsam (1998) uses 3439 firm-year observations over period 1990 to
1993 and finds a positive association between discretionary accruals and CEO cash
compensation after controlling for increase in share price and nondiscretionary components
of income. However, the association varies with the sign of discretionary accruals. CEO cash
compensation appears to be associated with positive discretionary accruals only. Further, this
association varies with the firm situation i.e., cash compensation is more strongly associated
with positive discretionary accruals when discretionary accruals are used to attain the firm’s
earnings goals. In addition, as level of discretionary accruals increases, the relationship
between CEO cash pay and reported net earnings gets stronger. However, at extreme level of
high discretionary accruals, the relationship appears to be insignificant. This is consistent
with the argument of putting a limit on the management’s ability to increase its pay using
discretionary accruals.
Relatedly, Carter et al. (2009) report that changes in the executive bonus compensation
structure are associated with changes in earnings management behavior. They examine the
relationship between executive bonuses and earnings during pre and post period of
implementation of Sarbanes-Oxley act and other related reforms. Using a sample of 11,113
CEOs and 8,814 CFOs for period 1996 to 2005, they document that as chances of earnings
management decrease after the reforms due to less flexible financial reporting environment,
firms start placing more weight on earnings variations (a greater effort incentive) in the bonus
contracts. The relation between earnings increases and bonuses is stronger in the post
implementation period. The most affected firms by the reforms that show the largest
60
reduction in positive discretionary accruals present the largest increase in the weight on
earnings variations in bonus contracts.
In the same vein, using a sample of 1407 Chinese listed firms during the period from
2002 to 2008 Ye (2014) examines the relationship between independent directors’ cash
compensation and earnings management. Contrary to incentive hypothesis advocated by
Adams and Ferreira (2008) that high cash compensation to independent directors provides
incentives for effective monitoring which in turn restrain earnings management, he
documents that directors’ cash compensation is positively related to the magnitude of
earnings manipulation. He maintains that his results are consistent with reciprocity hypothesis
which states that independent directors receiving higher cash compensation may compromise
their independence and reciprocate higher cash compensation by decreasing the level of
monitoring of insiders (Brick et al., 2006).
The performance manipulation by executives may also be influenced by stock
ownership, long term incentives and equity compensation (Ibrahim and Lloyd, 2011). For
example, using a sample of 266 restated firm-years and around 8,000 non-restated firm-years
during 1995 to 2001, Burns and Kedia (2006) show that CEO’s option portfolio sensitivity to
stock value is positively associated with tendency to misreport. However, they do not find
any significant impact of cash compensation, bonuses, restricted stock, equity and long term
incentive on tendency to misreport the financial statements. The stronger incentive to
misreport in case of stock option is claimed to due to 1) less downside risk on detection of
misreporting due to convexity in CEO wealth brought by stock options, 2) facilitation by
stock option in exit strategies for CEOs for example, CEOs may pool with other managers
that exercise options for diversification and liquidity reasons.
Similarly, Armstrong et al. (2010) examine the relationship of CEO compensation and
equity holdings with accounting irregularities. They use a broader sample relative to prior
studies consisting of 9,118 CEO-firm-year observations (4,559 matched pairs). By employing
a propensity-score matched-pair research design, argued to be more robust as compared to
traditional logistic regressions, they observe that level of CEO equity incentives is weakly
related to shareholder complaints of manipulating accounts, events of accounting related
restatement and Accounting and Auditing Enforcement Releases (AAERs). In fact, they
observe some evidence of lower number of incidents of accounting irregularities in firm with
CEOs receiving higher equity-based compensation compared to firms with CEOs receiving
lower equity-based compensation in a similar contracting environment. Therefore, equity
61
incentives are argued to be effective in aligning shareholder-management interests with
respect to financial reporting.
Using Standard & Poor's ExecuComp database for the period 1993-2000 Cheng and
Warfield (2005) examine the relationship between equity incentives (stock ownership and
stock based compensation) and earnings manipulation. They find that executives receiving
higher equity incentives are more likely to offload their stock in the year subsequent to
earnings announcement. Further, when executive compensation packages contain high
proportion of equity incentives, executives seem to be more involved in earning manipulation
to meet or beat analysts’ forecasts. However, they show that consistently higher equity
incentives to the executives lead to lower chances of reporting larger positive earnings shocks
which is consistent with earning smoothing hypothesis.
While study Cheng and Warfield (2005) suggest that equity incentives provide incentives
to executives to manage earnings, Duellman et al. (2013) document that equity incentives
lead to lower (higher) level of earnings management in the presence of higher (lower) level of
monitoring. They argue that intense monitoring has an interest alignment effect on the
relationship between equity incentives and earnings management, leading to lower level of
earnings management. On the other hand, as monitoring intensity decreases, executive would
likely to behave opportunistically and consequently possibility of earnings management
would increase. They use a sample of 4780 firm-year observations from 1074 distinct firms
during 2001 to 2007.
Timing of exercise the stock option is important especially when CEOs are rewarded in
form of stock options. Bartov and Mohanram (2004) examine the executives’ ability to time
stock option exercise and how that is related to earnings management. After analyzing more
than 1200 firms for the period 1992 to 2001, they document stock price reversal around the
year in which top executives unusually exercise large stock options. They further show that
managers use private information to time the large exercise of stock options. In addition,
earnings are found to be abnormally high in the pre-exercise period, whereas in the post-
exercise period earnings are abnormally low. Overall, evidences suggest that executives have
the ability to time the abnormally larger exercise of options and inflate earnings before
exercising these options to receive greater payouts.
In a similar study Bergstresser and Philippon (2006), using about 4,500 year-observation
during 1990s, show that when CEOs total compensation is more strongly linked to value of
shares and stock options, the use of discretionary accruals to manipulate reported income is
more prevalent. In addition, abnormally higher option exercises and sale of stocks by CEOs is
62
significantly linked to period in which high discretionary accruals make up large part of
reported earnings.
Likewise, Alves (2012) examines 33 non-financial Portuguese listed firms for the period
2003 to 2010 and finds that likelihood of earnings management increases when executives
hold stock options. This seems consistent with opportunistic financial reporting by the
executives. She suggests stock option grants to management appear to be affecting the
information quality, leading to lower value relevance of publically available information.
Since chief financial officer (CFO) is supposed to be responsible for financial reporting,
therefore equity incentive offered to CFO may induce earnings management. Taking on this
argument Jiang et al. (2010b) examine the relationship between CFO equity incentives and
earnings management. Using a sample of 17,542 firm-year observations on CEO and CFO
compensation available with ExecuComp database during 1993 to 2006, they find that CEO
and CFO equity incentives appear to be inducing earnings management however earnings
management as measured by the likelihood of beating analyst forecast and level of accruals is
more responsive to CFO equity incentives. In addition, influence of CFO equity incentives on
earnings management is independent of CEO equity incentives. These findings are claimed to
support new disclosure requirement on CFO compensation by Security and Exchange
Commission (SEC). Moreover, analysis shows that the association of equity incentives of
CEO and CFO with accrual-based earnings management seems to fade away in the post
Sarban Oxley Act (SOX) period.
2.2.4 Summary of Empirical Research
A profound review of relevant literature suggests that it is difficult to establish a straight
forward relationship between CEO compensation and firm performance. Overall, the
empirical literature seems to partially support both optimal contracting and managerial power
approaches (see, Bebchuk and Fried, 2003, Chen et al., 2011, Devers et al., 2007, Frydman
and Jenter, 2010, van Essen et al., 2012b). Importantly, pay-performance relationship is
largely conditioned to socio-economic and institutional peculiarities prevailing in the
economies under examination (Hüttenbrink et al., 2014, Sun et al., 2010, van Essen et al.,
2012a, van Essen et al., 2012b). Thus, the relationship in question is complex and somewhat
country specific.
In US where ownership is diffusely held and legal institutions are strong, the evidence
suggest that the relationship between CEO compensation and firm performance is statistically
significant and pay for performance sensitivities have increased over the last three decades
63
(see, e.g., Frydman and Saks, 2010, Guillet et al., 2012, Sun et al., 2013). This indicates the
utilization of pay for performance relationship to align the executive incentives with
shareholders’ interest. Further, pay for performance sensitivities are higher for top
management as compared to middle management (Aggarwal and Samwick, 2003). However,
although, the pay-performance link seems statistically significant but it is considered to be
weak economically (Tosi et al., 2000). Regarding governance factors, ownership
concentration appears to be reducing the CEO entrenchment effect and related to lower CEO
compensation (Amoako-Adu et al., 2011, Core et al., 1999). Board size, greater number
outside directors appointed by CEO, greater number of gray outside directors and CEO
duality are related to higher CEO compensation in US (Core et al., 1999).
In European countries, family ownership structure is more common than in US (Croci et
al., 2012) however laws of investor protection are well enacted and formal institutions are
effective. The empirical evidence on Europe suggests that there is weak or insignificant pay
for performance relationship. For example, in UK, while Ozkan (2007) finds insignificant
pay-performance relationship, Ozkan (2011) find statistically significant but lower pay for
performance sensitivity as compared to the sensitivity in US. Similarly, pay for performance
link appears to be weak in Italy, Portugal, Denmark and Scandinavian countries (Banghoj et
al., 2010, Brunello et al., 2001, Fernandes, 2008, Randøy and Nielsen, 2002) with exception
of Netherlands where Laan et al. (2010) find some strong evidence of pay-performance
relationship. Nevertheless, using data from 14 Continental European countries, Croci et al.
(2012) report that firm performance as measured by market adjusted stock return, does not
have any significant impact on CEO compensation. Regarding corporate governance
variables, it appears that family ownership, blockholders and institutional ownership are
related to lower CEO pay in European studies (Croci et al., 2012, Ozkan, 2007, Ozkan,
2011). However, Banghoj et al. (2010) show that ownership concentration is related to higher
CEO pay in private firms of Denmark. Higher board size is generally related to higher CEO
compensation in European countries. Higher percentage of non-executive director is also
related to higher CEO compensation (Ozkan, 2007). In addition, pay-performance sensitivity
is low in firms with greater number of non-executive directors sitting in the board
(Fernandes, 2008).
Most of the Asian economies are considered to be emerging. In these countries, firms
face heavy government intervention on business activities through taxation, regulation, and
state ownership (Fan et al., 2011). Governments influence and control various aspects of
business: from output, production process, to input such as labor, energy, land, mines,
64
infrastructures, and financing (Fan et al., 2011). In addition, many emerging markets share
weakness in their financial market development and many governments still control or
influence allocating financial resources based on not just market rules but also other strategic
criteria (La Porta et al., 2002). Overall, formal institutional setup is weak in these markets.
Contrary to US and European countries, executive compensation research in Asia has
come up with more complex findings. Although, the majority of studies find a statistically
significant positive association between pay and firm performance however, there are some
inconsistencies across different sectors and companies. For example, Firth et al. (2006) and
Firth et al. (2007) find that overall CEO compensation is positively related to firm accounting
performance as well as to stock market performance measures in China. However, firms that
have ownership dominated by State agency do not exercise performance based compensation
plans. Similarly, Kato and Long (2006) report positive pay-performance relationship in China
in general however, this association is weak for state owned firms, indicating ineffectiveness
of state-owned firms in resolving agency problems. Su et al. (2010) find significant positive
association between executive compensation and return on assets for both SOEs and non-
SOEs. Mengistae and Xu (2004) find that although pay for performance relationship is
positive however, CEO semi-elasticity to profitability is lower in China than in US. Conyon
and He (2011) show that sum of salary and bonus in US is nearly seventeen times greater
than the compensation in China. However, they find that executive compensation is
significantly positively associated with firm performance in China. In another study, Conyon
and He (2012) find that association of CEO compensation with firm performance is stronger
for accounting performance than market performance in China. Further, they show that
economic determinants of CEO compensation such as firm performance, firm size and CEOs’
human capital are more important than ownership and corporate governance after accounting
for pay persistence and unobserved firm level heterogeneity. Mitsudome et al. (2008) find
that CEOs in Japan and US are rewarded for contemporaneous firm performance however,
pay for performance relationship is more persistent in Japanese firms. Kato et al. (2007)
show in Korea that CEO cash compensation is significantly associated with stock
performance only and the magnitude of the relationship is quite comparable to Japan and US.
However, this pay for performance relationship is mainly driven by non-Chaebol firms. In
Indian market, Ghosh (2010) find positive association between pay-performance however the
magnitude is small.
Concentrated ownership appear to have significant influence on CEO compensation in
US, UK and other European countries. However, only handful of studies examine the
65
relationship between ownership concentration and executive compensation in Asia. These
studies report mixed results. For example, Conyon and He (2011) report that concentrated
ownership lead to lower CEO incentives and executive compensation in China. However,
Conyon and He (2012) find that concentrated ownership does not have significant influence
on CEO compensation. Similarly, Su et al. (2010) find no such evidence for SOEs in China
but find U-shaped association between concentrated ownership and CEO compensation in
non-SOEs. Wang and Xiao (2011) show that majority shareholding is an important
determinant of adopting performance based compensation schemes. Basu et al. (2007) report
that higher stock ownership by the board appears to be related to higher executive
compensation, suggesting an entrenchment effect of management. Surprisingly, influence of
family ownership on CEO compensation is rarely studied in Asia. However, the results
support the expropriation view. For example, Kato et al. (2007) examine influence of family-
group, Chaebol, on pay-performance relationship in Korea and report insignificant directors’
pay-performance relationship in Chaebol firms. Basu et al. (2007) investigate family
influence on level of executive compensation in Japan and report that executive receive
higher pay in family firms.
The studies examining influence of board structure variables on CEO compensation are
also limited in Asia and report mixed results. For example, in China, Firth et al. (2007) report
that higher board size leads to lower CEO compensation. While higher number of non-
executive directors has positive impact on CEO pay-performance sensitivity, CEO duality
tend to reduce pay-performance sensitivity. In contrast to this, Conyon and He (2011) show
that board size, number of independent directors and CEO duality do no significantly affect
level of executive compensation or executive pay-performance sensitivity. Similarly, using
more comprehensive data from China and robust methodology, Conyon and He (2012) find
that effects of these corporate governance variables on the level of CEO pay and CEO pay-
performance sensitivity are insignificant. In Japan, Basu et al. (2007) find that presence of
outside director reduces the executive compensation but board size does not have any effect.
Ghosh (2006) also reports for India that board size does not affect CEO compensation.
However, number of non-executive directors and CEO duality are positively related to CEO
compensation in India.
Although pay as reward for performance is extensively studied area however research on
influence of pay on future performance being as motivation to perform in future is important
and exists concurrently. Overall, the empirical literature reviewed here suggests that pay does
have influence on future firm performance directly (e.g., Adithipyangkul et al., 2011, Balafas
66
and Florackis, 2014, Hayes and Schaefer, 2000) and indirectly in response to stock options
included in the pay plans (e.g., Fich and Shivdasani, 2005, Hanlon et al., 2003, Kato et al.,
2005). However, similar to pay-for-performance relationship, performance-for-pay
relationship also seems to be moderated by the same institutional, socio-economic and socio-
cultural factors.
In sum, CEO compensation research is puzzling and lack consensus and generalization.
The pay-performance relationship is considerably different across the countries of different
regions. Similar differences exist for the effects of concentrated/family ownership and board
structure variables on the level of executive compensation and pay-performance sensitivity.
These differences appear to be due to the differences in formal (rule of law and legal investor
protection) and informal institutional (ownership concentration and code of good corporate
governance) settings that protect investors from executive underperformance and
overcompensation (van Essen et al., 2012a). Given the countrywide differences in the
institutional contexts, it is important to understand how firm performance and corporate
governance practices contribute towards CEO compensations contracts indigenously in
different regions. This study aims to examine how CEO compensation is influenced by firm
performance and corporate governance practices in an emerging market, Pakistan, which is
characterized by high concentrated ownership and family control environment and where
institutional environment is weak.
It is an established fact that Asian socio-economic and behavioral peculiarities, and
institutional settings (formal and informal) are different from Western World and studies
conducted in Western World have limited implications for Asian countries (Fan et al., 2011,
Ghosh, 2006, Hofstede, 1980, Sun et al., 2010, van Essen et al., 2012a). However, within
Asia, Pakistani context is different from other Asian countries like China, Japan and Korea
where most of the Asian studies on executive compensation are based. For example, first,
disclosure requirement about CEO compensation is stronger in Pakistan. Companies are
required to report all the components of CEO compensation. This is not the case for most of
the other Asian countries (see, e.g., Basu et al., 2007, Conyon and He, 2011, Kato et al.,
2007). Therefore, there is more reliable and accurate data available in Pakistan which can
provide better insight into CEO compensation relationships with firm performance and
corporate governance.
Second, concentrated and family ownership is more common in Pakistan than in Japan
and Korea. Although, in China, firms have more ownership concentration than in Pakistan
however that ownership concentration is due to high stake of the government. In Pakistan
67
concentrated ownership is maintained by non-government shareholders. Non-government
ownership concentration makes firms like private owned firms which may have different
implications for CEO compensation. However, the evidence on how concentrated/family
ownership affects CEO compensation almost does not exist in Pakistan.
Third, legal and political environment is weaker and overall governance is poor
(Rehman et al., 2012). Laws are often passed but their implementation is a big issue in
Pakistan due to political influences. In addition, there is more foreign influence on
governance and corporate environment. Pakistan has been under the influence of
International Monetary Fund (IMF) and other funding agencies for so long. Moreover,
Pakistani economy is plagued with more corruption than many Asian countries. According to
Transparency International, the Corruption Perception Index (CPI) never cross 30 for
Pakistan (100 shows no corruption), indicating highly corrupted economy. Also government
effectiveness index and regulatory quality index estimated by World Bank remained negative
in the last decade or so. All this makes executives more prone to unethical and opportunistic
behavior (Mujtaba and Afza, 2011) which might lead to opportunistic executive
compensation not justified against performance. Therefore, it is important to examine how
firm performance and corporate governance practices affect executive compensation in
Pakistan.
Given all the differences, Pakistani market provides a unique context to study CEO
compensation and its implications. Pakistani context is not only different from Western
World but also different from other Asian countries. Moreover, as reforms in corporate
governance deepen in Pakistan and all the responsibility of setting CEO compensation lies
with the board of directors, it is important to study how boards see firm performance as a
determinant of CEO compensation, what role concentrated/family ownership plays in setting
CEO pay, how board structure affect CEO compensation decisions and finally whether CEO
compensation motivate manipulation of firm performance. Thus, overall, this study adds to
the debate in international literature especially in Asian literature on CEO compensation, and
respond to the recent calls (e.g., Fan et al., 2011, Sun et al., 2010) for more research on
executive compensation in Asian context.
With regard to the relationship between executive compensation and earnings
management, several studies (e.g., Balsam, 1998, Carter et al., 2009, Guidry et al., 1999,
Healy, 1985), in general, seem to support the argument that managers do manipulate earnings
to maximize their bonus compensation. The empirical evidence also suggests that managers
receiving equity incentives involve in earnings management and tend to give signals to the
68
market that favor their trade of stock they hold and exercise of stock options. Nevertheless,
better monitoring and sensitivity of pay-performance relationship seem to be some of the
important factors that determine the intensity of the relationship between executive
compensation and earnings management (see, e.g., Burns and Kedia, 2006, Duellman et al.,
2013). However, as discussed earlier corporate governance practices and pay-performance
sensitivity varies with institutional settings under examination therefore may lead to different
results across the countries. The examination of the influence of CEO compensation is an
added contribution of this study in Asian context as there are very limited studies that focus
on this issue in Asia.
On the methodological front, the extant literature shows inconsistencies in the
methodological procedures while examining executive compensation. For example, many
executive compensation studies are cross sectional however in the last decade panel models
seem to replace the cross section models (Devers et al., 2007). But, most of the studies use
only pooled regressions and does not control for unobserved heterogeneity across industries
or firms. In addition, assumptions of heteroskedasticity and serial correlation seem to be
violated which may lead to biased results. Furthermore, there are very few studies (e.g.,
Conyon and He, 2012, Croci et al., 2012, O'Connor and Rafferty, 2010, Ozkan, 2011,
Yoshikawa et al., 2010) in executive compensation that take into account the problem of
endogeneity, indicating methodological limitations in the extant literature.
The existing evidence on CEO compensation in Pakistan is very limited. There are only
two published studies. However, these studies only examine the effect of current firm
performance on CEO compensation but not of the previous firm performance, therefore,
ignore the influence of an important factor. Further, there is no evidence in Pakistan on the
motivational aspects of CEO compensation, for example, 1) evidence related to influence of
CEO compensation on future firm performance and 2) CEO compensation induced earnings
management. In addition, many studies in Pakistan on corporate governance including studies
on CEO compensation use small sample size with relatively shorter time period. For
example, most of the Pakistani studies discussed in literature review section use sample data
consisting of 30 to 114 firms for period of maximum 5 years. Given that there are about 400
non-financial and 250 financial listed companies in Pakistan, the possibility of sampling
biased is greater in these studies. The current study controls for methodological difficulties
and attempt to examine CEO compensation using the most recent and robust methodologies.
Therefore, this study has important policy implications for investors, practitioners and
regulators.
69
Chapter 3
Methodology
70
3.1 Research Philosophy
A research philosophy is an idea or belief about the way in which data about a
phenomenon is collected, examined and interpreted (Saunders et al., 2012). The researchers
need to be aware of their philosophical commitments in terms of choice of research strategy
because this has significant impact not only on what they do but they understand what it is
they are researching (Johnson and Clark, 2006). The decision about an appropriate and
relevant research philosophy depends upon the ontological, epistemological and
methodological considerations. The following lines describe these considerations and
research philosophy adopted for this study.
3.1.1 Ontological Considerations
Ontology is defined as “…claims and assumptions that are made about the nature of
social reality, what exists, what it looks like, what units make it up and how these units
interact with each other. In short, ontological assumptions are concerned with what we
believe constitutes social reality” (Blaikie, 2000, p. 8). The primary point of concern here is
the question whether social entities can and should be regarded as objective entities having a
reality external to social actors, or whether they can and should be regarded as social
constructions built up from the perception and actions of social actors (Bryman, 2012). These
positions are often referred to as objectivism and constructivism respectively.
Objectivism implies that social phenomena and the categories that we come across have
an existence that is separate or independent from actors (Bryman, 2012). For example, as
Bryman (2012) states, an organization can be discussed as a tangible object. It has
standardized procedures to get things done. It has rules and regulation. There exists an
organizational hierarchy. It has goals and a mission statement. The existence of these features
vary from organization to organization, however in thinking we tend to view that an
organization has a reality which is independent and external to the people who inhabit it. The
organization corresponds to a social order in that people working in the organization are
under pressure to conform to the requirements set by the organization. They learn and follow
rules and regulations. They do their jobs according to their job descriptions. They follow
standardized procedures. They are made to learn and apply the organizational values. They
may be penalized or even fired if they do not work as per the organization requirements.
Thus, the organization shapes the people as per its own requirements.
Contrary to that, constructivism also known as subjectivism implies that social
phenomena and categories that we come across are not only created through social interaction
71
of actors but that they are constantly changing (Bryman, 2012). For example, a psychiatric
hospital as an organization can be thought of as social entity constructed through social
interactions of individuals because relatively little of the spheres of action of doctors, nurses,
and other personnel is generally prescribed, the social order of the hospital can be an outcome
of agreed-upon patterns of action that are themselves the products of negotiations between
the different parties involved and it can be in constant state of change (see, Bryman, 2012,
Strauss et al., 1973).
Given the objectives of current study, it is assumed that the relationships between CEO
compensation, firm performance and corporate governance (phenomena) do exist in real
(single reality) and are independent and external to social actors. This reality can be
observable and regarded as objective. Thus, ontologically, this study implies the objectivists
position.
3.1.2 Epistemological Considerations
Epistemology is concerned with the acquisition of knowledge and is related to the
question that how we know what we know. It is defined as “…the possible ways of gaining
knowledge of social reality, whatever it is understood to be” (Blaikie, 2000). Bryman (2012)
states that “An epistemological issue concerns the question of what is (or should be) regarded
as acceptable knowledge in a discipline”. Epistemology has two fundamental positions. One
is positivism and the other is interpretivism. “Positivism is an epistemological position that
advocates the application of the methods of the natural sciences to the study of social reality
and beyond” (Bryman and Bell, 2011). Positivism maintains that the knowledge can
genuinely be regarded as acceptable knowledge if it is confirmed by the senses in a scientific
way. The objective of theory is to generate testable hypotheses. These hypotheses are
confirmed or rejected through data collection and analysis and contribution is made towards
the explanation or development of theory. This requires the use of an optimal and well-
structured research methodology and techniques by the researcher (Bryman and Bell, 2011).
Positivism further stresses that the scientific research must be conducted objectively and
should be independent of values. In addition, positivism asserts to make clear distinction
between scientific and normative statements and advocates the supremacy of scientific
statements over normative ones (Bryman, 2012).
On the other hand as stated by Bryman (2012), “interpretivism is a term that usually
denotes an alternative to the positivist orthodoxy that has held sway for decades. It is
predicated upon the view that a strategy is required that respects the differences between
72
people and the objects of the natural sciences and therefore requires the social scientist to
grasp the subjective meaning of social action”. Interpretivism requires researchers to try to
understand and make sense of the surroundings. And this is done in a continuous way by
interacting with others and giving meaning to the perceptions about others based on this
interaction (Saunders et al., 2012).
Given the ontological position of the study as objectivist, epistemologically, this study
implies the positivist position. Different theories are used to develop hypotheses about the
reality then these hypotheses are tested using a well-structured methodology and inference is
made about the observed phenomena. The observer remains external to the reality and is not
influenced, or does not influence the observed phenomena. Overall, the research is conducted
in a scientific way.
3.1.3 Methodological Considerations
There are two research approaches, deductive and inductive. In deductive approach,
hypotheses are developed about the relationships between the concepts or variables from an
existing theory, a set of theories or existing literature. These hypotheses are expressed in
operational terms indicating how variables in questions are to be measured. Data is collected
and analyzed. Then based on the findings, hypotheses are confirmed or rejected. Finally,
theory(theories) is(are) updated if necessary. On the other hand, in inductive approach, data is
collected and analyzed first and then generalizable inferences are drawn from the analysis to
make a theory (Bryman, 2012).
Bryman and Bell (2011) suggest that the deductive approach in research is typically
related to positivist position. Therefore, given the ontological position of the study as
“objectivist” and epistemological position of the study as “positivist”, this study employs a
deductive research approach as a methodological consideration. The contribution of firm
performance and corporate governance practices towards CEO compensation is explored.
The arguments are derived mainly from agency theory and managerial power theory. The
extant theoretical and empirical literature provides a strong background for the relationships
under consideration. This enables the researcher to advance the research hypotheses. Then
data is collected and analyzed, and outcomes are used to confirm or reject the hypotheses.
And finally explanations are provided for confirming or rejecting the hypotheses.
Another challenge in methodological consideration is the choice of quantitative vs
qualitative approaches to data collection and analysis. The quantitative approach is referred to
as “a research strategy that emphasizes quantification in the collection and analysis of data”
73
(Bryman, 2012, p. 35). The quantitative approach represents objectivist ontological position
and view social reality as external, embodies positivist epistemological position and
incorporate the norms and practices of the natural scientific model, and involves a deductive
approach to research with emphasis on testing of theories (Bryman and Bell, 2011, Saunders
et al., 2012). On the other hand, qualitative approach is referred to as “a research strategy that
usually emphasizes words rather than quantification in the collection and analysis of data”
(Bryman, 2012, p. 36). The qualitative approach represents constructivist ontological
position, embodies interpretivist epistemological position and rejects the norms and practices
of the natural scientific model, and accentuates an inductive approach to research with
emphasis on the creation of theories (Bryman, 2012).
The study adopts a quantitative approach in data collection and analysis based on
objectivist ontology position, positivist epistemology position and a deductive approach to
research. A survey database methodology is used for data collection based on
longitudinal/panel data for eight time horizons through secondary data sources.
Ontology
Objectivist
Epistemology
Positivist
Methodology
Deductive Approach
Quantitative Analysis
Figure 3.1
Research Philosophy
74
3.2 Hypotheses Development
3.2.1 CEO Compensation and Firm Performance
Agency theory (Jensen and Meckling, 1976) is the standard approach to executive
compensation. In this theory, shareholders have two options to solve the agency problems.
First, they obtain more information about managers’ behavior and actions (effort) with better
monitoring and curtail self-interested behavior. Second, they design such compensation plans
that provide managers with incentives in a way that managers’ interests become aligned with
shareholders’ interests. Since shareholders can only imperfectly monitor managers’ actions,
or doing so incur heavy monitoring costs therefore, as a result, shareholders opt to design
optimal compensation plans that attract, retain and motivate executives to perform in the best
interest of the shareholders i.e. to maximize shareholders’ wealth. As managers are assumed
to be risk-averse and self-interested therefore, they would adhere to the incentives provided
by optimal compensation contracts and would make more effort, leading to an outcome that
is more desirable by the shareholders. This predicts a link between executive compensation
and firm performance.
