Political Connections, Ownership and Within-Firm Pay Gap

Abstract

The difference in wages between executives and employees reflects the class conflict in corporate governance. To investigate the political factors within the practice of corporate governance related to employees, this paper empirically tests the relationship among political connections, ownership and within-firm pay gaps. We take the A-share listed companies on the Shanghai and Shenzhen Stock Exchange as the example, design hypothesis tests and examine the effects of political connections on the pay gap in two distinctive groups of companies, the state-owned enterprises (SOEs) and the non-stated-owned enterprises (non-SOEs). The overall result indicates that political connections increase the average salary of executives and decrease the average salary of employees, thereby expanding the within-firm pay gap. Pay gaps in companies with political connections are 16% higher than companies without political connections. The further test results of distinguishing property rights show that in non-SOEs, political connections increase the executives’ compensation and decrease the average compensation of employees, resulting in an increase of the within-firm pay gap. Similar relationships appear in SOEs but without statistical significance. These findings expand the research on income distribution effects of political connections theoretically, and provide useful insights for SOEs’ reform and income distribution system reform in practice.

1. Introduction

Corporate governance relates to the structure of rights and responsibilities among the parties with a stake in the firm [1]. In practice, rights and responsibilities diverge across parties, and thus may incur class conflict when the interests of management oppose that of labor, particularly regarding distributional issues [2]. Accordingly, the income inequality measured by the pay gap between the firm executives and employees is an important aspect for corporate governance studies.

The income inequality has constantly enlarged during the last two decades in China, which has attracted more and more attention from both the academia and the public [3,4]. According to “the annual Global Salary Survey 2015” released by international recruiter Robert Walters, the world’s largest within-firm pay gap was observed in China. The excess of income inequality may incur severe losses to companies, as is stated by existing studies [5,6,7,8]. Employees are an important stakeholder for a corporation [2]. While the literature of stakeholder-oriented corporate governance discusses more about the economic consequences of the pay gap between employees, very few studies discuss about the determinants of the pay gap [2]. China is experiencing a reform of the income distribution system. Under this circumstance, it is important to study how the pay gap between ordinary staff and executives is formed.

In a perfectly competitive labor market, the level of employee salary is determined by market. In reality, some specific enterprise characteristics may also be an essential factor, such as the political connections. In the bribery election case of the Liaoning National People’s Congress (NPC) deputies which shocked the central government in recent years, 90% of the representatives involved were entrepreneurs, which once again calls into question the political connections of enterprises: how will the “politically connected” status exchanged for huge sums of money affect the internal income distribution and gap of the enterprise? Tang and Sun [9] show that executives of politically linked companies receive significantly higher salaries, and executives have an inevitability incentive to gain political identity. However, the study does not distinguish between different types of corporate political connections. The mechanism of these political connections may also be different, and the further question is whether more economic benefits obtained by executives of politically connected companies will exacerbate the inequality of income distribution between executives and employees within company. This question also needs further answers. These are the key points of this study.

A series of previous studies focus on the establishment of political connections, the costs of cultivating political connections, and the mechanism of political connections. However, they seldom conduct their studies by classifying companies by their ownership types [4,9,10,11,12,13,14]. This study examines the effects of political connections in state-owned enterprises (SOEs) and non-state-owned enterprises (non-SOEs) on the pay gap within the company, and finds completely different conclusions. We take A-share listed companies on the Shanghai and Shenzhen Stock Exchange as the research object, and study their financial performance during 2010–2014. We developed two hypotheses and four subsequent hypotheses. Through the hypothesis testing, we have results about the effects of political connections. First overall, political connections will significantly increase the average salary of senior executives and reduce the average salary of ordinary employees, thereby widening the pay gap between executives and employees within the company. Second, we find that this phenomenon is mainly caused by non-SOEs in which executives’ political connections significantly increase executive compensation, while at the same time significantly reducing the average compensation of employees, which ultimately leads to widening the within-firm pay gap.

