Appointment-Based CEO Connectedness and Employee Compensation: Empirical Evidence from China

Abstract

Employee compensation is an often-neglected but essential part of corporate social responsibility which emphasizes caring for the needs of all stakeholders, including employees. In order to address pressure from stakeholders to strengthen prosocial acts, CEOs might prefer to raise employee compensation. However, other top executives are often reluctant to do so due to the concern that it reduces firm profits. In this paper, we propose that appointment-based CEO connectedness (ABCC) has a positive effect on employee compensation as it facilitates CEOs gaining support from the top management team to raise employee compensation. We employ multivariate linear regression as our research approach and find supportive evidence using data from Chinese listed firms during 2011–2020. Our results are robust to endogeneity concerns and survive an array of robustness checks. Heterogeneity tests show that this impact is stronger for firms facing less market competition and firms with low financial constraints. We further show that greater ABCC is associated with higher CSR scores of non-shareholders responsibility dimensions. Overall, our study suggests ABCC is conducive to the fulfillment of corporate social responsibility towards non-shareholders.

1. Introduction

In recent decades, there has been a trend among firms to care about the interests of all stakeholders and engage in corporate social responsibility activities [1,2]. As employees are crucial stakeholders, corporate social responsibility entails concern for employee compensation. Employee compensation is a major measure for firms to prompt employee effort [3,4] as well as desired work behaviors [5,6]. In addition, raising employee compensation is conducive to firms’ reputation. By increasing employees’ compensation, firms signal that they are socially accountable.

In order to promote employee-friendly compensation policies in firms, it is important to understand the determinants of employee compensation. The existing literature has documented a battery of factors that affect employee compensation. These factors include employee ability, unionization [7,8] and firm characteristics such as firm size [9], firm age [10,11], ownership structure [12], the presence of FDI [13], capital structure [14] and so on. The most relevant study investigates how CEO formal power affects employee compensation, demonstrating that CEOs with absolute formal power based on voting rights pay their employees more [15]. The basic idea is that CEOs with absolute decision discretion are free to carry out firm decisions they prefer, even if other top executives may not agree with them.

The recent literature suggests that appointment-based CEO connectedness (hereafter ABCC), i.e., CEOs’ internal connections to other executives built through appointing executives, can be considered as informal network-based soft “power” [16]. The reason for this is that it enables CEOs acquire loyalty and support from the executive suite [17,18] and thus increase their social influence as well as managerial discretion. However, whether network-based informal “power” obtained through appointments has the same effect on firm employee compensation as formal power remains unknown, despite ABCC is found to affect firm decisions regarding investment, fraud, earnings management and executive compensation [1,2,3,4,5,6]. Given the prevalence of ABCC and the importance of employee compensation in firms, it would be useful to answer the research question regarding the impact of ABCC on employee compensation.

To fill the gap, we investigate the relationship between ABCC and employee compensation. Despite the fact that the employee-friendly compensation policy benefits firms in several ways, non-CEO top executives often are reluctant to adopt it. Raising employee compensation reduces current profit. Therefore, non-CEO top executives might be against it because firms care more about short-term financial outcomes [19]. In contrast, CEOs might be motivated to pay employees more in order to respond to external pressure from stakeholders requiring firms to strengthen prosocial acts [20].

Building on the prior literature, we predict that ABCC positively relates to employee compensation. Given that other executives are reluctant to pay employees more, CEOs who lack adequate support from the top executive team (hereafter TMT) may find it difficult to implement an employee-friendly compensation policy. CEO connectedness built through appointment decisions, however, enhances CEOs’ social influence and managerial discretion over executives appointed and thus enables the compliance or coordination needed to dominate employee compensation decisions. Given that CEOs play an essential role in hiring, nominating, as well as appointing top executives, executives who are appointed by a given CEO are inclined to hold similar preferences with, and are more loyal as well as beholden to the CEO [21,22]. Thus, CEOs could gain more support from the TMT as ABCC increases. Therefore, we expect that a higher ABCC is associated with higher employee compensation.

However, this relation is likely to be moderated by the availability of firm financial resources. Even if the TMT supports CEOs to pay employees more, financial resource constraints may limit their ability to increase employee compensation. Firms are likely to have more internally generated financial resources when faced with less product market competition [23,24] and more external financial resources through financing when faced with fewer financial constraints. Hence, we explore how market competition and financial constraints moderate the effect of CEO connectedness. Specifically, we propose that product market competition and financial constraints weaken the positive relation between ABCC and employee compensation.