Family and concentrated ownerships are common in Pakistani firms. Ownership
concentration and family ownership provide strong incentives to large shareholders to
oversee managers’ activities (Jensen and Warner, 1988). Therefore, concentrated ownership
generally suggests that shareholders are able to better protect their interests in their
companies, leading to mitigating potential CEO entrenchment and relatively strong
association between CEO compensation and firm performance.
Code of corporate governance in Pakistan makes board of directors responsible for
designing CEO compensation contracts. However, direct relations between controlling
shareholders/families and management in Pakistan, generally, lead to bypassed boards in
decision making (Kamran and Shah, 2014, World Bank, 2005), resulting in more powerful
CEOs than the boards of directors. In addition, formal institutions such as legal system and
investor protection are weak in Pakistan, making CEOs more prone to behave
opportunistically. Thus, given this context, CEOs are more likely set their own pay
opportunistically which is less likely to be correlated with firm performance. However, going
with standard approach i.e. agency theory, this study makes the main hypothesis that:
75
Hypothesis 1: Firm performance has a positive effect on CEO compensation.
3.2.2 CEO Compensation and Corporate Governance
Designing optimal contracts for executives and better monitoring go side by side in
solving agency problems. As an efficient corporate governance structure provides close
monitoring and oversight of management, therefore in the presence of strong corporate
governance, the total executive compensation would be low and be linked to firm
performance. On the other hand, weak corporate governance may make executives more
powerful relative to board of directors, leading to more influence on pay-setting process for
higher compensation which may not be linked to firm performance.
A number of corporate governance indicators have been reported to influence the CEO
compensation (see, e.g.,Core et al., 1999, Devers et al., 2007, van Essen et al., 2012b). In
Pakistan, several studies (e.g., Awan, 2012, Ibrahim et al., 2010, Javed and Iqbal, 2006, Javid
and Iqbal, 2008, Khatab et al., 2011, Nishat and Shaheen, 2006-2007, Yasser et al., 2011)
examine the impact of corporate governance on firm performance, however, none of these
examine the relationship between executive compensation and corporate governance. The
only exceptions are Shah et al. (2009) and Kashif and Mustafa (2012). Both suggest that
board size has positive effect on CEO compensation. However, Shah et al. (2009)
additionally suggest that ownership concentration is also positively while board independence
is negatively related to CEO compensation.
A number of corporate governance variables have been reported to influence the CEO
compensation, however, there influence seems to be conditioned to the context under
examination. (see, e.g.,Core et al., 1999, Devers et al., 2007, Sun et al., 2010, van Essen et
al., 2012b). Since concentrated and family ownership is an important feature of corporate
governance environment in Pakistan and board composition has been one of the major
focuses in the Code of corporate governance to mitigate the agency conflicts (see, SECP,
2002, SECP, 2012), therefore we focus on concentrated/family ownership and board structure
(Board size, Board Independence and CEO duality) in this study. The hypotheses related to
each variable are discussed in the following section.
3.2.2.1 Ownership Concentration, Family Ownership and CEO compensation
Concentrated and family ownership can affect CEO compensation contract in two
competing ways, 1) interest alignment effect and 2) entrenchment effect. According to
76
interest alignment effect, which relates to agency theory, large or family shareholders have
strong incentives to oversee agents’ activities because of being insiders, strong commitment
and better firm specific knowledge (Bertrand and Schoar, 2006, Harris and Raviv, 2008,
Jensen and Warner, 1988, Su et al., 2010). Therefore, concentrated and family ownership
generally suggests that shareholders are better able to protect their interests in their
companies, leading to reduced managerial opportunism, higher interest alignment and lower
CEO compensation.
However, according to entrenchment effect, family or controlling shareholders can
expropriate minority shareholders’ interests through many ways including excessive
compensation packages (Croci et al., 2012, Su et al., 2010, Wang and Xiao, 2011). Similarly,
CEOs in close relation with controlling shareholders or a family may set their own pay
opportunistically high, thereby expropriating the minority shareholders’ wealth. Although, a
few studies (e.g., Abdullah et al., 2011) suggest that there is likelihood of expropriation of
minority interests by controlling/family shareholders in Pakistan however going with interest
alignment effect, this study hypothesize that:
Hypothesis 2: Ownership concentration has a negative impact on CEO compensation.
Hypothesis 3: Family firms pay lower CEO compensation as compared to non-family
firms.
3.2.2.2 Board Size and CEO Compensation
Board size is considered to be an important determinant of board effectiveness, leading
to significant contribution to quality of governance (Jensen, 1993, Lipton and Lorsch, 1992).
Large board size is likely to have a greater level of expertise. However, it may become so
heavy, leading to ineffective executive monitoring (Jensen, 1993). Boards are less likely to
function effectively and are easier to be controlled by executive when board members exceed
seven or eight (Jensen, 1993). Similarly, Lipton and Lorsch (1992) suggest that larger board
size leads to problem of dysfunctional boards and recommend limiting board size to ten
directors. Larger boards may also suffer from communication and coordination difficulties,
leading to board ineffectiveness (Ozkan, 2007, Yermack, 1996). Thus, larger boards are
assumed to compromise their monitoring role and hence weaken the internal governance
structure. Consequently, executives gain more power over the internal control mechanisms,
77
leading to more influence on their own pay, resulting in higher executive compensation.
Many studies find that larger boards are related to higher executive compensation (e.g., Core
et al., 1999, Croci et al., 2012, Fahlenbrach, 2009, Kashif and Mustafa, 2012, Ozkan, 2011,
Shah et al., 2009, van Essen et al., 2012b).
In Pakistan, board size generally tends to be driven by directors appointed from the
controlling families or by proxy directors working on behalf of controlling shareholders.
Consequently, it is less likely that board size has any effective role in reducing agency
conflicts. In that case, board size should not be negatively correlated with CEO
compensation. Thus, this study makes the hypothesis that:
Hypothesis 4: Large board size leads to higher CEO compensation.
3.2.2.3 Board Independence and CEO Compensation
Agency theory suggests that independent directors are likely to play important role in
aligning shareholder-manager interests by providing adequate monitoring. Accordingly,
board independence is encouraged in the Code of corporate governance (2002)7 issued by
SECP to improve the quality of governance. If pressure of independent board reflects a
tendency for increased quality of corporate governance, we can expect to observe different
patterns of executive compensation and incentives. Independent outside directors are less
subject to collude with management and have reputation to protect shareholders in the labor
market (Core et al., 1999, Fama and Jensen, 1983), therefore, have strong incentives to
effectively monitor the CEO. On the other hand, inside directors are more obligated to CEO
and can be under greater CEO influence, leading to compromised CEO monitoring to get
personal benefits from CEO such as career opportunities (see, Bebchuk and Fried, 2003,
Weisbach, 2007). Nevertheless, external directors are also prone to have negative impact on
internal governance if they have some secret relationship with management (Core et al.,
1999). Thus, overall, board independence is expected to be related to efficient compensation
contracts and less managerial opportunism.
Empirically, available evidence is mixed over the relationship between board
independence and executive compensation. For example, Boyd (1994), Lambert et al. (1993)
and Core et al. (1999) find positive association, while others (e.g., Byrd and Cooperman,
7 New Code issued in March, 2012 makes a mandatory requirement to have at least one independent director on board.
78
2010, Conyon and He, 2011, Conyon and He, 2012) find no or negative relationship between
number of independent directors in board and executive compensation. Given the Pakistani
context where non-executive directors are generally hired from within the family or obligated
to work on behalf of controlling shareholders (Javid and Iqbal, 2008, World Bank, 2005), we
may expect so-called board independence to become irrelevant in corporate decision making,
leading to non-negative relation between CEO pay and board independence. However, an
earlier study in Pakistan, Shah et al. (2009), finds negative association between CEO
compensation and board independence. Thus, going with agency theory argument this study
hypothesizes that:
Hypothesis 5: Board independence reduces CEO compensation.
3.2.2.4 CEO Duality and CEO Compensation
Code of corporate governance 2012 in Pakistan requires companies to separate CEO
position from chairman of the board of directors, indicating that CEO duality (CEO as
chairman board of directors at the same time) is a potential agency problem in Pakistan. CEO
duality suggests `concentrated power in one person. This provides opportunity to “self-
interested” CEOs to influence the major decisions in order to maximize their own utilities
instead to maximizing shareholders’ wealth (Core et al., 1999, Jensen, 1993). Thus,
separation of CEO from chairman position is suggested to improve board effectiveness
(Jensen, 1993). CEO duality reduces the board independence and increases the executive
powers over control decisions including designing executive compensation contracts, leading
to higher executive compensation.
Empirically, several studies (e.g., Boyd, 1994, Brick et al., 2006, Conyon and He, 2012,
Core et al., 1999, Fahlenbrach, 2009, van Essen et al., 2012b) find positive association
between CEO duality and executive compensation, indicating CEO entrenchment and
excessive payment to CEO when he/she is also a chairman board of directors. In Pakistan,
Shah et al. (2009) and Kashif and Mustafa (2012) find no evidence of the relationship in
question. However, taking the view the Code of corporate governance in Pakistan that CEO
duality leads to ineffective governance structure this study makes the hypothesis that:
Hypothesis 6: CEO duality has positive impact on CEO compensation.
79
3.2.3 CEO Compensation and Future Firm Performance
A common feature of much of the literature is that it views CEO compensation as reward
for the materialized firm performance and examines the relationship of CEO compensation
with current and past performance. Nevertheless, the other view is that board may reward
executives for value-maximizing activities with outcomes that have not yet materialized, and
hence that are unobservable to outside shareholders (see, Balafas and Florackis, 2014, Cooper
et al., 2013, Hayes and Schaefer, 2000). The former perspective believes that a positive
relationship between CEO compensation and materialized firm performance is ascribed to the
fact that materialized performance signals the ability of CEO to effectively manage the firm
(see, Danker et al., 2013). While, latter suggests that efficient compensation packages
especially that includes incentive-based compensation may motivate executives to exert
costly effort to enhance the growth opportunities of their firms, leading to long term value
creation to shareholders and reduced shareholder-manager agency problems (see, Balafas and
Florackis, 2014).
Accordingly, under agency theory, if boards compensate CEOs for both observable and
unobservable performance measures and unobservable performance measures correlate with
future observable performance measures then current compensation that is unexplained by
current observable performance measures is likely to be correlated to future observable
performance measures (see, Balafas and Florackis, 2014, Cooper et al., 2013, Hayes and
Schaefer, 2000). This leads to the deduction that current CEO compensation can predict
future performance, leading to the hypothesis that:
Hypothesis 7: CEO compensation has positive effect on future firm performance.
Several studies find that current compensation packages influence future performance
(see, e.g., Balafas and Florackis, 2014, Buck et al., 2008, Cooper et al., 2013, Hanlon et al.,
2003, Hayes and Schaefer, 2000). However, in Pakistan, this study would be the first to
explore this relationship in concentrated/family ownership and weak legal environment.
3.2.4 CEO Compensation and Performance Manipulation
Corporate boards are presumed to restrain CEO opportunism and align the CEO actions
with shareholders’ interests by designing efficient compensation contracts that typically link
CEO compensation with firm performance. However efficient compensation contracts may
encourage managers to manipulate performance because their compensation is tied to some
80
pre-specified performance criteria (Sun, 2012). These performance criteria generally include
earnings targets, share price performance and some specific financial ratios that signal the
financial well-being of the company (Ibrahim and Lloyd, 2011, Puffer and Weintrop, 1991).
Therefore, managers tend to manipulate earnings through the choice of accounting policies
because their compensation is tied to firm performance (Watts and Zimmerman, 1978).
Through the choice of accounting methods, managers can increase incentive cash
bonuses or value of the firm share, leading to increased manager’s wealth. Therefore,
managers are prone to pick those accounting policies which transfer reported earnings from
future to current period, or vice versa, under specific situations (Watts and Zimmerman,
1986). Now, if executive in Pakistan are prone to exert a costly effort8 to manipulate earnings
in order to increase their pecuniary benefits then we can expect that CEO compensation
would be related to earnings manipulation.
Existing literature finds evidence that executive compensation is associated with earning
manipulation driven by a managers’ desire to meet the performance targets, leading to higher
compensation (see, Devers et al., 2007, Frydman and Jenter, 2010, Ibrahim and Lloyd, 2011,
Ye, 2014). However, in Pakistan, I could not trace any study that explores the motivation for
earnings management related to executive compensation. Given the institutional and
corporate governance structure in Pakistan (see, Cheema, 2003, Ibrahim, 2006, Kamran and
Shah, 2014, Rehman et al., 2012, World Bank, 2005), management may tend to window
dress the firm performance opportunistically in order to increase their compensation. CEO
compensation can be linked to firm performance on one hand in order to give the impression
that CEO is being rightly compensated while on the other hand performance is manipulated.
Thus, I make the hypothesis that:
Hypothesis 8: CEO compensation has positive impact on earnings management.
3.3 Measurement of Key Variables
3.3.1 CEO Compensation
Typically, CEO compensation has been reported to have the forms: base salary, annual
bonus, perks, long-term incentive plans, stock options, restricted stocks and post-employment
8 Costly effort in the sense that if managers get caught for manipulating earnings to gain pecuniary benefits, this would be fatal for their career
81
benefits (Frydman and Jenter, 2010, Murphy, 1999). The literate classifies executive
compensation into two broad classes i.e., cash compensation (short-term compensation) and
non-cash compensation (long-term compensation) based on nature and/or time-horizon of the
award. Cash compensation is the remuneration paid to the executives during the fiscal year. It
may include base salary and cash bonuses (see, e.g., Cooper et al., 2013, Core et al., 1999,
Croci et al., 2012, Ozkan, 2011, Shaw and Zhang, 2010) or may include base salary, cash
bonuses and other cash benefits (see, e.g., Balafas and Florackis, 2014, Conyon and He,
2011, Conyon and He, 2012, Ntim et al., 2013). Other forms are included in non-cash
compensation.
In Pakistan, long term incentive plans, stock options and restricted stocks are virtually
absent. CEOs are paid in the form of base salary, cash bonuses, perks and benefits and post-
employment benefits. These elements of pay can easily be located in the financial statements
of the companies because under International Accounting Standard (IAS 24), companies are
required to disclose all remuneration paid to CEOs during the year. However, companies use
many accountings heads in disclosure and nomenclature of the accounting heads is
inconsistent across the firms. The most used accountings heads include managerial
remuneration, bonus, leave encashment, house rent, utilities, medical expenses, gratuity,
provident fund and retirement benefits. These heads are clear in their nature that to which
category they would fall in such as base salary, bonuses, perks or retirement benefits.
Consistent with existing literature, this study uses two measures of CEO compensation i.e.,
cash compensation and total compensation. Cash compensation includes managerial
remuneration and bonuses while total compensation is the sum of all the components.
3.3.2 Firm Performance
Both accounting and market measure of firm performance is analyzed. Accounting
performance is represented by return on assets (ROA) estimated by ratio of income before
interest (EBIT) and taxes to total assets (see, e.g., Banghoj et al., 2010, Bebchuk et al., 2011,
Conyon and He, 2012, Matolcsy and Wright, 2011). Another popular measure of accounting
performance i.e. EPS (see, e.g., Chen et al., 2011, Laan et al., 2010) is also calculated but that
is for robustness purposes. EPS is estimated as earnings before taxes divided by number of
shares outstanding.
Market performance is measured as total return to shareholders calculated as current
market price share plus dividend for the current year divided by previous year market price
(see, e.g., Basu et al., 2007, Boschen et al., 2003, Guthrie et al., 2012, Liu and Stark, 2009).
82
In Pakistan, companies are required to lay annual financial statements in the annual general
meeting (AGM) which must be held within 4 months after the close of financial year9. For
example, if company’s financial year closes on 30th June, the AGM must be held by the end
of October. Thus, it takes no more than four months for shareholders to see the information
contained in the annual reports. Keeping this in view, market price per share is taken on the
date that is 4 months after the closing date10. One advantage of this measure is that no
adjustment is needed for the announcement of final dividends for the year as ex-dividend date
resides within the 4-month span. Nevertheless, market price on the closing date is also
collected for robust analysis.
3.3.3 Corporate Governance Variables
3.3.3.1 Family Ownership
Many definitions of a family firm exists (see, e.g., Anderson and Reeb, 2003a, Anderson
and Reeb, 2003b, Gómez-Mejía et al., 2007, Villalonga and Amit, 2006), however, following
Anderson and Reeb (2003a), Wang (2006) and Achleitner et al. (2014), this study define a
family firm that fulfills one of the following conditions.
1) a person or family group holds at least 25% of voting right as measured by
percentage of shares owned directly or indirectly. An indirect voting right can be
achieved through a trust or holding company.
2) two or more family members sit in the board of directors.
To identify the firm’s family status, initially, family members are identified in the board of
directors by their family names from company-information page in the annual reports. In the
next step, pattern of shareholding (part of annual report) is examined for family ownership. It
is important to note here that in some instances control is achieved through holding company
or trust. Therefore, holding company or trust is followed on the internet to know the
ownership status of the holding company. If that belongs to a family and its stake in the
company is more than 25% then the company is recognized as family firm.
3.3.3.2 Ownership Concentration
Consistent with existing literature on ownership concentration (see, Holderness, 2014, La
Porta et al., 1999), this study uses a simple measure of ownership concentration i.e.,
ownership of largest shareholder. However, ownership stake of three largest shareholders is
9 See section 233 of Companies Ordinance, 1984 10 Several studies (e.g., Callen et al., 2010, Jiang et al., 2010a, Lara et al., 2011) adopt similar approach
83
also collected for robustness purposes. Three largest shareholders and their ownership stakes
are obtained from pattern of shareholding included in the annual reports.
3.3.3.3 Board Size
Board size is measured as number of sitting directors on the board as mentioned in the
annual reports.
3.3.3.4 Board Independence
Although Code of corporate governance (2012) in Pakistan requires the representation of
independent director on the board however, this was a voluntary requirement until year 2013.
Further, disclosure regarding independent director was very much inconsistent across the
companies. Therefore, this study uses ratio of non-executive directors to board size as a
measure of board independence. The number of non-executive directors is mentioned in the
Statement of Compliance with Code of Corporate Governance annexed with annual reports in
Pakistan.
3.3.3.5 CEO Duality
CEO-duality is incorporated as a dummy variable taking the value one if CEO is also the
chairman of board of directors and zero otherwise.
3.4 Data
The study focuses on all the non-financial firms listed at Karachi Stock Exchange (KSE),
Pakistan for the period 2005 to 201211. The list of companies is taken from annual balance
sheet analysis (BSA) published by State Bank of Pakistan (SBP) in 2013. The financial firms
are excluded because they have different accounting and regulatory requirements and
structures of valuation ratios and profits. Similar exclusion is made in other studies in
Pakistan (see, e.g., Javid and Iqbal, 2008, Shah et al., 2009).
There are total of 399 non-financial listed companies in BSA classified in 12 economic
(industry) groups by SBP. In the first step 86 companies are dropped because they are
declared defaulted by KSE, newly listed, merged or demerged during the sample period,
leading to inconsistent and missing financial data in BSA. For remaining 313 companies,
attempt is made to collect data for the variables that are not included in BSA such as
11 Since new Code of corporate governance is issued in March 2012 (applicable from year 2013), therefore, to avoid any structural change in the variables due to amended requirements of the Code, the data period ends at year 2012.
84
corporate governance variables, cash flows and the main variable CEO compensation. Since
no database is available in Pakistan that covers these variables so data is extracted from hand
collected annual reports of the companies. The annual reports are collected from different
sources including SBP, KSE and companies’ websites. As I go farther from 2012, the
availability of annual reports decreases therefore sample period is restricted to start from
2005. One benefit to start data period from 2005 is that the Code of corporate governance is
implemented from 2003 therefore the choice of starting period as 2005 gives two-year
adjustment period to listed firms because provisions of the Code, 2002 are not mandatory.
Out of 313 companies 53 companies are dropped due to non-availability of annual
reports. For the remaining 260 companies, I manage to collect data for at least three
consecutive years, making an unbalanced panel data containing 1836 firm-year observations
from 260 companies categorized in 12 industrial groups. However, for compensation
analysis, out of 1836 firm-year observations, 328 observations are dropped because in those
observations CEO compensation is zero. Final sample contains 1508 firm-year observations
from 225 firms for the period 2005 to 2012. In the final step, market data is matched and
collected from official website of KSE.
Table 3.1 describes the industry wise distribution of observations. Industry wise
maximum observations (505 observations) are coming from textile group because it is the
biggest group in KSE. Food and, chemicals and pharmaceuticals groups are second and third
largest contributors to the dataset with 222 and 212 observations respectively. The last two
columns of Table 3.1 show the percentage representation of each group in the sample and the
whole population respectively. Despite some discrepancies, sample data has fair
representation of all the sectors.
Table 3.2 reports the observations by pay components. Out of 1508 observations, only
374 (24.80%) observations have bonus payments, suggesting that smaller number of firms
pay bonuses to CEOs. 657 (43.57%) observations have retirement benefits while 1399
(92.77%) observations have perks (such as club fee, house rent, utility bills and other
benefits). This suggests that CEOs are drawing maximum pay in the form of basic pay and
perks.
85
Table 3.1 Industry Wise Observations
Economic/Industry Group as per BSA Observations
Representation (%age)
Sample Population
Textiles 505 33.49 38.85
Food 222 14.72 13.53
Chemicals and Pharmaceuticals 212 14.06 10.78
Other manufacturing n.e.s 102 6.76 7.52
Other non-metallic mineral products 118 7.82 7.02
Motor vehicles, trailers and auto parts 86 5.70 5.51
Fuel & Energy 77 5.11 4.51
Information, Communication & transport Services 39 2.59 3.26
Coke and refined petroleum products 70 4.64 2.26
Paper, paperboard and products 25 1.66 2.26
Electrical machinery and apparatus 24 1.59 2.01
Other services activities 28 1.86 2.51
Total 1508 100 100
Table 3.2 Observations by Pay Components
Basic Pay Bonus Perks Retirement
Benefits Total
Observations
1508 374 1399 657 1508
100% 24.80% 92.77% 43.57% 100%
86
3.5 Model Specifications
3.5.1 CEO Compensation, Firm Performance and Corporate Governance
To test the association of CEO compensation with firm performance and corporate
governance (first six hypotheses), following model is estimated. Similar models are used in
the extant literature (e.g., Gallego and Larrain, 2012, Murphy, 1999, Sun et al., 2013, Wang
and Xiao, 2011)
𝐿𝑁𝐶𝑂𝑀𝑃𝑖𝑡 = 𝛽1𝑅𝑂𝐴𝑖𝑡 + 𝛽2𝑇𝑅𝐸𝑇𝑖𝑡 + 𝛽3𝐵𝐷𝑆𝐼𝑍𝐸𝑖𝑡 + 𝛽4𝐵_𝐼𝑁𝐷𝑖𝑡 + 𝛽5𝐹𝐴𝑀𝑂𝑊𝑁𝑖𝑡
+ 𝛽6𝑂𝑊𝑁𝐶𝑂𝑁𝑆𝑖𝑡 + 𝛽7𝐷𝑈𝐴𝐿𝐶𝐸𝑂𝑖𝑡 + 𝛽8𝐹𝐼𝑅𝑀𝑆𝐼𝑍𝐸𝑖𝑡 + 𝛽9𝐹𝐼𝑅𝑀𝑅𝑆𝐾𝑖𝑡
+ 𝛽10𝑀𝑇𝐵𝑖𝑡 + 𝛽11𝐹𝑀𝐴𝐺𝐸𝑖𝑡 + 𝛽12𝐶𝐸𝑂𝐶𝐻𝑁𝐺𝑖𝑡 + 𝛼𝑖 + 𝜔𝑡
+ 𝜀𝑖𝑡 (1)
Where
LNCOMPit Log of cash compensation and total compensation, cash compensation
includes managerial remuneration and bonuses while total compensation
includes cash compensation, perks and retirement benefits
ROAit Return on assets, earnings before interest and taxes divided by total assets
TRETit Total return to shareholder, sum of current year market value and dividend
divided by previous year market price
BDSIZEit Board size, number of directors on the board
B_INDit Board independence, ratio of non-executive directors to board size
FAMOWNit Family ownership, a dummy variable taking value 1 for family firms and zero
otherwise
OWNCONSit Ownership concentration, percentage of shares held by largest shareholder
DUALCEOit CEO Duality, a dummy variable taking value 1 if CEO is also chairman board
of directors and zero otherwise
FIRMSIZEit Firm size, log of total assets
FIRMRSKit Firm risk, standard deviation of monthly stock returns for the fiscal year
MTBit Market to book value, market value per share divided by book value per share
FMAGEit Firm age, as mentioned in the annual reports
CEOCHNGit Interaction term of a dummy variable taking value 1 if CEO is replaced during
the year and number of CEOs drawing remuneration during the year
αi group fixed effects for firm or industry
ωi Time fixed effects
87
Two important issues that must be addressed in regression analysis are heteroskedasticity
and serial correlation. Both problems lead to biased standard errors which in turn lead to
inefficient regression results. To test the heteroskedasticity in the panel data, two well-known
tests are performed. These tests include 1) Breusch-Pagan / Cook-Weisberg
heteroskedasticity test and 2) White heteroskedasticity test. In Table 3.3, both tests indicate
the presence of heteroskedasticity however Breusch-Pagan test’s p-value is slightly above the
conventional level of significance of 5%, indicating that heteroskedasticity may not be a
problem.
To test the serial correlation in error terms (testing independence of error term),
Wooldridge test, proposed by Wooldridge (2002), for serial correlation in panel data is
performed. In Table 3.3, the test rejects the hypothesis of no serial correlation, thereby
suggesting presence of serial correlation in error terms.
Table 3.3
Heteroskedasticity and Serial Correlation Tests
Test Statistic p-value
Breusch-Pagan / Cook-Weisberg test for heteroskedasticity 3.77 0.0520
White's test for heteroskedasticity 318.57 0.0000
Wooldridge test for autocorrelation in panel data 85.64 0.0000
As for now, there is statistical evidence that the data is prone to heteroskedasticity and
serial correlation in error terms which may lead to biased results therefore, standard errors
need to be adjusted to account for potential difficulties due to heteroskedasticity and serial
correlation. Accordingly, to account for these difficulties, Huber-White robust standard errors
clustered at firm level are used. This combination for adjustment in standard errors account
for both heteroskedasticity and serial correlation (Greene, 2011).
One advantage of panel data is that it allows for controlling variations due to unobserved
group-specific heterogeneity or omitted variables. Fixed effect or random effect models can
be used for this purpose in panel data analysis (see, Greene, 2011, Wooldridge, 2002). Fixed
effect models assume that effects of unobserved heterogeneity or omitted variables are time-
invariant within cross sectional groups and are correlated with independent variables. In fixed
effect models, these time-invariant effects can be captured or removed by using dummy
variables for cross sectional groups or by differencing. Fixed effect models capture much of
the error variance resulting from distortions due to omitted variables or unobserved
88
heterogeneity between cross sectional groups. In contrast to fixed effect models, random
effect models assume that effects due to omitted variables or unobserved heterogeneity
between groups are not fixed or constant over time rather randomly distributed within groups
and between groups and these effects are uncorrelated with independent variables (see,
Greene, 2011, Wooldridge, 2002).
Table 3.4 Hausman Specification test for random effect model
b = consistent under Ho and Ha; obtained from fixed effect model
B = inconsistent under Ha, efficient under Ho; obtained from random effect model
Test: Ho: difference in coefficients not systematic
chi2(12) = (b-B)'[(V_b-V_B)^(-1)](b-B) = 514.14
Prob>chi2 = 0.0000
(V_b-V_B is not positive definite)
The choice of fixed or random effect model depends upon the assumption of correlation
between group-specific effects and independent variables. If assumption of no correlation
between group-specific effects and independent variables holds, the random effect model is
more efficient and would be preferred over fixed effect model (see, Greene, 2011,
Wooldridge, 2002). However, if this assumption does not hold, the random effect model
would be inconsistent and would lose its preference over fixed effect model. Statistically,
Hausman (1978) test is widely used to check the validity of random effect assumption of no
correlation between group-specific effects and independent variables. This test compares
parameters estimated using fixed effect and random effect models, and suggests the use of
random effect model if the difference between two parameters is sufficiently small otherwise
use of fixed effect model is suggested. Table 3.4 reports the results of Hausman test. The
results strongly support the use of fixed effect model as p-value of the test is zero, rejecting
the null hypothesis of no systematic difference between parameters. Accordingly, fixed effect
models are used in this study. Both industry and firm fixed effects are used. Time (year) fixed
effects are also included in the model to control for macroeconomic shocks that affect the
entire cross sectional units.