This study has three main contributions. First, existing studies mostly explain the determinants of the pay gap through the internal governance variables of a company [15]. This study takes the political connections as the entry point, analyzes its impact on company’s internal pay distribution and pay gap, explains the reasons why political connections increase the pay gap from the perspective of agency theory, and further expands the achievements in the research field of pay gaps. Second, Yang [15] demonstrated the vicious circle triggered by political connections through model derivation. Corporates’ investment in political connections will weaken the building of its internal capacity, resulting in the lack of heterogeneous and high-quality product supply in the market. The fierce competition of homogeneous products has greatly reduced the profits of enterprises, and the limited profits will restrict the R&D capabilities of enterprises. In the frustration of being unable to expand living space, the enterprises’ dependence on government and control of costs will be further strengthened, which will cause the government departments to obtain large amounts of rent, while the average worker’s income grows slowly. That is, the vicious circle of “political connections prevails-income distribution is seriously unfair-corporates lack of capabilities-political connections is more prevalent…”. This study provides empirical evidence at the micro-enterprise level for this theoretical derivation, namely political connections may trigger unfair income distribution within company. Third, this study also forms a useful supplement to the literature that has used national institutional factors to explain the practice of corporate governance [2]. The study provides a new analytical perspective for rationally evaluating the governance effects of government compensation regulation at the micro-enterprise level, indicating that government interventions implemented by politically connected executives have not played a positive role in narrowing the income distribution gap. The evidence in this study provides strong policy implications for regulators. There is still a long way to deepen the reform of SOEs. To improve the income distribution system, it still needs to overcome many difficulties and move forward. To establish and improve the market mechanism and promote market-oriented reforms in an all-round way, both “effective market” and “promising government” are necessary. Thus, the market may provide good services on resource allocation for enterprise development and social progress. To make the government play an efficient role, we must use reform to stimulate market vitality, use policies to guide market expectations and use laws to regulate market behaviors.

The remaining content of this paper is organized as follows. Section 2 introduces China’s relevant institutional background, conducts theoretical analysis and proposes corresponding hypotheses. Section 3 sets up specific variables and provides the measurement model. Section 4 deals with quantitative investigation of political connections, Chinese listed companies’ pay gap between executives and employees and other enterprise characteristics. Section 5 reports the analysis results of empirical tests. Section 6 conducts a robustness test. Concluding remarks and implications are presented in Section 7.

2. Institutional Background, Theoretical Analysis and Hypothesis

The existing theoretical studies on the pay gap are mainly on the basis of the tournament theory [6,16], social comparison theory [17], relative exploitation theory [18], distribution preference theory [19] etc. Based on these theories, Chinese scholars have studied the influence of corporate governance structure, industry and ownership on within-firm pay gap [20], but these studies have not considered the China’s unique background-political connection. Next, we will elaborate the institutional background from two aspects: one is the political connection of Chinese companies, and the other is the development of China’s salary system.

In an increasingly competitive environment, the government and policies have become an important tool for enterprise competition [21]. Winning the government’s “helping hand” and making full use of this competitive tool has two important implications. First, from the “environmental perspective”, the political environment is an external environment that plays a key role in the survival and development of enterprises; especially in the context of China’s transitional economy, the government still plays a leading role in economic activities [9,22]. Changes in government policies can directly affect the survival of an industry. This political uncertainty exposes companies to huge risks [23], and good government–enterprise relations help companies to perceive the wind direction, make preparations early before “uncertainty” becomes “certainty”, fully respond [24], mitigate the impact of policies and actively adapt to changes in the political environment. Second, from the perspective of “resources”, the government has many public resources and disposal rights. These resources are related to the survival and sustainable development of enterprises. However, due to limited resources and high information costs, the government cannot provide all enterprises in need with a helping hand free of charge, and they cannot efficiently allocate resources to the enterprises most in need. So some specific political connection with the government [15] can help enterprises obtain key resources and establish competitive advantages. Just as the market competition strategy is used to occupy the product market, in order to obtain the government’s “helping hand”, enterprises must invest resources to formulate and implement a political competition strategy [21]. As a common political competition strategy used by enterprises, political connection has received extensive attention in the research. Political connection is an essential relationship resource with senior executives as the carrier [25]. Enterprises may encourage senior executives to seek political identities, such as members of the National People’s Congress (NPC) or the Chinese People’s Political Consultative Conference (CPPCC), or directly hire retired officials as corporate executives, to establish good relations with the government [12].

