Corps de l’article

Executive compensation has been an issue largely debated by the public and the academics. It is usually criticized for being excessive and for providing perverse incentives that could destroy the firm’s value (Bhagat and Romano, 2009). However, defining “excessive compensation” is still challenging and disputed. In, 1993, US congress defined it as “non-performance –based compensation of more than $1 million” (Harris and Livingstone, 2002). Nichols and Subramaniam (2001) argue that a CEO compensation is perceived “excessive” if: “(1) the levels of compensation are unfair compared to those of other workers or (2) the amounts are unjustified when compared to the firm’s or the CEO’s performance”. They detail arguments to which a compensation is considered excessive. First, the large disparity between CEO and average employee’s compensation could be a sign of excessive compensation. Second, comparing domestic and foreign compensation could be a lead in determining if the compensation is excessive. Nevertheless, it would be pointless to address curbing a CEO’s excessive compensation, without defining the extent to which a compensation is considered, in fact, excessive (Markham, 2007).

To prevent from such excessive disparities, laws and corporate governance codes have been published and modified in order to improve the efficiency of corporate governance mechanisms. In France, the law on New Economic Regulation introduced in 2001 mandates French listed firms to report on the level and components of their executive compensation. The board of directors is the core entity that issues strategic decisions. That is why its composition is very important as to how it affects corporate decision-making, especially those related to executive pay. In fact, the AFEP-MEDEF[1] codes advocate a certain balance in the composition of the boards of directors and their committees. These codes suggest that, at least half the board members should be independent and encourage the board gender diversity. More recently, the “Financial Markets Authority”[2] published its yearly report on corporate governance and CEO compensation (AMF, 2019) supporting a stronger participation of employees directors on boards. It is primarily through the May 22nd 2019 “PACTE” law that employee participation arrangements have been reinforced. This law amended the L. 225-37-3 article of the commercial codes, introducing new requirements, such as comparing CEO compensation with the employees’ median (or mean) salaries, and therefore asserting a CEO-worker pay ratio. The “PACTE” law also increases the number of employee representatives to at least 2 members for boards of directors with 8 (formerly 12) or more members and at least one representative for boards with less. In the French context, we note the important role of AMF and AFEP-MEDEF in promoting the employees board participation in the strategic decision-making such as CEO compensation policy. The renewed interest in employee board representation is seen as an important feature of European model of corporate governance (Gordon and Roe, 2004).

The rising interest in board’s composition suggests a positive outcome of board diversity towards strategic decisions. Scholars have demonstrated how board diversity improves decision-making and corporate performance (Huse et al., 2009). Bear, Rahman and Post (2010) determine how a diverse board empowers diversity of expertise of the board’s members, as they have distinct experience, knowledge, and connections. Therefore, it may enhance board’s innovation and improve its efficiency. It also allows for an effective CEO monitoring, as a more diverse board is more likely to benefit from diverse members’ skills and experience in managing their relationship with the CEO.

Previous literature has evaluated the effect of board composition on CEO pay, studying the influence of board structure characteristics such as independence and gender diversity. A significant limitation to this literature is that these papers adopt an agent-principal perspective, which leads to an incomplete analysis of the role of board of directors when other stakeholders are part of the corporate governance system. This paper takes this limited perspective further by analyzing the effect of employee representation on CEO pay. The paper contributes to the emerging interest in board diversity by noting that employee representatives may improve the board decision-making including CEO pay policy. Most previous studies have focused on the impact of employee ownership on corporate performance (Kim and Patel, 2017; Kruse and Blasi, 1995; Martes, 2012; O’Boyle et al., 2016). However, as far as we know, the link between BLER and CEO pay has not yet been extensively studied in the French context. Toe, Hollandts and Valiorgue (2017) demonstrate that “excessive” employee ownership affects negatively the firm performance and this negative effect is amplified by the presence of employee directors. In addition, the French context has several specific characteristics that could influence the relation between BLER and CEO pay. In fact, previous studies demonstrate a concentrated ownership with a high presence of French State, family, and institutional investors in the ownership structure in many companies (Dardour and Boussaada, 2015; Mard and Marsat, 2012; Broye et al., 2018). For example, Dardour and Boussaada (2015) report that the average voting rights of the first owner is about 36.53%. Taking in consideration these characteristics, our study aims to bridge the literature gap by including the ownership structure in the relation between BLER and CEO pay. We examine a sample of French firms listed on the SBF 120 index over the period 2010 to 2017. First, we found a negative relation between employee representation and CEO total compensation level. Second, employee representation has no effect on firm financial performance. Third, we found that French State strengthens the control of BLER over CEOs and their compensation, rather than simply mediate it.