To study our hypotheses, we utilize a sample which consists of 3828 Chinese A-share listed firms spanning from 2011 to 2020. Following prior studies, we measure ABCC as the fraction of non-CEO top executives appointed within the duration of a CEO’s tenure [5]. Employee compensation is measured as the firm-level average compensation of ordinary employees, which is the total employee expense (excluding top executives’ salaries) scaled by the number of ordinary employees [25,26]. Control variables include firm characteristics, CEO traits and GDP per capita of the province where the firms are registered.

Consistent with our predictions, our findings suggest that CEO connectedness built through appointing top executives positively impacts employee compensation in a statistically and economically significant way. Specifically, a one-standard-deviation increase in ABCC leads to an approximate rise of RMB 1487 in average employee compensation. Moreover, we show that both product market competition and financial constraints negatively moderate the positive influence of ABCC on employee compensation.

Since our findings might be plagued with endogeneity issues, we apply the two-stage least square (2SLS) method to tackle the potential endogeneity problem. In accordance with Khanna et al. [17], two instrument variables are employed. The first one is the deaths of non-CEO top executives during the tenure of the current CEO up to the current year. In addition, we also utilize the yearly average non-CEO executive turnover ratio in the industry that a firm belongs to (excluding each sample firm) as the second instrument variable. Our findings hold in 2SLS estimations, which further confirms the positive influence of ABCC on employee compensation.

The positive influence of ABCC on employee compensation survives several robustness tests. First, to control for potential unobserved firm-level heterogeneity that does not change over time, we validate our finding by including firm fixed effect in the OLS regression. Second, we run the main regressions with an alternative proxy for ABCC which is the proportion of directors appointed within the duration of a CEO’s tenure. We believe this alternative proxy also reflects CEOs’ influence on TMTs because CEOs are able to affect TMTs through their influence over the directors. In line with the primary findings, this robustness test validates that ABCC positively affects employee compensation.

At last, we analysis the impact of ABCC on corporate social responsibility (hereafter CSR). Using Hexun CSR scores, we find that greater ABCC is positively associated with the CSR score of the non-shareholder responsibility dimensions. This finding indicates that an increase in ABCC improves the CSR performance.

Overall, this empirical evidence supports the idea that greater CEO connectedness built through appointing top executives enables CEOs to make socially responsible firm decisions that benefit non-shareholders, such as paying employees higher compensation.

This paper makes three contributions to the existing literature. First, we add to the literature that empirically studies the determinants of employee compensation by identifying an important factor, namely, ABCC. Second, by showing that the influence of ABCC spreads down to employee compensation-setting processes, this study extends the literature regarding the effects of CEO connectedness on socially responsible behaviors in firms. Third, our study presents a boundary condition on the relation between ABCC and firm employee compensation by considering the moderating role of available financial resources.

The rest of the paper is organized as follows. Section 2 develops the hypotheses. In Section 3, we present the research methodology. The results of the empirical analysis are shown in the Section 4. Section 5 further illustrates the impacts on corporate social responsibility. At last, we discuss and conclude.

2. Literature Review and Hypotheses

2.1. ABCC and Firm Outcomes

The recent literature has highlighted the importance of ABCC in determining firm outcomes. In general, there are two conflicting perspectives regarding the impacts of ABCC. On the one hand, ABCC may lead to undesirable firm outcomes. The connection that CEOs build through appointments enables them to exert “social influence” on the incoming executives, given the direct involvement of CEOs in the processes of executive nomination, recruitment, and appointment [17]. According to Agha et al. [18], the newly-appointed top executives are inclined to hold similar preferences with the CEO who appointed them, which can weaken internal governance from independent executives. Consistent with this view, empirical evidence suggests that ABCC results in poorer firm performance [21], more corporate fraud and decreased likelihood of detection [17], more earnings management [27], lower investment efficiency [18] and more insider opportunism [28]. On the other hand, ABCC may benefit firms since it enhances coordination and TMT information transmission. As Li et al. [16] suggests, ABCC signifies a strong bond between CEO and top executives, thus reducing TMT coordination costs and enabling CEOs to make prompt decisions. In addition, appointment induced loyalty fosters implicit trust, which improves investment efficiency by facilitating TMT information transmission [29]. Overall, while prior studies have established that ABCC impacts many firm outcomes, whether it influences employee compensation remains unknown.