Due to the difference in the important nature of ownership, there are significant differences between China’s SOEs and non-SOEs in terms of political connection and salary system. In terms of political connection, from the perspective of the motivation of formation, the political connection of SOEs is the product of government intervention. The government controls the personnel appointment and removal of SOEs executives and grants political identities to company executives to implement political intervention to ensure the effective control [9] of SOEs. For the purpose of political competition, non-SOEs obtain the “helping hand” of the government by establishing political connections. From the perspective of the cost of cultivating political connections, there is a natural relationship between SOEs and the government, and most of the political connections are the result of government’s active intervention, so the cost of cultivating political resources for SOEs is very low; however, non-SOEs are significantly different because the establishment and maintenance of political connections require a lot of resources, coordination and communication with government departments or engage in more activities to meet government performance needs, resulting in higher non-productive expenditures [15,26]. The economic consequences of political connections are determined not only by motivation, but also by input costs. The so-called motive decision means that SOEs are subject to policy favoritism due to government intervention and may achieve more social goals [27,28], while private enterprises use political connections to win government attention, such as financing concessions [29,30,31,32,33,34], tax concessions [35], and government subsidies [36,37]. The so-called “cost-affected” refers to the private enterprise’s investment in political connections under resource-constrained conditions, which will result in insufficient investment in core capacity building, and have a long-run negative impact on the organization, incentive mechanism and cultural construction of the enterprise [15].

The internal income distribution of an enterprise is the last link in a series of activities such as the production and operation of an enterprise. The economic consequences of political connection for enterprises are naturally reflected in this link. How political connection affects the pay gap within an enterprise is different due to differences in the motivations and input costs of SOEs and non-SOEs in forming political connections.

For non-SOEs, political connection as an enterprise resource can improve or optimize the external environment of the enterprise and bring economic benefits to the enterprise. Then, non-SOEs driven by the goal of maximizing value naturally have a strong incentive to hire executives with political identities or backgrounds to seek external resources and interests. Under the mechanism of market competition, for executives with political connections, the high rent-seeking ability to obtain government resources or policies will be priced by the market [9]. At the same time, in order to ensure that executives with political connections are motivated to obtain external resources, the company will also provide corresponding incentives [12]. So, in politically connected enterprises, the salary level of senior executives will be higher [9,14]. The premium paid for soliciting politically related executives is the cost of investing in establishing political connections. As pointed out by Yang [15], enterprises with limited resources must make a trade-off between capacity building and political connections. Compared with SOEs, the existence and development space of non-SOEs are squeezed [38]. Under the constraints of salary resources, conflicts may arise between the interests of management and the interests of labor, and they are more likely to erode employee wages to earn profits [38,39]. In addition, Tian and Zhang [27] found that the political connection of a private holding company would reduce the total labor cost of the company. Therefore, the premium compensation of politically connected executives must be achieved by sacrificing the salaries of ordinary employees, which results in unequal income distribution between executives and employees, and further widens the gap between executives and employees. In summary, we have the first hypothesis:

H1. 

For non-SOEs, political connections increase the salary gap between executives and employees.

H1 can be further extended to two sub-hypotheses:

H1.1. 

For non-SOEs, political connections increase the executive compensation.

H1.2. 

For non-SOEs, political connections decrease the employee compensation.