The remainder of this paper is structured as follows. The second section presents the literature review and hypotheses development. Section 3 discusses data and sample selection. In section 4, we present and discuss regression results. We also present further additional results in section 5. We conclude our research in section 6.

Background and Hypotheses

Employee Representation in Continental Europe

Board-level employee representation (hereafter BLER) is not a common practice to all European countries. Fourteen countries have widespread BLER, which are Austria, Czech Republic, Germany, Denmark, Finland, France, Hungary, Luxembourg, Netherlands, Norway, Poland, Sweden, Slovenia and Slovak Republic. Four countries have limited participation rights, which are Spain, Greece, Ireland and Portugal. The remaining 12 countries[3] have no rights of BLER at all (Conchon, report 121, 2011, p. 11). The trade unions could nominate employee representatives in boardrooms. Besides, in some countries, the final appointment has to be validated by the shareholders annual general meeting, as in Netherlands, Hungary and Germany.

In France, employee representation was first introduced in 1983, and only concerned State controlled companies. In fact, the fifth article of the July 26th 1983 law n° 83-675 discussed board composition of State owned companies and required the election of employee representatives (six representatives in State owned industrial and commercial firms). Afterwards, BLER was extended to privately held companies. The July 2nd 1986 law n° 86-793 promoted the election of employee representatives (one employee/employee owners’ representative for boards with fifteen members or less, and two representatives for boards with more). In the privatization law of 1993, the participation of employees in private companies is further supported, as the board of directors has to elect two employee representatives and one representing employee shareholders (three representatives for boards with more than fifteen members). The laws have, in fact, introduced and defended the presence of employees in the board of directors, but the application of the laws was restrained to few companies. The election of employee representatives (with voting rights) was presented only recently for most French listed companies.

The nomination of employee representatives in the board of directors has been addressed in the June 14th 2013 Law (Loi relative à la sécurisation de l’emploi). In fact, this law mandates firms with head offices located in France that have at least 5,000 permanent employees, and firms with head offices in France and overseas that have at least 1,000 permanent employees, to have directors representing the employees in the board. More recently, the Rebsamen Law of August 17th 2015 focused on this matter and widened its application for firms (including their subsidiaries) with 1,000 employees in France, and 5,000 employees for firms operate in France and overseas.

As for employee-owners representation, it is mandatory for firms that have more than 3% of their shares owned by employees, to include in their board of directors a representative of these employee-owners, as stated in the article 32 of the December 30th 2006 Law n°2006-1770. Over one-fifth of the largest French companies have adopted employee representation in their board of directors, and that the laws and governance regulations implemented in France have “created significant cross-sectional variation in the extent and type of employee board representation” (Ginglinger et al., 2011). A recent study by Eres Group (2016) focusing on the extent of employee ownership in firms listed in the SBF 120 index, shows that 27% of these firms proceeded to at least a capital increase or equity transfer in favor of their employees. France is also first in Europe, as employees hold an average of 3.5% of shares, while the average employee ownership in Europe is 1.6% (1.5% in the UK for example).

Hypotheses Development

BLER and CEO Compensation

Our examination of the relationship between BLER and CEO compensation is rooted in both agency theory and stakeholder theory. Developed by Jensen and Meckling (1976), agency theory suggests that there is potential divergence of interests when a person or an entity (principal) hires another person (agent) to act on their behalf. According to this theory, BLER represents an internal governance mechanism that contribute to CEO monitoring. We build on Fauver and Fuerst (2006) who defend that employee representation provides a powerful means of managerial monitoring. Employee representatives act as monitoring agents, therefore reducing agency costs such as shirking practices, perk-taking and excessive executive pay (Fauver and Fuerst, 2006). Similarly, Germain and Lyon-Caen (2016) discussed employee representation from a principal-agent perspective. They were able to conclude that including employee representatives in the board of directors may be valuable to shareholders, as well as other stakeholders. They also confirm that BLER may solve problems of divergence of interests between CEOs and shareholders, as employees would provide valuable information, and therefore reduce agency costs. Overall, they conclude that employee representation is a valuable governance mechanism that improves corporate decisions and shareholders value. Thomsen, Rose and Kronborg (2016) state that employee representatives have varying degrees of influence on corporate decisions. By facilitating the firm internal information exchange, employee directors might help the board to make beneficial decisions for the shareholders as well as for employees. Also, employee directors have the opportunity to detect and prevent arrangements by shareholders or managers and mitigate worker’s exposure to decisions that might cause a deterioration of their rents (Gregoric and Poulsen, 2019). Thus, employees can interfere in CEO compensation policy and increase corporate governance efficiency. Employee shareholders can also affect board’s decisions regarding CEO compensation as they also have directors appointed to represent them on the board of directors.