2.2. Influencing Factors of Employee Compensation

There are many empirical studies that investigate the determinants of employee pay. We briefly review the main impacting factors in four categories. The first category centers on employee human capital. The general findings are that firms pay more to competent employees. For example, employees with higher education or having better work experience are found to earn more [15,30]. The second category studies the role of employee unionization in setting employee pay and generally concludes that unionization raises employee compensation [7]. The third category focuses on firm characteristics, such as firm size [9], age of firms [10,11], ownership structure [12], the presence of FDI [13] and capital structure [14]. The last category of the literature investigates how managerial power or discretion relate to employee pay. Managerial discretion is found to positively correlate to workers’ pay because of managers’ preference to pursue non-pecuniary private benefits such as pursuing a “quiet life” and lowering wage bargaining efforts [31,32]. In addition to formal managerial power, CEOs’ network-based CEO connectedness from appointing top executives may also impact employee compensation. This paper aims to empirically examine this conjecture.

2.3. ABCC and Employee Compensation

Notwithstanding that paying employees more could improve firm productivity and firms’ reputation as socially responsible corporates [3,4], non-CEO top executives often dislike such practices. Given that short-termism that focus on maximizing current profits usually dominates corporate decision-making [19], top executives might be reluctant to increase employee compensation as it decreases current profit. Specifically, non-CEO top executives may undertake efforts to discourage CEOs to raise overall employee compensation by persuasion or lobbying the board. On the contrary, CEOs might prefer to increase employee compensation in response to external pressure from stakeholders that requires firms to strengthen prosocial acts [20].

It can be challenging for CEOs to raise employee compensation, as non-CEO executives generally prefer not to pay employees more. In this case, greater ABCC may serve as an implementation tool that enables CEOs to gain support from the TMT and assert dominance over corporate employee compensation decisions. The underlying logic is that ABCC captures the “social influence” that CEOs have over the incoming executives [17]. As Kim and Lu [22] argues, executives are prone to hold the same preferences and beliefs with, and might be more loyal and beholden to those CEOs who appointed them. The reason is that CEOs typically play a vital role in hiring, nominating as well as appointing top executives [21]. In other words, if a CEO appoints more top executives during their tenure, the CEO would have higher social influence as well as managerial discretion, and thus their decisions would be more easily supported and carried out by the TMT.

Based on the foregoing arguments, Hypothesis 1 is developed as follows:

Hypothesis 1 (H1). 

ABCC has a positive impact on employee compensation.

2.4. The Moderating Effect of Financial Resource Constraints

While we propose that CEOs with higher appointment connectedness would increase employees’ compensation, this effect might be moderated by corporates’ financial condition. Specifically, we focus on how financial resource constraints moderate the nexus between ABCC and employee compensation. The basic idea is that financial resource constraints, namely a lack of financial resources in firms [33], impose constraints on firms’ ability to pay employees more. This is consistent with prior studies that suggest access to more financial resources can facilitate firms pursuing various strategies and opportunities [34]. Given that financial resources are generated either internally from operations or externally through financing, we further study the moderating role of market competition and financial constraints on the link between ABCC and employee compensation.

In a competitive market, corporates operating in industries that exhibit intense market competition tend to have lower profitability [23,24,35], and thus generate less financial resources. According to Yazdanfar [36], industry-specific characteristics, including the degree of market concentration, market barriers, and product differentiation help to increase firm profitability. Therefore, firms that operate in industries with intense market competition are less likely to generate substantial operational cash flow. With insufficient internally generated financial resources, CEOs of these firms are less able to improve employee compensation. Thus, we propose:

Hypothesis 2 (H2). 

Market competition weakens the positive effect of ABCC on employee compensation.

In addition to utilizing internally generated financial resources, corporations may also gain financial resources through external financing. However, firms may face financial constraints, which refer to their inability to raise external funds [37]. Financially constrained firms incur greater costs for external financing, and thus have more difficulties obtaining external funds [38]. Therefore, everything else being equal, financially constrained firms can obtain fewer external funds from the capital market [39] and thus have fewer financial resources available for CEOs to increase employee pay. Hence, we propose:

Hypothesis 3 (H3). 

Financial constraints weaken the positive effect of ABCC on employee compensation.

Following [40], we present the conceptual framework of this paper in Figure 1.

3. Research Methodology

3.1. Research Approach

This paper employs multivariate linear regression analysis to quantitatively examine our hypotheses. We also use two-stage least squares regression to tackle the potential endogeneity concern. With natural occurring open data derived from publicly listed firms, these methods enable us to ascertain whether ABCC has an impact on employee compensation, as well as the extent of this influence.

3.2. Data and Sample

We use Chinese A-share listed firms from 2011 to 2020 as our research sample. The average employee compensation variable was developed based on data sourced from the China Stock Market and Accounting Research (CSMAR) database. CEO characteristics, firm and industry data are also sourced from CSMAR. The educational background information of employees is sourced from the Wind financial database (WindDB). Data regarding the deaths of non-CEO top executives are collected from the RESSET financial research database. We supplement this data by reviewing the firm announcements from CNINFO, an online platform for listed firms to disclose information as required by the China Securities Regulatory Commission (CSRC). To avoid the effect of extreme values, we winsorize all variables (excluding dummy variables) at the first and last percentiles.