For SOEs, political connection is an important means of government intervention, and the impact on the pay gap within the enterprise reflects the government’s policy intentions. Therefore, the influence of the political linkages of SOEs on the pay gap may have some consistency with the direction of SOEs salary system reform. In the context of China’s unique “dual-track” economic pattern, China’s corporate salary system has a split development between different ownerships systems [40,41]. Due to clearer property rights and the pursuit of economic benefits, which are relatively simple business objectives, non-SOEs’ salary systems have always followed the market-oriented trend. The salary system of SOEs has gone through twists and turns with the pace of SOEs reform.

One of the consistencies lies in the trend of gradually linking the salary with performance presented by market-oriented corporate reforms. For example, from 2003 to 2004, the State-owned Assets Supervision and Administration Commission (SASAC) successively issued the “Interim Measures for the Performance Evaluation of the Heads of Central Enterprises” (SASAC Order No. 30), and “Interim Measures for the Compensation Management of the Heads of Central Enterprises” (SASAC Distribution (2004) No. 227), a request to promote the marketization of the income distribution of the heads of central enterprises, and link the compensation of executives of SOEs with the evaluation of operating performance. In 2005 and 2006, the “Notice on Linking Total Wages with Economic Benefits of Central Enterprises in 2005/2006” (SASAC Distribution (2005) No. 303, SASAC Distribution (2006) No. 266) was released year by year. In 2006, the “Notice on Printing ‘the Trial Measures for the Implementation of Equity Incentives for State-owned Listed Companies (Domestic)’” (SASAC (2006) No. 175) and “Notice on Printing ‘the Trial Measures for the Implementation of Equity Incentives for State-owned Listed Companies (Overseas)’” (SASAC (2006) No. 8) permitted SOEs to implement an equity incentive system to executives based on corporate performance. Under the guidance of such policies, the compensation of senior executives of SOEs is mainly set by the competent department of state-owned equity, and the compensation of the heads of enterprise is composed of three parts: base salary, performance salary and medium- and long-term incentive units, and linked with the performance appraisal results according to the provisions of the SASAC.

Other consistencies are reflected in the three “salary regulations” issued in response to the widening income distribution gap. For example, in February 2009, the Ministry of Finance issued the “Remuneration Management Measures for the Heads of State-owned and State-owned Financial Enterprises (Draft for Comment)”, which stipulates that the maximum annual salary of the heads of state-owned financial enterprises should not exceed RMB 2.8 million. In September 2009, with the consent of the State Council, the Ministry of Human Resources and Social Security, together with the Central Organization Department, Supervision Department, Ministry of Finance, Audit Office, and SASAC, jointly issued the “Guiding Opinions on Further Regulating the Remuneration Management of Heads of Central Enterprises” on the 16th, which stipulates that the basic annual salary of senior executives of enterprises should be related to the average salary of employees in the previous year, and the annual performance salary is determined according to the evaluation of operating performance. Since the annual performance salary cannot exceed a certain multiple of the basic annual salary, this article essentially restricts the salary level of SOEs executives by limiting the pay gap. In August 2014, the Political Bureau of the Central Committee reviewed and approved the “Remuneration System Reform Plan for the Heads of Centrally-Managed Enterprises”, which further requires strictly regulating the compensation level of executives of SOEs, and adjusting unreasonably high and excessive incomes, focusing on limiting high salaries of appointed executives of central SOEs and the heads of central SOEs in some monopolistic high-income industries. Promoting marketization is the general trend of the reform of SOEs. However, the twists and turns and contradictory policies will inevitably weaken the effect of reform of the SOEs’ salary incentive system [42]. Although studies have found that politically connected executives in SOEs are more likely to cater for government intervention [43,44], it is likely that such conflicting policies may be at a loss and cannot actively promote compensation regulation by narrowing the pay gap. Policy contradictions will weaken the effect of government intervention, leading to the blurring of the impact of political connections formed by political intervention on the pay gap within SOEs. However, it is worth noting that the executives of SOEs have the dual roles of “political man” and “economic man”. In the context of the lack of actual controllers and the prominent problems of insider control in SOEs, the behavior of senior management has an important impact on the income distribution within the enterprise. Studies have shown that the improvement of the economic performance of SOEs will increase the probability of promotion of SOEs executives and reduce the probability of leaving. In SOEs, operating performance has become an important evaluation indicator to determine the remuneration and political promotion of executives. Executives in SOEs have sufficient incentives to improve corporate performance, from the perspective of the “economic man”. On the other hand, in SOEs executives accumulate performance capital for their political promotion, to meet the demands of the “political man”. Thirdly, existing studies have found that, compared with non-SOEs, the salaries of ordinary employees in SOEs are significantly higher. It is beneficial for the executives of SOEs to take corporates’ residual income as their own. Therefore, the motivation and behavior of the politically connected executives in SOEs in terms of income distribution directly determine how the political connections affect the internal pay gap of the enterprise.