Ginglinger et al. (2011) suggest three arguments as to why trade unions involvement in corporate governance can enhance the firm value. First, BLER improves information transfer from employees to board members, as employees tend to provide detailed information of their company to the boardroom. Second, board members can also promote better and more precise information transfer to employees. Therefore, it can lead to more cooperation for employees (Freeman and Lazear, 1995). Finally, BLER can be a motivating factor for workers. Employees are often more motivated when they are granted board seats. As for employee-owners, Ginglinger et al. (2011) argue that when employees become shareholders, they have financial interests as the other shareholders. This alignment of interests between shareholders and employees is value maximizing, as motivation will be greater when employees purchase their company’s shares. Allen, Carletti and Marquez (2009) argue that firms that give importance to their employees is shown to “prosper in competition with purely shareholder-oriented firms” (Ginglinger et al., 2011). Besides, Germain and Lyon-Caen (2016) develop a theoretical model for boards of directors that includes employee representatives. The model portrays the importance of BLER, as it maximizes shareholder value, help in corporate strategies such as long-term investments, and help to align employees’ interests to those of the shareholders.

Alternatively, stakeholder theory suggests extending the agent-principal relationship to other stakeholders. Friedman and Miles (2006) define stakeholder theory as a management theory that focuses on stakeholders’ interests and relationships. It recommends fairness, honesty and generosity in treating all the stakeholders within the company (Harrison, Freeman and Abreu, 2015, p. 859). Harrison et al. (2015) suggest that a stakeholder-oriented strategy is efficient as it is motivating for stakeholders. Accordingly, stakeholders (in our case employees), would be motivated to work harder and be more loyal to the organization if the latter is more concerned about their interests. Consequently, executives’ role will be to manage stakeholders’ relationships and ensure the maximization of their interests. Stakeholder theory advocates the incorporation of personal values in the company’s strategic plans and decisions (Freeman, 2004, p. 234). Benson and Davidson (2010) use stakeholder theory to explain the impact of stakeholder management on firm value and CEO compensation. They suggest that managers would be rewarded in a way that serves stakeholders’ interests as well. They argue that CEOs who consider stakeholder management receive greater compensation than those who do not. However, they found no significant results. Ayuso and Argandona (2007) advocate the involvement of stakeholders in boards of directors, giving them a legitimate voice in respecting their rights, but also providing important resources to their company.

Finally, it is important to take in consideration that an opposing framework could be used to describe the BLER and CEO pay relation: Managerial entrenchment theory. In this perspective, entrenched CEOs would use their discretionary power to affect employees’ decisions regarding their compensation. In that case, the relation would be positive, as CEOs would actively support the employees’ causes and interests, such as granting them stock options. Thus, employees are less critical of the CEO pay issue and their representatives tend to limit their pressure level regarding CEO compensation decisions. Cronqvist et al. (2007) found that “CEOs with more control pay their workers more”. Using a managerial discretion approach implies that BLER would have a positive impact on CEOs’ compensation.

Both stakeholder and agency theories provide support to the role of BLER. The latter could be a governance mechanism that improves board efficiency (particularly regarding CEO pay), as well as CEO monitoring.

Overall, and consistent with the existing theory, BLER is often perceived as an effective CEO monitoring mechanism. We base our hypothesis on an agent-principal approach, but we do not deny that BLER can also be viewed from a stakeholder perspective and also from a managerial discretion approach.

Overall, the above development allows us to posit the following hypothesis:

H1. Board Employee representation has a negative effect on CEO total compensation.

Board Employee Representation and Firm Performance

In 1991, Barney’s paper “Firm Resources and Sustained Competitive Advantage” has led to distinguish the importance of a firm’s resources and capabilities in defining and implementing value-creating strategies, and their impact on the firm’s performance (Barney, 1991; Wernerfelt, 1984). The purpose of this framework is to view the firm from a resource perspective rather than from its industry structure. Considering that Wernerfelt (1984) defines a resource as “…anything which could be thought of as a strength or weakness of a given firm”, employee participation in board decisions could be viewed as a valuable resource. Employee are likely to have different experiences than other directors, and diversity arguments imply that this diversity increases the quality of discussions in the boardroom (Levinson, 2001).

Moreover, employee representatives contribute to affecting board decisions and effectiveness (Huse, Nielsen and Hagen, 2009). Balsmeier et al. (2011) support that employee representation improves relationships and information transfer between employees and top managers, as well as prevents conflicts. The improved communication is two-way, as it will not only help managers to access more detailed and operational information, but also, employees will be more informed and involved in board decision-making. This scenario helps to prevent conflicts between employees and managers and leads to enhance firm productivity and performance. In such case, employees portray a valuable resource to the firm by improving communication, hierarchical relationships, and productivity. Their representation on the board of directors would contribute to value-creating strategic decisions, and therefore, affect the firm’s profitability and performance.