Control Variables
Besides firm performance and corporate governance variables, there are contextual
factors that play an important role in devising CEO compensation contracts. Literature review
suggests that amongst many, firm size, firm risk, firm growth opportunities and firm age are
89
the most widely used control variables in executive compensation research. Accordingly, this
study introduces these variables in regression models.
Large and complex firms are hard to manage therefore these firms attract executives with
better managing abilities which in turn lead to higher executive compensation. In theorizing,
Jensen and Meckling (1976) argue that larger firms hire better performing executive to
maximize their productivity. Firm size is perhaps one of the most cited determinants of CEO
compensation across the world. Countless studies report that firm size is positively related to
executive compensation (see, e.g., Conyon and Murphy, 2000, Croci et al., 2012, Devers et
al., 2007, Frydman and Jenter, 2010, Ghosh, 2006, Kaplan, 1994, Ye, 2014). In Pakistan,
Shah et al. (2009) and Kashif and Mustafa (2012) also find positive association between CEO
compensation and firm size. Two popular proxies for firm size are total assets and net
revenues. Following extent literature (e.g., Gallego and Larrain, 2012, Wang and Xiao,
2011), this study uses total assets as proxy for firm size. However, data on net revenues is
also collected for robustness checks.
Firm risk is also an important constituent of CEO compensation contract. Managing a
risky company needs better managerial skills, leading to higher compensations. Following
several studies (e.g., Brick et al., 2006, Conyon and He, 2011, Core et al., 1999, Lippert and
Moore, 1994, Lippert and Porter, 1997), this study controls for firm risk as proxied by
standard deviation of monthly stock returns for the fiscal year.
Firms with greater growth opportunity are expected to hire the executives with better
skills who can exploit the available growth opportunities to maximize the shareholder value.
This leads to positive link between growth opportunity and CEO compensation. Consistent
with existing literature (e.g., Brick et al., 2006, Conyon and He, 2011, Conyon and He, 2012,
Ho et al., 2004), this study controls for firms’ growth opportunities as proxied by Market to
Book ratio.
Consistent with existing literature (Conyon and He, 2012, Ho et al., 2004), firm age is
another control variables introduced in this study. Firm age is related to maturity and
experience of the firm. Intuitively, mature firms are more likely to devise more efficient
compensation contracts. Firm age is measured in years as mentioned in the annual reports.
If CEO is replaced in a firm, the new contract is unlikely to be the same as the previous
one. New CEO might start with lower or greater compensation. Further, if CEO is replaced in
the middle of a year, there would be 2 persons drawing the compensation in a year.
Therefore, to account for such instances, this study controls for interaction term of a dummy
90
variable taking the value 1 for new CEO and another variable indicating number of CEOs
drawing compensation in a year.
3.5.2 CEO Compensation and Past Firm Performance To test association of CEO compensation with past firm performance, current
performance variables are replaced by their lagged values (see, e.g., Conyon and He, 2012,
Laan et al., 2010, Murphy, 1999). Thus, model (1) becomes:
𝐿𝑁𝐶𝑂𝑀𝑃𝑖𝑡 = 𝛽1𝑅𝑂𝐴𝑖𝑡−1 + 𝛽2𝑇𝑅𝐸𝑇𝑖𝑡−1 + 𝛽3𝐵𝐷𝑆𝐼𝑍𝐸𝑖𝑡 + 𝛽4𝐵_𝐼𝑁𝐷𝑖𝑡 + 𝛽5𝐹𝐴𝑀𝑂𝑊𝑁𝑖𝑡
+ 𝛽6𝑂𝑊𝑁𝐶𝑂𝑁𝑆𝑖𝑡 + 𝛽7𝐷𝑈𝐴𝐿𝐶𝐸𝑂𝑖𝑡 + 𝛽8𝐹𝐼𝑅𝑀𝑆𝐼𝑍𝐸𝑖𝑡 + 𝛽9𝐹𝐼𝑅𝑀𝑅𝑆𝐾𝑖𝑡
+ 𝛽10𝑀𝑇𝐵𝑖𝑡 + 𝛽11𝐹𝑀𝐴𝐺𝐸𝑖𝑡 + 𝛽12𝐶𝐸𝑂𝐶𝐻𝑁𝐺𝑖𝑡 + 𝛼𝑖 + 𝜔𝑡
+ 𝜀𝑖𝑡 (2)
The definitions of the variables are same as in model (1). The model also is estimated
using industry and firm fixed effects as well as time fixed effects.
3.5.3 Controlling for Endogeneity
3.5.3.1 Regressing on Lagged Values of Explanatory Variables
Several studies in executive compensation and corporate governance literature indicate
the potential endogeneity problem (see, e.g., Cornett et al., 2007, Croci et al., 2012, Gillan,
2006, Hartzell and Starks, 2003, Mura, 2007, O'Connor and Rafferty, 2010). Endogeneity
may arise from different sources, for example, presence of unobserved heterogeneity,
simultaneity or measurement errors. Unobserved determinants of executive compensation
may also be the determinants of firm performance (Himmelberg et al., 1999), leading to
endogeneity. Similarly, performance and corporate governance variables may have two way
causal effects (simultaneity problem), again leading to endogeneity. If endogeneity is not
controlled, the results may be biased and inconsistent (Conyon and He, 2012, Mura, 2007,
O'Connor and Rafferty, 2010).
A number of studies in Pakistan (e.g., Awan, 2012, Ibrahim et al., 2010, Javed and Iqbal,
2006, Javid and Iqbal, 2008, Khatab et al., 2011, Nishat and Shaheen, 2006-2007, Yasser et
al., 2011) show links between corporate governance variables, firm performance and other
firm specific contextual variables. Any shock that is unobserved may affect these variables
and compensation simultaneously, leading to potential problem of endogeneity. Therefore,
following a number of studies (e.g., Cornett et al., 2007, Croci et al., 2012, Hartzell and
91
Starks, 2003, Ozkan, 2011) current values of all independent variables except FIRMAGE and
CEOCHNG in Model (1) are replaced with their lagged values treating them as potential
cause of endogeneity. Thus, model (1) becomes:
𝐿𝑁𝐶𝑂𝑀𝑃𝑖𝑡 = 𝛽1𝑅𝑂𝐴𝑖𝑡−1 + 𝛽2𝑇𝑅𝐸𝑇𝑖𝑡−1 + 𝛽3𝐵𝐷𝑆𝐼𝑍𝐸𝑖𝑡−1 + 𝛽4𝐵_𝐼𝑁𝐷𝑖𝑡−1
+ 𝛽5𝐹𝐴𝑀𝑂𝑊𝑁𝑖𝑡−1 + 𝛽6𝑂𝑊𝑁𝐶𝑂𝑁𝑆𝑖𝑡−1 + 𝛽7𝐷𝑈𝐴𝐿𝐶𝐸𝑂𝑖𝑡−1
+ 𝛽8𝐹𝐼𝑅𝑀𝑆𝐼𝑍𝐸𝑖𝑡−1 + 𝛽9𝐹𝐼𝑅𝑀𝑅𝑆𝐾𝑖𝑡−1 + 𝛽10𝑀𝑇𝐵𝑖𝑡−1 + 𝛽11𝐹𝑀𝐴𝐺𝐸𝑖𝑡
+ 𝛽12𝐶𝐸𝑂𝐶𝐻𝑁𝐺𝑖𝑡 + 𝛼𝑖 + 𝜔𝑡 + 𝜀𝑖𝑡 (3)
The definitions of the variables are same as in model (1). This model also is estimated
using industry and firm fixed effects as well as time fixed effects.
3.5.3.2 Dynamic Panel Estimation using Generalized Method of Moments (GMM)
To control for endogeneity, instrument variable (IV) approach is also adopted.
Specifically, a linear dynamic panel model of the following form is estimated using GMM
technique. Similar models are used in the extant literature (see, e.g., Conyon and He, 2012,
Mura, 2007, O'Connor and Rafferty, 2010, Yoshikawa et al., 2010).
The model is:
𝐿𝑁𝐶𝑂𝑀𝑃𝑖𝑡 = 𝛼 + 𝛾𝐿𝑁𝐶𝑂𝑀𝑃𝑖𝑡−1 + 𝛽1𝑅𝑂𝐴𝑖𝑡 + 𝛽2𝑇𝑅𝐸𝑇𝑖𝑡 + 𝛽3𝐵𝐷𝑆𝐼𝑍𝐸𝑖𝑡 + 𝛽4𝐵_𝐼𝑁𝐷𝑖𝑡
+ 𝛽5𝐹𝐴𝑀𝑂𝑊𝑁𝑖𝑡 + 𝛽6𝑂𝑊𝑁𝐶𝑂𝑁𝑆𝑖𝑡 + 𝛽7𝐷𝑈𝐴𝐿𝐶𝐸𝑂𝑖𝑡 + 𝛽8𝐹𝐼𝑅𝑀𝑆𝐼𝑍𝐸𝑖𝑡
+ 𝛽9𝐹𝐼𝑅𝑀𝑅𝑆𝐾𝑖𝑡 + 𝛽10𝑀𝑇𝐵𝑖𝑡 + 𝛽11𝐹𝑀𝐴𝐺𝐸𝑖𝑡 + 𝛽12𝐶𝐸𝑂𝐶𝐻𝑁𝐺𝑖𝑡 + 𝜔𝑡
+ 𝜀𝑖𝑡 (4)
The definitions of the variables are same as in model (1).
The dynamic panel models estimated using GMM technique simultaneously control for
unobserved heterogeneity, serial correlation and endogeneity. The approach relies on internal
instrument variables that are based on lagged values of dependent and explanatory variables.
Three approaches have evolved for estimation of the dynamic panel model using GMM.
Arellano and Bond (1991) suggest estimating equation in differences (GMM_DIFF) to
control for unobserved heterogeneity and using the lagged levels of the variables as
instruments. Arellano and Bover (1995) suggest using first differences as instrument
variables for equation in levels that leaves the unobserved heterogeneity in errors (GMM-
Orthogonal). They also propose a new transformation of variables, forward orthogonal
deviations, that preserves orthogonality in error terms, leading to a computationally
convenient alternative for models that include predetermined variables. Blundell and Bond
92
(1998) propose a technique that uses both level equation with differenced values as
instrument variables and differenced equation with lagged values as instrument variables,
making a system GMM estimator (GMM-SYS). These techniques are basically developed to
account for 1) small number of periods i.e. small T and large number of cross section units
i.e. large N, 2) relaxing the assumption of strict exogeneity of independent variables i.e.
independent variables are allowed to correlate with current and past realization of the error
term, 3) fixed group effects and 4) heteroskedasticity and autocorrelation within groups but
not across the groups (Roodman, 2009).
This study reports the results using three GMM estimation techniques. Initially, the
dynamic panel model is estimated using GMM-DIFF following Arellano and Bond (1991)
then variables are transformed as proposed by Arellano and Bover (1995) and model is
estimated, and in the last, GMM-SYS estimation technique is used12. All models also include
time fixed effects to control macroeconomics shocks.
The choice of instrument variables is crucial in GMM estimation techniques. Two
conditions must be met for a valid instrument; 1) instrument variable must be uncorrelated to
error terms (instrument exogeneity) and 2) instrument variable must be correlated with
endogenous variables (instrument relevance). The validity of the instruments can be tested
using Sargan (1958) and Hansen (1982) tests of over-identification restrictions under the null
hypothesis of joint validity of instruments13. The Sargan/Hansen tests are standard for joint
validity of the instruments in GMM estimation. In addition, Arellano and Bond (1991)
develop a test for autocorrelation in error terms that may invalidate some lags as instruments.
For example, if there is an autocorrelation of the order 2 in error terms then second lag cannot
be used as instrument, thereby suggests using third or earlier lags as instruments. This study
uses Sargan/Hansen over-identification tests and Arellano and Bond autocorrelation test for
validity and suitability of the instrument variables.
12 GMM models are estimated using xtabond2 in STATA developed by David Roodman, see Roodman (2009) for details 13 See Roodman (2009) for details
93
3.5.4 CEO Compensation and Future Firm Performance
When unexpectedly high CEO compensation is due to the information that CEO effort
was greater than implied by current observable firm performance measures alone then one
should find future firm performance to be higher than implied by current firm performance
alone (Balafas and Florackis, 2014, Hayes and Schaefer, 2000). This can reasonably be tested
using three-step procedure as proposed by Hayes and Schaefer (2000):
1. regress CEO compensation on current performance and take residuals
2. regress future firm performance on current firm performance and take the residuals
3. regress residuals of (2) on residuals of (1)
This is equivalent to regressing future firm performance on current CEO compensation
and current firm performance. This way the effects of current firm performance would be net
out. Thus the model would be:
𝑅𝑂𝐴𝑖𝑡+1 = 𝛼 + 𝛽1𝑅𝑂𝐴𝑖𝑡 + 𝛽2𝑇𝑅𝐸𝑇𝑖𝑡 + 𝛽3𝐿𝑁𝐶𝑂𝑀𝑃𝑖𝑡 + 𝛼𝑖 + 𝜔𝑡 + 𝜀𝑖𝑡 (5)
Where
ROAit+1 Measure of future performance, return on assets, earnings before interest and
taxes divided by total assets
ROAit Measure of current performance, return on assets, earnings before interest and
taxes divided by total assets
TRETit Measure of current performance, total return to shareholder, sum of current
year market value and dividend divided by previous year market price
LNCOMPit Measure of current CEO compensation, log of cash compensation and total
compensation, cash compensation includes managerial remuneration and
bonuses while total compensation includes cash compensation, perks and
retirement benefits
Note here that in step 1 above, objective is to measure the degree of information
contained in CEO compensation that is not explained by current firm performance. Therefore,
two measures of current performance, ROA and total shareholder return, are incorporated in
the model. As measure of future performance, ROA is used as it is more convenient to be
used as a measure of future observable performance (see, Adithipyangkul et al., 2011,
Balafas and Florackis, 2014, Hayes and Schaefer, 2000). Group fixed effects and time fixed
94
effects are also incorporated to control for unobserved heterogeneity and macroeconomic
shocks.
In addition, corporate governance and firm specific variables also correlate with CEO
compensation and firm performance therefore to provide as much explanatory power as
possible, most widely used corporate governance and firm specific variables are incorporated.
Thus, following model is estimated.
𝑅𝑂𝐴𝑖𝑡+1 = 𝛼 + 𝛽1𝑅𝑂𝐴𝑖𝑡 + 𝛽2𝑇𝑅𝐸𝑇𝑖𝑡 + 𝛽3𝐿𝑁𝐶𝑂𝑀𝑃𝑖𝑡 + 𝛽4𝐵𝐷𝑆𝐼𝑍𝐸𝑖𝑡 + 𝛽5𝐵_𝐼𝑁𝐷𝑖𝑡
+ 𝛽6𝐹𝐴𝑀𝑂𝑊𝑁𝑖𝑡 + 𝛽7𝑂𝑊𝑁𝐶𝑂𝑁𝑆𝑖𝑡 + 𝛽8𝐷𝑈𝐴𝐿𝐶𝐸𝑂𝑖𝑡 + 𝛽9𝐿𝑁𝑆𝐴𝐿𝐸𝑆𝑖𝑡
+ 𝛽10𝑀𝑇𝐵𝑖𝑡 + 𝛽11𝐹𝐼𝑅𝑀𝑅𝑆𝐾𝑖𝑡 + 𝛼𝑖 + 𝜔𝑡
+ 𝜀𝑖𝑡 (6)
Where
ROAit+1 Measure of future performance, return on assets, earnings before interest and
taxes divided by total assets
ROAit Measure of current performance, return on assets, earnings before interest and
taxes divided by total assets
TRETit Measure of current performance, total return to shareholder, sum of current
year market value and dividend divided by previous year market price
LNCOMPit Log of cash compensation and total compensation, cash compensation
includes managerial remuneration and bonuses while total compensation
includes cash compensation, perks and retirement benefits
BDSIZEit Board size, number of directors on the board
B_INDit Board independence, ratio of non-executive directors to board size
FAMOWNit Family ownership, a dummy variable taking value 1 for family firms and zero
otherwise
OWNCONSit Ownership concentration, percentage of shares held by largest shareholder
DUALCEOit CEO duality, a dummy variable taking value 1 if CEO is also chairman board
of directors and zero otherwise
LNSALESit Log of net sales
MTBit Market to book value, market value per share divided by book value per share
FIRMRSKit Firm risk, standard deviation of monthly stock returns for the fiscal year
αi group fixed effects for firm or industry
ωi Time fixed effects
95
To control for unobserved heterogeneity and macroeconomic shocks group fixed effects and
time fixed effects are incorporated. Standard errors are adjusted to account for
heteroskedasticity and autocorrelation in error terms.
3.5.5 CEO Compensation and Performance Manipulation
3.5.5.1 Detecting Earnings Management
In earnings management (EM) research, the detection of EM is crucial. The use of
discretionary accruals as proxy for EM behavior has become a standard approach in EM
research. To detect discretionary accruals, this study uses Jones (1991) model which is
modified first by Dechow et al. (1995) and subsequently by Kothari et al. (2005). However,
for robust checks, results are also obtained using original Jones (1991) model and Dechow et
al. (1995) modification. The general model for detecting discretionary accruals is:
𝐷𝐴𝐶 = 𝑇𝐴𝐶 − 𝑁𝐷𝐴𝐶
Where
DAC Discretionary accruals
TAC Total accruals
NDAC Non-discretionary accruals
Jones (1991) proposes following model to detect discretionary accruals.
𝑇𝐴𝐶𝑖𝑡
𝐴𝑖𝑡−1= 𝛼 + 𝛽1 (
1
𝐴𝑖𝑡−1) + 𝛽2 (
∆𝑅𝐸𝑉𝑖𝑡
𝐴𝑖𝑡−1) + 𝛽3 (
𝑃𝑃𝐸𝑖𝑡
𝐴𝑖𝑡−1) + 𝜀𝑖𝑡
Where
TACit Total accruals, calculated as difference between income before extraordinary
items minus operating cash flow
ΔREVit Change in firm revenues
PPEit Property, plant and equipment for firm i at time t
Ait-1 Lagged total assets
Jones (1991) defines fitted values as non-discretionary accruals. Thus, residuals from this
model represent discretionary accruals.
Dechow et al. (1995) modify Jones (1991) model arguing that some adjustment needs to
be made for discretionary revenue management and deduct change in accounts receivable
from change in revenues as adjustment in the model. Thus, the model becomes:
96
𝑇𝐴𝐶𝑖𝑡
𝐴𝑖𝑡−1= 𝛼 + 𝛽1 (
1
𝐴𝑖𝑡−1) + 𝛽2 (
∆𝑅𝐸𝑉𝑖𝑡 − ∆𝐴𝑅𝑖𝑡
𝐴𝑖𝑡−1) + 𝛽3 (
𝑃𝑃𝐸𝑖𝑡
𝐴𝑖𝑡−1) + 𝜀𝑖𝑡
Where
ΔAR Change in accounts receivable
All other variables are same as above.
Kothari et al. (2005) show that in many research settings, performance-matched
discretionary accrual measures increase the reliability of inferences and suggest adding
performance measure in the model. Thus, modified model becomes:
𝑇𝐴𝐶𝑖𝑡
𝐴𝑖𝑡−1= 𝛼 + 𝛽1 (
1
𝐴𝑖𝑡−1) + 𝛽2 (
∆𝑅𝐸𝑉𝑖𝑡 − ∆𝐴𝑅𝑖𝑡
𝐴𝑖𝑡−1) + 𝛽3 (
𝑃𝑃𝐸𝑖𝑡
𝐴𝑖𝑡−1) + 𝛽4(𝑅𝑂𝐴𝑖𝑡−1) + 𝜀𝑖𝑡
Where
ROAit-1 Lagged performance measure, return on assets
All other variables are same as in above models.
This model is estimated cross-sectionally each year for each industry group as per SBP
classification and residuals of this model represent discretionary accruals – a proxy for
earnings management behavior. Since earnings can be managed in both sides therefore
following earnings management literature absolute value of discretionary accruals (|DAC|) is
used as proxy for earnings management.
3.5.5.2 Regression Model
After estimating discretionary accruals, in line with existing literature (e.g., Ibrahim and
Lloyd, 2011, Ye, 2014), the following model is run to examine the impact of CEO
compensation on Earnings Management (EM) behavior.
|𝐷𝐴𝐶|𝑖𝑡 = 𝛼 + 𝛽1𝐿𝑁𝐶𝑂𝑀𝑃𝑖𝑡 + 𝛽2𝐵𝐷𝑆𝐼𝑍𝐸𝑖𝑡 + 𝛽3𝐵_𝐼𝑁𝐷𝑖𝑡 + 𝛽4𝐹𝐴𝑀𝑂𝑊𝑁𝑖𝑡
+ 𝛽5𝑂𝑊𝑁𝐶𝑂𝑁𝑆𝑖𝑡 + 𝛽6𝐷𝑈𝐴𝐿𝐶𝐸𝑂𝑖𝑡 + 𝛽7𝐹𝐼𝑅𝑀𝑆𝐼𝑍𝐸𝑖𝑡 + 𝛽8𝐿𝐸𝑉𝐸𝑅𝐴𝐺𝐸𝑖𝑡
+ 𝛽9𝐶𝐹𝑂𝑖𝑡 + 𝛽10𝑅𝑂𝐴𝑖𝑡 + 𝛽11𝑀𝑇𝐵𝑖𝑡 + 𝛽12𝐿𝑂𝑆𝑆𝐷𝑈𝑀𝑖𝑡 + 𝛼𝑖 + 𝜔𝑡
+ 𝜀𝑖𝑡 (7)
Where
|DAC|it Absolute discretionary accruals, a proxy of Earnings Management
97
LNCOMPit Log of cash compensation and total compensation, cash compensation
includes managerial remuneration and bonuses while total compensation
includes cash compensation, perks and retirement benefits
BDSIZEit Board size, number of directors on the board
B_INDit Board independence, ratio of non-executive directors to board size
FAMOWNit Family ownership, a dummy variable taking value 1 for family firms and zero
otherwise
OWNCONSit Ownership concentration, percentage of shares held by largest shareholder
DUALCEOit CEO duality, a dummy variable taking value 1 if CEO is also chairman board
of directors and zero otherwise
FIRMSIZEit Firm size, log of total assets
LEVERAGEit Total debt divided by total assets
CFOit Cash flow from operations
ROAit Return on assets, net income divided by total assets
MTBit Market to book value, market value per share divided by book value per share
LOSSDUMit A dummy variable taking value one if the firm is in net loss, and zero
otherwise
αi group fixed effects for firm or industry
ωi Time fixed effects
Corporate governance variables are included to examine the effects of corporate
governance practices in Pakistan. Weak corporate practices may give room to managers to
manipulate earnings which in turn may depict a camouflaged picture of the firm to the
shareholders.
Other control variables are typically associated with incentives to manage earnings (see
Frankel et al., 2002, Ibrahim and Lloyd, 2011). There is tendency that larger firm would have
larger discretionary accruals. Further, larger firms tend to have more analysts following.
Debts are accompanied with debt covenants which cannot be avoided therefore firms with
higher leverage have more incentives to manage earnings. The firms with better performance
(CFO and ROA) tend to have more accruals and will be more likely to meet or just beat the
analyst expectations, suggesting performance as an important factor that influences earning
management. Similarly, firms having more growth opportunities have greater incentives for
earnings management. Another important factor that can influence the earnings management
is loss because firms with loss have different incentives to manage earnings.
98
To control for unobserved heterogeneity and macroeconomic shocks group fixed effects
and time fixed effects are incorporated. Standard errors are adjusted to account for
heteroskedasticity and autocorrelation in error terms.
99
Chapter 4
CEO Compensation, Firm Performance and Corporate
Governance: Empirical Results and Discussion
100
4.1 Descriptive Statistics
Table 4.1 presents the mean, median and standard deviation of the variables used.
4.1.1 CEO Compensation
Cash compensation includes managerial remuneration (base salary) and bonus while
total compensation includes cash compensation, perquisites and retirement benefits. Over the
sample period, both average total and cash compensations have nearly tripled from Rs4.817
million to Rs13.060 million and Rs3.353 million to Rs9.378 million respectively.
Consistently lower median value than mean value indicates that the distribution of
compensation is positively skewed. Greater number of CEOs is receiving pay that is less than
average compensation. Increased standard deviation over time shows more disperse
distribution of compensation in later years. This may be an indication of firms using their
own criteria to set CEO pay rather than following peer groups or overall market. Figure 4.1
shows overall trend in both cash and total compensation. The increased difference between
cash and total compensation over time indicates that ratio of non-cash to total compensation
has increased. Also, the slop of both the compensations is slightly greater after year 2010.
This might be an indication of revised pay packages after financial crisis and political crunch.
4.1.2 Firm Performance
Return on assets (ROA) is calculated as earnings before interest and taxes divided by
total assets. Mean ROA of Pooled sample is 10.99 percent with standard deviation of 13.39
percent while median is 9.51%. Lower median value than mean indicates that the distribution
of ROA is positively skewed. Average ROA does not seem to vary abnormally across the
years. ROA decreases from 12.52% in year 2006 to 10.17% in year 2007 followed by 9.78%
in year 2008 and 9.26% in year 2009. This downward trend seems to reflect the effects of
financial crisis of 2008. After 2009 recovery takes place and ROA again rises to over 12%
mark followed by a steady decrease again to 10.54% till year 2012.
Total return to shareholders (TRET) is calculated as current year market price plus
dividend divided by previous year market price. Average TRET of Pooled sample stays at
16.74% with standard deviation of 80.04% while median is -0.84%. There exists more
fluctuation in TRET than ROA. In year 2005 average TRET is 27.70% which suddenly
decrease to 3.49% in year 2006 followed by an increase to 24.87% in year 2007. Negative
TRETs (both mean and median) in year 2008 and 2009 are potential indication of financial
crisis. Recovery seems to start after that with positive TRETs ending with mean TRET at
101
86.46% level in year 2012 with standard deviation of 114.89% and median value of 47.84%.
we can see overall downward trend in firm accounting and market performance in 2007 to
2009 reflect effects of unrest due to political issues and general elections, start of energy
crisis in Pakistan and global financial crisis.
4.1.3 Corporate Governance
Board size (BDSIZE) is measured as number of directors on board. Average board size
of pooled sample is slightly above 8 with standard deviation of 1.57. The median board size
in the pooled sample is 7 which is less than mean value, meaning that in more than 50%
observations board size is less than or equal to 8. Recently, similar average board size slightly
above 8 is reported in India however median value is reported exactly as 8 (see, Jameson et
al., 2014). This might be due to resembling institutional setting of India and Pakistan with
concentrated and family ownership structure. However, this average board size is slightly
lower than the board size in US (mean 9.54 and median 9) where firms are known to be
diffusely held and China (mean 9.372 and median 9) where state ownership is common (see,
Conyon, 2014, Huang and Wang, 2015). The average board size seems to have decreased
over time as mean of board size in year 2005 is 8.13 followed by 8.15, 8.08, 8.04 and 8.01 in
year 2006, 2007, 2008 and 2009 respectively and then stays at 8 afterwards. This is perhaps
in line with the developments in corporate governance system in Pakistan emphasizing on
effective boards.
Board independence (B_IND) is measured as ratio of non-executive directors to board
size. Pooled average of B_IND indicates that on average firms keep 63% non-executive
directors on the boards. The median value of pooled sample i.e. 66.67% is greater than mean
value, indicating that more 50% values lie above 63.91% (the mean value). Thus distribution
of B_IND is negatively skewed. Both mean and median of B_IND have moderately
downward trend over time. This might be due to reduction in board size over time as non-
executive directors are more likely to be dropped when board size needs to be reduced. The
ratio of non-executive directors in the board in Pakistan is quite similar to the ratio in India as
reported by Ghosh (2006) but greater than in some European countries and China (Fernandes,
2008, Firth et al., 2007, Ozkan, 2007).
Family ownership (FAMOWN) is a dummy variable taking value 1 if a firm qualifies as
family firm under conditions mentioned in research strategy section and zero otherwise.
Mean of 0.75 shows that 75% observations are coming from family firms. This ratio seems to
be greater than in Europe as Croci et al. (2012) report this to be about 42% in their dataset
102
from continental Europe. Similarly, two studies in Asia, from Korea and Japan, report about
40% and 20% respectively in their datasets (Basu et al., 2007, Kato et al., 2007). The family
firm ratio is stable overtime as family firm in year 2005 is very likely to remain a family firm
in year 2012. Therefore, within firms, no variation is expected over the sample period for this
variable. Thus, to see the impact of family ownership on CEO compensation, variations
between firms should be considered.
