1
Yet another perspective on power:
Digging deep into the relationship between CEO bonding and corporate
political activities of a firm.
Thesis M.Sc. Business Administration - International Management Track University of Amsterdam - Faculty of Economics and Business
Name of Student: Jyotsana Rankawat Student Number: 11244658
Supervisor: Dr. Ilir Haxhi
2
Statement of Originality
This document is written by me, Jyotsana Rankawat, and I declare to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
3
Abstract
In recent years, growing interest in corporate governance practices has enlightened every related field to it. Therefore, like many other authors, it is right to say market strategies as well as non-market strategies to increase profits by a firm are highly important to study. There are many studies done on CEO compensation, which is a highly important factor in corporate governance, and on the other hand many research papers are written about CEOs pay linked to firm’s financial performance. Despite this attention to CEO compensation and its implications, previous research is limited in explaining the effect of CEO pay linked with firm’s financial performance (pay bonding) on a firm’s political activities because non-market strategies (CPA) are also proven to be instrumental in a firm’s profits. This current study addresses this gap by examining the link between CEO pay bonding and corporate political activities. My core argument is that political activities like lobbying and PAC contributions are perceived to affect a firm’s financial performance due to several factors and therefore, I first argue that CEO pay linked to financial performance of a firm positively affects such political activities, proving that higher efforts are put by CEO on the grounds of self-interest. Second, I argue that there are positive moderating effects of diversity in a firm’s board on the CEO-CPA relationship. My results show that while CEO pay bonding doesn’t have significant effects on firm’s CPA overall or lobbying in particular, it does have a positive relationship with PAC contributions made by a firm. At the same time, board diversity positively moderates the relationship between CEO bonding and PAC contribution. I contribute to the hundreds of arguments that favor CEO pay bonding as a solution to agency problem as well as another ground to the studies of CPA and firm performance. My practical contribution substantiates the need for restructuring top executives’ pay due to its high integration with non-market strategies for firms that generate higher financial returns eventually.
Keywords: Corporate political activity (CPA); Corporate non-market strategies; Lobbying; Political action committee (PAC); Corporate financial performance (CFP); Top management teams (TMT); Board diversity.
4
Table of contents:
1. Introduction 2. Literature review
2.1. Top management teams and corporate political activity (CPA) 2.2. Core argument 1:
2.2.1. CPA and corporate financial performance (CFP) 2.3. Core argument 2:
2.3.1. CFP and CEO compensation
2.4. Board Diversity (Importance for market and non-market strategies) 3. Theoretical Framework
3.1. Explanation and hypotheses 3.2. Conceptual model diagram 4. Data and method
4.1. Data collection and sample 4.2. Data sources 4.3. Variables 4.3.1. Dependent Variables 4.3.2. Independent Variable 4.3.3. Moderator Variable 4.3.4. Control Variables 4.4. Methodology
5. Analysis and results 5.1. Descriptive statistics
5.2. Correlations and multicollinearity 5.3. Regression
6. Discussion 6.1. Findings 6.2. Contribution
5 6.2.2. Practical contributions
6.3. Limitations
6.4. Directions for future research 7. Conclusion
8. References
9. Appendices (including tables and figures throughout the paper) 9.1. Summary of regression models
9.2. Descriptive statistics 9.3. Tolerance and VIF
9.4. Correlation and multicollinearity 9.5. Regression results
6
Introduction:
A basic function of a firm’s CEO is to manage and monitor the exposure of a firm with external environment and therefore, a substantial amount of research and application is done on the factors that decide a CEOs actions for firm’s performance (Finkelstein and Hambrick, 1996; Cannella, 2009). But firms also operate in non-market environments i.e. political relationship of a firm is also one of the responsibilities of a firm (Baron, 1995). Research most recently has concentrated upon the CEOs impact on a firm’s on-market strategies and his influence on the success of such strategies for firm’s better results. Like, CEOs who are more young or Female CEOs are more investing into CSR initiatives of a firm (Borghesi, Houston and Naranjo, 2014). Or, CEOs with political ideologies invest more into PAC donations as well as CSR initiatives (Chin, Hambrick and Trevino, 2013). Yet, little research has been done to study the factors that influence a CEO to invest more into corporate political activity of a firm. Some of the demographic factors like CEO’s age and tenure and other characteristics have been studied in detail by Rudy and Johnson very recently in 2016. To go in deep of the CEOs relationship with CPA of a firm, a small introduction is discussed as follows on the importance of CPA in today’s corporate world.
Organizations are usually more creative and responsive when they expose themselves to turbulences in order to operate efficiently; yet, a significant amount of turbulence management is required (Lynn, 2005) by activities that relate to market or non-market, such as, social and political activities. According to Baysinger (1984), organizations engage in corporate political activities (CPAs) to fulfill three objectives i.e. to gain special monetary and anticompetitive benefits from government, to manage environmental threats by government to the legitimacy of organization’s goals (Castello and Galang, 2014). Moreover, in order to manage similar threats to an organization’s core business. These corporate political activities are of two types, “buffer” and “bridge” (Meznar and Nigh, 1995). Where former behave like airbag to the dangers of social and political environment and later bridges a business with that environment. It is important to know when does a business engage itself in two of these CPAs. At this stage only, I should also clarify that the “political” role that a business plays here is conceptually related to its participation in corporate partnership, rule-setting process and rule-finding discourses (Pies et al, 2014). Among researches, many are done on what all aspects of this “new governance” play an important role in economy and politics of a country or how does politics affect businesses.
7 Most of such studies are executed on USA and it will be not entirely wrong to say that political activities in the United States affect the globe (Held et al, 2000, Kim, J. H. 2008). Due to a large economic impact of MNCs in the USA on global business world, every single factor that may or may not affect a business is important to study. From global meltdown to high economic volatility in the United States, mass effects are visible everywhere. Through lobbying, influence on a government increases and this eventually results in policy changes that are helpful in fulfilling goals of a business (through non-market strategies). When both the demand and supply side attributes are attractive of a political market, firms engage in CPA (McGrath et al, 2010; Bonardi et al, 2005). Research on how lobbying expenditure influences company performance (Lord, 2000 & Chen H. et al, 2010, 2015) has found some very interesting results. With increasing non-market activity of firms, literature advances the requirement of integration of market and non-market strategies to get the best corporate performance results and efficiency (Rodriguez et al, 2006; Baron, 2001; Mellhahi, et al, 2016; Den Hond et al, 2014; Frynas and Stephens, 2015). However, research on effects of lobbying from corporate governance perspective is limited.
Vast research tells us about the positive impact of firm size (Meznar and Nigh, 1995), firm age (Kim and Baysinger, 1988; Hillman, 2003) and market share (Schuler, 1996) on CPA but researches that focus on CPA and corporate governance have found that different ownership structures influence the intensity of CPA using agency theory (Ozer & Alakent, 2013). Despite studying all the aspects of CPA from firm’s point of view, very little talks about CEO compensation, may affect firm’s propensity of CPA. A decent amount of research has provided empirical evidence on pay for performance link of CEO and firm’s performance (Brick, Palmon and Wald, 2006; Devers, Cannella, Reily and Yoder, 2007; Ozer, 2010; Gupta et al, 2017) or the effect of bonus in CEO compensation that is on shareholders returns (Michaud and Gai, 2009). I move a step further and dig deeper into the relationship between the pay for performance structure of CEO compensation and CPA.