Our sample period begins in 2011 because WindDB started to provide employee education background data this year. We drop financial firms and firms in unusual listing conditions (ST, *ST, PT) as these firms tend to have idiosyncratic characteristics such as capital structure. We also eliminate firms that only report the number of headquarters employees because we are unable to calculate average employee compensation in these firms. In addition, we drop those abnormal firm-year observations where employee average compensation is negative or larger than top executive average compensation (We calculate top executive average cash compensation as the total top executive cash compensation including cash salary and bonuses divided by the number of salaried top executives). There are 28,278 firm-year observations in the final sample.

3.3. Variable Description

3.3.1. Employee Compensation

Following prior works [14,26,41], we proxy average employee compensation ($AEP$) as the total ordinary employee expense scaled by the number of ordinary employees. More specifically, $AE{P}_{i,t}$ of firm $i$ in year $t$ is measured as in Equation (1). Total employee expense is calculated as firm total salary expense minus the compensation of top executives. We derive the number of ordinary employees by subtracting the number of top executives from the total number of employees. In order to reduce the right-skewness, we take the natural logarithm of $AE{P}_{i,t}$.

$$\begin{gathered} LnAE{P}_{i,t}=ln ((Employee payroll payabl{e}_{i,t}-Employee payroll payabl{e}_{i,t-1} \\ +Cash paid to and for employee{s}_{i, t}-Total top executive cash pa{y}_{i, t}/ \\ \left(Total number of employee{s}_{i, t}-Number of top executive{s}_{i, t}\right)) \end{gathered}$$

(1)

3.3.2. ABCC

We use the fraction of non-CEO top executives appointed during the tenure of a given CEO ($FTE{A}_{i, t}$) in firm $i$ as of year $t$ to proxy ABCC [17,42]. A higher $FTE{A}_{i, t}$ value means that a given CEO has appointed more top executives during his/her tenure.

3.3.3. Moderating Variables

Herfindahl–Hirschman Index ($HHI$). Following Dai et al. [43], we use the Herfindahl–Hirschman Index (HHI) based on total assets as the measure of market competition in a given industry. The larger the HHI index, the smaller the market competition in an industry.

Financial constraints ($FC$). Financial constraints of a given firm is proxied by the FC index introduced by Hadlock and Pierce [44]. A larger value of FC index denotes that the firm is more financially constrained.

3.3.4. Control Variables

We follow Chemmanur et al. [14] and Cronqvist et al. [15] to control for a set of firm-level factors, CEO traits and province-level GDP per capita that might impact employee compensation. The definition of variables is presented in Appendix A.

In terms of firm characteristics, $SOE$ is the state ownership dummy variable that takes value 1 if the ultimate controlling owner is the state, and zero otherwise. $LnSize$ denotes the natural logarithm of firms’ book value of total assets. $Growth$ is the yearly operating revenue growth, derived as the increase of operating revenue scaled by the operating revenue as of the end of last year. $ROA$ is the return on assets, computed as the net profit scaled by the book value of total assets. $Lev$ is firm leverage, computed as the book value of total debt scaled by the book value of total assets. $Initial\_Cash$ is the initial cash holding of a firm, calculated as the beginning cash balance divided by the book value of total assets in a given year. $Fixed\_assets$ is the ratio of the book value of fixed assets to the book value of total assets. $Age\_firm$ is the years since the foundation of a firm. To measure non-CEO blockholders’ financial incentives, we control for $Tophold$ which denotes the fraction of shares held by the biggest non-CEO shareholder. We also control for two firm human capital variables. The first one is the natural logarithm of the number of ordinary employees ($LnNum\_workers$). The second is $Pct\_grad$ which is the fraction of ordinary employees with postgraduate education. In addition, we also control for $LnGDP$, which is the natural logarithm of GDP per capita of the province where firms are registered.

The CEO traits that we control for in the model are as follows. $CEO\_own$ is the fraction of shares the CEO holds. $Dual$ is a dummy variable that has a value of 1 if the CEO is also the chairman, and 0 otherwise. $LnTenure$ is the natural logarithm of the months after a CEO took office. $Male$ is a dummy variable that takes a value 1 if the CEO is a male. We also control for $LnAge$ which is the natural logarithm of CEO age. Lastly, we control for $LnSalary$, namely the natural logarithm of CEOs’ cash compensation, which includes base salary and bonuses.