Ownership concentration (OWNCONS) represents the number of voting shares held by
the largest shareholder. The average voting shares held by the largest shareholder is 33.67%,
indicating a high concentrated ownership environment – consistent with other studies in
Pakistan (e.g., Awan, 2012, Ibrahim et al., 2010, Javed and Iqbal, 2006, Javid and Iqbal,
2008, Khatab et al., 2011, Nishat and Shaheen, 2006-2007, Yasser et al., 2011). The median
value of 26.66% is less than the mean value, suggesting that less than 50% firms have
OWNCONS more than 33.67% (positively skewed distribution). There is slightly increase in
OWNCONS in later years of 2011 and 2012 with values 34.39% and 35.51% respectively
otherwise OWNCONS is stable around 33% in early years of sample period.
CEO duality (DUALCEO) is a dummy for CEO holding the position of chairman as
well. On average about 34% CEOs also hold the position of chairman board of directors. The
trend of DUALCEO is increasing initially till year 2010 but after that may be due to more
emphasis on strong corporate governance practices, duality has downward trend. This is
similar to the UK and China where emphasis in on separating the post of CEO from the
chairman, and unlike the US where it is usual to combine these two positions (Conyon and
He, 2012).
103
Table 4.1 Descriptive Statistics
Variables Statistic 2005 2006 2007 2008 2009 2010 2011 2012 Pooled Sample
Total Mean 4817 5530 6512 7562 8396 9449 10628 13060 8475 Compensation Median 3156 3384 4153 4553 4999 5459 6060 7519 4800 In Rs.’000 S.D 5854 7030 9094 11028 11278 12519 13493 19935 12538
Cash Mean 3353 3918 4751 5529 6004 6577 7523 9378 6042 Compensation Median 2000 2303 2715 2836 3435 3450 4001 5186 3174 In Rs.‘000 S.D 4626 5840 7786 9268 8867 9690 10460 14766 9733
ROA Mean 11.93 12.52 10.17 9.78 9.26 12.28 11.74 10.54 10.99 %age Median 9.14 9.58 7.86 7.96 9.53 11.20 10.84 10.32 9.51 S.D 11.07 10.67 11.23 11.52 19.31 11.73 12.96 14.80 13.39
TRET Mean 27.70 3.49 24.87 -14.35 -1.73 8.37 -0.47 86.46 16.74 %age Median 18.97 -1.01 7.38 -27.20 -18.04 -6.14 -9.99 47.84 -0.84 S.D 52.18 39.08 71.94 80.74 67.19 62.89 72.12 114.89 80.04
BDSIZE Mean 8.13 8.15 8.08 8.04 8.01 8.00 8.01 8.00 8.05 Median 7.00 7.00 7.00 7.00 7.00 7.00 7.00 7.00 7.00 S.D 1.68 1.68 1.62 1.57 1.53 1.51 1.51 1.51 1.57
B_IND Mean 64.32 64.31 63.90 63.65 62.85 63.23 62.58 63.91 63.54 %age Median 70.00 70.71 70.00 66.67 66.67 66.67 62.50 66.67 66.67 S.D 20.37 20.72 20.71 20.59 20.94 20.64 20.31 19.96 20.49
FAMOWN Mean 0.75 0.75 0.76 0.75 0.74 0.75 0.75 0.74 0.75 Median 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 S.D 0.44 0.43 0.43 0.43 0.44 0.43 0.43 0.44 0.43
OWNCONS Mean 32.72 33.01 33.07 33.23 33.36 33.61 34.39 35.51 33.67 %age Median 26.46 26.61 26.71 26.25 26.13 26.13 27.45 29.40 26.66 S.D 21.01 21.22 20.78 20.48 20.44 20.30 20.74 21.32 20.74
DUALCEO Mean 0.30 0.33 0.33 0.36 0.36 0.36 0.35 0.33 0.34 Median 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 S.D 0.46 0.47 0.47 0.48 0.48 0.48 0.48 0.47 0.47
TASSETS Mean 7173 8607 9827 10898 11636 12675 14490 17215 11833 Rs in Median 2385 2603 2970 3042 3034 3068 3544 3831 3061 Millions S.D 14281 16821 18677 22158 26033 31098 35893 45860 29092
SALES Mean 8080 10277 11273 12872 13944 15895 18639 21949 14493 Rs in Median 1856 2407 2486 2469 2840 3355 4239 4452 3007 Millions S.D 217689 28719 32536 41939 48335 57547 64179 80479 52015
Net Income Mean 716 771 726 605 576 756 966 1161 790 Rs in Median 88 79 64 53 53 108 138 132 84 Million S.D 2858 4046 4168 5061 4714 4968 5773 8022 5240
FIRMRSK Mean 15.03 12.15 12.70 13.24 24.29 17.88 14.07 15.97 15.88 %age Median 13.54 11.20 12.13 11.66 19.32 13.18 11.15 13.29 13.01 S.D 8.00 8.51 6.63 7.74 23.59 15.63 11.00 11.37 13.59
MTB Mean 1.79 2.35 2.17 1.74 1.56 1.31 1.30 1.84 1.73 Times Median 1.19 0.99 1.13 0.86 0.62 0.52 0.43 0.65 0.75 S.D 2.37 8.17 4.76 4.77 4.81 6.03 4.58 5.62 5.35
FMAGE Mean 31.03 31.84 32.56 32.87 34.01 35.06 36.14 37.50 34.06 Years Median 27.00 28.00 29.00 29.00 30.00 31.00 32.00 33.00 30.00 S.D 16.65 16.75 16.48 16.43 16.61 16.56 16.46 16.59 16.65
104
.
4.1.4 Control Variables
Total assets and net sales are taken as proxies for firm size. Average total assets
(TASSETS) and average net sales (SALES) have increased over time. Larger mean than
median suggests that mean value is driven by large firms. In the pooled sample, average
TASSETS are Rs11,833 million with standard deviation of Rs29,092 million but median is
only Rs3,061 million-nearly four times lower than the mean. This is possibly due to the
reason that family firms in Pakistan are in larger number but they are smaller in term of total
assets than non-family firms like state owned firms. Similarly, average of SALES is also
driven by large firms. The median of SALES in pooled sample is Rs3,007 million while
mean is about five times larger than the median. High standard deviation in TASSETS and
SALES suggests much dispersion in firms in term of size.
Operating cash flows and net income have mean values of Rs926 million and Rs790
million with standard deviations of Rs4,617 million and Rs5,240 million respectively. Note
that there is big difference in mean and median for these two variables. This indicates that
smaller number of firms is making huge profits and distributions of these variables are
positively skewed. Given that textile sector is the biggest sector in Pakistan in term of number
of firms however profits are greater for other sectors like electronic and petroleum sectors
which have smaller number of firms. This possibly leads to positively skewed distribution.
0
2
4
6
8
10
12
14
2005 2006 2007 2008 2009 2010 2011 2012
Rs in Million
Years
Average CEO Compensation Over the Sample Period
Cash Compensation
Total Compensation
Figure 4.1
105
Firm risk (FIRMRSK) is proxied by standard deviation of monthly stock returns in the
fiscal year. Average FIRMRSK is highest (24.29%) in year 2009 while it is lowest (12.15%)
in year 2006 however pooled average is 15.88% with standard deviation of 13.59% while
median stays 13.01% level. The highest FIRMRSK depicts the effects of financial crisis with
higher variations in the stock prices when uncertainty is high in late 2008. It also represents
political situation in Pakistan after the resignation of President of Pakistan during the same
period.
Market to book value (MTB) is used as proxy for firm growth opportunities. Lowest
MTB prevails in year 2010 and 2011 while highest is in year 2006. Being highest in year
2006 MTB gradually decreases to 1.30 in year 2011 then it rises to 1.84 in 2012. This is
perhaps due to political situation in Pakistan. As the economy was booming till 2007 but
recession hits due to energy crisis in 2008.
Average firm age is about 34 years with standard deviation of 16.65 years while median
is 30 years. This shows concentration of mature firms in Pakistan.
4.2 CEO Compensation and Realized Firm Performance
4.2.1 Correlation Analysis
Table 4.2 presents the correlation matrix among compensation, performance and
corporate governance variables. Other control variables are also included in the matrix. Log
transformation is performed for compensation variables and firm size proxies i.e. total assets
and sales. Many significant correlations exist in Table 4.2, for example, high correlation
(0.9722) between cash compensation and total compensation. As evident from Table 3.2, in
Pakistan, not all firms have cash bonuses and retirement benefits plans for CEOs however,
perks are common in the form of house rent, club fee, utilities and others benefits. The perks
are generally linked to the base salaries as some percentage of base salary. Therefore, a
strong correlation is quite expected between cash compensation and total compensation.
Both cash and total compensation have positive correlation with accounting performance
(ROA) with values 0.3088 and 0.3098 respectively but correlation with market performance
(TRET) is not significant although sign is positive. Board structure variables BDSIZE and
B_IND are also positively correlated with compensation, indicating the potential
ineffectiveness of boards in monitoring and reducing CEO entrenchment and hence CEO
compensation.
106
Family firms seem to pay lower compensation to CEO as depicted by negative
correlation between FAMOWN and both cash and total compensation with values -0.3775
and -0.3760 respectively. Ownership concentration is positively correlated with
compensation which is inconsistent with the argument of agency theory that concentrated
ownership provides better incentives to monitor CEOs and mitigates agency problems and
manager’s self-serving behavior. Ownership concentration is also positively and significantly
correlated with BDSIZE and B_IND, indicating that, in Pakistan, concentrated ownership
does not seem to contribute towards board effectiveness.
Surprisingly, CEO duality is negatively correlated with CEO compensation which is
inconsistent with managerial power hypothesis. Apparently, CEO duality looks to reduce the
board independence as correlation between DUALCEO and B_IND is -0.1863, thereby
reducing the board effectiveness. But on the other hand CEO duality has positive effects on
board effectiveness as it has negative correlation with board size with value on -0.2605.
LNASSETS, LNSALES, MTB and FIRMAGE are positively correlated with compensation
as expected. However, CEO compensation reduces as firm risk increases which is
inconsistent with the argument that risky firms need to pay higher compensation to their
CEOs.
4.2.1.1 Addressing Multicollinearity
In regression models, the problem of multicollinearity may lead to higher standard errors
of the coefficients and makes the inference difficult and biased. Therefore, to trace down the
multicollinearity in Model 1 & 2, two widely used techniques 1) correlation matrix and, 2)
variance inflation factor (VIF) and tolerance are used. Table 4.2 shows that none of the
absolute values of correlation coefficients is greater than normally used threshold level of
0.70 for potential serious problem of multicollinearity. To further confirm that no
multicollinearity problem exists, variance inflation factor (VIF) and tolerance are estimated.
Table 4.3 indicates that none of the VIF is greater than generally used threshold level of 10
and mean of VIFs is 1.31 which is not considerably different from one. Similarly, all the
tolerance levels are above the threshold level of 0.10. These diagnostics suggest that problem
of multicollinearity is unlikely to make the regression results unreliable and variables can be
included in one regression model.
107
Table 4.2 Correlation Matrix (Variables Used in Model 1,2,3 & 4)
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15)
LNTCOMP (1) 1.0000
LNCASHCOMP (2) 0.9722* 1.0000
ROA (3) 0.3088* 0.3098* 1.0000
TRET (4) 0.0148 0.0225 0.1734* 1.0000
BDSIZE (5) 0.3215* 0.3138* 0.1333* -0.0038 1.0000
B_IND (6) 0.1872* 0.1438* 0.0326 -0.0242 0.3538* 1.0000
FAMOWN (7) -0.3775* -0.3760* -0.2202* 0.0188 -0.4532* -0.2273* 1.0000
OWNCONS (8) 0.3555* 0.3651* 0.2218* 0.0497 0.1007* 0.0638* -0.5435* 1.0000
DUALCEO (9) -0.2944* -0.2777* -0.1671* 0.0108 -0.2605* -0.1863* 0.1730* -0.0841* 1.0000
LNTASSETS (10) 0.5781* 0.5614* 0.1706* -0.0443 0.3795* 0.1320* -0.2750* 0.2495* -0.2164* 1.0000 LNSALES (11) 0.6102* 0.5912* 0.3240* -0.0194 0.3590* 0.1285* -0.3206* 0.2610* -0.2380* 0.8818* 1.0000
FIRMRSK (12) -0.2040* -0.2060* -0.1895* 0.1968* -0.1272* -0.0691* 0.1155* -0.0418 0.1978* -0.2403* -0.2789* 1.0000 MTB (13) 0.0932* 0.1197* 0.1750* 0.0691* 0.0803* -0.1189* -0.1730* 0.2793* -0.0949* 0.0802* 0.1210* -0.0913* 1.0000
FMAGE (14) 0.2469* 0.2380* 0.0670* 0.0354 0.1897* 0.0538* -0.2164* 0.2026* -0.0167 0.1158* 0.1379* -0.0334 0.0389 1.0000 CEOCHNG (15) 0.1359* 0.1178* 0.0290 -0.0013 0.2176* 0.1406* -0.2644* 0.1834* -0.1039* 0.2192* 0.1939* -0.0224 0.0222 0.0621* 1.0000
(*) p-value < 0.05
108
Table 4.3 Variance Inflation Factors (Variables Used in Model 1,2,3 & 4)
Variable VIF Tolerance
FAMOWN 1.91 0.52433
OWNCONS 1.67 0.59708
BDSIZE 1.66 0.60337
FIRMSIZE 1.33 0.75329
B_IND 1.21 0.82805
ROA 1.18 0.84446
FIRMRSK 1.18 0.8483
MTB 1.15 0.8683
DUALCEO 1.15 0.86999
CEOCHNG 1.12 0.89144
TRET 1.1 0.90679
FMAGE 1.08 0.92238
Mean VIF 1.31
4.2.2 Regression Results
4.2.2.1 Total Compensation
Table 4.4 describes the empirical results of models 1 and 2 developed in chapter 3 for
total compensation as dependent variable. Columns 1,3 and 5 presents results of Model 1 that
includes current firm performance while columns 2,4 and 6 provide insight of the relationship
between total compensation and lagged firm performance i.e. Model 2. Column 1 and 2
exhibit results of pooled regression. Unobserved heterogeneity is controlled using fixed
effects as described in chapter 3. Columns 3 and 4 report results with industry and time fixed
effects while columns 5 and 6 present results with firm and time fixed effects. Note that R-
Squared has increased when fixed effects are used. F-tests (unreported) indicate that fixed
effects capture significant variation in CEO total compensation thus, endorse the use of fixed
effects.
In Table 4.4, consistent with hypothesis 1, CEO total compensation is positively related
to current firm accounting performance as measured by ROA. The results are consistent
qualitatively over pooled regression (column 1), industry and time fixed effects model
(column 3), and firm and time fixed effects model (column 5). However, coefficient of ROA
in pooled model is the highest with value of 1. 3484 followed by 1.3156 in industry and time
fixed effect model and 0.2460 in firm and time fixed model. The positive association is
consistent with several studies in Asia (e.g., Conyon and He, 2011, Conyon and He, 2012,
Ghosh, 2006, Kato and Kubo, 2006, Mitsudome et al., 2008). However, it is in contrast to the
109
findings of Shah et al. (2009) and Kashif and Mustafa (2012) study in Pakistan which find no
relationship between CEO compensation and ROA. This might be due to difference in data
period and sample used or methodological difficulties in these studies. In addition Shah et al.
(2009) and Kashif and Mustafa (2012) use only pooled regression models which might not
have captured the true picture.
Current total returns to shareholder do not significantly contribute to pay setting process
as none of the models in columns 1, 3 and 5 reports coefficient for TRET that is different
from zero significantly. Since CEO compensation in Pakistan rarely includes any restricted
stocks, stock options and other stock based bonuses, therefore weak link between CEO total
compensation and market performance is expected. Complement to that, like many other
emerging markets, Pakistani bourses are considered to be highly volatile (Sheikh and Riaz,
2012), therefore using market performance as benchmark for setting CEO compensation may
not be a good choice for board of directors, leading to an insignificant relationship between
CEO total compensation and firm market performance. This is consistent with the findings of
(Aggarwal and Samwick, 1999b) that pay to stock performance link is weak when stock
volatility is high and vice versa. The insignificant coefficients of TRET are also consistent
with studies in China and Japan (see, e.g., Basu et al., 2007, Conyon and He, 2012) where use
of stock base compensation is not very prominent.
With regard to influence of lagged firm performance on current CEO total compensation,
columns 2, 4 & 6 show that previous year ROA also significantly influences CEO total
compensation in positive direction. The coefficient is highest with value 1.6911 in pooled
regression followed 1.5490 and 0.2575 in other two models that incorporate industry and firm
fixed effect along with time fixed effects respectively. The previous year ROA seems to
influence more than the current year ROA as coefficients of previous year ROA is greater in
all models. Similar findings are reported in India where Ghosh (2006) find that lagged ROA
has larger impact than current ROA. Thus, past firm performance has more contribution in
pay setting process in Indian subcontinent.
Surprisingly, previous year total return to shareholders (market performance) has
negative relationship with current CEO total compensation in column 2 and 4. But in column
6 the relationship becomes insignificant with positive sign. One possible reason for negative
relationship may be that since Pakistani markets are highly volatile and it is difficult to base
compensation on market performance. Thus, during the period of bad market performance,
next year compensation may have increased due to positive accounting performance, leading
to negative relationship between current compensation and previous year market
110
performance. In addition, in times of good market performance, increase in current
compensation may not correspond to extra ordinary good market performance in last year,
leading to weak positive or negative relationship between current compensation and previous
year market performance. Therefore, it is presumed that this negative association is by
chance.
Overall, the results suggest that the first hypothesis that firm performance influences
CEO compensation is accepted for firm accounting performance only as both current and
previous year ROA positively contributes towards CEO pay setting process. This implies that
the boards of directors in Pakistan reward CEOs against accounting performance only and do
not give weight to market performance due to opaque nature of Pakistani stock markets. This
is consistent with studies in China and India where too only accounting performance affect
CEO compensation decisions. The positive impact of firm performance on CEO
compensation in a family ownership environment in Pakistan suggests that family ownership
contributes towards alignment of shareholder-manager interests. Existing evidence, however,
on family ownership environment from Asia indicates expropriation view of family
ownership as Kato et al. (2007) find in Korea that accounting performance (ROA) does not
have any impact on CEO compensation in family firms. This may imply that behavior of
family firms in Pakistan is different.
Ownership concentration seems to have positive impact on CEO total compensation
which is inconsistent with agency theory argument that concentrated ownership has better
incentives to monitor and curb managerial opportunism, leading to lower CEO total
compensation. The coefficients of OWNCONS are significantly positive in column 1 through
4. Interestingly, within firm estimation (column 5 & 6) does not depict significant
relationship between CEO total compensation and ownership concentration. This might be
due to the fact that less variation exists for ownership concentration within the firms, leading
to insignificant coefficients. Overall these results reject hypothesis 2 that concentrated
ownership has a negative effect on CEO compensation. The positive impact of ownership
concentration on CEO compensation supports the view that large shareholders expropriate
minority interests by paying excessive compensation to CEOs. Controlling shareholders seem
to be colluding with management to get personal benefits at the expense of minority
shareholders thus overlooking CEO compensation. This finding is consistent with an earlier
study, Shah et al. (2009), in Pakistan but inconsistent with studies in US (Core et al., 1999),
Canada (Amoako-Adu et al., 2011), Europe (Croci et al., 2012, Ozkan, 2007, Ozkan, 2011)
except Denmark (Banghoj et al., 2010) and a few instances in Asia (Conyon and He, 2011).
111
Thus, in term of CEO compensation, concentrated ownership in Pakistan generally behaves
differently as compared to North American and European countries.
In pooled regression, CEOs seem to receive lower total compensation in family firms.
This is inconsistent with Basu et al. (2007) in Japan which finds higher executive
compensation in family firms. However, when industry and time fixed effects are introduced
in the model, coefficients of FAMOWN do not appear to be significant and sign of the
coefficients changes from negative to positive (see column 3 & 4). In addition, introduction
of firm fixed effects along with time fixed effect makes the results significant with positive
sign. These results may be interpreted as CEOs receive less total compensation in family
firms as compared to non-family firms but within firm if family takes over it revises
compensation contracts upward. However, estimates of FAMOWN must carefully be
interpreted in the presence of firm fixed effects. When firm fixed effects are used there must
be reasonable variations within the firm for FAMOWN. However, this is unlikely as family
firm in 2005 is very likely to remain a family firm throughout the sample period of 8 years.
Within industry CEOs in family and non-family firms seem to receive equal compensation on
average. Overall, the third hypothesis that family firms pay less CEO compensation finds
weak support from the results.
Board size is normally used as proxy for board effectiveness. Larger boards beyond
certain size limits become ineffective. Such larger boards can easily be influenced by CEOs,
leading to excessive CEO compensation and hence positive association between CEO pay
and board size. The results do not support this argument as coefficient of board size is
consistently insignificant in all models. These results suggest that board size does not
influence CEO total compensation in either side positive or negative. Thus, hypothesis 4 that
board size is positively associated with CEO compensation is rejected. This implies that
board size is irrelevant in Pakistan. This finding contradicts with the finding of several
studies (e.g., Basu et al., 2007, Conyon, 2014, Conyon and He, 2012, Core et al., 1999,
Fernandes et al., 2013, Shah et al., 2009) however, it corresponds with the institutional
context of Pakistan discussed earlier. Groups or family firms use proxy directors who are
obligated to controlling groups or families only. These proxy directors don’t have any say or
they support the decisions coming from controlling group or family. This makes board size
irrelevant. Similarly, state owned firms and foreign owned firms bypass boards in taking
major decisions which also leads to no influence of board size on CEO compensation
decisions.
112
The insignificant coefficients for B_IND in all models indicate no effect of board
independence in CEO pay process. Thus, hypothesis 5 that board independence contributes
effectively to reduce managerial opportunism and hence reduce CEO compensation is
rejected. The results seem to support the managerial power approach that independent
directors can be under influence of CEO as they wish to be re-nominated and re-appointed
and may engage in secret contracts with CEOs, which in turn may lead to overlooking of
CEO compensation contracts or approving excessive CEO compensation contracts. The
finding implies that non-executive directors are neutral in CEO compensation decisions. They
do not give their inputs. This is due to the reason that non-executive directors are generally
hired from within controlling families and cannot go against the family decisions. Further it
implies that CEO compensation decisions are not made within the board. The boards just
approve CEO pay decisions which have been taken elsewhere14.
While this finding is consistent with Kashif and Mustafa (2012) in Pakistan and some
other studies outside Pakistan (e.g., Guthrie et al., 2012, Ozkan, 2011) however, it is in
contrast with Basu et al. (2007) in Japan and Shah et al. (2009) in Pakistan which find
negative relationship between CEO total compensation and board independence. This
contradicting finding from Shah et al. (2009) in Pakistan may be due to difference in data
period and sample size used. However as discussed earlier Shah et al. (2009) also suffers
from methodological difficulties which might be another reason for contradicting results.
CEO duality appears to be significantly negatively related to CEO total compensation.
The coefficients of DUALCEO are consistently negative and significant in column 1 through
4. However, with firm fixed effects they lose their significance but sign remains negative.
Again within firms, there might not be reasonable variations in CEO duality over sample
period which might have led to insignificant coefficients when firm fixed effect models are
used. Nevertheless, overall, hypothesis 6 that CEO duality leads to ineffective corporate
governance and results in higher CEO compensation is rejected. Thus, the argument that
more concentrated power in one person lead to expropriation resulting in higher CEO
compensation is not supported. This finding is inconsistent with most of the studies inside
and outside Pakistan (e.g., Chen et al., 2011, Conyon and He, 2012, Core et al., 1999, Kashif
14 One can argue that principle component analysis (PCA) should be warranted as insignificant coefficients of board size and board independence may be due to potential multicollinearity issues in board structure variables. However, correlation matrices and VIFs suggest that there is no issue of multicollinearity in the studied variables. In addition, the primary objective of the study is to examine the individual influences of corporate governance variables on CEO compensation. If PCA is performed, much of the information in the variables would be lost as PCA would reduce the number of variables to one or two components which are themselves difficult to interpret (see, Jolliffe, 2002). Therefore, PCA should not be warranted in this case.
113
and Mustafa, 2012, Shah et al., 2009). This is an unexpected finding which implies that
Pakistani firms behave differently from rest of the world in that when CEO also holds the
position of chairman, he/she agrees upon receiving lower compensation. One reason for this
finding might be that since family ownership is common in Pakistan and the position of CEO
or chairman is generally held by founding father or elder sibling therefore being steward of
the family and holding all the powers single handedly, he might not give weight to higher
compensation against power he enjoys or his social stature, resulting in lower compensation
(see, e.g., Gomez-Mejia et al., 2003). A split sample analysis (unreported) confirms that
negative impact of CEO duality on CEO compensation is driven by family firms. Similarly,
mean comparison of CEO compensation shows that family firms and firms with CEO duality
have lower average CEO compensation. Overall, this finding warrants investigation on the
impact of internal family issues and conflicts on CEO compensation as mentioned by
Bertrand and Schoar (2006). However, this is beyond the scope of this study. In addition, the
dataset used in this study does not allow to investigate effects of internal family issues on
CEO compensation.
Firm size (FIRMSIZE) as measured by log of total assets is appeared to be significantly
positively related to CEO total compensation in all models which implies that large firms pay
higher compensation to their CEOs. This supports the argument that larger firms are complex
and difficult to run and hence require quality CEOs with higher compensations. Firm risk
(FIRMRSK) is negatively associated with CEO total compensation however coefficient is
significant only in the models with firm fixed effects. Negative association indicate that risky
firms pay less to their CEOs which is inconsistent with existing studies (e.g., Brick et al.,
2006, Conyon and He, 2011, Core et al., 1999, Lippert and Moore, 1994, Lippert and Porter,
1997). MTB appears to have no significant relationship with CEO total compensation. Thus,
it can be assumed that MTB does not influence CEO total compensation in Pakistan. Aged
and matured firms seem to pay higher compensation to their CEOs as most of the coefficients
of FMAGE are positive and significant.
4.2.2.2 Cash Compensation
Table 4.5 describes the empirical results of models 1 and 2 developed in chapter 3 for
cash compensation as dependent variable. Similar to Table 4.4, columns 1,3 and 5 presents
the results of Model 1 that includes current firm performance while columns 2,4 and 6
provide insight of the relationship between cash compensation and lagged firm performance
i.e. Model 2. Column 1 and 2 exhibit results of pooled regression. Unobserved heterogeneity
114
is controlled using fixed effects as described in chapter 3. Columns 3 and 4 report results with
industry and time fixed effects while columns 5 and 6 present results with firm and time fixed
effects.
In Table 4.5, again, consistent with the main hypothesis, CEO cash compensation also
seems to be positively related to current firm accounting performance as measured by ROA.
The results are consistent qualitatively over pooled regression (column 1), industry and time
fixed effects model (column 3) but not over firm and time fixed effects model (column 5).
The coefficient of ROA in pooled model is the highest with value 1. 3644 followed by 1.3315
in industry and time fixed effect model and 0.2497 in firm and time fixed model. The
closeness of coefficients of the models with cash compensation and total compensation
indicates high correlation between cash compensation and total compensation in Pakistan.
The positive association of cash compensation and firm accounting performance is consistent
with other studies in Asia (e.g., Ghosh, 2006, Kato and Kubo, 2006, Kato and Long, 2006)
Current total returns to shareholder do not significantly contribute to cash compensation
as none of the models in columns 1, 3 and 5 reports coefficient for TRET that is significantly
different from zero in both tables. This indicates that firms in Pakistan do not link base salary
and cash bonuses to market performance. The reason might be existence of high volatility in
stock markets of Pakistan which may drive away the attention of directors to use market
performance as benchmark for setting CEO cash compensation (Shaw and Zhang, 2010).
With regard to influence of lagged firm performance on current CEO cash compensation,
columns 2, 4 & 6 show that previous year ROA also significantly influences CEO cash
compensation in positive direction. The coefficient is highest with value 1.7678 in pooled
regression followed 1.6122 and 0.3101 in other two models that incorporate industry and firm
fixed effect along with time fixed effects respectively. The previous year ROA seems to
influence more than the current year ROA as coefficients of previous year ROA are greater in
all models. Surprisingly, previous year total return to shareholders (market performance) has
negative relationship with current CEO cash compensation in column 2 and 4. But in column
6 the relationship becomes insignificant with positive sign. This is similar situation as in case
of total compensation as dependent variable in Table 4.4. The possible reasons for this
finding are the same as discussed earlier
Ownership concentration seems to have positive impact on CEO cash compensation also
like on total compensation which is inconsistent with agency theory arguments. This might
again be an indication of expropriation of minority interests by controlling shareholders.