Organizational results are an outcome of top management team’s decisions (Hambrick and Mason, 1984). In this thesis paper, I follow an argument that firm performance is affected by CPA and if a CEO has his compensation linked with performance, he tends to invest and strategize more into different corporate political strategies, both lobbying expenditures on the lobbyists for policy making favors and informational strategies i.e. PAC donations. Early 2000s witnessed a trend that
8 lobbying expenditures are tripled and PAC donations of US publics firms quadrupled (Opensecrets, 2017), therefore, it will be justified to say that both the kinds of political activities are equally important not just for their impact but for numbers. Lobbying is a lightly regulated activity and firms spend way more in lobbying that in associated PACs donations (Milyo et al, 2000; Cooper et al., 2010). One also needs to know hereby that according to a recent research by Aslan and Grintein (2011), firms’ PAC contributions are also positively associated with its performance.
From agency theory viewpoint, it is not necessary CPA increases shareholder’s wealth but for sure it increases the agency costs borne by Shareholders (Hill, Kelly, Lockhart and Ness, 2013). CPA affects firm’s performance (Chen et al., 2015) but on the other hand, it increases securing bailout assistance (Alexander et al, 2009), lesser corporate taxes and higher level of accounting performance (Richter et al, 2009; Chen et al, 2015), so I can conclude by this that it increases shareholder’s interest. According to the existing research on CEO pay and lobbying by Skaife et al. (2013), the CEOs earn greater compensation in lobbying firms as compared to their counterparts in non-lobbying firms after controlling for standard economic determinants of pay. They argue that the CEO pay and lobbying relation strengthens as the intensity of firm increases in lobbying. With both of these points of view, one can think of a gap that is in the study of impact of CEO on lobbying expenditure of a firm.
My study focusses on the reverse relationships of both the activities with CEO bonding. Paper of Skaife et al (2013) provides an evidence that CEOs engage in rent extraction through lobbying and this is a governance characteristic of a firm. Integrating corporate governance and political activities of a company and finding out that the CEO bonding, firm performance and political activities are interrelated in a cycle, a gap is seen in the literature that is interesting to study. CEO bonding can be a possible solution for principle-agent problem according to Amzaleg et al. (2014). Agency theory suggests that high pay-performance sensitivity (PPS) of CEO’s compensation is an important motivation mechanism to the CEO to improve corporate performance. They argue that boards should maximize the bonding to help ensuring that CEO goals, such as maximizing his/her own pay, align with shareholder’s goals i.e. maximizing the firm value. This thesis tries to find the quantitative link between the CEO compensation bonded with shareholder return and his motivation to spend more on lobbying and PAC contribution.
9 Previous theories suggest here that a personal benefit pushes an individual to work more on reimbursing the cost incurred from his own pocket rather than company costs. In CPAs, high agency costs are incurred and this can possibly be a highly effective principle-agent issue solution. (Bryson, A., Forth, J., & Zhou, M. 2014). To sum the first part of this research study, it is plausible to say that the research question hereby is:
Does CEO bonding affect corporate political activity of a firm?
A growing body of literature challenges the relationship between CPA and corporate performance by showing that lobbying is a feature of managerial agency problems (Agrawal and Knoeber, 2012; Hadani and Schuler, 2013). This study presents a counter challenge to display that agency problem’s renowned solution i.e. PPP link of executive compensation can actually benefit a company’s performance due to political strategy grounds too.
Second part of this research focuses upon the diversity in top management team and its relationship with political activities. It reflects on how the relationship of CEO bonding and lobbying is moderated by board diversity. Literature in corporate governance and CPA emphasizes on the need for firms to pay more attention to social and environmental issues (Donaldson & Preston, 1995; Korn Ferry publication year 2007, 2010). Firms have altered the composition and structure of boards in order to effectively tackle such macro issues (Korn Ferry publication year 2007, 2010) and the overall objective of this moderation study is to find out whether the board composition (diversity, for this study) can be classified as shareholder oriented and whether it has any impact on lobbying expenditure when CEO pay is linked to shareholder returns because the key determinants of CEO pay are the board members. Essentially, the board of directors of a firm are the most responsible for the decisions in an organization because they are the governing body of a company (Dalton et al, 1999; Tricker, 2009). The board develops strategies on corporate level, plans, budgets and their responsibilities go beyond operations. CPA is not a routine activity but is highly categorized in a financial expense of a substantial amount and large sum of capital is allocated in direct lobbying, as well as other expenses like backing up campaign issues, PACs, advertising and bribery (Rehbein & Schuler, 1999; Austen-Smith & Wright 1996; Hansen & Mitchell 2000; Getz, 2002 and more). It requires careful consideration and is an important responsibility of board (Ozer, 2010). The composition of board is subject to heterogeneity and it can be very different from country to country or industries (Tricker, 2009). Therefore, diversity is
10 an important determinant in CPA for examining the level of involvement and CPA orientation. Firms can use various CPA strategies on resource level, institutional level and political level (Lawton et al, 2013). But deep review of role of directors in CPA has been neglected. it is crucial to discover the deep roots of diversity’s effects of a firm (Baysinger et al, 1990). This paper does not focus on the how the compensation of CEO is affected by board diversity. Rather, it tests how board diversity strengthens or weakens the lobbying expenditure and CPA, when CEO pay is linked with firm performance. Board diversity in this study is considered as a whole factor including gender, nationality and other characteristics of board members. I propose to study board diversity as a diverse board is more common in today’s era. Certainly, by the gap found in the literature and my curiosity in far-sighting effects of board diversity, the second research gap that I find is:
Does board diversity positively moderate the relationship between CEO bonding and CPA? While there is an evidence of relationship between board diversity and strategic decision-making process of a company like organizational growth and innovation (Bantel and Jackson, 1989, Boarker, 1997), there is a little evidence of how it relates to corporate political activities. Previous researches conclude that diverse top management executives can interpret and perceive the importance of CPA for a firm (Chataway et al., 2004; Ozer and Markoczy, 2010). Ozer in his paper from 2010 argues that Top management team’s (TMT) heterogeneity negatively moderates the relationship between TMT involvement in political activity and firm’s CPA. But my research dives deep into a relationship between condition when the responsibility of firm’s performance goes in the hands of CEO through performance linked pay and a diverse TMT.
I expect in this paper that board diversity has some moderation in the relationship of first research question but the direction of moderation can be mixed like previous researches. Still, I expect a diverse board to strengthen the relationship between CEO bonding and CPA because a diverse board is more inclined towards increasing shareholder’s wealth and higher firm performance, on the grounds of an argument that like market-based strategies CPA is also an important strategy to be studied in this world of corporate political entrepreneurs (Yoffie & Bergenstein, 1985). If this research is able to find out the direction of this moderation, future researches can dig more into the challenges or benefits of the changing era of allegedly non-diverse boards of the USA. With the increase in more inclusion of different nationalities, race and gender
11 (Siciliano, 1996), this study may throw light towards future research on fallacies of more diversity inclusion in boards.