This perspective is reflected in corporate governance literature. In fact, some authors examine the effect of the employee involvement in board tasks on firm outcomes. Previous empirical studies showed mixed results. In the context of the German mandatory model of BLER, Fauver and Fuerst (2006) provide a strong evidence for a positive and significant relationship between BLER and firm market value. On the contrary, Ginglinger et al. (2011) suggest that trade unions directors do not have any significant influence on firm value and profitability, while directors elected by employee shareholders have a significant and positive impact on firm value and profits. Other studies found no significant correlation between employee representation and corporate performance (Berglund and Holmén, 2016; Wagner, 2011; Baums and Frick, 1998). Finally, some authors argue that the divergence of the results can be explained by firm size rather than employee representation (Wagner, 2011).

Therefore, we expect a positive relationship between BLER and firm performance.

H2. Board Employee representation has a positive impact on firm performance.

The Mediating Role of Ownership Types

In this study, we explore the mediating role of ownership structure in the relationship between BLER and CEO compensation. Balsmeier et al. (2011) debate the importance of ownership concentration over employee power, and state that depending on the structure, employee power may differ. They found that the power of employees decreases with the power of the largest shareholder. In France, ownership is mostly concentrated. Mard et al. 2014 found that the major shareholder hold an average of 38.8% of shares of firms listed on the SBF 250 from 2004 to 2008. The concentrated ownership, therefore, may alter the supposed direct effect of employee representation on CEO pay. The major shareholder monitoring might reduce opportunistic managers’ behavior (Nikolić and Babić, 2016, Ishtiaq et al., 2017). Consequently, the nature of the largest owner can help our understanding of the emphasis of such monitoring. In the case of the presence of a largest shareholder, BLER could have different effects on the level of CEO pay. Moreover, we have to consider the nature of the largest shareholder (family, French State or other institutional) to better determine the possible mediation effect. We test the effect of the nature of the major shareholder as a mediator to the employee representation-CEO compensation relation. Therefore, we suggest the following hypothesis:

H3. The effect of BLER on CEO compensation is mediated by the nature of the largest shareholder.

Data and Sample Selection

The study sample is comprised of an unbalanced plan of 112 firms listed on the SBF 120 index during 2010-2017 period. We choose the SBF 120 index for it holds 75% of the market capitalization of Euronext Paris. The study period 2010 to 2017 seems to be in line with the recent laws and the AFEP-MEDEF Corporate governance code which advocating the board employees representation. CEO compensation and corporate governance data were collected manually from the registration documents and annual reports published by the firms of our sample. Financial and accounting information comes from Orbis database. Consequently, the final sample consists of an unbalanced panel of 112 firms and 896 firm-year observations.

Variables and Model

Dependent Variable: Measure of CEO Compensation

For this study, CEO total compensation is the main dependent variable. CEO’s total compensation corresponds to the sum of the base salary, the potential value of stock options, performance shares, and annual bonus.

Independent and Control Variables

Employee Representation Variables

As an independent variable, we use employee representation on the board. Following Dardour et al. (2015), we measure the employee representation by the ratio of the number of seats held by employee representatives by the board size (BLER).

Control Variables

To avoid biased results, and consistent with previous empirical literature, we include the following control variables. Board size refers to the total number of directors on the board. A large board is more beneficial to firms with greater organizational complexity and dependence on external resources (Coles et al. 2008). Larger firms are more likely to institute a larger board in order to provide diversified expertise and to bring in key external resources. However, a small board also avoids the free-rider problem and facilitates decision-making productivity of each board member (Yermack, 1996). Hermalin and Weisbach (2003) argued the possibility that larger boards can be less effective than small boards.

Outside directors is measured as the number of independent directors divided by the board size. Jensen (1993) argue that the presence of outside directors enhances the board’s effectiveness. It also plays an important role in CEO monitoring (Core et al., 1999; Ozkan, 2007). As for CEO’s individual characteristics, we include CEO duality, which equals 1 if the CEO is also chairman of the board, and 0 otherwise. The duality (or separation) of the CEO-Chairman functions have been used in explaining board efficiency, and in particular, compensation decisions (Boyd, 1994; Chen, Yi, and Lin, 2013; Irani et al., 2017). CEO age also seems to be an important factor affecting CEO pay (Al-Najjar, 2017). The CEO’s age is associated to more power (Brockman et al., 2016), wisdom, knowledge, and experience (Kogan and Shelton, 1960; Taylor and Walker, 1998). Older CEOs are usually paid more in terms of short-term compensation plans (Adhikari et al., 2015). Finally, CEO tenure is often perceived as an entrenchment tool for CEOs (Ozkan, 2007). Entrenched CEOs tend to demand higher levels of compensation that benefit their own interests (Allgood and Farrell, 2000). Thus, controlling for CEO tenure seems to be important. Firm size is also an important factor in determining CEO compensation (Jensen and Murphy, 1990). It is measured as the natural logarithm of total asset (Martínez-Ferrero et al., 2015). The firm size is expected to have a significant impact on CEO pay.