115
Thus, hypothesis 2 that concentrated ownership is negatively related CEO compensation is
rejected for cash compensation also.
Like in case of total compensation, influence of family ownership on CEO cash
compensation seems inconclusive. However, since within firm variation in FAMOWN
variable is not expected therefore ignoring firm fixed effect model, the hypothesis that family
firms pay less to their CEO is partially accepted with a need to explore it further by
incorporating variables related to family characteristics such as family CEO, number of
family members on board etc.
The coefficients of BDSIZE and B_IND stay insignificant in all models. Thus, Board
size and board independence has no impact on setting CEO cash pay. These results reject
hypotheses 4 and 5 and seem to support the managerial power approach.
Similar to CEO total compensation, CEO duality appears to be significantly negatively
related to CEO cash compensation also. The results are quite similar to as in case of total
compensation. Thus, hypothesis 6 that CEO duality leads to ineffective corporate governance
and results in higher CEO compensation is also rejected for cash compensation.
All other control variables such as FIRMSIZE, FIRMRSK, MTB and FMAGE have
same relationship with CEO cash compensation as they have with CEO total compensation.
FIRMSIZE and FMAGE have positive, FIRMRSK has negative while MTB has no
relationship with CEO cash compensation with similar possible reasons as described earlier.
This is mainly due to high correlation between the two measures of CEO compensation.
116
Table 4.4
Total Compensation and Firm Performance Dependent variable is log of total compensation as measured by sum of all pay components, ROA = return on assets measured
by earnings before interest and taxes divided by total assets, TRET = total return to shareholders as measured by current market
price plus dividend divided by previous year market price minus one, OWNCONS = concentrated ownership as measured by
voting shares held by the largest shareholder, FAMOWN = a dummy variable taking value one for family firms and zero
otherwise, BDSIZE = board size as measured by number of sitting directors on board, B_IND = board independence as
measured by ratio of non-executive directors to board size, DUALCEO = CEO duality a dummy variable taking value one if
CEO is also chairman board of directors, FIRMSIZE = firm size as measured by log of total assets, FIRMRSK = firm risk as
measured by standard deviation of monthly stock returns over the fiscal year, MTB = market to book ratio as measured by
market value divided by book value per share, FMAGE = firm age as mentioned in annual reports, CEOCHNG = Interaction
term of a dummy variable taking value 1 if CEO is replaced during the year and number of CEOs drawing remuneration during
the year
Log of Total Compensation
(1) (2) (3) (4) (5) (6)
VARIABLES Pooled Pooled Fixed Effects
(Industry and
Time)
Fixed Effects
(Industry and
Time)
Fixed Effects
(Firm and
Time)
Fixed Effects
(Firm and
Time)
ROAit 1.3484*** 1.3156*** 0.2460*
(0.3927) (0.3429) (0.1268)
TRETit 0.0132 -0.0406 0.0053
(0.0262) (0.0246) (0.0189)
ROAit-1 1.6911*** 1.5490*** 0.2575**
(0.3965) (0.3296) (0.1289)
TRETit-1 -0.0849*** -0.0576** 0.0137
(0.0316) (0.0262) (0.0162)
OWNCONSit 0.7634** 0.7818** 0.5726* 0.5797** 0.0384 0.1224
(0.3298) (0.3247) (0.2937) (0.2910) (0.4862) (0.4174)
FAMOWNit -0.2767* -0.2640* 0.0137 0.0095 0.5938*** 0.4843***
(0.1486) (0.1494) (0.1465) (0.1461) (0.0488) (0.0482)
BDSIZEit 0.0016 0.0133 0.0029 0.0083 -0.0145 0.0056
(0.0425) (0.0418) (0.0382) (0.0379) (0.0351) (0.0393)
B_INDit 0.3561 0.2843 0.2463 0.2020 0.2489 0.2712
(0.2844) (0.2862) (0.2536) (0.2562) (0.2291) (0.1954)
DUALCEOit -0.3416*** -0.3489*** -0.2217** -0.2156* -0.0483 -0.0587
(0.1217) (0.1221) (0.1095) (0.1105) (0.1156) (0.1262)
FIRMSIZEit 0.3546*** 0.3396*** 0.3503*** 0.3450*** 0.2255*** 0.2967***
(0.0353) (0.0356) (0.0363) (0.0359) (0.0622) (0.0708)
FIRMRSKit -0.1918 0.0228 -0.0783 0.0749 -0.2950** -0.2913**
(0.2106) (0.2151) (0.2110) (0.2103) (0.1339) (0.1432)
MTBit -0.0080 -0.0087 -0.0075 -0.0082 -0.0005 0.0009
(0.0086) (0.0086) (0.0092) (0.0088) (0.0040) (0.0031)
FMAGEit 0.0097** 0.0090** 0.0072** 0.0071** -0.0218 -0.0716
(0.0037) (0.0037) (0.0032) (0.0030) (0.0468) (0.0524)
CEOCHNGit -0.1296 -0.1395* -0.1196* -0.1288* 0.0389 0.0296
(0.0805) (0.0815) (0.0711) (0.0732) (0.0366) (0.0406)
Observations 1,508 1,277 1,508 1,277 1,508 1,277
R-squared 0.4686 0.4863 0.6060 0.6120 0.9301 0.9377
Industry FE NO NO YES YES NO NO
Firm FE NO NO NO NO YES YES
Time FE NO NO YES YES YES YES
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
117
Table 4.5
Cash Compensation and Firm Performance Dependent variable is log of cash compensation as measured by sum of basic salary and cash bonuses, ROA = return on assets
measured by earnings before interest and taxes divided by total assets, TRET = total return to shareholders as measured by
current market price plus dividend divided by previous year market price minus one, OWNCONS = concentrated ownership as
measured by voting shares held by the largest shareholder, FAMOWN = a dummy variable taking value one for family firms
and zero otherwise, BDSIZE = board size as measured by number of sitting directors on board, B_IND = board independence
as measured by ratio of non-executive directors to board size, DUALCEO = CEO duality a dummy variable taking value one if
CEO is also chairman board of directors, FIRMSIZE = firm size as measured by log of total assets, FIRMRSK = firm risk as
measured by standard deviation of monthly stock returns over the fiscal year, MTB = market to book ratio as measured by
market value divided by book value per share, FMAGE = firm age as mentioned in annual reports, CEOCHNG = Interaction
term of a dummy variable taking value 1 if CEO is replaced during the year and number of CEOs drawing remuneration during
the year
Log of Cash Compensation
(1) (2) (3) (4) (5) (6)
VARIABLES Pooled Pooled Fixed Effects
(Industry and
Time)
Fixed Effects
(Industry and
Time)
Fixed Effects
(Firm and
Time)
Fixed Effects
(Firm and
Time)
ROAit 1.3644*** 1.3315*** 0.2497
(0.4029) (0.3611) (0.1572)
TRETit 0.0244 -0.0304 0.0066
(0.0267) (0.0258) (0.0224)
ROAit-1 1.7678*** 1.6122*** 0.3101**
(0.4083) (0.3513) (0.1502)
TRETit-1 -0.0872** -0.0594** 0.0115
(0.0336) (0.0277) (0.0180)
OWNCONSit 0.8819*** 0.9131*** 0.7061** 0.7242** 0.0172 0.0938
(0.3363) (0.3318) (0.3114) (0.3095) (0.5338) (0.4787)
FAMOWNit -0.2915* -0.2765 -0.0421 -0.0428 0.6186*** 0.4294***
(0.1657) (0.1705) (0.1649) (0.1658) (0.0566) (0.0578)
BDSIZEit 0.0160 0.0291 0.0253 0.0317 -0.0127 0.0054
(0.0485) (0.0490) (0.0437) (0.0445) (0.0391) (0.0395)
B_INDit 0.1132 0.0403 0.0292 -0.0175 0.2313 0.3415
(0.2965) (0.3032) (0.2725) (0.2801) (0.2475) (0.2265)
DUALCEOit -0.3210** -0.3297** -0.2191* -0.2121* -0.0040 0.0089
(0.1269) (0.1279) (0.1150) (0.1167) (0.1254) (0.1421)
FIRMSIZEit 0.3530*** 0.3390*** 0.3532*** 0.3490*** 0.2712*** 0.3298***
(0.0368) (0.0371) (0.0385) (0.0381) (0.0679) (0.0753)
FIRMRSKit -0.2886 -0.0559 -0.1914 -0.0144 -0.2849** -0.2640*
(0.2126) (0.2215) (0.2044) (0.2059) (0.1365) (0.1437)
MTBit -0.0040 -0.0049 -0.0052 -0.0062 -0.0002 0.0000
(0.0081) (0.0081) (0.0088) (0.0084) (0.0036) (0.0028)
FMAGEit 0.0092** 0.0085** 0.0062* 0.0062* -0.0116 -0.0571*
(0.0040) (0.0040) (0.0034) (0.0033) (0.0249) (0.0330)
CEOCHNGit -0.1733** -0.1898** -0.1583** -0.1757** 0.0127 -0.0086
(0.0858) (0.0854) (0.0767) (0.0775) (0.0391) (0.0411)
Observations 1,508 1,277 1,508 1,277 1,508 1,277
R-squared 0.4474 0.4666 0.5707 0.5779 0.9135 0.9226
Industry FE NO NO YES YES NO NO
Firm FE NO NO NO NO YES YES
Time FE NO NO YES YES YES YES
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
118
4.2.2.3 Regression on Lagged Explanatory Variables
To control for potential problem of endogeneity, CEO compensation is regressed on
lagged values of explanatory variables as discussed in chapter 3. Table 4.6 reports the results
for Model 3 in chapter 3 for both cash and total compensation.
Firms accounting performance as measure by ROA appears to influence both CEO cash
and total compensation even after controlling for endogeneity. The coefficients of ROA are
significant across pooled and fixed effects models. However, market performance as
measured by total return to shareholders, surprisingly, seems to be negatively related to both
cash and total compensation. The reason for this negative association is assumed to
preference of board to not using highly volatile market performance as a benchmark for pay
setting process as discussed earlier. Thus, in Pakistan, the only firm performance that matters
in setting CEO compensation is accounting performance.
Concentrated ownership appears to positively affect both CEO cash and total
compensation in pooled and industry fixed effect models. Thus, it can fairly be assumed that
concentrated ownership positively influence CEO compensation and large shareholders
oblige CEOs by setting CEO compensation high. Given the Pakistani context, this is mainly
due to get personal benefits. Influence of family ownership on CEO compensation is again
mixed. Again comparison between firms without considering industry (pooled regression)
shows that CEO in family firms receive less pay than their counterparts but within firm
change of family ownership status from non-family to family positively influence CEO
compensation.
BDSIZE and B_IND do not appear to influence CEO compensation as indicated by
insignificant coefficients for these variables. CEOs who are also chairman board of directors
appear to receive less compensation than their counterparts as coefficients of DUALCEO are
negative and significant except in column 3 and 6. This indicates that within firm change of
status from separate chairman to CEO as chairman does not seem to affect CEO
compensation on average.
Coefficients of FIRMSIZE are highly significant with positive sign in all models for cash
and total compensation, indicating that firm size is perhaps one the most important factors
that contributes towards CEO compensation. FIRMRSK and MTB do not appear to have
convincing contribution towards devising CEO compensation packages. FMAGE does have
some positive influence on CEO compensation.
Overall, regression on lagged explanatory variables confirms the previous results.
119
Table 4.6
CEO Compensation and Firm Performance, controlling for Endogeneity (Regression on lagged explanatory
variables) Dependent variables are log of compensation, ROA = return on assets measured by earnings before interest and taxes divided by
total assets, TRET = total return to shareholders as measured by current market price plus dividend divided by previous year
market price minus one, OWNCONS = concentrated ownership as measured by voting shares held by the largest shareholder,
FAMOWN = a dummy variable taking value one for family firms and zero otherwise, BDSIZE = board size as measured by
number of sitting directors on board, B_IND = board independence as measured by ratio of non-executive directors to board size,
DUALCEO = CEO duality a dummy variable taking value one if CEO is also chairman board of directors, FIRMSIZE = firm
size as measured by log of total assets, FIRMRSK = firm risk as measured by standard deviation of monthly stock returns over
the fiscal year, MTB = market to book ratio as measured by market value divided by book value per share, FMAGE = firm age as
mentioned in annual reports, CEOCHNG = Interaction term of a dummy variable taking value 1 if CEO is replaced during the
year and number of CEOs drawing remuneration during the year
Cash Compensation Total Compensation
(1) (2) (3) (4) (5) (6)
VARIABLES Pooled Fixed Effects
(Industry and
Time)
Fixed Effects
(Firm and
Time)
Pooled Fixed Effects
(Industry and
Time)
Fixed Effects
(Firm and
Time)
ROAit-1 1.8433*** 1.6888*** 0.4430*** 1.7596*** 1.6158*** 0.3792***
(0.4097) (0.3522) (0.1338) (0.4006) (0.3325) (0.1245)
TRETit-1 -0.0752** -0.0556* 0.0120 -0.0773** -0.0573** 0.0158
(0.0335) (0.0288) (0.0166) (0.0317) (0.0275) (0.0150)
OWNCONS it-1 0.9074*** 0.7132** 0.3700 0.7796** 0.5749* 0.4696
(0.3462) (0.3213) (0.4289) (0.3380) (0.3011) (0.3959)
FAMOWNit-1 -0.2700 -0.0448 0.3306*** -0.2530* 0.0133 0.3422***
(0.1717) (0.1679) (0.0535) (0.1511) (0.1483) (0.0513)
BDSIZEit-1 0.0322 0.0305 -0.0258 0.0160 0.0072 -0.0110
(0.0496) (0.0447) (0.0451) (0.0428) (0.0385) (0.0455)
B_INDit-1 0.0312 -0.0228 0.1582 0.2701 0.1902 0.0826
(0.2954) (0.2760) (0.1926) (0.2804) (0.2543) (0.1795)
DUALCEOit-1 -0.3234** -0.2168* 0.0725 -0.3580*** -0.2329** -0.0232
(0.1298) (0.1180) (0.0995) (0.1244) (0.1123) (0.0899)
FIRMSIZEtt-1 0.3439*** 0.3529*** 0.2735*** 0.3451*** 0.3490*** 0.2601***
(0.0373) (0.0386) (0.0830) (0.0357) (0.0361) (0.0815)
FIRMRSKit-1 -0.0665 0.0075 0.0833 0.0468 0.1100 0.0584
(0.2231) (0.2129) (0.0985) (0.2191) (0.2191) (0.0913)
MTBit-1 -0.0055 -0.0071 -0.0036 -0.0093 -0.0090 -0.0022
(0.0083) (0.0088) (0.0037) (0.0089) (0.0093) (0.0043)
FMAGEit 0.0084** 0.0062* -0.0661* 0.0089** 0.0071** -0.0798
(0.0040) (0.0033) (0.0380) (0.0037) (0.0031) (0.0577)
CEOCHNGit -0.1809** -0.1674** -0.0043 -0.1252 -0.1165 0.0354
(0.0853) (0.0774) (0.0406) (0.0803) (0.0722) (0.0404)
Observations 1,277 1,277 1,277 1,277 1,277 1,277
R-squared 0.4645 0.5765 0.9211 0.4856 0.6119 0.9362
Industry FE NO YES NO NO YES NO
Firm FE NO NO YES NO NO YES
Time FE NO YES YES NO YES YES
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
120
4.2.2.4 Dynamic Panel Model – GMM Estimation
Table 4.7 exhibits the results of dynamic panel model (Model 4 in chapter 3) estimated
using GMM approach. As described in chapter 3 the model is estimated using three GMM
techniques. Column 1 & 4 present results of the model estimated using of GMM-Diff for cash
compensation and total compensation respectively. Similarly, columns 2 & 4 report results of
GMM-Orthogonal technique and column 3 & 6 show results of GMM-System technique for
cash compensation and total compensation respectively. Arellano-Bond serial correlation
tests m1 & m2 and instrument over-identification tests are also reported at the bottom of the
table. Arellano-Bond test for second order (m2) validates the use second and earlier lags of
dependent variables as instruments in all the columns. None of the values of m2 rejects the
hypothesis of no second order correlation in error terms. Similarly, over-identification tests
Sargan test and Hansen J test show that the instruments used in the GMM estimation
techniques are jointly valid. All the values of Sargan test statistic and Hansen J test statistic
are insignificant thus, accepting the hypothesis of joint validity of the instruments used. Note
that GMM models’ results are within firm estimation and comparable to firm fixed and time
fixed effect models only which are discussed earlier.
The GMM estimation confirms the finding of positive relationship between both
measures of CEO compensation and firm accounting performance. However, although signs
of the coefficients of TRET are negative but they remain insignificant in all three techniques
for both measures of CEO compensation, confirming no influence of market performance on
CEO compensation. Thus, it is confirmed that only firm accounting performance contribute
towards CEO pay setting process in Pakistan. The boards of director ignore market
performance while setting CEO compensation. This is believed to be due to highly volatile
and opaque Pakistani stock markets.
With regard to corporate governance variables, there is little evidence that concentrated
ownership has impact on total CEO compensation (see column 6) as compared to firm fixed
effect models where such evidence does not exist (see Table 4.4). Thus, dynamic panel model
that controls for endogeneity problem along with unobserved heterogeneity and serial
correlation seems to confirm that large shareholders support setting CEO compensation high
as they are in direct relation with CEOs to get personal benefits.
Family ownership appears to be related to lower CEO cash compensation as the
coefficient of FAMOWN is negative and significant in GMM-Orthogonal technique (see
column 1 to 3). On the other side, FAMOWN appears to have positive influence on CEO
total compensation in GMM-DIFF technique. As cash compensation includes only base
121
salary and cash bonuses while total compensation includes all the components therefore,
these results indicate that CEOs in family firms receive lower base salary and cash bonuses
but higher perks and retirement benefits.
The coefficients of BDSIZE and B_IND are consistently insignificant which is consistent
with fixed effect models. Thus, dynamic panel model confirms that board size and number of
non-executive directors in the board in Pakistan are irrelevant in CEO compensation
decisions. DUALCEO appears to have significantly negative relationship with both CEO
cash and total compensations. This confirms that after controlling endogeneity, CEOs who
are also chairman board of directors receive lower cash as well as total compensation.
Overall, the results of corporate governance have improved using GMM approach. This is
due to better control for endogenous nature of corporate governance variables.
The coefficients of FIRMSIZE are consistently positive and significant across all three
techniques for both cash and total compensation. FIRMRSK does not have any significant
coefficient in any of three techniques used. Earlier FIEMRSK shows negative effect on both
cash and total CEO compensation in firm fixed effect models (see Table 4.4 & 4.5). MTB
does not seem to have any convincing relationship with CEO total compensation however
there is some sign of positive effect on cash compensation in GMM-Orthogonal approach.
FMAGE appears to have decreasing effect on both CEO cash compensation and CEO total
compensation as GMM-DIFF and GMM Orthogonal approaches report significantly negative
coefficients for FMAGE (see columns 1, 2, 4 and 5).
One important feature of dynamic panel model is that it considers the dynamic
adjustment of CEO pay towards the target levels. Conyon and He (2012) argue that the
boards of directors have incomplete information about CEOs’ capabilities and this
information has to be updated over time. Thus, boards of directors adjust CEO compensation
to target levels as they learn more about CEO capabilities over time. However, the boards
cannot adjust CEO pay quickly to the target levels rather they try to smooth the adjustments
process, leading to serial correlation in CEO compensation.
Consistent with the above arguments coefficients of lagged CEO pay are positive and
significant in all three techniques for both cash and total compensation, indicating that CEO
pay is highly persistent and takes time to adjust to its long-term equilibrium level. This is an
important finding of this study as most of the literature on CEO compensation (see, e.g.,
Croci et al., 2012, Devers et al., 2007, Kashif and Mustafa, 2012, Ozkan, 2007, Shah et al.,
2009) ignores the dynamic nature of CEO compensation and estimate static pay models
122
considering that pay is in equilibrium, thus, ignoring the CEO pay adjustment to long run
equilibrium.
It is important to note here that very few studies have adopted dynamic panel model
estimation while analyzing CEO compensation. Conyon and He (2012) is the only study in
Asian context of which results are directly comparable to this study. Consistent with this
study, Conyon and He (2012) also find in China that the boards of directors pay more
attention to firm accounting performance than to firm market performance while setting CEO
compensation. This may be because of the fact that stock based CEO compensation is rare in
both countries which may eliminate the option of making market performance as benchmark
for CEO compensation. Further, Conyon and He (2012) also find that CEO compensation is
highly persistent and dynamically adjusted towards target equilibrium level. However,
inconsistent with this study, ownership and board structure variables (board size, board
independence and CEO duality) do not have significant impact on CEO pay process in their
study after controlling for pay persistence, unobserved heterogeneity and endogeneity. In
Pakistan, as results suggest, ownership concentration and CEO duality have important
contribution towards pay setting process even after controlling for pay persistence,
unobserved heterogeneity and endogeneity. The differences in results of corporate
governance variables of both studies are presumably due to differences in institutional
contexts of China and Pakistan.
4.2.3 Robustness Checks
Although results presented are robust across different model specifications however, to
further reconfirm, three robustness checks are conducted. First, all the continuous variables
are winsorized using 1% level at both tails to eliminate potential outliers and all models are
re-estimated. But, the results do not change qualitatively therefore it is decided to report the
original data results. Second, alternative measures of firm accounting performance, firm size
and ownership concentration as log of net sales, EPS and voting shares held by three largest
shareholders respectively, are incorporated. Again, the results remain qualitatively similar to
as reported above. Third, median regression is estimated using STATA but that also does not
lead to change the inferences.
Fourth, ownership concentration and family ownership variables are incorporated in
separate models as they may shadow each other. However, the results remain qualitatively
similar to as reported. Fifth, two dummy variables are constructed to account for
effectiveness of board size and board independence in another way. One, considering board
123
size between 8 and 10 as optimum board size, a dummy variable is constructed taking value 1
if board size is between 8 and 10, and zero otherwise. The other, considering one of the
requirements of the Code 2002 that the ratio of the number of executive directors to board
size should not exceed 75%, a dummy variable is constructed taking value 1 if this ratio is
less than or equal to 75%, and zero otherwise to see if a company meets the requirement.
These variables are incorporated in the model instead of original variables of board size and
board independence. However, both dummy variables appear to have no impact on CEO
compensation in any direction. Thus, this confirms the original results that board size and the
presence of non-executive directors in the board do not contribute towards CEO
compensation decisions. Thus, findings of this study are robust to different model
specifications and alternative measures of important variables.
124
Table 4.7
CEO Compensation and Firm Performance, Dynamic Panel Model (GMM Estimation) Dependent variables are log of compensation, ROA = return on assets measured by earnings before interest and taxes
divided by total assets, TRET = total return to shareholders as measured by current market price plus dividend divided by
previous year market price minus one, OWNCONS = concentrated ownership as measured by voting shares held by the
largest shareholder, FAMOWN = a dummy variable taking value one for family firms and zero otherwise, BDSIZE = board
size as measured by number of sitting directors on board, B_IND = board independence as measured by ratio of non-
executive directors to board size, DUALCEO = CEO duality a dummy variable taking value one if CEO is also chairman
board of directors, FIRMSIZE = firm size as measured by log of total assets, FIRMRSK = firm risk as measured by
standard deviation of monthly stock returns over the fiscal year, MTB = market to book ratio as measured by market value
divided by book value per share, FMAGE = firm age as mentioned in annual reports, CEOCHNG = Interaction term of a
dummy variable taking value 1 if CEO is replaced during the year and number of CEOs drawing remuneration during the
year
Cash Compensation Total Compensation
(1) (2) (3) (4) (5) (6)
VARIABLES GMM
Differenced
GMM
Orthogonal
GMM
System
GMM
Differenced
GMM
Orthogonal
GMM
System
LNCOMPt-1 0.4281*** 0.3732*** 0.8995*** 0.3547*** 0.3908*** 0.8037***
(0.1356) (0.1215) (0.0413) (0.1185) (0.1286) (0.0577)
ROAit 0.2723* 0.4127*** 0.4194*** 0.2687* 0.3765*** 0.5282***
(0.1528) (0.1506) (0.1404) (0.1480) (0.1228) (0.1567)
TRETit -0.0480 -0.0507 -0.0275 -0.0289 -0.0302 -0.0447
(0.0397) (0.0429) (0.0465) (0.0343) (0.0404) (0.0615)
OWNCONSit -0.0541 0.2543 0.1173 -0.3847 0.0807 0.1486*
(0.6606) (0.3469) (0.0910) (0.4114) (0.2765) (0.0831)
FAMOWNit 0.0322 -1.1709*** -0.0163 0.1859** -0.2949 -0.0339
(0.0856) (0.4462) (0.0432) (0.0754) (0.6523) (0.0380)
BDSIZEit -0.0168 0.0031 0.0069 0.0045 0.0142 0.0087
(0.0387) (0.0286) (0.0114) (0.0249) (0.0304) (0.0093)
B_INDit 0.0982 0.2273 0.0160 0.2362 0.1820 0.0300
(0.1735) (0.1478) (0.0612) (0.3773) (0.1341) (0.0707)
DUALCEOit -0.2422* -0.0427 -0.0651* -0.2400* -0.1257* -0.0750*
(0.1471) (0.0866) (0.0353) (0.1278) (0.0738) (0.0386)
FIRMSIZEit 0.1785*** 0.2548*** 0.0365** 0.1721*** 0.2023*** 0.0664***
(0.0590) (0.0600) (0.0146) (0.0550) (0.0549) (0.0204)
FIRMRSKit 0.1119 0.0136 0.0493 0.0067 -0.0520 -0.0763
(0.1908) (0.2300) (0.1222) (0.1493) (0.2155) (0.1580)
MTBit 0.0034 0.0027* -0.0004 0.0024 0.0022 0.0010
(0.0023) (0.0014) (0.0035) (0.0029) (0.0019) (0.0032)
FMAGEit -0.0585*** -0.0507** -0.0004 -0.0799*** -0.0612* 0.0008
(0.0157) (0.0254) (0.0008) (0.0223) (0.0313) (0.0008)
CEOCHNGit 0.0229 0.0263 0.0186 0.0595* 0.0487 0.0292
(0.0387) (0.0372) (0.0419) (0.0335) (0.0340) (0.0382)
Observations 1,051 1,057 1,277 1,051 1,057 1,277
Number of firms 215 215 220 215 215 220
Arellano-Bond
AR(1) in Diff. (m1) -2.90*** -2.82*** -4.44*** -3.32*** -3.36*** -5.67***
AR(2) in Diff. (m2) -1.37 -1.33 -0.99 -0.95 -0.84 -0.33
Over identification tests
Sargan test 61.43 54.49 93.99 91.14 57.05 63.43
Hansen J test 64.48 61.98 86.42 70.83 56.72 42.38
(df) (60) (60) (79) (80) (60) (52)
Windmeijer-corrected Standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
125
4.3 CEO Compensation and Future Firm Performance
4.3.1 Correlation Analysis
Table 4.8 presents the correlation matrix of the variables used in Model 6 developed in
research strategy section. Current firm performance seems to be positively correlated with
future firm performance with correlation coefficient of 0.5697. This seems to indicate that
currently well performing firms are also likely to perform well in future. Both cash and total
compensations (LNCASHCOMP and LNTCOMP) also appear to have positive correlation
with ROAt+1 and coefficients are significant with values of 0.2720 and 0.2738 respectively.
This indicates some motivation aspects of CEO compensation to perform good in future.
Current total returns to shareholders also seems to have significant positive correlation with
ROAt+1 (coefficient is 0.1707). With regard corporate governance variables, ownership
concentration (OWNCONS) and board size (BDSIZE) are positively correlated with future
firm performance with correlation coefficients of 0.2144 and 0.1155 respectively while
family ownership (FAMOWN) and CEO duality are negatively correlated with future
performance with coefficients of -0.2064 and -0.1607. However, board independence does
not seem to have any significant relationship with future firm performance as indicated by its
insignificant correlation coefficient with ROAt+1. As far as firm specific variables are
concerned, LNSALES and MTB have significant positive correlation while FIRMRSK is
significantly negatively correlated with future firm performance.