3.4. The Benchmark Model

We test the impact of ABCC on employee compensation with a model as follows:

$$AE{P}_{ i,t}={\beta }_0+{\beta }_{1}FTE{A}_{i, t}+\sum {\beta }_j{\left(CVs\right)}_{i, t}+\sum Industry+\sum year+\sum province+{\epsilon }_{i, t}$$

(2)

In the model, $AE{P}_{ i,t}$ refers to the logarithm of average employee compensation of firm $i$ in year $t$. $FTE{A}_{i, t}$ is the proxy of ABCC. $C{V}_{i, t}$ is a battery of control variables. Moreover, year, industry and province fixed effects are also controlled for in our regression model.

4. Empirical Results

4.1. Descriptive Statistics

Table 1. Summary Statistics.

Variables	N	Mean	SD	Min	Max
AEP (in thousands)	28,278	117.10	63.492	20.99	364.75
LnAEP	28,278	11.54	0.498	9.95	13
FTEA	26,100	0.53	0.335	0.00	1
FDA	26,100	0.84	0.251	0.11	1
SOE	27,765	0.34	0.475	0.00	1
Size (in billions)	28,272	13.02	38.678	0.15	392.38
Growth	28,258	0.17	0.451	−0.69	3.50
ROA	28,272	0.04	0.064	−0.27	0.20
Lev	28,272	0.42	0.208	0.05	0.89
Initial_Cash	28,101	0.15	0.125	0.01	0.63
Fixed_assets	28,272	0.21	0.159	0.00	0.71
Age_firm	28,273	17.45	5.692	2.00	32
Tophold	28,277	34.14	15.151	7.49	75
Num_workers	28,278	4835.11	9206.436	107	63,169
Pct_grad	20,383	4.56	5.747	0.12	35
GDP (in thousands)	28,274	77.17	32.820	16.41	164.89
CEO_own	26,820	0.06	0.121	0.00	0.53
Dual	27,953	0.30	0.457	0.00	1
Tenure (in months)	27,926	49.48	39.884	1	169
Male	27,957	0.93	0.249	0.00	1
Age (in years)	27,953	49.72	6.475	33	65
Salary (in thousands)	27,907	837.20	81.686	0	4998.40
HHI	25,108	0.15	0.138	0.04	1
FC	25,108	0.48	0.284	0	0.99

displays the summary statistics of variables. On average, each firm in in our sample has 4835.11 ordinary employees. In terms of employee compensation, the average is about RMB 117.10 thousand per firm-year. The mean value of FTEA is approximately 0.53, indicating that about 53% of the executives are appointed by the same CEO. As for CEOs’ ages, the average is 49.72 years. Among the CEOs, about 93% of them are males. The average of the Herfindahl–Hirschman Index $\left(HHI\right)$ is 0.15, indicating a quite strong competition among the Chinese listed firms. The mean value of financial constraints ($FC$) is 0.48. This implies that the firms in our sample do not experience substantial financial constraints.

4.2. Main Results

The results of the regression analysis are displayed in

Table 2. Impact of appointment-based CEO connectedness on employee compensation.

	(1)	(2)	(3)
FTEA	0.0612 ***	0.0333 ***	0.0376 ***
	(0.0134)	(0.0119)	(0.0125)
SOE		0.1270 ***	0.1378 ***
		(0.0127)	(0.0131)
LnSize		0.1908 ***	0.1694 ***
		(0.0084)	(0.0082)
Growth		−0.0132 **	−0.0134 **
		(0.0066)	(0.0066)
ROA		0.2527 ***	0.0789
		(0.0711)	(0.0695)
Lev		0.0130	0.0240
		(0.0316)	(0.0310)
Initial_cash		0.2350 ***	0.1839 ***
		(0.0360)	(0.0343)
Fixed_assets		0.0035	−0.0121
		(0.0410)	(0.0411)
Age_firm		0.0011	0.0006
		(0.0010)	(0.0010)
Tophold		0.0013 ***	0.0014 ***
		(0.0004)	(0.0003)
LnNum_workers		−0.1689 ***	−0.1736 ***
		(0.0084)	(0.0083)
Pct_grad		0.0270 ***	0.0259 ***
		(0.0012)	(0.0012)
LnGDP		0.0423	0.0315
		(0.0392)	(0.0406)
CEO_own			0.0867 *
			(0.0462)
Dual			−0.0310 ***
			(0.0116)
LnTenure			−0.0148 ***
			(0.0035)
Male			0.0227
			(0.0159)
LnAge			0.0627 *
			(0.0354)
LnSalary			0.0861 ***
			(0.0067)
_cons	11.0849 ***	7.4270 ***	6.7425 ***
	(0.0676)	(0.4688)	(0.5020)
N	26,100	18,278	16,895
Adj. R2	0.387	0.614	0.626
Industry FE	Yes	Yes	Yes
Year FE	Yes	Yes	Yes
Province FE	Yes	Yes	Yes

Note: Robust standard errors clustered by firm in parentheses. ***, ** and * indicate significance at 1%, 5% and 10% levels, respectively.