Ownership structure was also included in our model. In France, most of the companies have a major shareholder (Dardour and Boussaada, 2017; Mard et al., 2014). Therefore, we use the largest Shareholder’s voting rights. Concerning the nature of the largest shareholder, we implement dummy variables that equal one if the largest shareholder is a Family, the French State, or other institutional investor, respectively. Previous studies found a negative relationship between the power of the largest shareholder and the CEO compensation (Dardour et al. 2015; Dardour and Boussaada, 2017). Similarly to Carpenter and Sanders (2002), Finkelstein and Hambrick (1996), and Mehran (1995), we control for firm performance by using Return on Assets (ROA) which is calculated as the earnings before taxes, divided by total assets. Finally, we have included ten industry dummy variables: health equipment and services, consumer goods, distribution, electronic and electric equipment, community services: industrial transport, manufacturing, software and IT services, travel and leisure, financial services, and media and advertising.

Models

We use the following baseline model to test our first hypothesis.

The second hypothesis suggests a positive relationship between firm performance and employee representation. Therefore, to test this hypothesis, we use a second model with the return of assets (ROA) as a dependent variable instead of CEO total compensation.

Results

Descriptive Statistics

Table 1 reports descriptive statistics of the variables. The independent variable BLER shows that they represent only 2.76% of board members. The mean value of board size is 11.87. The largest shareholder has on average more than 35% of voting rights which confirms the higher ownership concentration in France context (Dardour et al. 2015). Outside directors represent 51.45% of board members.

Table 2 reports the Pearson correlation coefficients. Because none of the correlation coefficients is high enough (> 0.80), we conclude that multicollinearity is not a concern in our analysis. As a supplement to the information presented, the Variance Inflation Factor (VIF) test is provided. The highest VIF value is 2.41 (mean VIF 1.70), which is below the tolerance value of 10, indicating that the results are not biased due to multicollinearity (Kutner et al., 2005, p. 409). As reported below, we found no correlation between the proportion of employee directors on board (BLER) and CEO compensation level (CEO_comp). However, CEO compensation is negatively correlated to the voting rights of the major shareholder, and to the presence of the French State or institutional investors as a major shareholder.

Table 3 reports the evolution of the ratio of board-level employee representatives before and after the June 14th 2013 Law (Loi relative à la sécurisation de l’emploi). In the first window [2010-2013], the part of BLER has a mean of 1.51%. This mean increases to 4.18% in the second period [2014-2017]. The mean-comparison test confirms a difference in the BLER means (significant at 1%).

Multivariate Analysis

In Table 4, the estimations for testing our research hypotheses are displayed. To test the first hypothesis, Model 1 is provided in column 1, which is divided in two estimations techniques (GLS and GMM). Using lagged values of the dependent variable is efficient in dynamic panel models, as GMM estimators are recommended in case of heteroskedasticity issues (Arellano and Bond, 1991). GLS estimators are generally biased, we therefore present GMM as robust estimators.

Table 1

Descriptive statistics

Descriptive statistics

Mean, standard deviation, minimum and maximum values of main variables. CEO_Compensation is the dependent variable, measured as the logarithm of the sum of the CEO base salary, bonus and equity-based compensation attributed over the fiscal year; BLER is the proportion of employee representatives, measured as the number of employee representatives divided by the board size. Outside directors is the proportion of outsiders in the board of directors. CEO tenure is the number of year at this position. CEO age. Firm size measures the company’s size as the logarithm of total assets. Major shareholder is the percentage of voting rights controlled by the largest shareholder. Family ownership is a dummy variable, equal to 1 if the largest shareholder is a Family, or 0 if otherwise. State ownership is a dummy variable equal to 1 if the largest shareholder is the French State, or 0 if otherwise. Institutional ownership is a dummy variable equal to 1 if the largest shareholder is an institutional shareholder other than French State, or 0 if otherwise. Widely held ownership is a dummy variable equal to 1 if the largest and second largest shareholder have less than 20% of voting rights, or 0 if otherwise. *** Significant at 1%, ** at 5% and * at 10%.