4.3.1.1 Addressing Multicollinearity
The correlation matrix (see Table 4.8) shows that none of the absolute values of
correlation coefficients is greater than normally used threshold level of 0.70 for potential
serious problem of multicollinearity. To further confirm that no multicollinearity problem
exists, variance inflation factor (VIF) and tolerance are estimated. Table 4.9 indicates that
none of the VIF is greater than generally used threshold level of 10 and mean of VIFs is 1.45
which is not considerably different from one. Similarly, all the tolerance levels are above the
threshold level of 0.10. Thus, the tests suggest that variables of Model 6 can be used
simultaneously in one regression model without the fear of multicollinearity problem.
126
Table 4.8 Correlation Matrix (Variables used in Model 6)
Variables
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13)
ROAt+1 (1) 1.0000 ROA (2) 0.5697* 1.0000
TRET (3) 0.1707* 0.2096* 1.0000 LNTCOMP (4) 0.2720* 0.3264* -0.0511 1.0000
LNCASHCOMP (5) 0.2738* 0.3261* -0.0452 0.9721* 1.0000 OWNCONS (6) 0.2144* 0.2339* 0.0482 0.3693* 0.3802* 1.0000
FAMOWN (7) -0.2064* -0.2269* -0.0001 -0.3852* -0.3861* -0.5578* 1.0000 BDSIZE (8) 0.1155* 0.1393* 0.0233 0.3205* 0.3134* 0.1107* -0.4468* 1.0000
B_IND (9) 0.0240 0.0356 -0.0320 0.1762* 0.1330* 0.0588* -0.2278* 0.3578* 1.0000 DUALCEO (10) -0.1607* -0.1670* 0.0044 -0.3048* -0.2859* -0.0823* 0.1690* -0.2594* -0.1807* 1.0000
LNSALES (11) 0.2704* 0.3313* -0.0340* 0.6130* 0.5926* 0.2748* -0.3240* 0.3560* 0.1231* -0.2328* 1.0000 MTB (12) 0.1625* 0.1560* 0.0853* 0.1031* 0.1289* 0.2814* -0.1717* 0.0836* -0.1050* -0.0971* 0.1196* 1.0000
FIRMRSK (13) -0.1236* -0.1777* 0.1742* -0.1956* -0.1989* -0.0335 0.1213* -0.1289* -0.0462 0.1877* -0.2591* -0.0905* 1.0000
(*) p-value < 0.05
127
Table 4.9 Variance Inflation Factor (Variables used in Model 6)
Variable VIF Tolerance
FAMOWN 1.90 0.525556
LNTCOMP 1.88 0.530793
LNSALES 1.78 0.561353
OWNCONS 1.72 0.582379
BDSIZE 1.57 0.63807
ROA 1.28 0.779931
B_IND 1.20 0.830764
DUALCEO 1.18 0.848947
FIRMRSK 1.16 0.862592
MTB 1.14 0.874929
TRET 1.12 0.890896
Mean VIF 1.45
4.3.2 Regression Results: Compensation and Future Firm Performance
Table 4.10 reports the results of Model 6 for future firm performance as dependent
variable and current CEO compensation and corporate governance variables as explanatory
variables. The standard errors are adjusted for heteroskedasticity and serial correlation in
error terms.
There seems to be no evidence that CEO compensation has any influence on future firm
performance as coefficients of compensation are insignificant consistently in all models. So,
it can confidently be assumed that CEO compensation unexplained by current observable
firm performance has no influence on observable future firm performance. However, it may
also be inferred that CEOs in Pakistan are compensated only against realized firm
performance but not against current unobservable performance. The results are inconsistent
with other studies ((e.g., Adithipyangkul et al., 2011, Balafas and Florackis, 2014, Hayes and
Schaefer, 2000). These inconsistent results might be due to higher unpredictability of future
performance in Pakistan given the political and economic situation. Nevertheless, the results
do not support hypothesis 7 that CEO compensation is positively related to future firm
performance, leading to rejection of hypothesis 7. This implies that CEO pay packages are
not designed to compensate for unobservable future firm performance. The boards of
directors in Pakistan reward CEOs for realized firm accounting performance only.
Ownership concentration does not influence future firm performance as measured by
ROA. Similar finding is reported by Balafas and Florackis (2014) in UK. However, there is
weak evidence that family ownership has some negative influence on future firm
128
performance as indicated by significant negative coefficients in column 3 & 6. Board size and
board independence do not seem to have any influence on future ROA however, CEO duality
does seem to have weak negative relationship with future ROA in pooled regression in both
tables. In a similar study in UK context, Balafas and Florackis (2014) find strong negative
effect of board size, moderate negative effect of ratio of non-executive director to board size
and no effect of CEO duality on future performance as measured by ROA. Overall the current
study finds that corporate governance variables seem to have very little influence on future
ROA.
Other control variables such as LNSALES and MTB do not seem to have convincing
relationship with future ROA however negative relationship is found between FIRMRSK and
future ROA in industry fixed model.
4.3.3 Robustness Checks
Although results presented are robust across different model specifications however, to
further reconfirm, certain robustness checks are conducted. First, all the continuous variables
are winsorized using 1% level at both tails to eliminate potential outliers and all models are
re-estimated. But, the results do not change qualitatively therefore it is decided to report the
original data results. Second, alternative measures of firm size and ownership concentration
as log of net sales and voting shares held by three largest shareholders respectively, are
incorporated. Again, the results remain qualitatively similar to as reported above.
129
Table 4.10
CEO Compensation and Future Firm Performance Dependent variables is one year forward ROA as measured by earnings before interest and taxes divided by total assets,
LNCASHCOMP = log of cash compensation as measured by sum basic salary and cash bonuses, LNTCOMP = log of total
compensation as measured by sum of all the pay components, TRET = total return to shareholders as measured by current
market price plus dividend divided by previous year market price minus one, OWNCONS = concentrated ownership as
measured by voting shares held by the largest shareholder, FAMOWN = a dummy variable taking value one for family firms
and zero otherwise, BDSIZE = board size as measured by number of sitting directors on board, B_IND = board independence
as measured by ratio of non-executive directors to board size, DUALCEO = CEO duality a dummy variable taking value one if
CEO is also chairman board of directors, LNSALES = log of net sales, MTB = market to book ratio as measured by market
value divided by book value per share, FIRMRSK = firm risk as measured by standard deviation of monthly stock returns over
the fiscal year
One-year forward Return on Assets (ROA)
(1) (2) (3) (4) (5) (6)
VARIABLES Pooled Fixed effect
Industry and
Time
Fixed effect
Firm and
Time
Pooled Fixed effect
Industry and
Time
Fixed effect
Firm and
Time
LNCSHCOMPit 0.0052 0.0067 -0.0022
(0.0056) (0.0064) (0.0127)
LNTCOMPit 0.0053 0.0073 -0.0033
(0.0058) (0.0069) (0.0127)
ROAit 0.5138*** 0.4952*** 0.0244 0.5140*** 0.4949*** 0.0245
(0.1332) (0.1328) (0.1487) (0.1329) (0.1326) (0.1487)
TRETit 0.0137 0.0141 0.0215** 0.0138 0.0142 0.0215**
(0.0102) (0.0101) (0.0103) (0.0102) (0.0101) (0.0103)
OWNCONSit 0.0178 0.0246 0.1930 0.0183 0.0250 0.1934
(0.0244) (0.0249) (0.1989) (0.0244) (0.0249) (0.1976)
FAMOWNit -0.0124 -0.0119 -0.1136*** -0.0125 -0.0123 -0.1129***
(0.0098) (0.0119) (0.0160) (0.0098) (0.0120) (0.0154)
BDSIZEit -0.0019 -0.0001 0.0131 -0.0018 -0.0000 0.0131
(0.0030) (0.0030) (0.0094) (0.0029) (0.0030) (0.0094)
B_INDit -0.0068 -0.0006 -0.0373 -0.0081 -0.0022 -0.0370
(0.0164) (0.0166) (0.0383) (0.0166) (0.0168) (0.0382)
DUALCEOit -0.0130* -0.0107 -0.0043 -0.0129* -0.0106 -0.0046
(0.0076) (0.0076) (0.0170) (0.0076) (0.0077) (0.0171)
LNSALESit 0.0044 0.0045 0.0162 0.0043 0.0043 0.0164
(0.0037) (0.0044) (0.0131) (0.0037) (0.0045) (0.0132)
MTBit 0.0013 0.0015 0.0006 0.0013 0.0015* 0.0006
(0.0010) (0.0009) (0.0005) (0.0010) (0.0009) (0.0005)
FIRMRSKit -0.0101 -0.0423* 0.0310 -0.0107 -0.0432* 0.0308
(0.0207) (0.0228) (0.0460) (0.0207) (0.0228) (0.0458)
Observations 1,277 1,277 1,277 1,277 1,277 1,277
R-squared 0.3475 0.3660 0.5638 0.3473 0.3660 0.5639
Industry FE NO YES NO NO YES NO
Firm FE NO NO YES NO NO YES
Time FE NO YES YES NO YES YES
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
130
Chapter 5
CEO Compensation and Earnings Management: Empirical
Results and Discussion
131
5.1 CEO Compensation and Earnings Management
5.1.1 Descriptive Statistics
Table 5.1 presents the descriptive statistics of variables used in Model 7. Pooled average
of absolute discretionary accruals (|DAC|) is 7.93 percent with standard deviation of 12.19.
Average |DAC| reaches to its maximum level in year 2008 with value of 11.55 percent and
standard deviation of 28.10 percent. This is possibly because of three reasons: 1) effects of
financial crisis, 2) unrest due to political issues and general elections and 3) start of energy
crisis in Pakistan, leading to more camouflaged earnings. Apart from that, overall trend is
mixed over the sample period.
Both average total and cash compensation have gradually increased and nearly tripled
over the sample period. Total compensation has increased from Rs4.817 million in 2005 to
Rs13.060 in 2012 while cash compensation has increased from Rs3.353 million in 2005 from
Rs.9.378 million in 2012. Consistently lower median value than mean value indicates that the
distribution of compensation is positively skewed meaning that greater number of CEOs is
receiving pay that is less than overall average pays.
Average board size of pooled sample is slightly above 8 with standard deviation of 1.57.
Recently, similar average board size of 8 is reported for Indian firms (see, Jameson et al.,
2014). This is possibly due to resemblance in institutional setting in India and Pakistan which
is characterized by concentrated and family ownership structure. The average board size in
Pakistan is lower than the board size recently reported for US (mean 9.54 and median 9) and
China (mean 9.372 and median 9) where ownership is either widely held or state has the
major stake in the firms (see, Conyon, 2014, Huang and Wang, 2015). On average, firms
seem to keep 63% non-executive directors on the boards. However, B_IND has slightly
downward trend over time. This is seemingly due to decrease in board size over time. The
non-executive directors are more likely to be an easy target when board size needs to be
reduced.
In our sample, about 75% observations are from family firms and, quite expectedly, this ratio
is stable overtime as family firm in year 2005 is very likely to remain a family firm in year
2012. In addition, more than 30% average voting shares held by the largest shareholder
indicates a highly concentrated ownership environment. Interestingly, OWNCONS has
slightly increasing trend over time which could have implications for earnings management
behavior. On average about 34% CEOs also hold the position of chairman board of directors.
Interestingly, DUALCEO shows maximum value i.e. 36% in 2008, 2009 and 2010, the time
132
characterized by financial crisis, energy crisis and political change. However, recent
downward trend in 2011 and 2012 is consistent with more emphasis on strong corporate
governance practices in Pakistan. This is similar to the UK and China where emphasis in on
separating the post of CEO from the chairman, and unlike the US where it is usual to
combine these two positions (Conyon and He, 2012).
5.1.2 Correlation Analysis
Table 5.2 shows the correlation matrix for the variables included in Model 7 in chapter 3.
The correlation between absolute total accruals (|TAC|) and absolute discretionary accruals
(|DAC|) is positive with value of 0.8266. Only firm size as measured by log of total assets
and leverage as measured by debt to total assets ratio have some significant correlation with
both |TAC| and |DAC|. FIRMSIZE is negatively correlated with |TAC| and |DAC| with
coefficient values of -0.0712 and -0.0895 respectively, indicating that larger firms are more
likely to not engage in earnings management activities. LEVERAGE has some significantly
positive correlation with proxies of earnings management with values of 0.0701 and 0.0557
for |TAC| and |DAC| respectively. This gives the impression that higher debt firms have more
incentives to manage earnings to show better picture of the company to debt financers who
require companies to achieve certain landmarks.
Surprisingly, all other variables including log of cash compensation and log of total
compensation do not appear to have any significant correlation with absolute total accruals or
absolute discretionary accruals.
5.1.2.1 Addressing Multicollinearity
The correlation matrix in Table 5.2 shows that none of the absolute values of correlation
coefficients is greater than normally used threshold level of 0.70 for potential serious problem
of multicollinearity. To further confirm that no multicollinearity problem exists, variance
inflation factor (VIF) and tolerance are estimated. Table 5.3 indicates that none of the VIF is
greater than generally used threshold level of 10 and mean of VIFs is 1.51 which is less than
2. Similarly, all the tolerance levels are above the threshold level of 0.10. These diagnostics
suggest that Model 7 can be estimated without worrying about problem of multicollinearity.
133
Table 5.1 Descriptive Statistics (Earnings Management and other variables)
Variables Statistic 2005 2006 2007 2008 2009 2010 2011 2012 Overall
|DAC| Mean N/A 7.16 7.36 11.55 7.47 7.22 7.75 7.52 7.93
%age Median N/A 5.28 5.66 6.45 4.99 5.20 5.87 5.20 5.49
S.D N/A 6.13 8.44 28.10 8.58 6.29 7.48 8.43 12.19
Total Compensation Mean 4817 5530 6512 7562 8396 9449 10628 13060 8475
In Rs.’000 Median 3156 3384 4153 4553 4999 5459 6060 7519 4800
S.D 5854 7030 9094 11028 11278 12519 13493 19935 12538
Cash Compensation Mean 3353 3918 4751 5529 6004 6577 7523 9378 6042
In Rs.‘000 Median 2000 2303 2715 2836 3435 3450 4001 5186 3174
S.D 4626 5840 7786 9268 8867 9690 10460 14766 9733
BDSIZE Mean 8.13 8.15 8.08 8.04 8.01 8.00 8.01 8.00 8.05
Median 7.00 7.00 7.00 7.00 7.00 7.00 7.00 7.00 7.00
S.D 1.68 1.68 1.62 1.57 1.53 1.51 1.51 1.51 1.57
B_IND Mean 64.32 64.31 63.90 63.65 62.85 63.23 62.58 63.91 63.54
%age Median 70.00 70.71 70.00 66.67 66.67 66.67 62.50 66.67 66.67
S.D 20.37 20.72 20.71 20.59 20.94 20.64 20.31 19.96 20.49
FAMOWN Mean 0.75 0.75 0.76 0.75 0.74 0.75 0.75 0.74 0.75
Median 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00
S.D 0.44 0.43 0.43 0.43 0.44 0.43 0.43 0.44 0.43
OWNCONS Mean 32.72 33.01 33.07 33.23 33.36 33.61 34.39 35.51 33.67
%age Median 26.46 26.61 26.71 26.25 26.13 26.13 27.45 29.40 26.66
S.D 21.01 21.22 20.78 20.48 20.44 20.30 20.74 21.32 20.74
DUALCEO Mean 0.30 0.33 0.33 0.36 0.36 0.36 0.35 0.33 0.34
Median 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
S.D 0.46 0.47 0.47 0.48 0.48 0.48 0.48 0.47 0.47
TASSETS Mean 7173 8607 9827 10898 11636 12675 14490 17215 11833
In Rs. Millions Median 2385 2603 2970 3042 3034 3068 3544 3831 3061
S.D 14281 16821 18677 22158 26033 31098 35893 45860 29092
LEVERAGE Mean 0.63 0.60 0.64 0.65 0.66 0.63 0.62 0.57 0.62
Times Median 0.63 0.64 0.63 0.64 0.64 0.62 0.60 0.56 0.62
S.D 0.31 0.38 0.44 0.39 0.35 0.32 0.31 0.42 0.37
CFO Mean 911 799 721 632 1024 1266 941 1020 926
Rs in Median 73 154 108 31 144 175 81 182 113
Millions S.D 3434 3982 3417 4951 4668 5057 5522 4932 4617
ROA Mean 6.96 6.36 4.42 4.01 2.29 5.76 5.18 4.42 4.84
%age Median 5.12 4.46 2.96 2.79 1.93 4.97 4.75 4.38 3.80
S.D 8.52 8.61 9.31 9.09 18.26 9.23 11.22 13.25 11.64
MTB Mean 1.79 2.35 2.17 1.74 1.56 1.31 1.30 1.84 1.73
Times Median 1.19 0.99 1.13 0.86 0.62 0.52 0.43 0.65 0.75
S.D 2.37 8.17 4.76 4.77 4.81 6.03 4.58 5.62 5.35
LOSSDUM Mean 0.11 0.16 0.23 0.27 0.31 0.21 0.23 0.24 0.22
Median 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
S.D 0.31 0.37 0.42 0.45 0.46 0.41 0.42 0.43 0.42
134
Table 5.2 Correlation Matrix (Variables used in Model 7)
Variables (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14)
|TAC| (1) 1.0000
|DAC| (2) 0.8266* 1.0000
LNCASHCOMP (3) -0.0209 -0.0258 1.0000
LNTCOMP (4) -0.033 -0.0409 0.9722* 1.0000
OWNCONS (5) 0.0223 -0.0255 0.3651* 0.3555* 1.0000
FAMOWN (6) -0.0252 0.0087 -0.3760* -0.3775* -0.5435* 1.0000
BDSIZE (7) 0.0211 0.0186 0.3138* 0.3215* 0.1007* -0.4532* 1.0000
B_IND (8) 0.0161 -0.0002 0.1438* 0.1872* 0.0638* -0.2273* 0.3538* 1.0000
DUALCEO (9) -0.0459 -0.0279 -0.2777* -0.2944* -0.0841* 0.1730* -0.2605* -0.1863* 1.0000
FIRMSIZE (10) -0.0712* -0.0895* 0.5614* 0.5781* 0.2495* -0.2750* 0.3795* 0.1320* -0.2164 1.0000
LEVERAGE (11) 0.0701* 0.0557* -0.1944* -0.1912* -0.0873* 0.0889* -0.0435 -0.0205 0.1877* -0.1164* 1.0000
CFO (12) -0.0402 -0.0534 0.1331* 0.1499* 0.2235* -0.1975* 0.2435* 0.1382* -0.0983* 0.3226* -0.1184* 1.0000
ROA (13) -0.0455 -0.0438 0.2893* 0.2886* 0.1948* -0.1996* 0.1304* 0.0346 -0.1816* 0.1578* -0.2351* 0.2838* 1.0000
MTB (14) 0.0061 0.0083 0.1197* 0.0932* 0.2793* -0.1730* 0.0803* -0.1189* -0.0949* 0.0802* -0.0436 0.0452 0.1613* 1.0000
135
Table 5.3 Variance Inflation Factor (Variables used in Model 7)
Variable VIF Tolerance
LNTCOMP 1.87 0.5335
OWNCONS 1.71 0.5831
FAMOWN 1.88 0.532
BDSIZE 1.65 0.6056
B_IND 1.22 0.8217
DUALCEO 1.19 0.843
FIRMSIZE 1.75 0.5727
LEVERAGE 1.15 0.872
CFO 1.28 0.7793
NROA 1.74 0.5763
MTB 1.15 0.8729
LOSSDUM 1.57 0.6355
Mean VIF 1.51
5.1.3 Regression Results: Compensation and Earnings Management
Table 4.13 exhibits the results for compensation driven earnings management (EM)
behavior taking absolute discretionary accruals as proxy for EM estimated by Jones (1991)
Model modified by Dechow et al. (1995) first and then Kothari et al. (2005). The results of
pooled and fixed effect models are reported using Huber-White heteroskedasticity consistent
standard errors clustered by firms.
There seems to be no evidence that earnings management is driven by CEO cash or total
compensation. The coefficients of LNCASHCOMP and LNTCOMP are insignificant in all
six columns. Thus, the results do not seem to support the hypothesis 8 developed in chapter 3
that CEO compensation is positively related to earnings management behavior, resulting in
rejection of hypothesis 8. These results are inconsistent with other studies on the relationship
between executive pay and earnings management such as Healy (1985), Holthausen et al.
(1995), Guidry et al. (1999), Balsam (1998) and Ye (2014). One possible reason for this
might be the less use of cash bonuses in compensation contracts. In Table 3.2, only 24.80%
observations have cash bonuses as part of CEO compensation plan. Therefore, earnings
management does not seem to provide extra pecuniary benefits to CEOs in term of extra
bonuses presumably based on earnings.
Although the results in previous chapter suggest that CEO compensation is tied to firm
accounting performance however, despite that, this study does not find any evidence of
136
higher compensation as a motive to manage earnings in either direction (upward or
downward).
With regard to corporate governance variables, Ownership concentration, family
ownership and board independence do not seem to have any influence on earnings
management behavior. This finding is consistent over all six columns as coefficients of
OWNCONS, FAMOWN and B_IND are consistently insignificant. This means that large
shareholders or family owners are not significantly engaged in earnings management
activities in Pakistan. Kamran and Shah (2014) find similar results in Pakistan for
OWNCONS using different proxy for ownership concentration. Insignificant impact of
concentrated/family ownership seems to be contradicting with existing evidence in Asian
context. For example, Fan and Wong (2002) find that ownership concentration is related to
low earnings informativeness in 7 east Asian countries while Ye (2014) find increased
earnings management with the increase in ownership concentration.
Board size has some positive influence on earnings management. Firms with larger
boards seem to be engaged in earnings management more than their counterparts. This might
be indicating ineffectiveness of larger boards. DUALCEO has some significant coefficients
with negative sign for both cash and total CEO pay when industry and time fixed effects are
used. This means CEO duality reduces earnings management behavior. The negative
association between CEO duality and EM is more pronounced (unreported) when
discretionary accruals are estimated using Dechow et al. (1995) modification of Jones model.
The results of BDSIZE and DUALCEO are inconsistent with Kamran and Shah (2014) as
they find no relationship of these variables with discretionary accruals. This might be due to
different sample size and data period used15.
Larger firms appear to have less earnings management activities as coefficients of
FIRMSIZE are significant with negative sign in all models except when industry fixed effects
are used. This is consistent with the argument that larger firms are more visible and more
people follow them, therefore engage in less earnings management as compared to small
firms (see, Kamran and Shah, 2014). LEVERAGE has some positive association with EM in
pooled regression for both cash and total compensation, indicating some evidence of debt
covenant driven EM. All other variables such as CFO, ROA, MTB and LOSSDUM do not
have significant influence on EM behavior.
15 Kamran and Shah (2014) use 986 firm-year observations for period from 2003 to 2010.
137
To get more insight of the relationship between CEO compensation and earnings
management, absolute discretionary accruals are divided into two components, income
increasing and income decreasing discretionary accruals, on the basis of the sign of estimated
discretionary accruals. Then regression models are run taking income increasing and income
decreasing discretionary accruals as dependent variables separately. Table 5.5 reports the
results for income increasing discretionary accruals while Table 5.6 shows the results for
income decreasing discretionary accruals.
5.1.3.1 Income Increasing Discretionary Accruals
In Table 5.5, higher LNTCOMP appears to be negatively related to income increasing
discretionary accruals, indicating that CEOs with higher total compensation avoid costly
decision of exaggerating the firm performance. However, coefficients are significant at 10%
level of significance, showing moderate level of relationship. LNCASHCOMP appears to
have no relationship with income increasing EM behavior in any model. Again the results are
inconsistent with existing studies (e.g., Balsam, 1998, Holthausen et al., 1995, Ye, 2014).
This might be an indication that costs are higher than the benefits associated to higher income
increasing discretionary accruals for the firm, leading to negative impact on executive
compensations16.
With regard to corporate governance variables, only DUALCEO seem to have some
relationship with income increasing EM behavior with negative sign, meaning that CEO
duality leads to lower exaggerated earnings. This is seemingly contradicting with managerial
power approach. No other corporate governance variable has significant impact on income
increasing EM.
Higher amount of debt leads to higher reported earnings as indicated by strong positive
relationship between LEVERAGE and income increasing discretionary accruals. This is an
indication of debt convents driven earnings management. Larger firms do not appear to
report exaggerated earnings as coefficients of FIRMSIZE are insignificant in all models.
Higher amount of debt leads to higher reported earnings as indicated by strong positive
relationship between LEVERAGE and income increasing discretionary accruals. This is an
indication of debt convents driven earnings management. Larger firms do not appear to
report exaggerated earnings as coefficients of FIRMSIZE are insignificant in all models.
16 The costs may include inability to set aside reserves due to maximizing reported income, tax costs and higher future targets due higher current performance (Balsam, 1998)
138
5.1.3.2 Income Decreasing Discretionary Accruals
In Table 5.6, firms with higher cash compensation seem to have greater income
decreasing discretionary accruals, inconsistent with Ye (2014) for independent directors. The
coefficients LNCASHCOMP are statistically significant in pooled and industry fixed effect
models. The significant positive relationship between income decreasing accruals and total
compensation can also be seen but the coefficient of LNTCOMP is significant only in pooled
regression. This might be an indication of more conservative accounting in firms with highly
paid CEOs. Another possible reason for this relationship can be deduct from contextual
setting of Pakistan. Given the concentrated family ownership environment in Pakistan, the
possibility of conflicts between controlling family and non-family shareholders is high.
Therefore, highly paid executives, being under family obligation, may engage in
manipulating earnings through income decreasing discretionary accruals in order to meet
long-term family expectations instead of achieving short-term objectives.
Family owned firms and firms with larger boards appear to have more income decreasing
discretionary accruals, consistent with Ali et al. (2007) in the US context. However, the
coefficients of FAMOWN are significant only in pooled regression. Ali et al. (2007) find
more negative (income decreasing) discretionary accruals for family firms in US. The family
firms may be motivated to increase income decreasing discretionary accruals by their desire
to minimize taxes or reduce political cost (Ali et al., 2007). Firms with larger boards also
appear to have more income decreasing accruals as the coefficients of BDSIZE are
significant in the pooled and industry fixed effect models. Other variables of corporate
governance have no impact on income decreasing discretionary accruals.
The coefficients of FIRMSIZE are significantly negative across all models, indicating
that larger firms do not encourage income decreasing accruals. LEVERAGE also does have
some significantly negative coefficients, suggesting that high debt leads to higher reported
income, leading to lower income decreasing discretional accruals. This is consistent with the
finding in case of income increasing discretionary accruals where LEVERAGE is
significantly positive.
5.1.4 Robustness Checks
Certain robustness checks are conducted to further confirm the results. First, all the
continuous variables are winsorized using 1% level at both tails to eliminate potential outliers
and all models are re-estimated. But, the results do not change qualitatively therefore it is
decided to report the original data results. Second, discretionary accruals are also estimated
139
using original Jones (1991) model and modification of Jones Model by Dechow et al. (1995).
However, the results remain qualitatively similar to the results reported above except that the
negative relationship between CEO duality and absolute discretionary accruals is more
pronounced. This relatively more pronounced result has been discussed earlier as well.
Table 5.4
CEO Compensation and Earnings Management Dependent variables is absolute discretionary accruals estimated using Kothari et al. (2005) modification of Jones Model,
LNCASHCOMP = log of cash compensation as measured by sum basic salary and cash bonuses, LNTCOMP = log of total
compensation as measured by sum of all the pay components, OWNCONS = concentrated ownership as measured by voting
shares held by the largest shareholder, FAMOWN = a dummy variable taking value one for family firms and zero otherwise,
BDSIZE = board size as measured by number of sitting directors on board, B_IND = board independence as measured by ratio
of non-executive directors to board size, DUALCEO = CEO duality a dummy variable taking value one if CEO is also
chairman board of directors, FIRMSIZE = firm size as measured by log of total assets, LEVERAGE = leverage as measured by
debt to total asset ratio, CFO = operating cash flows taken from cash flow statements, ROA = return on assets as measured by
net income divided by total assets, MTB = market to book ratio as measured by market value divided by book value per share,
LOSSDUM = a dummy variable taking value one if company is in loss.