To address the two research questions, I apply a quantitative approach. For my sample of 295 S&P Fortune 500 firms in the period of 2012-2015, I test the relationship between CEO pay linked with shareholder returns and CPA as well as the moderation effects of Board diversity on it. I test four hypotheses using hierarchical regression analysis, CEO pay is operationalized by its linkage to shareholder returns in dummy variables and CPA follows Hillman and Hitt (1999) typology of CPA. Given the results that I find, CEO bonding has positive effect on CPA but I cannot find substantiate results of moderation of board diversity on CEO-CPA relationship.
Once, these gaps are studied in this paper, this will enrich existing theory that discusses CEO power on non-market strategies by its effects on CPA. Existing literature describes the effects of executive compensation on overall firm performance. However, no research has yet explored into one such resource of higher firm performance, lobbying expenditure. It is a common practice for improving firm performance in the USA but what can possibly increase or decrease it has a gap. Furthermore, this study will find out if diverse board structure emphasizes or decreases the effects of CEO pay on CPA. Rationale behind this study lies in the cross-examination of extent by which diverse boards can affect the firm performance. This study will provide new insights on the relationship of shareholder returns and lobbying through the lens of CEO compensation linked to it and the regression amount will depict whether it motivates the top executives to engage into non-market strategies for personal benefit, if only linked with company’s benefits. Practically, this study can be used as a tool in issuing performance bonds or in determining how much of an executive’s pay is to be linked to performance of a company when the firm engages in political activities. This is especially true in the firms where government related projects or national policies are highly instrumental. Behind studying the moderation, purpose of this study is to explore another effect of board diversity which may not be visible directly. Different studies analyze board characteristics and their impact on CEO compensation but this study will present a new perspective to bridge a gap between Corporate governance and corporate political activities. The board of directors is a pivotal governance mechanism that plays a significant role in strategic decision-making and monitors a firm’s policies, such as its compensation policy (Ntim, 2015; Ntim et al., 2017). Building on these thoughts, firms may find additional strengthening factors to pay
12 performance link that have been a solution to agency problems as per many authors since years (Bebchuk & Fried, 2003; Morgan and Poulsen, 2001; Dyl, 1988) with my paper.
Remainder of this paper is organized as follows: next chapter reviews the literature in this field of research while further chapters propose framework to study the relationship between CEO-CPA-Board diversity and present a set of four hypotheses, choice of methods and data, discussion of findings of data analysis, concluding remarks, implications as well as limitations of this study.
Literature Review:
It is not to debate that the companies are highly benefited by an influence on policy making in the United States (Lord, 2000) but any variables that affect such influence are highly important to study. By definition (Hillman et al, 2004), corporate political activity is a corporate attempt to shape government policy in ways favorable to the company. Firms that are able to gain access to the political process may benefit from a reduction in uncertainty, reduced transaction costs and increased survival. Given the elusive nature of measuring the effect of CPA and the continued and escalated use of such strategies by firms, the study confirms that CPA engagement is not without justification. CPA in this paper consists of PAC contributions by a firm as well as lobbying expenditure made. Typically, corporate lobbying is larger in magnitude and are direct, than the spending on PACs by a firm (Chen et al, 2014). Therefore, the literature studied to establish my theory has been from an antecedent point of view of CPA. The antecedent here is CEO compensation. Furthermore, effects and factors of diversity in a board that affect this antecedent is studied.
Top management teams and corporate political activity:
Study by Ozer (2010) explores that senior executives affect firms’ propensity to engage in political activity. Senior executive’s involvement in a particular political activity affects their firm’s commitment to that activity contingent on CEO tenure and TMT heterogeneity. In America, a survey results show that nearly one hour per day of a corporate leader is engaged in national political activity. Politically active leaders and politically active firms are strongly related to each other. CEOs office is an additional locus of political power within business firms (Nownes and Aitalieva, 2013). Management is expected to pursue effective political mechanisms in order to
13 create value as a part of their non-market strategy (Mathur et al, 2013). Lobbying is also an effective risk-aversion strategy and executives tend to ward off potential threats through political power because lobbying induces policy changing power to a business leader (Grinstein and Hribar (2004); Rapetto, 2007). Politically active stakeholders may be a firm's most potent political resource according to Lord (2000). Politically connected firms are more likely to be bailed out due to presence of political resources in the TMT and tend to receive preferential treatment like debt financing (Faccio, Masulis and McConnell, 2006).
Agency conflict says that more powerful management may pursue personal objectives to detriment of shareholders but lobbying may serve a twofold interest where it is in managerial interest as well increases shareholders’ value in a firm (Mathur et al, 2013). Corporations ideally operate in a socially responsible manner (Ayuso and Argandoña, 2007) to create social welfare. But managers may engage in CPA for other reasons that are not profit driven and that influence financial gains (Hadani and Schuler, 2013) and evidently CPA may create information asymmetry between agents and managers. An important issue of corporate governance is to understand the bridge between these asymmetrical interests (Becht, Bolton and Roell, 2003). This can be done by monitoring the practices of CEO (manager) as well as by putting a reward component (bonding) to avoid opportunistic behavior (Aguilera et al., 2008). Many research articles have focused on these two practices implying that the agency issue is due to a misalignment between agents and managers and in this case, CPA could be a source of this.
Corporate political activities and financial performance of firms:
Lobbying has a positive effect on the firm’s equity returns relative to the market as per a study by Kim (2000) on S&P 500 firms between 1998-2004. Evidently, Faccio and Parsley (2009) suggest that firm loses about two percent in shareholder’s value at the time of an event that disrupts political connection of the firm. CPAs are to influence legislators and encourage favorable policies/outcomes. Market performance significantly improves as compared to the previous benchmarks with higher lobbying intensity (Chen et al, 2015). Recent researches also explored direct effects of CPA on financial performance of firms (Shaffer, 1995). Shaffer et al. (2000) provide empirical evidence to suggest the importance of political actions for firm performance. They studied the effect of both market and CPA actions of airlines on performance variables such as gross profit margin, load factors, and changes in market share. In their sample, no significant
14 effect was found for market actions, but CPA actions significantly affected performance for all measures. Firms with high growth opportunities benefit more from lobbying than low-growth firms. Lobbying seems to provide limited tangible benefits in terms of helping a firm obtain government contracts or successfully pass the congressional bills lobbied. These results suggest that agency costs dominate the strategic benefits of lobbying activities. Nonmarket actions have a positive and significant impact on performance, regardless of whether the latter is measured as profits, market share or capacity utilization (Shaffer, 2000). However, there is some evidence that firms benefit when there is political alignment between the firm and the party in power (Cao Z. et al, 2017). CPA influences firm’s financial performance when political activities influence the government and actions are taken in favor of the firm (to take or not to take an action) Increase in obtaining rate, earmarks, new import tariffs, petitions to get anti-dumping protection, government sales are a few ways policy change affect firm performance directly or indirectly (Bonardi et al, 2006; De Figueiredo & Silverman, 2006; Schuler, 1996; Epstein 1969; Marsh, 1998). Firms whose TMT are elected to federal office get greater results in shareholder returns (Hillman et al, 1999).