. The coefficient presented in Column (1) shows that FTEA is positively related to employee compensation. We further control for firm characteristics, province-level GDP per capita, and CEO characteristics in Columns (2)–(3). Likewise, the coefficients of FTEA, as shown in Columns (2)–(3), are positive and statistically significant. According to Column (3), if ABCC goes up by one standard deviation (0.335, as shown in Table 2), the natural logarithm of employee compensation increases by 0.0376 × 0.335 = 0.0126, translating to a 1.27% rise in average employee compensation (RMB 1487, 117,102.6 × 0.0127). Note that the average number of ordinary employees in our sample firms is 4835.11, as shown in Table 2. Hence, on average, such a rise in average employee compensation results in each sample firm paying RMB 7.1898 million (1487 × 4835.11) more to ordinary employees, which is economically significant. Overall, these findings support Hypothesis 1.

To examine Hypothesis 2, we categorize all firms into high and low market competition groups, utilizing the Herfindahl–Hirschman Index (HHI) based on firm total assets [43]. Specifically, we categorize firms into the high (low) market competition group if HHI of the sub-industry to which a firm belongs exceeds (falls below) the yearly median HHI of all sub-industries (CSMAR only provides HHI index in each sub-industry. As an illustration, the letter “C” denotes the manufacturing industry. “C13” to “C43” denote the sub-industries of the manufacturing industry). In order to examine Hypothesis 3, likewise, our sample is partitioned into a more (less) financially constrained group if a firm’s financial constraints index (FC) developed by Hadlock and Pierce [44] is above (below) the yearly industry median.

Table 3. Heterogeneity tests.

	HHI		FC
	(1)	(2)	(3)	(4)
	High	Low	High	Low
FTEA	0.0458 **	0.0299 *	0.0169	0.0500 ***
	(0.0191)	(0.0157)	(0.0153)	(0.0184)
Controls	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes
Industry FE	Yes	Yes	Yes	Yes
Obs.	7394	9394	8083	8135
Adj. R2	0.609	0.651	0.615	0.641
p value (Permutation test)	0.074 *		<0.001 ***

Note: (a) We use Fisher’s Permutation test to examine the coefficient difference between high/low groups. The p values are generated using the bootstrap method (500 times). (b) Robust standard errors clustered by firm in parentheses. ***, ** and * indicate significance at 1%, 5% and 10% levels, respectively.

presents the estimated coefficients of FTEA from subsamples. In Columns (1)–(2), we show the estimated coefficient is less significant and smaller for the high market competition (low HHI) group, suggesting a comparatively weaker impact of FTEA for firms with high market competition. Therefore, Hypothesis 2 is supported. Consistent with Hypothesis 3, the positive impact of FTEA on employee compensation concentrates in firms with less financial constraints. As shown in Columns (3)–(4), the estimated coefficient is much bigger and statistically significant in the less financially constrained group. These results support Hypothesis 3.

4.3. Addressing Endogeneity Concerns

Although we explicitly control for firm-level characteristics and CEO traits in the main regression, there is still a possible issue of endogeneity. To mitigate the endogeneity concerns, we follow Khanna et al. [17] to construct instrument variables (IVs) and then run the two-stage least squares (2SLS) regression. Our IVs are $Exe\_Death$ and $Ind\_ExeTurnover$. In particular, $Exe\_Death$ is the count of non-CEO top executives who died during the tenure of the current CEO up to the current year. We also use $Ind\_ExeTurnover$ as the second instrument variable, which is the annual average non-CEO executive turnover rate in the industry that a firm belongs to. For each sample firm, we exclude the firm when calculating $Ind\_ExeTurnover$ in order to prevent capturing firm-specific factors that are linked to employee compensation. According to Khanna et al. [17], $Ind\_ExeTurnover$ could reflect industry shocks, such as industry business cycles. These shocks might affect executive turnovers.

A rise in $Exe\_Death$ and $Ind\_ExeTurnover$ leads to an increase in executive vacancies. To maintain operational efficiency, an executive vacancy usually needs to be filled in the same year. Therefore, we contend that our instrument variables increase the number of top executives recruited by the current CEO and thus result in a greater FTEA.