Judging from the motivation of the “economic man”, on the one hand, politically-connected executives in SOEs can obtain more subsidies and help from the government, and increase the performance basis of their salary evaluation; on the other hand, based on their political background, they often have a closer relationship with the SASAC, and have a stronger ability to bargain on the executive compensation performance linking program^, so that they can obtain higher compensation. From the perspective of a “political person”, the government is more likely to intervene in companies through politically connected executives [44]. Politically related executives may cause companies to be more frequently intervened in by the government and need to play the role of “political person”, to bear more of the policy burden. However, precisely because of the complexity of business objectives, in the case of poor business performance, it can blur the relationship between corporate performance and executive efforts and therefore create rationalized excuses [45] to strive for higher salary for executives. In summary, executives of SOEs with political connections may receive significantly higher salaries. Regarding employee compensation, although SOEs may bear part of the social burden or because of the soft budget constraints, SOEs may not have too many restrictions on labor costs, and even under the regulation. When restricting of executives’ salary multiples companies may also increase employee salaries at the same time. However, due to the Geometrid effect of the remuneration system [46,47], the remuneration of executives in China’s enterprises is “sticky”, and the remuneration of employees does not have the characteristic of stickiness. Executives of SOEs can share the benefits of increased remuneration brought about by the rise in corporate performance, while it is impossible to obtain the corresponding salary increase equally, which will inevitably lead to an increase in the salary gap.

In summary, we have the second hypothesis:

H2. 

For SOEs, political connections increase the salary gap between executives and employees.

H2 can be further extended to two sub-hypotheses:

H2.1. 

For SOEs, political connections increase the executive compensation.

H2.2. 

For SOEs, political connectionsmay decrease the employee compensation.

3. Variables and Model

In this chapter, we will give the definitions and measurement methods of independent variables, dependent variables and controlled variables, and introduce our research model in detail.

3.1. Dependent Variables

In this study, we have four dependent variables.

Following Zhang [8] and Hambrick and Siegel [48], this paper measures the pay gap between executives and employees by two indicators, the absolute pay gap and the relative pay gap, as two dependent variables. The absolute pay gap is the natural logarithm of difference between executive compensations and ordinary employee compensations, as described in (1). The relative pay gap is the ratio of the average salary of executives to the average salary of ordinary employees, as described in (2).

$$absolute pay gap=ln(executives pay-employees pay)$$

(1)

$$relative pay gap=\frac{ executives average pay}{ employees average pay}$$

(2)

In addition, we include the average salary of ordinary employees as the third dependent variable.

$$ordinary employees average pay=\frac{ Cash paid to and for employees-pay of superiors}{ the number of employees-the number of senior superiors}$$

(3)

The average pay of executives is the fourth dependent variable. It includes two parts: monetary compensation and equity incentives. However, due to the late implementation of the equity incentive plan in China, the majority of executives have a large proportion of monetary compensation in total compensation. The phenomenon of zero holdings is relatively common, and it is difficult to judge whether the stocks held by executives are from company rewards or self-purchase according to public data [46]. Following Fang [46] and Tang and Sun [9], we measure the executive compensation by the average of the “total compensation of the top three executives with the highest remuneration” of listed companies.