CFP and CEO compensation:
Issue of CEO pay is a known matter and many scholars write about its factors, pay-for-performance principle (PP link) or related effects (Milbourn, 2003). Jensen and Murphy (1990) are among the firsts who examined the predictions of solution to agency problem in the light of PP link. They found a significant link between CEO compensation and firm performance. In the recent years, executive pay has drastically increased in the US (1990s and 2000s) and justification of this is looked into, by researchers, through the change in firm size, performance, industry growth and many more (Bebchuk & Grinstein, 2005). According to Gabaix and Landier (2008), the sixfold increase of CEO pay between 1980 and 2003 can be completely attributed to market capitalization of large US firms. They could find a very small dispersion in CEO talent through these years. Scholars examine the difference in patterns of CEO pay and there is a large amount of heterogeneity due to company’s size, industry, etc. (Gabaix et al. 2008; Tosi et al., 2000; Tosi Jr. et al, 1989 and many more). Many researchers criticize the CEO pay to be linked to company performance (Carpenter & Sanders, 2002) by exploring that there does no relationship exist. It is a controversial issue and has given rise to corporate governance questions about companies in the USA (Core et al., 1999; Kang et al, 1995). According to agency theory, contracts ought to motivate
15 the agent to aim for higher profits of principle and it should share the risk between the two parties to avoid a total loss situation for principle by agent’s mistakes (Eisenhardt, 1989). This ensures that CEO’s willingness to take more risks only in the best interest of the company.
CEO pay has many complex components and aim behind such a complex salary structure is to make sure the executive works in the interest of shareholders. Top managers are hired to maximize financial performance of a firm but a CEO is also a human who is concerned of his own utility of pay. Their concern should be aligned with the firm to decrease the agency cost (Sigler & Sigler, 2015). Accounting based measures do not yield very good evidence of relation between CEO compensation and financial performance but economic and market returns see substantial increase in the consecutive years of bonuses given to CEOs based on after-tax gross profits and good market performance of a company, respectively (Abowd J., 1990).
According to a study by Wade et al. in 1997, companies that pay their CEOs a very huge amount of salary are more likely to be large firms with many stakeholders. This creates a relationship between such companies and external environment more strained through the light of benefits to its stakeholders. Pay of CEOs should reflect their efforts and often it is found false (Fombrun and Rindova, 1998). Some scholars (Jensen and Murphy,1990) suggest that such high salaries of CEOs are justified by the financial impact they have on companies. But other studies (Holman & Jenkins, 2002) claim to have no relationship between the two. The most dominant approach in financial economists is that the managers pay can be a solution to agency problem (Bebchuk & Fried, 2003). Also, CEOs in public firms are paid 30% more than in private firms and at the same time, the pay-performance link is weaker in private firms than in public firms (Gao Huasheng and Li Kai, 2015).
But this complex pay structure also reflects the board’s decisions and information that is processed within it. Due to a limited information processing capacity, outcome-based pay is preferred by most of the boards (Seo et al., 2015). Their paper reviews a longitudinal sample of US firms from 1998 to 2005 and proves that when boards are less complex and are in less monitoring contexts, they tend less to link pay of CEO with firm performance. A large control over CEO pay is in the hands of board (Boyd, 1994). According to Murphy (1999), aspect of information asymmetry in the board due to lack of access to information or advice with other members is another reason for them to not to be able to argue on the pay of CEO. Hence, board of
16 directors is the authority to design CEO pay but at the same time, in the board CEO has enough power to influence his own pay plan. As a result, the compensation packages end up to being high (Bebchuk & Grinstein, 2005; Edmans & Gabaix, 2009). The principle determining the amount and frequency of CEO compensation is the relationship between board of directors and CEO which is best described as agency relationship in publicly listed firms. When one party to a transaction, the principle, becomes an authority and assigns the responsibility of handling the work of this transaction to other party, the agent, there occurs an agency relationship (Barney and Hesterly, 1996). Thereby, agent is responsible for the welfare of the principle and his choices affect the transaction’s profits.
My paper expects the relationship of CEO pay linked with shareholder returns to be a solution in order to claim more theoretical as well as managerial contribution. But two agency problems can exist and they are to be understood here. First, the desired goals between CEO (agent) and board (principle) may be in conflict with each other. And second is that the CEO is more risk taking than the board themselves. CEO may not be able to measure the loss that company can incur if such a loss is not his own (Eisenhardt, 1989). This agency problem characterizes the corporate governance choices of the board of directors of firms (Jensen and Meckling, 1976). In order to attempt a resolution to this agency problem, motivation in the agent is the main goal for any company so that they act in the best interest of the principle or here, board. The motivation on this level is mostly aligned to monetary compensations to the CEO (Bosse and Phillips, 2016). Such compensation may include salaries, commissions, stock, options and performance bonds (Eisenhardt, 1989). Motivational theory shows that the agents, if not paid sufficiently or have no personal losses on failure in corporate performance, will not act in the best interest of the company shareholders. According to the literature, there are two approaches of finding the best compensation system for CEOs (Bebchuk and Fried, 2003). The first approach, optimal contracting approach’ states that it possible to design a pay where the CEO compensation is performance linked (pay for performance model, Bebchuk and Fried, 2003). But in reality, the magnitude of CEO pay that is unlinked with performance is so huge and this principle doesn’t work as it should (Edmans & Gabaix, 2009; Holman J, 2002). Second approach of this model focusses upon the CEO power in order to let him influence over the nomination of board as well as the designing of compensation. As long as the compensation is justified visually, board of directors will incline towards the approval of CEO pay arrangement proposed by him. This
17 managerial power approach also says that the board members lose an opportunity to be a board member in other companies if they do not favor the decisions of CEO (Jensen and Murphy, 1990). Westphal and Stern (2006, 2007) find that TMT executive who are closer to CEO are more likely to receive board appointments in other firms where CEO is appointed as a director. In a paper by Choe Chongwoo et al (2014), authors flesh out the rest extraction view of CEO compensation put forward my managerial power theory and test its main implications on relationship between CEO power and compensation structure.
As an illustration, CEO bonds is a prominent kind of compensation structure in China where around one-tenth of corporations deploy performance bonds. It is a prototypical solution to the principle-agent problem where a bond is like an up-front promise made by CEO as an individual which is recoverable as per his performance (Bryson et al, 2014). CEO puts his personal wealth in the hands of performance measurement and therefore, the firm may not face any negative consequences at the cost of shareholder’s wealth. Theory also says that a potential gain is lesser effective that a potential loss when a CEO incentive compares to CEO bonding (Kahneman & Tversky, 2013, revised paper on Prospect theory).
Board Diversity (the moderator)
There exist two varied views in the literature about heterogeneity in board and corporate political activities of a company. A number of researches suggest that diversity in board is an important factor in strategic decision-making process (Cho, 2006, Kor, 2006 and others). Research also suggests that diversity of board brings in more emphasis on creativity and innovation (Bantel and Jackson, 1989). Basically, diversified views lead to heterogenous ideas, new solutions and more comprehensiveness in decision-making (Wiersema and Bantel, 1992). On the other hand, a number of researches state that the cost of board diversity is high and there exist many conflicts, struggles and communication issues due to different attitudes and values (McCain et al, 1983, Pfeffer, 1983, Simons T et al, 1999). As a result, it can hinder effective and speedy decision making.