The validity of the 2SLS estimates hinges on the assumption that $Exe\_Death$ and $Ind\_ExeTurnover$ are not directly linked to employee compensation. We contend that both $Exe\_Death$ and $Ind\_ExeTurnover$ are independent from employee compensation. At the end of a year, it is common practice that the firms’ human resources department makes employee compensation budgets for the subsequent year. These budgets are typically formulated by considering the current year’s operating income, total employee compensation, and the projected business plan for the upcoming year. These budgets establish the firm’s total fixed compensation, total performance-based awards, and the salary base for each position grades. The CEOs and the board then evaluate and approve the employee compensation budgets. As for ordinary employees, their employee compensations in the current year are indeed determined by the employee compensation plan established at the end of last year, rather than the decision of non-CEO top executives in the current year. Therefore, the death of non-CEO top executives, which is an unanticipated random exogenous event, does not change average employee compensation in the current year. The same reasoning applies to the instrument variable of $Ind\_ExeTurnover$.

The results of the first and second stages are shown in Columns (1)–(2), respectively, in

Table 4. Results of IV regression.

	(1)	(2)
	FTEA (Stage 1)	AEP (Stage 2)
ExeDeath	0.1160 ***
	(0.0187)
Ind_ExeTurnover	0.1065 ***
	(0.0394)
FTEA		0.2960 **
		(0.1437)
N	16,841	16,841
Adj. R2	0.176	0.600
Industry FE	Yes	Yes
Year FE	Yes	Yes
Province FE	Yes	Yes
F-statictics(IVs)	348.69	-
Prob > F(IVs)	0.001	-

Note: (a) We use the two-stage least squares (2SLS) method. In Stage 1, the instrument variable is the number of top non-CEO executives who were dead during the current CEO’s tenure up to the current year and the industry average non-CEO executive turnover ratio, excluding each sample firm. (b) Robust standard errors in parentheses, *** and** indicate significance at 1% and 5% levels, respectively.

. It is shown in Column (1) that FTEA increases with executive deaths and the industry executives turnover ratio. In Column (2), we find the predicted values of FTEA increase employee compensation at a significance level of 5%. Thus, our baseline results are not likely to suffer from endogeneity bias.

4.4. Robustness Checks

We further validate the robustness of our findings in this part. First, we conduct firm-year fixed effect regression to control for unobservable firm characteristics that do not change over time. As column (1) of

Table 5. Robustness tests.

	(1)	(2)
FTEA	0.0218 **
	(0.0099)
FDA		0.0346 **
		(0.0147)
N	16,895	16,895
Adj. R2	0.665	0.626
Controls	Yes	Yes
Year FE	Yes	Yes
Industry FE	No	Yes
Firm FE	Yes	No

Note: Robust standard errors clustered by firm in parentheses. ** indicates significance at 5% level.

suggests, the positive association between ABCC and employee compensation still holds, which is in line with our main findings.

Second, we utilize the fraction of directors who are appointed during a CEO’s tenure ($FDA$) as an alternative proxy of ABCC [17]. We contend that CEO connectedness through appointing directors captures CEOs’ influence on the boards, which further empowers CEOs to exert their influence on the TMT. Thus, a larger $FDA$ denotes CEOs who have more influence on the TMT. With $FDA$, we conduct the main regressions again and display the regression result in column (2) of Table 5. The estimated coefficient of $FDA$ reveals that it positively affects employee compensation. These results validate the robustness of our main findings.

5. The Impact of ABCC on Corporate Social Responsibility

Employees are one of the key stakeholders in firms. The above evidence supports the idea that CEOs with greater ABCC cater to employees’ needs by paying their employees more. Thus, it is natural to ask whether CEOs would make socially responsible decisions that benefit stakeholders. In this section, we address this question by investigating the relation between ABCC and CSR performance.

Following prior studies [45,46], we use Hexun CSR scores as a proxy of CSR performance in Chinese listed firms.The Hexun website (http://www.hexun.com (accessed on 25 May 2020)) offers professional comprehensive CSR ratings that cover all Chinese listed firms. We define variable $CSR\_NonShareholder$ which is the sum of the rating scores of responsibility towards employees, obligation toward suppliers, consumers and customers, responsibility toward the environment and responsibility toward government.

We then estimate the impact of ABCC on firm CSR performance ($CSR\_NonShareholder$) using multivariate linear regression. Year fixed effect, industry fixed effect and province fixed effect are controlled for.

Table 6. Impact of ABCC on firm CSR performance.