-> Voir la liste des tableaux

Table 2

Correlation Matrix

Correlation Matrix

***, **, and * indicate statistical significance at 1%, 5%, and 10% significance level, respectively

-> Voir la liste des tableaux

Table 3

Mean-Comparison test before and after the June 14th 2013 Law (Loi relative à la sécurisation de l’emploi)

Mean-Comparison test before and after the June 14th 2013 Law (Loi relative à la sécurisation de l’emploi)

-> Voir la liste des tableaux

It shows that the results agree with our predictions, presenting a negative coefficient, namely, that employee representatives on board negatively affect the CEO pay level. The second column of Table 3 demonstrates results for Model 2. We also conduct two estimation techniques for Model 2. However, regressions provide no significant results for the predicted relation.

With regard to control variables, Board size, outside directors, CEO Tenure and Age, ROA, and Firm Size, all appear to significantly affect the CEO’s total compensation. Likewise, Major shareholder has a significant coefficient. However, with robust estimations, only Board size, outside directors, ROA, and Firm Size are actually statistically significant.

Board Size and Outside Directors both show a positive coefficient. A large board is usually associated with low effectiveness and therefore high CEO pay (Ntim et al., 2015; Ozkan, 2011). As for appointing outside directors in the board, our results accord with previous studies documenting a positive relationship between outside directors and CEO compensation (Boyd, 1994; Core et al., 1999).

Return on assets (ROA in Model 1) and firm size still present a significant positive coefficient, as predicted. In line with the previous studies, the first model provides evidence that executive compensation increases with firm size (Barkema and Gomez-Mejia, 1998; Balkin and Gomez-Mejia, 1990; Galbraith and Kazanjian, 1986; Gomez-Mejia, 1992; Kerr, 1985; Kroll, Wright, Toombs, and Leavell, 1997; Tosi et al., 1999). With regard to the positive relation between firm performance and CEO compensation, we confirm the results of previous studies (Ángel and Fumas, 1997; Brunello et al., 2001; Jensen and Murphy, 1990; Kang and Shivdasani, 1995; Kaplan, 1994; Kato, 1997; Ntim et al., 2015; Poulain-Rehm, 2000).

As for the second model, we change our dependent variable for corporate performance (ROA).

Unfortunately, we are unable to confirm our second hypothesis. Using GLS and robust estimations, the second model does not provide any interesting significant results.

Extension Analysis

In this section, we conduct additional analysis in order to explore the effect of employee representatives, coupled with the nature of the major shareholder, on CEO compensation. First, we will use the nature of the major shareholder as an interactional variable. Then, we will test it as mediation to the BLER and CEO pay relation.

Moderation and Interaction Effects

Interaction Effect between Ownership types, Employee Representation and CEO Pay

Testing for interaction is equivalent to testing that the effect of a variable X on a variable Y is dependent of a third variable Z (Baron and Kenny, 1986; Jaccard et al. 1990; Hayes and Matthes, 2009). Baron and Kenny (1986) define a moderator as a “variable that affects the direction and/or strength of the relation between an independent variable and a dependent variable”. The moderator effect is tested by regressing the dependent variable on the independent variable, the moderator, and the product of the independent variable and the moderator. Interaction effect is detected if the coefficient of this product is statistically significant. In corporate governance studies, several authors have included the effect of a moderator. For instance, Pan, Huang and Gopal (2018) examine the relationship between board independence and firm performance, and argue that this relationship is moderated by new entry threats in the IT industry. Accordingly, we test whether the nature of the largest shareholder influence the BLER on CEO’s compensation. Thereby, the nature of the largest shareholder is, here, the moderator.

For this purpose, we conduct three regressions for each interaction variable, BLER*Family, BLER*State, and BLER*INST, with the dependent variable: the CEO pay level. Results are reported in Table 5.

Table 4

Employee representation, CEO total pay, and firm performance

Employee representation, CEO total pay, and firm performance

Notes: In this table, we report panel data regressions results with random effects for model 1 and model 2. Robustness tests using the Generalized Moments Method (GMM) are also reported. Coefficients are reported with z-values (z stat) in parentheses. ***, **, and * denote significance at the 1%, 5%, and 10% level, respectively.

-> Voir la liste des tableaux

Table 5

The Moderation effect between ownership structure and employee representation on CEO’s total compensation

The Moderation effect between ownership structure and employee representation on CEO’s total compensation

-> Voir la liste des tableaux

Table 5 shows that only under institutional ownership are we able to prove interaction effects. In fact, the interaction variable BLER* Institutional is statistically significant (with a coefficient of -0.032). This confirms that, indeed, institutional ownership and employee representation do interact to curb executive pay. Sahut and Gharbi (2010) in their review of institutional investors of French listed firms, suggest that when an institutional investor owns a rather important part of a company’s shares, they are more involved with CEO monitoring and strategic decisions.

However, this result is no longer significant with robust estimations. Under GMM regressions, only the control variables show interesting results. In fact, outside directors, Firm Size, and ROA present a positive and significant coefficient, therefore corroborating once again the positive impact these variables have on CEO compensation.