18
Theoretical Framework
Until now, literature has no been developed so much in the area of CEO compensation and its relationship with CPA keeping corporate governance in mind. Linkage of corporate governance with CPA is a relatively new topic in recent research. Several controversial concerns have encompassed CEO pay in public firms and agency problem is a visible issue for which pay for performance linkage is a popular method suggested by many authors. But performance itself has many antecedents and non-market strategies are some of those. According to principal agent theory, firms seek to achieve two objectives while making a compensation structure for CEOs. First, they want to give incentives to managers by tying the rewards with firm’s success. And since, agents may risk a firm’s investments, firms want to tie it to manager’s personal benefits that keeps him on toes before risking the firm’s performance. Therefore, such bonding is positively correlated to CEO’s control on firm’s performance. The major concern is the empirical gap and absence of studies of the antecedent of firm’s performance i.e. CEO compensation in terms of his motivation that may trigger him more into many strategies. These strategies include corporate political activities also and therefore, corporate political activities, as exemplified in LR, are positively correlated to firm’s performance. To date, most comprehensive study of CPA antecedents is given by Lux, Crook, and Woehr in 2011. Separately, CEO compensation related studies are done by many authors but among the popular cited studies is Coombs and Gilley (2005). Both the papers have a common link i.e. firm’s performance. By cross-sectioning the antecedents from these papers, I picked out the main motivation for CEO if the PP link is considered is firm’s performance (Abowd and Kaplan, 1999). One of the studies on Chinese firms by Bryson, Forth and Zhou (2012) puts forward some descriptive insights on companies that post corporate bonds in CEO compensation but deeper studies on USA public firms are still missing that link CPA and corporate governance, especially, CEO compensation. Here is where my dependent variables of CPA and independent variables of CEO bonding come forward in one link as to how much variance can occur in firm’s CPA if CEO’s pay is bonded with shareholder returns. This link will compliment the studies that are done on CEO compensation and its effects on firms if it is linked with shareholder returns complimentary to other studies like those of Coombs and Gilley (2005). This
19 paper digs deeper into all the political activities by a firm and therefore, it takes CPA total, lobbying and PAC contribution separately in consideration as per the following hypothesis: H1: CEO bonding positively affects firm’s cumulative CPA (sum of lobbying expenditure and PAC contribution)
H2: CEO bonding positively affects firm’s lobbying expenditure. H3: CEO bonding positively affects firm’s PAC contribution.
As per Chen et al (2014), lobbying expenditure is higher than PAC contribution but this study focuses on change in both the CPAs and overall CPA because CEO compensation is an important variable in corporate governance and its components relate to PAC contributions to different issues. I expect that CEO bonding significantly and positively affect lobbying as well as PAC contributions. Here we do not consider the conditions of when CEO bonding affects CPA more but the quantifiable magnitude of it that affects the variance in CPA. This eventually changes firm’s performance which mathematically increases or decreases compensation of CEO.
Secondly, board structure and CEO compensation have a vast literature written linking them together by many authors (Boyd, 1994; Zajac, 1990; Chhaochharia, V., & Grinstein, Y., 2009; Brick, Palmon and Wald, 2006; Coco and Christian, 2006 and many more). Also, board’s diversity has been explored for its relationship with firm performance by Erhardt and Werbel, 2003 and others but effects of this on the corporate political activities of a firm are still not explored (Upadhyay & Zeng, 2014; Sealy, Doldor and Vinnicombe, 2009; Campbell and Minguez-Vera, 2008 on gender diversity in boards, etc.). This could be due to the allegedly less diverse boards of USA and largely CPA literature belongs to the population of firms from the USA. But since this is a new venue to discover, it is interesting to see how diverse boards have effects on relationship between CEO compensation and CPA of a firm.
H3: Board diversity positively affects the relationship between CEO bonding and firm’s CPA overall i.e. cumulative CPA.
H4: Board diversity positively affects the relationship between CEO bonding and firm’s lobbying expenditure.
20 H5: Board diversity positively affects the relationship between CEO bonding and firm’s PAC contribution.
There is an evidence that board diversity is related with innovation and other strategic decisions like growth (Bantel and Jackson, 1989) in a company but there is not a very large research database on how it is associated with CPA. Diverse teams may have diverse sources through which interpretations and perspective may vary about CPA and thereby, top management teams do not recognize the importance of political activity of a firm in a consensus (Chataway et al., 2004). Ozer (2010) argues that TMT heterogeneity negatively moderates the relationship between TMT involvement in political activity and firm’s CPA. Our paper will explore that when the responsibility of firm’s performance goes in the hands of CEO through performance linked pay, does board diversity still affect the relationship. Gordon et al. (2007) says that senior executives have a higher impact on CPA decisions. So, Top management teams that are more interested in governmental activities will be more into CPA (Blumentritt, 2003). Literature doesn’t specify the TMT characteristics that affect CPA but as per Ozer (2010) firms differ in CPA depending on CEO and TMT characteristics.
Diagram of the MODEL:
CEO pay linked with Shareholder’s returns
(Bonding)
Lobbying Expenditure
Total PAC contribution Board Diversity
Firm Size (No. of
employees) Industry regulations
Dependence on Govt. H3 H5 H6 H2 CPA Overall H1 H4
21
Data and method:
This section provides an overview of the methodology that I used in analysis of the data to support my hypotheses and it covers all the variables, data sources, and definition of the sample that I selected. This study relies on quantitative database research and is exploratory. The data is collected from multiple resources and any missing data is completed through the help of corporate annual reports of the cases in the sample.
Data collection and sample:
The dataset I use in my research is cross-section and it consists of S&P 500 index firms from year 2012 to 2015. The data is collected and cleaned by students working under the supervision of Dr.I. Haxhi on Triple-C project. It involves a dedicated participation of students to collect more than 450 cases that comprised of public traded companies in the US. These firms participated in lobbying as well as made PAC contributions during the time period. The data for PAC contributions that I use is from two years 2012 and 2014. The S&P index is known to be the best measure of big firms in the US capital market. Also, its represents most of the capital market and must capture majority of political activities by public firms in the US.
Therefore, for my research, this data set justifies the patterns of all the firms in the US. The data is extracted from opensecrets.org which is maintained by Centre for responsible politics (CRP). The Lobbying disclosure act of 1995 requires the registration and reporting for those who want to affect policy making in the US or implementation of Federal programs. Those who register are required to file semi-annual reports of lobbying expenditure they make and who they lobby for. CRP categorizes date on this website by lobbying, PAC donations, and soft money. CRP sums up the mid-year and year-end data for the total amount of expenses reported by firms to get the final annual figure for each firm. Hence, this website provides the most comprehensive data in terms of CPA figures. It retrieves data from corporate disclosure reports collected by Senate’s office of public records as declared by firms themselves. For figures of CEO bonding and finances, I have used reputed resources like COMPUSTAT, Worldscope and Datastream that provide firm’s financial data. The data is cleaned with the help of financial reports of companies listed in the sample, removal of missing values if not available and few missing data is filled in from reliable resources like corporate websites of the firms. The final complete dataset that is tested in this
22 analysis consists of 295 cases. Some of the criteria for moderating variable will be based on previous researches with references. An OLS regression analysis will be done in order to test the first two hypothesis. Regression determines the change in dependent variable with respect to changes in one or more independent variables.