Absolute Discretionary Accruals
(1) (2) (3) (4) (5) (6)
VARIABLES Pooled Fixed effect
Industry and
Time
Fixed effect
Firm and
Time
Pooled Fixed effect
Industry and
Time
Fixed effect
Firm and
Time
LNCSHCOMPit 0.00421 0.000715 0.00440
(0.00317) (0.00285) (0.0114)
LNTCOMPit 0.00173 -0.00217 -0.00150
(0.00359) (0.00360) (0.0117)
OWNCONSit -0.00319 0.00257 0.0137 0.000279 0.00551 0.0146
(0.0230) (0.0182) (0.0600) (0.0214) (0.0169) (0.0602)
FAMOWNit 0.00316 0.000637 0.0197 0.00257 0.000560 0.0227
(0.00983) (0.00928) (0.0169) (0.00981) (0.00931) (0.0169)
BDSIZEit 0.00472* 0.00519** -0.00705 0.00488* 0.00528** -0.00699
(0.00257) (0.00225) (0.0103) (0.00262) (0.00227) (0.0102)
B_INDit -0.00773 0.00116 -0.0476 -0.00789 0.00174 -0.0453
(0.0150) (0.0138) (0.0439) (0.0148) (0.0139) (0.0433)
DUALCEOit -0.0128 -0.0147* -0.0117 -0.0133 -0.0151* -0.0120
(0.00868) (0.00811) (0.00803) (0.00903) (0.00829) (0.00828)
FIRMSIZEit -0.0100*** -0.00370 -0.0345* -0.00908*** -0.00262 -0.0326*
(0.00322) (0.00289) (0.0190) (0.00313) (0.00291) (0.0185)
LEVERAGEit 0.0164** 0.00431 0.00746 0.0156** 0.00368 0.00704
(0.00670) (0.00613) (0.00792) (0.00693) (0.00653) (0.00784)
CFOit -4.16e-10 -3.12e-10 -1.79e-09 -5.30e-10 -4.02e-10 -1.78e-09
(8.14e-10) (8.12e-10) (1.72e-09) (8.03e-10) (8.03e-10) (1.71e-09)
ROAit -0.0410 -0.0694 -0.0627 -0.0364 -0.0653 -0.0615
(0.0731) (0.0623) (0.0962) (0.0742) (0.0631) (0.0959)
MTBit 0.000323 -0.000121 0.00117 0.000320 -0.000155 0.00120
(0.000959) (0.000506) (0.00153) (0.000966) (0.000494) (0.00156)
LOSSDUMit -0.00166 0.000298 -0.00324 -0.00167 0.000209 -0.00336
(0.0176) (0.0161) (0.0174) (0.0177) (0.0160) (0.0172)
Observations 1,247 1,247 1,247 1,247 1,247 1,247
R-squared 0.018 0.090 0.252 0.018 0.090 0.252
Industry FE NO YES NO NO YES NO
Firm FE NO NO YES NO NO YES
Time FE NO YES YES NO YES YES
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
140
Table 5.5
CEO Compensation and Income Increasing Discretionary Accruals Dependent variables is absolute income increasing discretionary accruals estimated using Kothari et al. (2005) modification of
Jones Model, LNCASHCOMP = log of cash compensation as measured by sum basic salary and cash bonuses, LNTCOMP =
log of total compensation as measured by sum of all the pay components, OWNCONS = concentrated ownership as measured
by voting shares held by the largest shareholder, FAMOWN = a dummy variable taking value one for family firms and zero
otherwise, BDSIZE = board size as measured by number of sitting directors on board, B_IND = board independence as
measured by ratio of non-executive directors to board size, DUALCEO = CEO duality a dummy variable taking value one if
CEO is also chairman board of directors, FIRMSIZE = firm size as measured by log of total assets, LEVERAGE = leverage as
measured by debt to total asset ratio, CFO = operating cash flows taken from cash flow statements, ROA = return on assets as
measured by net income divided by total assets, MTB = market to book ratio as measured by market value divided by book
value per share, LOSSDUM = a dummy variable taking value one if company is in loss.
Absolute Income Increasing Discretionary Accruals
(1) (2) (3) (4) (5) (6)
VARIABLES Pooled Fixed effect
Industry and
Time
Fixed effect
Firm and
Time
Pooled Fixed effect
Industry and
Time
Fixed effect
Firm and
Time
LNCSHCOMPit -0.00648 -0.00737 -0.00839
(0.00436) (0.00471) (0.0133)
LNTCOMPit -0.00864* -0.00912* -0.0152
(0.00471) (0.00471) (0.0165)
OWNCONSit -0.0159 -0.00119 0.0601 -0.0147 -0.000993 0.0628
(0.0222) (0.0211) (0.0804) (0.0219) (0.0207) (0.0819)
FAMOWNit -0.00592 -0.00403 -0.0154 -0.00675 -0.00403 -0.0341
(0.0138) (0.0145) (0.0617) (0.0137) (0.0144) (0.0683)
BDSIZEit 0.00423 0.00383 -0.0117 0.00417 0.00370 -0.0116
(0.00303) (0.00299) (0.0114) (0.00300) (0.00299) (0.0114)
B_INDit -0.0182 -0.0102 -0.0214 -0.0152 -0.00753 -0.0196
(0.0180) (0.0175) (0.0426) (0.0179) (0.0174) (0.0426)
DUALCEOit -0.0145* -0.0170** -0.0117 -0.0152** -0.0174** -0.0140
(0.00751) (0.00787) (0.0211) (0.00752) (0.00782) (0.0220)
FIRMSIZEit -0.00263 0.00191 0.0132 -0.00180 0.00259 0.0144
(0.00374) (0.00418) (0.0261) (0.00377) (0.00417) (0.0265)
LEVERAGEit 0.0372*** 0.0278*** 0.0372* 0.0369*** 0.0279*** 0.0375*
(0.00716) (0.00678) (0.0202) (0.00713) (0.00675) (0.0203)
CFOit -6.07e-09** -5.87e-09** -9.81e-09* -6.12e-09** -5.91e-09** -9.74e-09*
(2.53e-09) (2.46e-09) (5.67e-09) (2.52e-09) (2.45e-09) (5.63e-09)
ROAit 0.302*** 0.262*** 0.423*** 0.309*** 0.266*** 0.421***
(0.0701) (0.0725) (0.0905) (0.0694) (0.0720) (0.0895)
MTBit -0.00137*** -0.00192*** 0.000720 -0.00142*** -0.00195*** 0.000827
(0.000503) (0.000707) (0.00263) (0.000512) (0.000689) (0.00270)
LOSSDUMit 0.00593 0.00603 0.00516 0.00651 0.00642 0.00550
(0.0119) (0.0119) (0.0167) (0.0119) (0.0118) (0.0166)
Observations 617 617 617 617 617 617
R-squared 0.140 0.235 0.552 0.142 0.237 0.553
Industry FE NO YES NO NO YES NO
Firm FE NO NO YES NO NO YES
Time FE NO YES YES NO YES YES
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
141
Table 5.6
CEO Compensation and Income Decreasing Earnings Management
Dependent variables is absolute income decreasing discretionary accruals estimated using Kothari et al. (2005)
modification of Jones Model, LNCASHCOMP = log of cash compensation as measured by sum basic salary and
cash bonuses, LNTCOMP = log of total compensation as measured by sum of all the pay components,
OWNCONS = concentrated ownership as measured by voting shares held by the largest shareholder, FAMOWN =
a dummy variable taking value one for family firms and zero otherwise, BDSIZE = board size as measured by
number of sitting directors on board, B_IND = board independence as measured by ratio of non-executive
directors to board size, DUALCEO = CEO duality a dummy variable taking value one if CEO is also chairman
board of directors, FIRMSIZE = firm size as measured by log of total assets, LEVERAGE = leverage as measured
by debt to total asset ratio, CFO = operating cash flows taken from cash flow statements, ROA = return on assets
as measured by net income divided by total assets, MTB = market to book ratio as measured by market value
divided by book value per share, LOSSDUM = a dummy variable taking value one if company is in loss.
Absolute Income Decreasing Discretionary Accruals
(1) (2) (3) (4) (5) (6)
VARIABLES Pooled Fixed effect
Industry and
Time
Fixed effect
Firm and
Time
Pooled Fixed effect
Industry and
Time
Fixed effect
Firm and
Time
LNCSHCOMPit 0.0122*** 0.00651* 0.0321
(0.00334) (0.00351) (0.0246)
LNTCOMPit 0.00834** 0.000604 0.0221
(0.00406) (0.00670) (0.0182)
OWNCONSit 0.0207 0.0176 0.148 0.0273 0.0253 0.149
(0.0338) (0.0297) (0.119) (0.0301) (0.0258) (0.117)
FAMOWNit 0.0192* 0.0113 -0.0183 0.0178* 0.0104 -0.0152
(0.0104) (0.00900) (0.0434) (0.0100) (0.00917) (0.0410)
BDSIZEit 0.00502* 0.00687** -0.00334 0.00553* 0.00728*** -0.00254
(0.00280) (0.00276) (0.00849) (0.00288) (0.00276) (0.00845)
B_INDit -0.000201 0.0115 -0.0379 -0.00142 0.0116 -0.0337
(0.0169) (0.0162) (0.0463) (0.0167) (0.0165) (0.0452)
DUALCEOit -0.0165 -0.0200 -0.0247 -0.0170 -0.0209 -0.0250*
(0.0144) (0.0158) (0.0159) (0.0151) (0.0164) (0.0151)
FIRMSIZEit -0.0172*** -0.00908*** -0.0440*** -0.0158*** -0.00697* -0.0418**
(0.00356) (0.00327) (0.0167) (0.00323) (0.00394) (0.0167)
LEVERAGEit -0.0120 -0.0255** -0.00312 -0.0136* -0.0272** -0.00465
(0.00735) (0.0124) (0.00949) (0.00772) (0.0131) (0.0103)
CFOit 1.95e-09*** 2.04e-09*** 6.34e-09** 1.69e-09** 1.78e-09*** 6.21e-09**
(7.20e-10) (6.23e-10) (2.60e-09) (6.93e-10) (6.17e-10) (2.75e-09)
ROAit -0.172*** -0.189*** -0.237*** -0.166*** -0.183*** -0.233***
(0.0509) (0.0441) (0.0621) (0.0535) (0.0463) (0.0648)
MTBit 0.000992 0.000630 0.00107 0.000999 0.000547 0.00105
(0.00135) (0.000520) (0.00201) (0.00134) (0.000533) (0.00200)
LOSSDUMit 0.0116 0.0173 0.0252 0.0117 0.0171 0.0258
(0.0278) (0.0280) (0.0322) (0.0278) (0.0277) (0.0324)
Observations 628 628 628 628 628 628
R-squared 0.045 0.103 0.405 0.042 0.101 0.403
Industry FE NO YES NO NO YES NO
Firm FE NO NO YES NO NO YES
Time FE NO YES YES NO YES YES
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
142
Chapter 6
Conclusions and Policy Implications
143
6.1 Conclusions
Family and concentrated ownerships are common in Pakistan and corporate governance
systems are weak but evolving. Given that there are chances of expropriation of minority
shareholders’ interest and a number of steps have been taken to reform the corporate
governance environment in Pakistan, it is important to examine how executives are rewarded
and incentivized. CEO compensation literature focused in developed economies may not
have implications for emerging economies due to differences in institutional contexts and
corporate governance practices.
Given the Pakistani context, the primary objective of the study is to examine the
relationship between CEO compensation and firm performance. In addition, the study also
explores how concentrated/ family ownership and board structure variables influence CEO
compensation, and how CEO compensation affect future firm performance. Moreover, the
relationship between CEO compensation and earnings management is also examined.
In Pakistan, long term incentive plans, stock options and restricted stocks are virtually
absent. CEO pay can be identified to have four forms: base salary, cash bonuses, perks and
benefits and post-employment benefits. Consistent with existing literature, this study uses
two measures of CEO compensation i.e., cash compensation and total compensation. Cash
compensation includes managerial remuneration and bonuses while total compensation is the
sum of all the components. Two measures of firm performance are used; one represents
accounting performance i.e. ROA and other represents market performance i.e. total return to
shareholders. Corporate governance variables include board size, board independence, CEO
duality, ownership concentration and family ownership. Other main control variables include
firm size, firm growth opportunities and firm risk.
The study uses an unbalanced panel data consisting of 1508 year-observations from 225
listed firms in KSE for period 2005 to 2012. Multiple regression models with fixed-effects
which account for unobserved heterogeneity are used. In addition, standards errors are
adjusted to account for heteroskedasticity and serial correlation in error terms. Moreover,
endogeneity is controlled by regressing CEO compensation on lagged explanatory variables
and by estimating a dynamic panel model using GMM techniques.
The study finds that current and past firm accounting performance as measured by ROA
positively contributes towards CEO compensation. This finding is consistent over different
model specifications and estimation techniques including a dynamic panel model estimated
using GMM. This is the main finding of this study. However, market performance as
144
measured by total return to shareholder does not have any effect on CEO pay setting process.
Although, there is some evidence that past market returns have some decreasing effect on
CEO compensation however, it is presumed that this negative association is by chance due to
high returns volatility and occasionally extra ordinary bearing and bullish trends in capital
markets of Pakistan. Overall, results suggest that firm accounting performance is an
important determinant of CEO pay process in Pakistan. Thus, despite the institutional context
characterized by powerful CEOs, expropriation of interests of minority shareholders by large
shareholders and weak legal environment, the boards of directors do account for accounting
performance while devising CEO compensation contracts.
Although CEOs in Pakistan are rewarded for current and past accounting performance
however this study does not find any conclusive evidence that CEOs are rewarded for
unobserved future firm performance. This implies that future expected firm performance does
not have any impact on CEO compensation until that future performance is realized. This is
believed to be the result influenced by higher unpredictability of future performance in
Pakistan because of prolonged volatile political and economic conditions.
Similarly, although CEO compensation is positively related to current firm accounting
performance however, the study does not find any convincing evidence that earnings
management is driven by any measure of CEO compensation. Thus, CEOs in Pakistan do not
manipulate earnings to have impact on their compensation.
An important by-product finding of this study is that boards of directors adjust CEO
compensation to target levels as they learn more about CEO actions over time. In dynamic
panel model estimation, CEO pay appears to be highly persistent and takes time to adjust to
its long-term equilibrium level. Most of the literature on CEO compensation ignores the
dynamic nature of CEO compensation and estimate static pay models considering that pay is
in equilibrium, thus, ignoring the CEO pay adjustment to long run equilibrium.
Pertaining to the influence of concentrated/family ownership on CEO compensation,
there is evidence that the largest shareholders have positive impact on CEO compensation.
This is an indication of some collusion between the largest shareholders and the executives to
get personal benefits at the expense of minority shareholders. CEOs in family firms seem to
receive lower compensation as compared to their counterparts. However, this finding is not
fully supported by all model specifications used in this study.
With regard to the board structure which is the primary focus in the Code of corporate
governance in Pakistan, this study finds that board size and board independence do not have
any impact on CEO compensation. This is consistent with Pakistani corporate governance
145
environment in that non-executive directors are generally hired from within family or they
are gray directors who do not have any influence on major corporate decisions. As the
number of non-executive director increases, board size increases, since non-executive
directors have no influence therefore board size remains ineffective.
Contrary to the expectations, this study finds that when CEOs also hold the position of
chairman, they agree upon receiving lower compensation. This is contrary to theoretical as
well as empirical literature in the area. Both agency theory and managerial power hypothesis
view that CEO duality should lead to higher CEO compensation. As these theories assume
that managers are self-interested therefore when managers gain power they behave
opportunistically and set their own pay higher. Nevertheless, this finding is considered to be
due to high family ownership in Pakistan. In family owned firms, the joint position of
chairman/CEO is generally held by founding father or the eldest sibling in the family because
they are considered to be the most respectful persons in the family. In return, being the eldest
and considered to be the most respectful person within the family, CEO acts as steward of the
family and works on collective basis while holding all the powers single handedly. Therefore,
it is against his social stature to negotiate for higher compensation or he might not give
weight to higher compensation against the power he enjoys. A split sample analysis confirms
that negative impact of CEO duality on CEO compensation is driven by family firms.
Firm size appears to be an important determinant of CEO compensation in Pakistan. This
supports the view that larger firms are complex and difficult to run and hence require quality
CEOs with higher compensations. The results of other control variables are not consistent
over different models and estimation techniques. At some instances, coefficients of these
variables change their signs and in other instances they lose their statistical significance.
Regarding corporate governance and future firm performance, ownership concentration,
board size and board independence do not have influence over future firm performance
however, there is weak evidence that family ownership and CEO duality have some negative
effect on future firm performance.
Concerning corporate governance and earnings management, ownership concentration,
family ownership and board independence do not have any influence on earnings
management behavior. However, firms with larger boards are more likely to engage in
earnings management which is an indication of ineffectiveness of larger boards. CEO duality
does appear to have influence in resisting earnings management behavior. Larger firms
appear to have less earnings management activities than their counterparts.
146
6.2 Policy Implications
Given the results, the study provides some important policy and practical implications.
First the study documents that base salary and perquisites are the most common forms in
which the CEOs are being paid in Pakistan. This indicates lack of incentive base
compensation in Pakistan. Internationally, CEO compensation is being shifted to take the
form of an optimal mix of base salary and incentive based remuneration. Short term
incentives include bonuses and other performance based benefits while long term incentives
include equity compensation in the form of stocks and stock options. There is evidence that
some firms are paying bonuses to CEO however equity compensation is virtually absent in
Pakistan. Equity compensation in the form of stocks and stock options is an important
incentive mechanism to align shareholder-manager interests in Western economies. Some of
the Asian countries have also become part of the global shift towards equity compensation as
a mechanism of reward and motivation. Regulators and policy makers in Pakistan may
consider the global shift as they develop optimal pay policies for CEOs. The introduction of
bonus plans and equity compensation is more likely to induce managers to perform better in
future as many studies report that after incentive based compensation are offered to
managers, the future firm performance increases.
Second, the negative relationship between CEO duality and CEO compensation is quite
unexpected in the Pakistani context. Considering that CEO duality is an inappropriate
corporate governance practice, recently, the Code of corporate governance has made
separation of two positions (chairman board of directors and CEO) mandatory in Pakistan.
However, the finding of this study indicates that CEO duality is consistent with protection of
minority shareholders’ interests in that it leads to less expropriation through excessive CEO
compensation. Therefore, this study suggests that policy makers and regulatory authorities
should consider conducting complete cost-benefit analysis of mandatory separation of the
two positions in revised Code of corporate governance in Pakistani context.
Third, the study finds that the number of non-executive directors and board size do not
have any influence on CEO compensation decisions. Given the Pakistani context, these
findings are quite expected as most of the non-executive directors are from within the
controlling family or they are gray directors having no influence on major corporate
decisions. Therefore, this study supports the recent steps taken by the regulatory authorities
emphasizing and making it mandatory to have more independent directors on the boards who
fulfill the recently issued definition of independent director. The mandatory board
147
independence would help reduce CEO entrenchment and probability of excessive executive
compensation.
Fourth, the study finds potential indications of expropriation of minority shareholders’
interests by the largest shareholders. Although recent reforms in corporate governance
environment of Pakistan are aimed at reducing all kinds of expropriations by the large
shareholders however this study suggests that besides ensuring best corporate governance
practices in the firms, the policy makers should also notice the decision making process
within the boards and the factors that influence decision making process including the largest
shareholders and family directors. This would help devise effective policies to reduce CEO
entrenchment, align mutual interests of large and small shareholders and reduce potential
expropriation by controlling shareholders.
Finally, the study finds that CEO pay is related to past pay levels and takes time to adjust
because of learning about CEO actions. In practice, boards should take into account the
incomplete information about CEO abilities/actions and adjustment costs along with firm
performance while designing CEO pay plans.
6.3 Limitations and Scope of Future Research
The study is conducted systematically with continuous review process contributed by
qualified and specialized supervisor. However, it is acknowledged that there are potential
theoretical and methodological limitations of the study which need to be recognized while
interpreting the findings.
Theoretically, there are many diverse and contrasting theories, hypotheses and
approaches (see, e.g., Otten, 2008) that explain executive pay process. However, it is not
possible to comprehend all the theories and hypotheses in a single study. Therefore, this study
develops its hypotheses mainly within the framework of most dominant approaches (optimal
contracting and managerial power) derived from agency theory. The reader may find
conflicts in the hypotheses developed in this study and the hypotheses presented by other
competing theories. For example, CEO duality is assumed to be a symptom of ineffective
internal corporate governance system in agency theory but at the same time, it is encouraged
in stewardship theory because of better control of CEO over the system to act in the best
interests of the shareholders without any hindrance in decision making. Nevertheless, the
empirical results will remain same regardless of competing theoretical arguments.
Methodologically, because of unavailability of data, the sample of firms used in this
study does not include unlisted, defaulted, delisted and merged/demerged firms. Thus, this
148
study might be prone to sampling and survivorship biases and warrant careful interpretation
when generalizing. In addition, including only listed firms in sample may lead to firm size
bias as only firms with certain size can be listed in the stock exchange. Furthermore,
investment and financial firms are not part of the sample due to difference in regulations,
accounting practices and performance parameters, leading to problem of generalizing the
results to these sectors. Moreover, although, this study controls for firm related unobserved
heterogeneity however it neither controls for CEO related unobserved heterogeneity such as
CEO ability nor it controls for CEO characteristics such as age, education and experience etc.
due to difficulty in operationalizing and unavailability of data.
Although this study provides evidence on how CEO compensation is related to firm
performance, important corporate governance variables and earnings management, however,
many questions remain unanswered. Since research on executive compensation in Pakistan
has not flourished therefore there are plenty of unanswered questions which can be addressed
in future research. Given the limitations of the study, future research may be aimed at
examining the implications of theories other than agency theory. The impact of a bundle of
other corporate governance variables that are not examined in this sturdy can be explored
once data availability increases. The impact of CEO’s characteristics on his compensation is
another avenue that needs to be focused in future research. The research could also be aimed
at how family characteristics interact with CEO compensation within family firms since most
of the companies in Pakistan are family owned. In Pakistan type of concentrated ownership is
considered to be more important than concentrated ownership, therefore further research may
be directed to how types of ownership interact with CEO compensation. Many firms in
Pakistan have some foreign directors in their boards. Influence of these foreign directors on
executive compensation decisions is another interesting area to explore in Pakistan.
Interaction of CEO compensation with capital structure decisions, dividend policy decisions,
investment decisions and asset management decisions are some broader areas on which future
research can be aimed at.
An important extension to CEO compensation research would be qualitative research
based on structured interviews for in depth analysis of the subject matter. Because most of the
research in CEO compensation relies of quantitative analysis, thus supplementing empirical
findings with elite interviews would be a valuable contribution towards understanding
executive pay setting mechanisms.
149
References
150
Abdullah, F., Shah, A., Iqbal, A. M. & Gohar, R. 2011. Investors’ power and the dividend cost
minimization model: Which one better explains the dividend policy in Pakistan? African
Journal of Business Management, 5, 10747-10759.
Achleitner, A.-K., Günther, N., Kaserer, C. & Sicilianb, G. 2014. Real Earnings Management and
Accrual based Earnings Management in Family Firms. European Accounting Review, 23,
431-461.
Adams, R. B. & Ferreira, D. 2008. Do directors perform for pay? Journal of Accounting and
Economics, 46, 154-171.
Adithipyangkul, P., Alon, I. & Zhang, T. Y. 2011. Executive perks: Compensation and corporate
performance in China. Asia Pacific Journal of Management, 28, 401-425.
Aggarwal, R. K. & Samwick, A. A. 1999a. Executive compensation, strategic competition, and
relative performance evaluation: Theory and evidence. The Journal of Finance, 54, 1999-
2043.
Aggarwal, R. K. & Samwick, A. A. 1999b. The other side of the trade-off: The impact of risk on
executive compensation. Journal of Political Economy, 107, 65-105.
Aggarwal, R. K. & Samwick, A. A. 2003. Performance Incentives within Firms: The Effect of
Managerial Responsibility. The Journal of Finance, 58, 1613-1649.
Ali, A., Chen, T. Y. & Radhakrishnan 2007. Corporate disclosures by family-ownership-controlled
firms. Journal of Accounting and Economics, 44, 238-286.
Alves, S. 2012. Executive Stock Options and Earnings Management in the Portuguese Listed
Companies. Revista de Contabilidad, 15, 211-235.
Ameer, B. 2013. Corporate Governance- Issues and Challenges in Pakistan. International Journal of
Academic Research in Business and Social Sciences, 3, 79-96.
Amoako-Adu, B., Baulkaran, V. & Smith, B. F. 2011. Executive compensation in firms with
concentrated control: The impact of dual class structure and family management. Journal of
Corporate Finance, 17, 1580-1594.
Anderson, R. C. & Reeb, D. M. 2003a. Founding family ownership and firm performance: Evidence
from the S&P 500. Journal of Finance, 58, 1301-1329.
Anderson, R. C. & Reeb, D. M. 2003b. Founding family ownership, corporate diversification, and
firm leverage. Journal of Law and Economics, 46, 653-684.
Arellano, M. & Bond, S. 1991. Some Tests of Specification For Panel Data: Monte Carlo Evidence
and an Application to Employment Equations. Review of Economic Studies, 58, 277-297.
Arellano, M. & Bover, O. 1995. Another look at the instrumental variable estimation of error-
components models. Journal of Econometrics, 68, 29-51.
Armstrong, C. S., Jagolinzer, A. D. & Larcker, D. F. 2010. Chief executive officer equity incentives
and accounting irregularities. Journal of Accounting Research, 48, 225-271.
Awan, S. H. 2012. Effect of board composition on firm performance: A case of Pakistan listed
companies. Interdisciplinary Journal of Contemporary Research in Business, 3, 853-863.
Balafas, N. & Florackis, C. 2014. CEO compensation and future shareholder returns: Evidence from
the London Stock Exchange. Journal of Empirical Finance, 27, 97-115.
Balsam, S. 1998. Discretionary accounting choices and CEO compensation. Contemporary
Accounting Research, 15, 229-252.
Banghoj, J., Gabrielsen, G., Petersen, C. & Plenborg, T. 2010. Determinants of executive
compensation in privately held firms. Accounting and Finance, 50, 481-510.
Barontini, R. & Bozzi, S. 2011. Board compensation and ownership structure: empirical evidence for
Italian listed companies. Journal of Management and Governance, 15, 59-89.
Bartov, E. & Mohanram, P. 2004. Private information, earnings manipulations, and executive stock-
option exercises. The Accounting Review, 79, 889-920.
Basu, S., Hwang, L.-S., Mitsudome, T. & Weintrop, J. 2007. Corporate governance, top executive
compensation and firm performance in Japan. Pacific-Basin Finance Journal, 15, 56-79.
Bebchuk, L. A., Cremers, K. J. M. & Peyerd, U. C. 2011. The CEO Pay Slice. Journal of Financial
Economics, 102, 199-221.
151
Bebchuk, L. A. & Fried, J. A. 2003. Executive compensation as an agency problem. Journal of
Economic Perspectives, 17, 71-92.
Bebchuk, L. A. & Fried, J. M. 2004. Pay without performance: The unfulfilled promise of executive
compensation, Harvard University Press.
Bebchuk, L. A., Fried, J. M. & Walker, D. I. 2002. Managerial power and rent extraction in the design
of executive compensation. University of Chicago Law Review, 69, 751-846.
Becker, G. 1981. Altruism in the family and selfishness in the market place. Economica, 48, 1-15.
Bergstresser, D. & Philippon, T. 2006. CEO Incentives and Earnings Management. Journal of
Financial Economics, 80, 511-529.
Bertrand, M. & Schoar, A. 2006. The Role of Family in Family Firms. Journal of Economic
Perspectives, 20, 73-96.
Blaikie, N. 2000. Approaches to Social Enquiry, Cambridge, Polty Press.
Blundell, R. & Bond, S. 1998. Initial Conditions and Moment Restrictions in Dynamic Panel Data
Modeis. Journal of Econometrics, 87, 115-143.
Boschen, J. F., Duru, A., Gordon, L. A. & Smith, K. J. 2003. Accounting and stock price performance
in dynamic CEO compensation arrangements. Accounting Review, 78, 143-168.
Boyd, B. K. 1994. Board Control and Ceo Compensation. Strategic Management Journal, 15, 335-
344.
Bratton, W. 2005. The Academic tournament over executive compensation. California Law Review,
93., 1557-1576.
Brick, I. E., Palmon, O. & Wald, J. K. 2006. CEO compensation, director compensation, and firm
performance: Evidence of cronyism? Journal of Corporate Finance, 12, 403-423.
Brunello, G., Graziano, C. & Parigi, B. 2001. Executive compensation and firm performance in Italy.
International Journal of Industrial Organization, 19, 133-161.
Bryman, A. 2012. Social Research Methods, Oxford: Oxford University Press.
Bryman, A. & Bell, E. 2011. Business Research Methods, Oxford: Oxford University Press.
Buck, T., Liu, X. & Skovoroda, R. 2008. Top Executive Pay and Firm Performance in China. Journal
of International Business Studies, 39, 833-850.
Burkart, M., Panunzi, F. & Shleifer, A. 2003. Family Firms. The Journal of Finance, 58, 2167-2201.
Burki, U. 2012. Institutional Voids and Corporate Governance: A Conceptual Understanding.
International Journal of Business and Management, 7, 99-107.
Burns, N. & Kedia, S. 2006. The impact of performance-based compensation on misreporting.