The Mediation Effect

In the second section, we hypothesize that the nature of the largest shareholder could mediate the influence of BLER on CEO compensation. Therefore, the mediator will appear as three variables: Family, State, and Institutional ownership. In order to test this hypothesis, we conduct further regressions using Baron and Kenny’s method on mediation analysis. Baron and Kenny (1986) suggest estimating three regressions. In the first step, we regress the dependent variable (CEO_comp) on the independent variable (BLER); in a second step, we regress the mediator variable on the independent variable. Since we used three mediators, the second step consists of three regressions as well, each for the nature of the largest shareholder (Models 7a, 7b and 7c); and in a third step, we estimate the dependent variable with both the independent variable and the mediator (Model 8). In this last regression, the three mediators are used jointly as controls. After conducting these regressions, results should meet the following requirements in order to establish mediation: First, the independent variable must affect the dependent variable; then the independent variable must affect the mediator; and last, the mediator must affect the dependent variable. If these conditions are verified, then the effect of the independent variable on the dependent variable should be reduced in the third equation in presence of the mediator.

Results are reported in Table 6 below. In this table, we only present GMM regression results.

Model 6 shows a significant and negative sign of the BLER variable with a coefficient of -0.022. The first equation (step) of Baron and Kenny’s mediational model is then verified.

Only Model 7b shows significant results, therefore confirming the second step for State ownership. Finally, State ownership remains significant in the last step (Model 8). Therefore, mediation conditions are established. In presence of the mediators, BLER still has a significant coefficient (-0.019). Accordingly, and based on Baron and Kenny’s method on mediation analysis (1986), we confirm that French State ownership mediates the negative impact of BLER on the CEO’s total pay.

However, the effect of BLER on CEO pay is not reduced (from -0.022 to -0.019) after introducing the mediator. Therefore, and to further strengthen our result on the mediation effect of the nature of the major shareholder on the BLER-CEO pay relation, we conduct the Sobel test that verifies whether the effect of an independent variable on a dependent one via a mediator is significantly different from zero (Baron and Kenny, 1986; Sobel, 1982). Sobel (1982) provided the following formula, where a is the path coefficient from the independent variable to the mediator, b is the path coefficient from the mediator to the dependent variable, and Sa and Sb are respectively the standard errors of a and b:

We find a Test statistic t that equals -2.005 (significant at 5%) with a standard error of 0.003.

The indirect effect of the mediator on the independent and dependent variables relation is, therefore, significant. The mediation effect of the State control on the BLER-CEO compensation relation is once again verified, despite an unusual increased effect.

We consider that result as a consequence of the 2012-915 decree of July 26th 2012 capping the compensations of the CEOs of French Firms under State control. Therefore, compensation is already curbed by the French State considering this decree. We argue that State ownership can strengthen the control of BLER over CEOs and their compensation, rather than simply mediate it. That could interestingly justify why the effect is increased in our mediation analysis.

Table 6

Employee representation and CEO compensation: the mediation effect of ownership structure

Employee representation and CEO compensation: the mediation effect of ownership structure

Notes: In this table, we report regressions results using the Generalized Moments Method (GMM). Coefficients are reported with z-values (z stat) in parentheses. ***, **, and * denote significance at the 1%, 5%, and 10% level, respectively.

-> Voir la liste des tableaux

Discussion

This study tests the effect of employee board-level representation on CEO compensation. Our results show a negative and significant effect of BLER on CEO compensation level. Through their representatives on the board, employees are able to contribute to the CEO pay monitoring. Employee involvement in corporate governance may enhance efficiency by improving board decision-making (Huse et al., 2009), and has, then, an impact on CEO compensation policy. Therefore, the board employee representation could be a relevant mechanism to reduce problems of “excessive” CEO compensation and its damage to the firm reputation. Our findings support the results of Huang et al. (2017) confirming the influence of employees - through trade unions - on CEO compensation. They suggest that employees can use their right to strike to exert pressure on firms concerning their executive compensation policy. In other words, trade unions use their power of pressure to curb CEO compensation levels. In the same vein, Lin, Schmid and Xuan (2018) advocate the use of direct employee influence, i.e. employee representation on boards, on corporate governance. Although their work focuses on employee codetermination[4] and financial leverage, it has brought insight to the power of employees in corporate governance structures. Proving that employees have a direct influence on a firm’s policy otherwise reflects the importance of such power over boards’ decisions. Based on stakeholder theory, the BLER also favor the establishment of a better communication between the firm and his stakeholders, which is expected to positively affect the board’s function of internal control and reduce information asymmetry.