Dependent variable, Independent variables and moderating variables as defined by above diagram are controlled by Industry type and firm size. Since firm size is an important determinant of lobbying expenditure (Hansen and Mitchell, 2000), it is important the hypotheses are controlled by it. Industry type is another control variable because every industry has different regulations on Lobbying and that’s why, it is important to view industry to be controlled since it affects lobbying behavior of a firm (Schuler et al, 2002).
Dependent variables:
This study has three dependent variables to analyze. Lobbying expenditure by a firm in a year is the first variable which is an important variable to determine firm’s financial performance (Chen et al, 2015; Hillman and Hitt, 1999). Second variable that I use to analyze CEO bonding is PAC contribution and according to Aggrawal et al, 2014, it is suggested to use PAC contribution to examine the financial performance of a firm. For this stewardship-based relationship of CEO with CFP and CPA, PAC contributions can add up to the conclusions more strongly (Schuler, Rehbein & Cramer, 2002).
I use total CPA as the third variable to conclude the best effects of CEO bonding on political (market strategies) of a firm overall (Hillman and Hitt, 2004). In a firm, the two non-market strategies, lobbying and PAC donations, are combined due to a large number of factors affecting it, like industry concentration, its size, politically active leaders, unionization, congressional caucus, free cash flow and government dependency (Shuler et al, 2002). Therefore, a large set of firms visibly invest in CPA as a whole. I aggregate the two activities into one CPA overall by using a log operation.
Independent Variable:
For the left-hand side of the model, I use CEO bonding i.e. CEO pay linked to the firm performance (Edmans and Gabaix, 2009; Fong, 2010; Albuquerque, 2009). Here, I use dummy variables for whether the CEO pay is linked to the firm performance or not. Due to limitation of
23 data collection, I use shareholders wealth as a measure of firm performance. If a CEO pay is linked, the variable will take place as 1 and if not, it will obtain a value of 0.
Control variables:
Due to nature of my research, I control the variables that can vastly affect firm’s performance and board characteristics. Also, the theoretical literature predicts that differences in firm characteristics could lead to differences in CEO compensation structure (Smith and Watts, 1992; Core et al, 1999). First variable that I control for is Firm size which is most commonly controlled due to its impact on results of a varied dataset. Bigger firms and smaller firms may have different economies (X. Zhou, 2000, Chao and Kumar, 2010). CEO compensation is conditional upon firm size (Finkelstein & Hambrick, 1989; Lambert, Larcker and Weigelt, 1991; Hallock and Torok, 2010) and large firms tend to engage in CPA most actively and highly in order to avoid downsizing cost if they lose market share (Schuler, 1996). For firm size to measure, number of employees is used in this study. Firm size is the most popular antecedent of CPA whether measured by sales (Bhuyan, 2000; Hansen & Mitchell, 2000; Schuler, Rehbein & Cramer, 2002), assets (Meznar & Nigh, 1995), market share (Schuler, 1996), or number of employees (Bhuyan, 2000; Hillman, 2003; Meznar & Nigh, 1995). This prominence and high correlation can also be seen later in the data analysis in this paper. Larger firms tend to be more politically active due to larger resources, ability of the firm to be politically engaged (Schuler and Rehbein, 1997), more stakeholders and high economic power to influence more public policy (Hillman and Hitt, 1999). Although a few studies (Cook and Fox, 2000) have also found that smaller firms are highly politically active to my surprise, with they also tend to join others to participate in CPA because logically, CPA is directly proportional to influence on public policies that are favorable to companies.
Other control variable is Government dependence of a company (Stigler 1971, Peltzman 1976, McChesney 1987). It is very crucial to control for this variable because CPA, as explained before, is directly proportional to influence of a firm on public policies that are favorable to them. High government dependence results into more benefits if a firm has influence over public policy changes and vice-versa. Therefore, government dependence is measured in numbers and is controlled for my hypothesis testing.
24 Last control variable that I find relevant to control with reference to studies by Schuler, Rehbein and Cramer (2002) or Macher, Mayo and Schiffer (2011) is industry regulation. In order to isolate the effects of regulated or non-regulated industry type, I use industry dummies of 0 or 1. A regulated industry takes up the value of 1 and an unregulated industry takes up the value of 0. This is important to control in order to get the corrected linkage variance between the cyclical relationship of CPA, CFP and CEO compensation. According to Lenway and Rehbein (1991), regulation of industry and concentration of industry determine a firm’s decision to engage in CPA strategies. Contribution in CPA results in abnormal positive returns around the disclosure of such an event and such positive returns are more in heavily regulated industries as compared to non-regulated or loosely non-regulated industries (Bonardi, Holburn and Bergh, 2006).
Moderating variable:
Diversity in the board including gender, nationality, age, composition of board in terms of independent vs other members, etc. (Benkraiem R et al, 2017) is an important variable that affects the selection of management. Pfeffer and Salancik (1978) provide an overview of a wide range of resources that directors bring to a firm and its antecedent is diversity of the board. Each resource in the board has its direct effect of firm performance (Westphal, 1999; Hillman and Dalziel, 2003) and this moderator is a required variable for my study because the social ties (in addition to corporate ties) between board and CEO should increase board’s increase in the interest of CEO’s actions. It results in a mutual motivation to provide more market as well as non-market resources (Westphal, 1999). It fits the model because TMT has an important stand in deciding the compensation structure of CEO (Chhaochharia, V., & Grinstein, Y., 2009; Boyd, 1994; Guthrie, K., Sokolowsky, J., & WAN, K. M., 2012; Core, J. E., Holthausen, R. W., & Larcker, D. F., 1999 for relationship of firm performance in the linkage) and it also has an important effect of corporate political strategy of a firm (Ozer, 2010; Lux, Crook and Woehr, 2011).
Method:
I plan to test all the hypotheses by performing multivariate multiple linear regression on each dependent variable represented by the following equation:
25 Where Y is the dependent variable. CPA overall is measured as the total of base-10 logarithm of lobbying expenditure and PCA contribution since the two constituents were not normally distributed. Lobbying expenditure (2012-15) and PAC contributions (2012 and 2014) are averages of the years in consideration taken individually. X1 is the independent variable CEO bonding which is the dummy variable for CEO pay linked with shareholders return and X2 is the moderator variable, board diversity operationalized in dummy variables. β1 represents direct effect of CEO bonding on dependent variable (each) and β2 indicates the direct effect of board diversity on dependent variable. β3 represents the interaction effect of CEO bonding and board diversity. X3, X4, X5 are the control variables and their direct effect on CEO bonding is expressed by β4, β5 and β6. β0 indicates the intercept and ε eventually is the residual error that accounts for effects on regression that may not be caused by any of the variables in this regression analysis. The model in entirety is visualized in the table as follows that displays the different models used in the subsequent analysis.