	$\mathit{C}\mathit{S}\mathit{R}\_\mathit{N}\mathit{o}\mathit{n}\mathit{S}\mathit{h}\mathit{a}\mathit{r}\mathit{e}\mathit{h}\mathit{o}\mathit{l}\mathit{d}\mathit{e}\mathit{r}$
FTEA	0.8142 ***
	(0.3087)
N	24,712
Adj. R2	0.227
Controls	Yes
Year FE	Yes
Industry FE	Yes
Province FE	Yes

Note: Controls are a set of control variables including: SOE, LnSize, Growth, ROA, Lev, Dual, Male, LnAge. Robust standard errors clustered by firm in parentheses. *** indicates significance at 1% level.

presents the regression results. The results suggest that ABCC has a positive and significant effect on responsibility towards non-shareholders.

6. Conclusions and Discussions

6.1. Conclusions

According to the existing literature, ABCC is considered as an informal network-based soft “power” [16]. Does CEO connectedness built through appointing top executives facilitate CEOs raising employee compensation? While formal CEO power based on voting rights increase employee compensation [15], whether ABCC, the informal “power”, has the same impact on employee compensation remains uninvestigated. In this paper, we address this question using data from Chinese listed companies during 2011–2020.

We started by emphasizing that CEOs face resistance from executives to increasing employee compensation. With greater ABCC, CEOs are able to gain more loyalty and support from other executives [17]. Thus, we predict that ABCC positively affects employee compensation. In addition, this study also examines the moderating effect of market competition and financial constraints faced by firms. We further address the potential endogeneity issues and test the robustness of our main findings. Finally, we examine the nexus between ABCC and CSR performance in order to determine whether the positive impact of ABCC can be extended to social responsibility dimensions other than the responsibility towards employees. The conclusions of this paper are summarized as follows.

First, ABCC has a significantly positive impact on employee compensation, suggesting that ABCC enables CEOs to implement employee-friendly compensation decisions. These findings hold true to robustness checks and the consideration of potential endogeneity issues. Second, the impact of ABCC on employee compensation concentrates in firms with less market competition and low financial constraints. Third, ABCC is positively related to the aggregated score of non-shareholder responsibility dimensions.

Overall, the takeaway message of this paper is that ABCC, an informal network-based “power”, is sufficient for CEOs to increase employee compensation and implement socially responsible firm decisions that benefit non-shareholders.

6.2. Discussions

In this study, we hypothesize and find that greater ABCC enables CEOs to pay their employees more. Moreover, firms with greater ABCC are found to exhibit better CSR performance for non-shareholders. These results can be explained by the managerial discretion theory. Managerial discretion refers to the latitude that executives possess to affect the corporate activities [47]. Existing studies indicate that managerial discretion serve as a mechanism for increasing CEO compensation [47,48]. Consistent with this notion, it is possible that greater ABCC leads to higher managerial discretion. Hence, CEOs with greater ABCC are able to pay their employees more, despite the fact that some stakeholders may feel reluctant to do so.

Our findings are broadly in line with the conclusions drawn in the most relevant papers. Cronqvist et al. (2009) find that CEOs with absolute formal power pay higher wages to their workers [15]. This paper demonstrates that the informal power that CEOs obtain through appointments rather than voting rights is sufficient to enable them to do so. In another paper, Coles et al. (2014) show that ABCC is positively associated with CEO compensation [49]. We show the influence of ABCC extends beyond CEO compensation and spreads down to the employee compensation-setting processes.

This paper contributes to the literature in several ways. First, this study enriches our understanding of the determinants of employee compensation. Previous work generally focuses on the influence of employee and corporate characteristics [7,14,50,51]. Considering the contradictory preferences among CEOs and other executives regarding employee compensation policy, this paper highlights the importance of CEOs’ internal connections with top executives in determining employee compensation. Second, this paper adds to recent works that explore the impact of ABCC on firm decisions by showing that it also affects employee compensation. Third, we show that market competition and financial constraints moderate the influence of ABCC. The identified boundary conditions enable a more comprehensive understanding of the implications of our findings.

This study aims to enhance our understanding of the determinants of employee compensation. By investigating the role of ABCC in shaping employee compensation, we can gain valuable insights into the informal network-based “power” dynamics within firms and their effects on compensation policies. This research has practical implications for firms seeking to promote employee-friendly compensation practices and improve employee morale. Additionally, it contributes to the broader discourse on corporate social responsibility and the fair distribution of resources within organizations.

This paper also exhibits some limitations. While this paper shows that CEOs with higher appointment-based connectedness pay their employees more, whether this effect holds to other CEO connections built through common employment, educational or social histories is unknown. It would be useful to extend our work to other types of CEO connections. In addition, due to the prevailing collective culture in China that places greater emphasis on interpersonal relationships, the causal association between ABCC and employee compensation might be stronger in Chinese listed firms. As such, we encourage researchers to further investigate whether cultural differences matter.