In a second hypothesis, we test the relationship between employee representation and firm financial performance. Insignificant result was found. This result supports the findings of Jones, Mygind and Sen (2019). In the institutional context of Estonia, these authors found no impact of employee involvement in firm decision-making on firm performance. They explain the lack of result by the fact that Estonia is an emerging economy and that employees are not fully trusted by firm executives. However, they find that employee ownership significantly increases firm performance. This suggests that employee participation on boards, individually, may not have an impact on performance. Although empowering employees through board membership may increase employee effort levels, reduce workers’ voluntary exists and allow for more efficient use of private information on the production process by increasing employee discretion (Baron and Kreps, 1999; Benabou and Tirole, 2003). However, the sole use of the mechanism is not sufficient to align firm financial interests with those of its employees. In fact, BLER is apt to be considerably stronger if significant incentives exist for employees to influence firm performance. The investigation of synergies between BLER and human resources practices (performance related pay, employee ownership through stock-options plan, etc.) may be a valuable approach to explore the relation between BLER and firm performance.

Concerning the third hypothesis, that test mediation and moderation effect of the nature of the largest shareholder on the BLER-CEO compensation relation. First, using moderation, we confirm the interaction effect between institutional ownership*BLER and CEO pay. In their review of institutional investors of French listed firms, Sahut and Gharbi (2010) suggest that when an institutional investor owns an important part of a company’s shares, he is more involved in CEO monitoring and strategic decisions. Institutional investors, thereby would participate more in the process of determining CEO pay, and would require reducing it. Second, we explore mediation and set ownership structure as mediator to the CEO pay-BLER relation. We find that the negative effect of BLER on CEO compensation is mediated by the French State (when the latter is a major shareholder). In other words, employee representatives curb executive compensation through the intervention of the French State. The State can act as a defender of employees’ rights and help their representatives to adjust the board’s decisions, i.e. executive compensation. Fauver and Fuerst (2006) argue that “when ownership becomes sufficiently concentrated, the control rights of the firm become aligned with the cash flow rights and the monitoring incentives of the large ownership block largely supplant the monitoring benefits of trade.”

Concluding Remarks

This paper explores the effects of BLER on CEO compensation in French listed companies. First, the results confirm a negative impact of employee representation on CEO incentive pay level. Second, we were not able to provide evidence for the positive effect of BLER on corporate performance. We provide new insights into the employee-CEO pay debate by establishing a moderator effect of institutional ownership and fully mediation effect by French State in the employee representation-CEO pay relation. We contribute to the corporate governance literature by including ownership structure as an important piece to the BLER-CEO compensation relation. Our results show that including employee representation on the board may help the CEO pay monitoring. Our work demonstrates that trade unions have a strong influence over corporate policy in French context. Therefore, trade-union representation gives a more direct voice to employees such as executive pay (Ahlquist, 2017). This brings insight to French employees’ voice and power to ensure fairness in remuneration. A coalition between French government and trade unions may improve the monitoring of CEOs activities and their compensation.

This research has some limitations that could be overcoming with future research. First, we hypothesise that trade union representatives are homogenous group. However, in France, trade unions are diverse and are composed of five confederations recognized by the state as negotiating partners. This suggests a rather fragmented unionism (Andolfatto and Labbé, 2016). Second, unionism in France has encountered a downturn. Trade unions gather very few members, which makes it “weak” and therefore, may have limited repercussions on employees’ real situation. Finally, unions’ curtailment of CEO compensation appears to be for the stock-options part, solely. As unionized firms are usually associated to a reduction in their market values (Ruback and Zimmerman, 1984), stock-related CEO compensation will consequently be as well reduced (Gomez and Tzioumis, 2006). Further research could include firm performance as a mediator to employee representation and CEO pay. Moreover, adding employee ownership to employee representations as measures of their participation could interestingly lead to different results. Therefore, future studies could focus on different outcome variables that are primarily in the interest of employees, such as risk taking behavior, employment growth and stability, investment in the research and development of new products, or the reduction of the CEO-to- worker pay-gap. Future work using larger data and other countries could also help to build further knowledge and to promote theoretical discussions to shed more light on the relationship between BLER and CEO compensation. Furthermore, our study focused on the effect of the largest shareholder on the BLER-CEO pay relation. Nevertheless, another complementary work could examine agency conflicts between the largest and the second shareholder. In fact, the second shareholder owns an average of 9% of the company’s shares, and accordingly may influence corporate decisions. Finally, a recent research by Dardour et al. (2018) has explored the profiles of independent board members and how it affects corporate social disclosure. They found a strong impact of BLER on corporate social disclosure. Moreover, we can investigate the effect of the combination of BLER and human resources practices on firm performance as well as CEO pay. Thus, some promising research avenues could focus on the profiles of employee representatives and this combination on the CEO pay policy.