As shown below (in Table 1), I use a stepwise approach for resting the relationships in hypothesis stated before. Model 1 expresses solely the control variables i.e. firm size expressed in the number of employees, industry regulation, government dependence and dependent variable, CPA in totality. Model 2 adds the independent variable, CEO bonding to the control variables. Model 3 eventually adds moderator, board diversity to the interaction of control variables with dependent and independent variables. These models are subsequently run for two dependent variables, lobbying expenditure and PAC contribution and a total of 9 models are shown (Model 4-9). I conducted further analysis in SPSS using the PROCESS tool given by A. Hayes (Hayes, A. F., 2017).
26
Table 1: Summary of Regression Models
Control variables Dependent variables Independent
variable Moderating variables Firm size Industry regulation Government dependence CPA Overall Lobbying expenditure PAC Contribution CEO bonding Board diversity Model 1 X X X X Model 2 X X X X X Model 3 X X X X X X Model 4 X X X X Model 5 X X X X X Model 6 X X X X X X Model 7 X X X X Model 8 X X X X X Model 9 X X X X X X
Results and Analysis:
Before analyzing the hypothesis using the data, running a normality test is important for the model to show “goodness of fit” and it is an underlying assumption that regression is to be run on normal data. (Judge et al, 1982; D’Agostino, R. B., Belanger, A., & D'Agostino Jr, R. B., 1990). Since the data was not normal, log10 transformation have been used to transform the data that was positively skewed data. The variable Government dependence (control variable) was majorly skewed and hence the P-P plot was first log transformed similarly. All the P-P plots are observed for normality and log transformed variables passed the tests. Transformed results of variables successfully passed numerally performed Shapiro-Wilk tests (explore) and histograms visually showed a normal data.
Descriptive statistics
The descriptive statistics are shown below in table 2 for independent, dependent and control variables. All three dependent variables and one of the control variables are transformed into log10 function, results are required to be transformed back to exponential numbers raising the log by 10. Mean of average CEO bonding is .773 and has a standard deviation of .367 which shows that a smaller variance exists in CEO bonding and data is around the mean overall. Since the
27 variable takes only two values 0 and 1 respectively for compensation not linked with shareholder returns and those linked with shareholder returns, 77% of the sample firms have CEO compensation bonded with shareholder returns. The mean of lobbying expenditure is USD 1341837.73 (106.1277) from year 2012-2015 while the mean of PAC contribution is USD 162554.88 (105.2110) of years 2012 and 2014, which implies a significantly high investment in lobbyists and lesser PAC donations made by the firms. Standard deviation in both the dependent variables is approximately .61. Minimum lobbying expenditure is seen to be USD 7585.78 and maximum is USD 17378008.28 which displays a vast range of lobbying resources allocated by firms. Similarly, minimum PAC donations are USD 758.58 and maximum PAC contribution is USD 3090295.43. Average of the moderating variable, that is, board diversity is 19.22 with a high standard deviation of 8.6 and this number was expected as the boards in USA are less diverse and cases are vastly different from the mean showing there is still a high variance in the diversity among boards across the firms in the sample (Zahra and Pearce, 1989). Industry regulation scores a mean of .35 which is a binary variable that should take up a value of 0 if industry is not regulated and 1 if an industry is regulated. This means that 35% of the firms of this sample are from industries that are regulated and remaining 65% are not from industries that are regulated.
Table 2: Descriptive statistics
Descriptive Statistics
N Minimum Maximum Mean Std. Deviation Lobbying expenditure* 295 3.88 7.24 6.1277 .61912 PAC contribution* 295 2.88 6.49 5.2110 .61746 CPA Overall* 295 7.12 13.60 11.3387 1.13977 AV_CEO_Bonding 295 .00000000 1.000000000 .77288135593220 .366631856535168 Av_Board_Diversity 295 .00000000 61.79750000 19.221929378531 8.60695960826045 Firm Size* 294 -.48 .52 .1440 .16733 Government dependence* 41 -3.21 -.09 -1.5103 .94299 Industry regulation 295 0 1 .35 .478 *Log10 functions
28
Correlations and Multicollinearity
Multicollinearity is most important to test when two or more independent variables are correlated to each other but in my thesis, there is only one independent variable, CEO bonding. Yet, I carry out Pearson’s test for correlation to identify correlations between all the variables in consideration in my dataset. A threshold of .7 is called to be sever multicollinearity and any conclusion drawn from severe correlated variables can be biased to prove the power of β coefficient’s value (Field, 2013) unless the logical reasoning behind those variables proves otherwise. Multicollinearity, tolerance and VIF figures are shown in the subsequent tables. VIF values should be lower than 5 while tolerance should exceed 3 (Bowerman and O’Connell, 1990; Field, 2013). Average of VIF should be more than 1 substantially (Field, 2013).
As table 3(i) shows, the tolerance cutoff is exceeded by all the variables except for two logically correlated variables Lobbying expenditure and PAC contribution. Tolerance of Lobbying expenditure and PAC contribution almost reached the minimum threshold and the same is mirrored in their VIF figures. None of the values exceed the cutoff (Field, 2013) as mentioned above and therefore, I was not required to remove any variables from my regression model. For all the other variables that are CEO bonding, Industry regulation, Government dependence, Firm size and board diversity, the tolerance lies between .5-.9 showing a low to medium linear relationship between the variables. Average VIF as shown in the table comes out to be approximately 2, I infer that multicollinearity is not a major issue in this research.
Table 3 (i): Tolerance and VIF
Coefficients
Model Collinearity Statistics
Tolerance VIF 1 AV_CEO_Bonding .671 1.491 LogLob .259 3.858 LogPAC .252 3.963 INDUS_REG .781 1.281 Av_Board_Diversity .681 1.468 LogLGEMP .727 1.375 LogGDEP .583 1.716
29
Column1 LogLob LogPAC LogCPAoverall AV_CEO_Bonding Av_Board_Diversity LogLGEMP LogGDEP INDUS _REG LogLob LogPAC .699** LogCPAoverall .922** .921** AV_CEO_Bonding .179** .196** .203** Av_Board_Diversity .199** 0.059 .140* 0.018 LogLGEMP .263** .332** .323** -0.064 .162** LogGDEP -0.114 -0.044 -0.087 -.390* -0.215 -0.242 INDUS _REG -.136* -0.076 -.115* -.132* 0.053 -0.035 -0.230 **. Correlation is significant at the 0.01 level (2-tailed).
*. Correlation is significant at the 0.05 level (2-tailed).
Table 3 (ii): Correlations and Multicollinearity
Table 3(ii) shows the correlations between all the dependent, independent, moderating and control variables. All the variables show low to medium correlation with an exception of three dependent variables i.e. Lobbying expenditure and PAC contribution and there cumulation, where r> .699, r> .922 and r>.921. According to statistical literature, the value is high but a logical reason behind these values is that the firms engage in corporate political activities as a non-market strategy at the same time with similar or closer intensity. It is very likely and was expected to show a correlation. Since, CPA overall is a cumulation of Lobbying and PAC contribution, it is logical to say the correlation will exist and it will not result in any biases whatsoever. This is one of the limitation of this study and conclusion is to be drawn keeping this limitation in mind. Since the correlation study is not all of this analysis, further models will explain further about the relationship of these variables. It will be not justifiable to remove them on the basis of these figures.