Journal of Accounting and Economics, 79, 35-67.
Byrd, J. & Cooperman, E. 2010. Director Tenure and the Compensation of Bank CEOs. Managerial
Finance, 36, 86-102.
Callen, J. L., Segal, D. & Hope, O.-K. 2010. The pricing of conservative accounting and the
measurement of conservatism at the firm-year level. Review of Accounting Studies, 15, 145-
178.
Carter, M. E., Lynch, L. & Zechman, S. 2009. Changes in bonus contracts in the post-Sarbanes–Oxley
era. Review of Accounting Studies, 14, 480-506.
Chalmers, K., Koh, P. & Stapledon, G. 2006. The determinations of CEO compensation: rent
extraction or labour demand? British Accounting Review, 38, 259-275.
Cheema, A. 2003. Corporate governance in Pakistan: issues and concerns. The NIPA Journal, 8, 7-19.
Cheema, A., Bari, F. & Saddique, O. 2003. Corporate Governance in Pakistan: Ownership, Control
and the Law. In: SOBHAN, F. & WERNER, W. (eds.) Comparative Analysis of Corporate
Governance in South Asia: Charting a Road Map for Bangladesh. Bangladesh Enterprise
Institute.
Chen, J., Ezzamel, M. & Cai, Z. 2011. Managerial power theory, tournament theory, and executive
pay in China. Journal of Corporate Finance, 17, 1176-1199.
Cheng, Q. & Warfield, T. 2005. Equity incentives and earnings management. The Accounting Review,
80, 441-476.
Cheung, Y., Rau, P. & Stouraitis, A. 2006. Tunneling, propping, and expropriation: evidence from
connected party transactions in Hong Kong. Journal of Financial Economics, 82, 343-386.
Clarke, T. 2007. Theories of corporate governance, Glasgow: , Bell & Bain Ltd.
152
Conyon, M. J. 2006. Executive compensation and incentives. Academy of Management Perspectives,
20, 25-44.
Conyon, M. J. 2014. Executive Compensation and Board Governance in US Firms. The Economic
Journal, 124, F60-F89.
Conyon, M. J. & He, L. 2011. Executive compensation and corporate governance in China. Journal of
Corporate Finance, 17, 1158-1175.
Conyon, M. J. & He, L. 2012. CEO Compensation and Corporate Governance in China. Corporate
Governance: An International Review, 20, 575-592.
Conyon, M. J. & Murphy, K. J. 2000. The prince and the pauper? CEO pay in the United States and
United Kingdom. The Economic Journal, 110, F640-F671.
Cooper, M., Gulen, H. & Rau, P. R. 2013. Performance for pay? The relation between CEO incentive
compensation and future stock price performance. SSRN Working Paper Series.
Core, J. E., Guay, W. R. & Larcker, D. F. 2003. Executive equity compensation and incentives: a
survey. FRBNY Economic Policy Review, 27-44.
Core, J. E., Holthausen, R. W. & Larcker, D. F. 1999. Corporate governance, chief executive officer
compensation, and firm performance. Journal of Financial Economics, 51, 371-406.
Cornett, M. M., Marcus, A. J., Saunders, A. & Tehranian, H. 2007. The impact of institutional
ownership on corporate operating performance. Journal of Banking & Finance, 31, 1771-
1794.
Croci, E., Gonenc, H. & Ozkan, N. 2012. CEO compensation, family control, and institutional
investors in Continental Europe. Journal of Banking & Finance, 36, 3318-3335.
Cuevas-Rodríguez, G., Gomez-Mejia, L. R. & Wiseman, R. M. 2012. Has Agency Theory Run its
Course?: Making the Theory more Flexible to Inform the Management of Reward Systems.
Corporate Governance: An International Review, 20, 526-546.
Daily, C. M., Dalton, D. R. & Cannella, A. A. 2003. Corporate Governance: Decades of Dialongue
and Data. Academy oí Management Review, 28, 371-382.
Danker, R. D., Darrough, M. N., Huang, R. & Plehn-Dujowich, J. M. 2013. The relation between
CEO compensation and past performance. Accounting Review, 88, 1-30.
Dechow, P., Sloan, R. & Sweeney, A. 1995. Detecting earnings management. The Accounting Review,
70, 193-225.
Dechow, P. M. 2006. Asymmetric sensitivity of CEO cash compensation to stock returns: A
discussion. Journal of Accounting & Economics, 42, 193-202.
Devers, C. E., Cannella, A. A., Reilly, G. P. & Yoder, M. E. 2007. Executive compensation: A
multidisciplinary review of recent developments. Journal of Management, 33, 1016-1072.
Dharwadkar, R., George, G. & Brandes, P. 2000. Privatization in emerging economies: an agency
theory perspective. Academy of Management Review, 25, 650-669.
Duellman, S., Ahmed, A. S. & Abdel-Meguid, A. M. 2013. An empirical analysis of the effects of
monitoring intensity on the relation between equity incentives and earnings management.
Journal of Accounting and Public Policy, 32, 495-517.
Eisenhardt, K. M. 1989. Agency Theory: An Assessment And Review. The Academy of Management
Review, 14, 57-74.
Eriksson, T. & Lausten, M. 2000. Managerial Pay and Firm Performance: Danish Evidence.
Scandinavian Journal of Management, 16, 269-286.
Faccio, M., Lang, L. & Young, L. 2001. Dividends and expropriation. American Economic Review,
91, 54-78.
Fahlenbrach, R. 2009. Shareholder Rights, Boards, and CEO Compensation. Review of Finance, 13,
81-113.
Fama, E. F. 1980. Agency problems and the theory of the firm. Journal of Political Economy, 88,
288-307.
Fama, E. F. & Jensen, M. C. 1983. Separation of ownership and control. Journal of Law &
Economics, 26, 301-325.
Fan, J. P. H., Wei, K. C. J. & Xu, X. 2011. Corporate finance and governance in emerging markets: A
selective review and an agenda for future research. Journal of Corporate Finance, 17, 207-
214.
153
Fan, J. P. H. & Wong, T. J. 2002. Corporate ownership structure and the informativeness of
accounting earnings in East Asia. Journal of Accounting and Economics, 33, 401-425.
Fernandes, N. 2008. EC: Board compensation and firm performance: The role of “independent” board
members. Journal of Multinational Financial Management, 18, 30-44.
Fernandes, N., Ferreira, M. A., Matos, P. & Murphy, K. J. 2013. Are U.S. CEOs Paid More? New
International Evidence. Review of Financial Studies, 26, 323-367.
Fich, E. M. & Shivdasani, A. 2005. The impact of stock-option compensation for outside directors on
firm value. Journal of Business, 78, 2229-2254.
Firth, M., Fung, P. M. Y. & Rui, O. M. 2006. Corporate performance and CEO compensation in
China. Journal of Corporate Finance, 12, 693-714.
Firth, M., Fung, P. M. Y. & Rui, O. M. 2007. How ownership and corporate governance influence
chief executive pay in China's listed firms. Journal of Business Research, 60, 776-785.
Frankel, R. M., Johnson, M. F. & Nelson, K. K. 2002. The relation between auditors’ fees for
nonaudit services and earnings management. The Accounting Review, 77, 71-105.
Frydman, C. & Jenter, D. 2010. CEO Compensation. Annual Review of Financial Economics, 2, 75-
102.
Frydman, C. & Saks, R. E. 2010. Executive Compensation: A New View from a Long-Term
Perspective, 1936-2005. Review of Financial Studies, 23, 2099-2138.
Frye, M. B. 2004. Equity-based compensation for employees: firm performance and determinants.
Journal of Financial Research, 27, 31-54.
Gallego, F. & Larrain, B. 2012. CEO compensation and large shareholders: Evidence from emerging
markets. Journal of Comparative Economics, 40, 621-642.
Gaver, J., Gaver, K. & Austin, J. 1995. Additional evidence on bonus plans and income management.
Journal of Accounting and Economics, 19, 3-28.
Ghosh, A. 2006. Determination of Executive Compensation in an Emerging Economy: Evidence from
India. Emerging Markets Finance and Trade, 42, 66-90.
Ghosh, S. 2010. Firm performance and CEO pay: Evidence from Indian manufacturing. Journal of
Entrepreneurship, 19, 137-147.
Gillan, S. L. 2006. Recent Developments in Corporate Governance: An Overview. Journal of
Corporate Finance, 12, 381-402.
Gomez-Mejia, L. R., Berrone, P. & Franco-Santos, M. 2010. Compensation and organizational
performance, Armonk, NY, M.E. Sharpe.
Gómez-Mejía, L. R., Haynes, K. T., Núñez-Nickel, M., Jacobson, K. J. L. & Moyano-Fuentes, J.
2007. Socioemotional Wealth and Business Risks in Family-controlled Firms: Evidence from
Spanish Olive Oil Mills. Administrative Science Quarterly, 52, 106-137.
Gomez-Mejia, L. R., Larraza-Kintana, M. & Makri, M. 2003. The determinants of executive
compensation in family-controlled public corporations. Academy of Management Journal, 46,
226-237.
Gomez-Mejia, L. R. & Wiseman, R. M. 1997. Reframing Execufive Compensation: An Assessment
and Outlook. Journal of Management, 23, 291-374.
Greene, W. H. 2011. Econometric Analysis, Pearson Education.
Guidry, F., Leone, A. & Rock, S. 1999. Earnings-based bonus plans and earnings management by
business-unit managers. Journal of Accounting and Economics, 26, 113-142.
Guillet, B. D., Kucukusta, D. & Xiao, Q. 2012. An examination of executive compensation in the
restaurant industry. International Journal of Hospitality Management, 31, 86-95.
Guthrie, K., Sokolowsky, J. & Wan, K. M. 2012. CEO Compensation and Board Structure Revisited.
Journal of Finance, 67, 1149-1168.
Hall, B. J. & Liebman, J. B. 1998. Are CEOs really paid like bureaucrats? Quarterly Journal of
Economics, 113, 653-691.
Hanlon, M., Rajgopal, S. & Shevlin, T. 2003. Are executive stock options associated with future
earnings? Journal of Accounting & Economics, 36, 3-43.
Hansen, L. P. 1982. Large sample properties of generalized method of moments estimators.
Econometrica, 50, 1029-1054.
Harris, M. & Raviv, A. 2008. A theory of board control and size. Review of Financial Studies, 21,
1797-1832.
154
Hartzell, J. C. & Starks, L. T. 2003. Institutional investors and executive compensation. Journal of
Finance, 58, 2351-2374.
Hausman, J. 1978. Specification Tests in Econometrics. Econometrics, 46, 1251-1273.
Hayes, R. M. & Schaefer, S. 2000. Implicit contracts and the explanatory power of top executive
compensation for future performance. The RAND Journal of Economics, 31, 273-293.
Healy, P. 1985. The effect of bonus schemes on accounting decisions. Journal of Accounting and
Economics, 7, 85-107.
Healy, P. M., Kang, S. H. & Palepu, K. G. 1987. The Effect of Accounting Procedure Changes on
Ceos Cash Salary and Bonus Compensation. Journal of Accounting & Economics, 9, 7-34.
Hermalin, B. E. & Wallace, N. E. 2001. Firm performance and executive compensation in the savings
and loan industry. Journal of Financial Economics, 61, 139-170.
Himmelberg, C., Hubbard, R. G. & Palia, D. 1999. Understanding the determinants of managerial
ownership and the link between ownership and performance. Journal of Financial economics,
53, 353-384.
Ho, S. S. M., Lam, K. C. K. & Sami, H. 2004. The investment opportunity set, director ownership,
and corporate policies: Evidence from an emerging market. Journal of Corporate Finance,
10, 383-408.
Hofstede, G. H. 1980. Culture’s consequences: International differences in work-related values,
Beverly Hills, CA, Sage.
Holderness, C. G. 2014. Culture and the ownership concentration of public corporations around the
world. Journal of Corporate Finance, Article in Press.
Holthausen, R. W., Larcker, D. F. & Sloan, R. G. 1995. Annual bonus schemes and the manipulation
of earnings. Journal of Accounting and Economics, 19, 29-74.
Huang, Y. S. & Wang, C.-J. 2015. Corporate governance and risk-taking of Chinese firms: The role of
board size. International Review of Economics and Finance, 37, 96-113.
Hüttenbrink, A., Oehmichen, J., Rapp, M. S. & Wolff, M. 2014. Pay-for-performance - Does one size
fit all? A multi-country study of Europe and the United States. International Business Review,
23, 1179-1192.
Ibrahim 2006. Corporate governance in Pakistan: Analysis of current challenges and
recommendations for future reforms. Washington university global studies law review, 5,
323-332.
Ibrahim, Q., Rehman, R. u. & Raoof, A. 2010. Role of Corporate Governance in Firm Performance: A
Comparative Study between Chemical and Pharmaceutical Sectors of Pakistan. International
Research Journal of Finance and Economics, 50, p7.
Ibrahim, S. & Lloyd, C. 2011. The association between non-financial performance measures in
executive compensation contracts and earnings management. Journal of Accounting and
Public Policy, 30, 256-274.
Jabeen, S., Kaleem, A. & Ehsan, S. 2012. Financial Performance of Family Firms (Evidence from
Selected Manufacturing Sectors of Pakistan). Journal of Basic and Applied Scientific
Research, 2, 10303-10313.
Jameson, M., Prevost, A. & Puthenpurackal, J. 2014. Controlling shareholders, board structure, and
firm performance: Evidence from India. Journal of Corporate Finance, 27, 1-20.
Javed, A. Y. & Iqbal, R. 2006. Corporate Governance and Firm Performance: Evidence from Karachi
Stock Exchange. The Pakistan Development Review, 45, 947-964.
Javid, A. Y. & Iqbal, R. 2008. Ownership Concentration, Corporate Governance and Firm
Performance: Evidence from Pakistan. The Pakistan Development Review, 47, 643-659.
Javid, A. Y. & Iqbal, R. 2010. Corporate Governance in Pakistan: Corporate Valuation, Ownership
and Financing. PIDE Working Papers
Jensen, M. 1993. The modern industrial revolution. Exit, and the failure of internal control systems.
Journal of Finance, 48, 831-880.
Jensen, M. & Meckling, W. 1976. Theory of the firm: Managerial behavior, agency costs, and
ownership structure. Journal of Financial Economics, 3, 305-360.
Jensen, M. C. 1983. Organization theory and methodology. Accounting Review, 58, 319-339.
Jensen, M. C. & Murphy, K. J. 1990a. CEO Incentives - It's Not How Much You Pay, But How.
Harvard Business Review, 68, 138-153.
155
Jensen, M. C. & Murphy, K. J. 1990b. Performance Pay and Top-Management Incentives. Journal of
Political Economy, 98, 225-264.
Jensen, M. C. & Warner, J. B. 1988. The distribution of power among corporate managers,
shareholders, and directors. Journal of Financial Economics, 20, 3-24.
Jiang, G., Lee, C. & Yue, H. 2010a. Tunneling through intercorporate loans: The China experience.
Journal of Financial Economics, 98, 1-20.
Jiang, J. X., Petroni, K. R. & Wang, I. Y. 2010b. CFOs and CEOs: Who have the most influence on
earnings management? Journal of Financial Economics, 96, 513-526.
Johnson, P. & Clark, M. 2006. Mapping the terrain: an overview of business and management
research methodologies. In: JOHNSON, P. & CLARK, M. (eds.) Business and Management
Research Methodologies. London: Sage.
Jolliffe, I. T. 2002. Principle Component Analysis, New York, Springer.
Jones, J. 1991. Earnings management during import relief investigations. Journal of Accounting
Research, 29, 193-228.
Kamran & Shah, A. 2014. The Impact of Corporate Governance and Ownership Structure on Earnings
Management Practices: Evidence from Listed Companies in Pakistan. The Lahore Journal of
Economics, 19, 27-70.
Kaplan, S. N. 1994. Top executive rewards and firm performance: a comparison of Japan and the
United Statesx. The Journal of Political Economy, 102, 510-546.
Kashif, S. & Mustafa, K. 2012. The Determinants of CEO Compensation: Evidence from Family-
Owned Listed Corporations in Karachi Stock Exchange. Middle East Journal of Economics
and Finance, 5, 45-61.
Kato, H. K., Lemmon, M., Luo, M. & Schallheim, J. 2005. An empirical examination of the costs and
benefits of executive stock options: Evidence from Japan. Journal of Financial Economics,
78, 435-461.
Kato, T., Kim, W. & Lee, J. H. 2007. Executive compensation, firm performance, and Chaebols in
Korea: Evidence from new panel data. Pacific-Basin Finance Journal, 15, 36-55.
Kato, T. & Kubo, K. 2006. CEO compensation and firm performance in Japan: Evidence from new
panel data on individual CEO pay. Journal of the Japanese and International Economies, 20,
1-19.
Kato, T. & Long, C. 2006. Executive compensation, firm performance, and corporate governance in
China: Evidence from firms listed in the Shanghai and Shenzhen Stock Exchanges. Economic
Development and Cultural Change, 54, 945-983.
Khatab, H., Masood, M., Zaman, K., Saleem, S. & Saeed, B. 2011. Corporate Governance and Firm
Performance: A Case study of Karachi Stock Market. International Journal of Trade,
Economics and Finance, 2, 39-43.
Khilji, S. E. 2003. To adapt or not to adapt, exploring the role of national culture in HRM - A study of
Pakistan. International Journal of Cross Cultural Management, 3, 109-132.
Kothari, S. P., Leone, A. & Wasley, C. 2005. Performance-matched discretionary accrual measures.
Journal of Accounting and Economics, 39, 163-197.
Kotowitz, Y. 1987. Moral Hazard. In: EATWELL, J., MILGATE, M. & NEWMAN, P. (eds.) The
New Palgrave – A Dictionary of Economics. London: The Macmillan Press.
Kubo, K. 2005. Executive compensation policy and company performance in Japan. Corporate
Governance-an International Review, 13, 429-436.
La Porta, R., Lopez-De-Silanes, F. & Shleifer, A. 1999. Corporate Ownership Around the World. The
Journal of Finance, 54, 471-517.
La Porta, R., Lopez-de-Silanes, F., Shleifer, A. & Vishny, R. 2000. Investor protection and corporate
governance. Journal of Financial Economics, 58, 3-27.
Laan, G. v. d., Ees, H. v. & Witteloostuijn, A. v. 2010. Is pay related to performance in The
Netherlands? An analysis of Dutch executive compensation, 2002-2006. Economist-
Netherlands, 158, 123-149.
Lambert, R., Larcker, D. & Weigelt, K. 1993. The Structure of Organizational Incentives.
Administrative Science Quarterly, 38, 438-461.
Lara, J. M. G., Osma, B. G. & Penalva, F. 2011. Conditional conservatism and cost of capital. Review
of Accounting Studies, 16, 247-271.
156
Leone, A. J., Wu, J. S. & Zimmerman, J. L. 2006. Asymmetric sensitivity of CEO cash compensation
to stock returns. Journal of Accounting & Economics, 42, 167-192.
Lippert, R. L. & Moore, W. T. 1994. Compensation Contracts of Chief Executive Officers -
Determinants of Pay-Performance Sensitivity. Journal of Financial Research, 17, 321-332.
Lippert, R. L. & Porter, G. 1997. Understanding CEO pay:a test of two pay to performance sensitivity
measures with alternative measures of alignment and influence. Journal of Business
Research, 40, 127-138.
Lipton, M. & Lorsch, J. W. 1992. A modest proposal for improved corporate governance. Business
Lawyer, 48, 59-77.
Liu, L. S. & Stark, A. W. 2009. Relative performance evaluation in board cash compensation: UK
empirical evidence. The British Accounting Review, 41, 21-30.
Malmendier, U. & Tate, G. 2009. Superstar CEOs. The Quarterly Journal of Economics, 124, 1593-
1638.
Matolcsy, Z. & Wright, A. 2011. CEO compensation structure and firm performance. Accounting and
Finance, 51, 745-763.
Mehran, H. 1995. Executive-Compensation Structure, Ownership, and Firm Performance. Journal of
Financial Economics, 38, 163-184.
Mengistae, T. & Xu, L. X. C. 2004. Agency theory and executive compensation: The case of Chinese
state-owned enterprises. Journal of Labor Economics, 22, 615-637.
Mitsudome, T., Weintrop, J. & Hwang, L. S. 2008. The relation between changes in CEO
compensation and firm performance: A Japanese/American comparison. Journal of the
Japanese and International Economies, 22, 605-619.
Mujtaba, B. G. & Afza, T. 2011. Business Ethics Perceptions of Public and Private Sector
Respondents in Pakistan. Far East Journal of Psychology and Business, 3, 1-11.
Mura, R. 2007. Firm Performance: Do Non-Executive Directors Have Minds of their Own? Evidence
from UK Panei Data. Financial Management, 36, 81-112.
Murphy, K. J. 1999. Executive compensation. In: ASHENFELTER, O. & CARDS, D. (eds.)
Handbook of Labor Economics. New York: North Holland.
Nishat, M. & Shaheen, R. 2006-2007. Corporate Governance and Firm Performance in an Emerging
Market - An Explanatory Analysis of Pakistan. Journal of Corporate Ownership and Control,
4, 216-225.
Ntim, C. G., Lindop, S., Osei, K. A. & Thomas, D. A. 2013. Executive Compensation, Corporate
Governance and Corporate Performance: A Simultaneous Equation Approach. Managerial &
Decision Economics, 36, 67-96.
Nyberg, A. J., Fulmer, I. S., Gerhart, B. & Carpenter, M. A. 2010. Agency Theory revisited: CEO
Return and Shareholder Interest Alignment. Academy of Management Journal, 53, 1029-
1049.
O'Connor, M. L. & Rafferty, M. 2010. Incentive effects of executive compensation and the valuation
of firm assets. Journal of Corporate Finance, 16, 431-442.
O'Reilly, C. A. & Main, B. G. M. 2010. Economic and psychological perspectives on CEO
compensation: a review and synthesis. Industrial and Corporate Change, 19, 675-712.
Otten, J. A. 2008. Theories on executive pay. A literature overview and critical assessment. MPRA
Working Paper No. 6969.
Ozkan, N. 2007. Do corporate governance mechanisms influence CEO compensation? An empirical
investigation of UK companies. Journal of Multinational Financial Management, 17, 349-
364.
Ozkan, N. 2011. CEO Compensation and Firm Performance: an Empirical Investigation of UK Panel
Data. European Financial Management, 17, 260-285.
Pepper, A. & Gore, J. 2015. Behavioral agency theory: new foundations for theorizing about
executive compensation. Journal of management, 41, 1045-1068.
Puffer, S. M. & Weintrop, J. B. 1991. Corporate Performance and CEO Turnover: The Role of
Performance Expectations. Administrative Science Quarterly, 36, 1-19.
Randøy, T. & Nielsen, J. 2002. Company Performance, Corporate Governance, and CEO
Compensation in Norway and Sweden. Journal of Management and Governance, 6, 57-81.
157
Rehman, R., Hasan, M., Mangla, I. & Sultana, N. 2012. Economic Reforms, Corporate Governance
and Dividend Policy in Sectoral Economic Growth in Pakistan. The Pakistan Development
Review, 51, 133-146.
Roodman, D. 2009. How to do xtabond2: An introduction to difference and system GMM in Stata.
The Stata Journal, 9, 86-136.
Sargan, J. D. 1958. The estimation of economic relationships using instrumental variables.
Econometrica, 26, 393-415.
Saunders, M., Lewis, P. & Thornhill, A. 2012. Research methods for business students, Essex,
England, Pearson Education Limited.
SECP 2002. Code of Corporate Governance.
SECP 2012. Code of Corporate Governance.
Shah, S. Z. A., Javed, T. & Abbas, M. 2009. Determinants of CEO Compensation Empirical Evidence
from Pakistani Listed Companies. International Research Journal of Finance and Economics,
32, 148-159.
Shapiro, S. P. 2005. Agency Theory. Annual Review of Sociology, 31, 263-284.
Shaw, K. W. & Zhang, M. H. 2010. Is CEO Cash Compensation Punished for Poor Firm
Performance? Accounting Review, 85, 1065-1093.
Sheikh, M. F. & Riaz, K. 2012. Overconfidence Bias, Trading Volume and Returns Volatility:
Evidence from Pakistan. World Applied Science Journal, 18, 1737-1748.
Shleifer, A. & Vishny, R. 1986. Large shareholders and corporate control. Journal of Political
Economy, 94, 461-488.
Strauss, A., Schatzman, L., Ehrich, D., Bucher, R. & Sabshin, M. 1973. The Hospital and its
Negotiated Order. In: SALAMAN, G. & THOMPSON, K. (eds.) People and Organizations.
London: Longman.
Su, Z., Li, Y. & Li, L. 2010. Ownership concentration and executive compensation in emerging
economies: evidence from China. Corporate Governance: An International Review, 10, 223-
233.
Sun, B. 2014. Executive Compensation and Earnings Management under Moral Hazard. Journal of
Economic Dynamics & Control, 41, 276-290.
Sun, F., Wei, X. & Huang, X. 2013. CEO compensation and firm performance Evidence from the US
property and liability insurance industry. Review of Accounting and Finance, 12, 252-267.
Sun, L. 2012. Executive Compensation and Contract-Driven Earnings Management. Asian Academy
of Management Journal of Accounting and Finance, 8, 111-127.
Sun, S. L., Zhao, X. & Yang, H. 2010. Executive compensation in Asia: A critical review and
outlook. Asia Pacific Journal of Management, 27, 775-802.
Tosi, H. L., Werner, S., Katz, J. P. & Gomez-Mejia, L. R. 2000. How Much Does Performance
Matter? A Meta-Analysis of CEO Pay Studies. Journal of Management, 26, 301-339.
van Essen, M., Heugens, P. P. M. A. R., Otten, J. & Van Oosterhout, J. 2012a. An institution-based
view of executive compensation: A multilevel meta-analytic test. Journal of International
Business Studies, 43, 396-423.
van Essen, M., Otten, J. & Carberry, E. J. 2012b. Assessing Managerial Power Theory: A Meta-
Analytic Approach to Understanding the Determinants of CEO Compensation. Journal of
management, 41, 164-202.
Villalonga, B. & Amit, R. 2006. How do family ownership, control, and management affect firm
value? Journal of Financial Economics, 80, 385-417.
Villalonga, B. & Amit, R. 2010. Family control of firms and industries. Financial Management, 39,
863-904.
Walsh, J. & Seward, J. 1990. On the Efficiency of Internal and External Corporate Control
Mechanisms. The Academy of Management Review, 15, 421-458.
Wang, D. 2006. Founding family ownership and earnings quality. Journal of Accounting Research,
44, 619-656.
Wang, K. & Xiao, X. 2011. Controlling shareholders' tunneling and executive compensation:
Evidence from China. Journal of Accounting and Public Policy, 30, 89-100.
Watts, R. L. & Zimmerman, J. L. 1978. Towards a Positive Theory of the Determination of
Accounting Standards. The Accounting Review, 53, 112-134.
158
Watts, R. L. & Zimmerman, J. L. 1986. Positive accounting theory. Contemporary Topics in
Accounting Series, Englewood Cliffs, NJ., Prentice-Hall.
Weisbach, M. S. 2007. Optimal executive compensation versus managerial power: A review of
Lucian Bebchuk and Jesse Fried's pay without performance: the unfufilled promise of
executive compensation. Journal of Economic Literature, 45, 419-428.
Wooldridge, J. M. 2002. Econometric Analysis of Cross Section and Panel Data, Cambridge, MA.,
MIT Press.
World Bank 2005. Pakistan : Report on the Observance of Standards and Codes (ROSC) : Corporate
Governance Country Assessment.
Yasser, Q. R., Entebang, H. & Mansor, S. A. 2011. Corporate governance and firm performance in
Pakistan: The case of Karachi Stock Exchange (KSE)-30. Journal of Economics and
International Finance, 3, 482-491.
Ye, K. 2014. Independent director cash compensation and earnings management. Journal of
Accounting and Public Policy, 33, 391-400.
Yermack, D. 1996. Higher market valuation for firms with a small board of directors. Journal of
Financial Economics, 40, 185-211.
Yeung, H. 2006. Change and continuity in Southeast Asian Chinese business. Asia Pacific Journal of
Management, 23, 229-254.
Yoshikawa, T., Rasheed, A. A. & Del Brio, E. B. 2010. The impact of firm strategy and foreign
ownership on executive bonus compensation in Japanese firms. Journal of Business Research,
63, 1254-1260.
Young, M. N., Peng, M. W., Ahlstrom, D., Bruton, G. D. & Jiang, Y. 2008. Corporate Governance in
Emerging Economies: A Review of the Principal–Principal Perspective. Journal of
Management Studies, 45, 196-220.