This table further shows that dependent variables, lobbying expenditure, PAC contribution and CPA overall are positively correlated to independent variable, CEO bonding at r= .179, .196, .203 at p<.01 indicating a significant positive relationship between the dependent and independent variables. This shows that the firms that have CEO compensation linked to shareholder returns tend to spend more in corporate political activities than the firms that do not keeping all other factors static. This is in line with my hypothesis that I try to analyze in this thesis. Though, this is only an initial level where I can discuss the association but in further models I will analyze the causal relation between these variables.
This table also shows that if CEO compensation is bonded with shareholder returns, a firm tends to invest a little higher in PAC contribution than in lobbying. Therefore, it is most important
30 to test further the direction of this relationship of CEO bonding to know how different strategy (lobbying or PAC donation) is affected differently by CEO compensation. In the table, furthermore one can see that Board diversity is positively correlated with both the dependent and independent variables but a caution is to be exercised hereby since the relationship is not significant and at this stage no cause-effect relationship can be proved. At p<.01 it is no significantly correlated to CPA overall but only to lobbying expenditure. But with CEO bonding it is not having any significant relationship. A positive relationship of board diversity indicates the assumption that it may co-evolve with lobbying. For moderating variable’s effect, it is an early stage to conclude any relationship. As logically expected, regulation of industries and lobbying has a negative relationship at p<.05 with r= -.136 while one of the surprising result is that there is no correlation that exists between government dependence and CPA (any). Firm size, which is explained previously to be an important factor in CPA, significantly correlated to lobbying expenditure (r= .263), PAC donations (r=.332), Overall CPA (r=.323) and board diversity (.162) at p<.01.
I explain these results by stating that firms that grow in size tend to spend highly on non-market strategies like CPA. And for the relation between firm size and board diversity, a simple fact can be stated for contemporary years is that as the size of the firm increases, a diverse board is seen more than in smaller firms with lesser number of employees.
Regression analyses:
This section provides the analysis of the proposed hypotheses by performing three hierarchical linear regressions on three dependent variables (CPA overall, lobbying expenditure and PAC contribution). It draws conclusions regarding the hypothesized relationships as stated before. Overall fit of the model is to be tested to know the models can account for some variance (Field, 2013) and this is denoted by R2 and adjusted R2 as shown in the table 4. PROCESS (A. Hayes) of SPSS doesn’t use adjusted R2 so I use R2 for interpretation of models 3, 6 and 9 when
moderating variable is to be studied. R2 falls between .2 to .3 for CPA overall, around .2 for lobbying expenditure and between .19 to .3 for PAC contributions which is moderate. None of the R2 values are low or negative for any of the models. Purpose of regression is to study the difference between lobbying expenditure, PAC donations and overall CPA as corporate political activities’
31 Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8 Model 9 Firm S ize 3.771** 3.75** 3.452** 2.113 2.063 1.923** 1.658 1.687 1.529 Government dependence -0.019 -0.027 0.111 -0.041 -0.061 0.004 0.022 0.033 0.106 Industry Regulation -0.399 -0.403 -0.223 -0.302 -0.313 -0.228 -0.097 -0.090 0.005 CEO bonding -0.050 0.294 -0.119 0.043 0.070 0.252
CEO Bonding x Board diversity 0.043 0.200 0.023 Constant 11.059 10.577 10.346 5.979 5.791 5.592 5.401 5.354 4.885 R square 0.229** 0.229 0.310 0.233** 0.236 0.287 .194** 0.196 .306** Adjusted R square 0.166** 0.144 0.211 0.170** 0.151 0.186 .129** 0.107 .207** F-S tatistic 3.663** 0.009 4.096 3.738** 0.158 2.529 2.972** 0.207 5.55** **. Regression is significant at the 0.05 level (2-tailed).
three different levels where the third is the cumulative CPA as a non-market strategy. I keep the control variables constant for all the dependent variables to study the relation without biased results. Here now, more details of lobbying expenditure and PAC contribution models are provided. R2 values of lobbying expenditure as a dependent variable are consistent while PAC contribution ones are ranging from .194 to .306 at significant level p<.5. While the R2 value of lobbying expenditure’s models 4-6 ranges around .25 (lower or higher), R2 value of PAC contribution ranges from .194 to .306. It is also important to note that model 9’s R2 value is significant at p<.5 with a value = .306. This shows that fit for model with PAC contribution as a dependent variable is better than that for lobbying, as those models account up to 30% of variance. Yet, up to 29% of variance can be accounted for model of lobbying expenditure in models 4-6. Adjusted R2 values are almost the same for all the models but for Model 7 and Model 8, that is, 12% and 10%. In this thesis study, as already shown in equations of regression, β coefficients indicate the effect of independent variable on dependent variables. When control variable Government dependence is being used for any model in regression, many missing values have been removed and a smaller N of 41 is being used. While analyzing with mean on place of missing values in government dependence, the results have not been massively different to change the conclusion. Results of all the models 1-9 are presented in a consolidated table 5 and at the end of this paper, further appendices are presented to explain in further.
Table 4: Regression analyses (β) results for CPA overall, lobbying expenditure and PAC contribution
32
Linear regression 1: For β of CPA Overall
The first round of regression is conducted to test the two hypotheses that relate to total CPA overall i.e. the sum of lobbying expenditure and PCA contribution. As hypothesized in H1, CEO bonding is sought to have a positive effect on CPA of a firm. This relationship is tested in Model 2 and results indicate no support for this relationship. However, the result is non-significant at p<.05 level. Therefore, no variance in CPA cumulative can be explained by CEO bonding exclusively. I can infer with 95% certainty that by CEO bonding there will be no positive significant effect on firm’s CPA overall which is why it is more interesting to study both the types of CPA activities in a firm. Similarly, as hypothesized in H4, a diverse board was sought to have positive effect on relationship between CEO bonding and CPA overall. As shown in the table above, β=.043 (not significant at p=.05) shows no such evidence that proves in favor of this hypothesis. At best, I can say that is a minor effect of board diversity but it is insignificant rejecting my hypothesis 1 and 4.
Linear regression 2: For β of lobbying expenditure
Second round of regression is conducted to test the two hypotheses that relate to lobbying expenditure. Results are shown in the consolidated table as shown above and model 4-6 relate to this variable. Model 5 results show the insights of relationship between CEO bonding and lobbying average expenditure of a firm and the resulting β is negative, however, insignificant at 95% level of significance. Neither adjusted R2 nor F-statistic are significant for Model 5 (R2 = 151; F-stat =
.156). I can thus conclude with these results that there is no significant relationship proven between lobbying expenditure and CEO bonding which fails hypothesis 2. Yet, when I analyze the moderation effect of diversity in board for a firm, it shows a positive β of .20 but the results are insignificant at p<.05. It can, with no significance, be said that 20% variance of relationship whatsoever between lobbying and CEO bonding is explained by diversity in board of directors of a firm.
Linear regression 3: For β of PAC contribution
Similarly, model 7-9 results are derived from the regression analysis conducted on dependent variable PAC contribution that gave the most significant results in favor of the hypothesis that I tested to find out if there is a positive effect of CEO bonding on PAC donations