3.2. Explanatory Variable

Existing studies usually use the political background of executives as a variable to measure the corporate political connections. For example, Yu and Pan [32] took the chairman or CEO with a political background as an explanatory variable. Luo and Tang [49] used the proportion of board members with political backgrounds as the measure of political connections. Deng and Zeng [50] use the proportion of politically connected executives among directors and senior executives as the measure of political connections. In our study, the research target is the pay gap between executives and employees, which is a typical principal-agent problem. The CEO is one of the most important agents of a company [39]. Thus, we use the political relationship of CEO as the variable of the political connection. We define PC in (4), to measure whether the CEO of the company has experience as a representative of the NPC or CPPCC (representative-type connection), or has work experience in government departments (official-type connection).

$$PC=\{\begin{array}{ll}1,& CEO of the company has been or is currently a representative of the NPC or CPPCC,or has served in government departments\\ 0,& no politcal connection\end{array}$$

(4)

3.3. Control Variables

In addition, we also include a series of control variables classified in two types: the company’s basic characteristics and financial status, and the characteristics of corporate governance.

In the first class, variables related to the company’s basic characteristics and financial status include:

• Firm size. Following Zhang [8] and Wei, Dong and Liu [51], we use the natural logarithm of the total number of employees of the company to measure the firm size.
• Operating performance. Following Lu et al. [38], we use the return on assets (ROA) to measure the company’s operating performance.
• Market performance. Following Hwang and Kim [52], we use the annual return of the company’s stock.
• Operating risk. We use the volatility of the company’s stock yield, to measure the company’s operating risk, following some studies [53,54];
• Company growth. According to Tang and Sun [9], company growth should be proportional to executive compensation;
• Corporate financial risk. Following Lu et al. [38], we use the asset-liability ratio to measure the financial risk of enterprises.

In addition, because there are some differences in the average wage level between different industries and regions, and different years may also cause salary fluctuations due to different economic environment, this paper controls the year, the industry and region of the company.

In the second class, variables related to corporate governance include:

• Size of the board of directors [52,54].
• Ratio of independent directors [55].
• Dual positions of chairman and CEO. If the chairman and CEO is the same person, it may lead to a stronger control by executives and thus increase their own compensation levels [56].
• The top shareholding ratio. The higher the proportion of the largest shareholder holds, the stronger its ability to supervise the company’s executives and the weaker the executives’ ability to set their own compensation [9], so the pay gap may be smaller.

We create these variables above and describe them in

Table 1. Definition of specific variables.

Variables	Symbols	Definitions and Descriptions
Pay gap and Pay variables
Executive-employee relative pay gap	Paygap_rel	Executive average pay/Ordinary employees average pay
Executive-employee absolute pay gap	Paygap_abs	Natural logarithm of difference between executives’ average pay and ordinary employees’ average pay
Executive average pay	Exepay	natural logarithm of the average salary of senior executives
Ordinary employees average pay	Ordpay	(Cash paid to and for employees—pay of superiors)/(the number of employees–the number of senior superiors)
Political connections variable
Political connections or not	PC	Virtual variable of whether the executive is political connection, if yes, 1; if not, 0
Company’s basic characteristics and financial status
Firm size	Firmsize	The natural logarithm of the company’s total number of employees
Operating performance	ROA	Net profit/total asset at year end
Annual excess return of company stock	Return	Annual return on company stock average annual market return weighted by market capitalization during the same period
Volatility of company’s stock returns	Volatility	The variance of the monthly return of company stock for that year
Company growth	MB	The ratio of the total market value of company stocks at the end of the year to the book value of equity
Corporate financial risk	Lev	Company total liabilities/Company total assets
Industry	Industry	Based on the 2012 revised version of the industry classification guidelines for listed companies issued by the CSRC as the classification standard, the industry is divided into categories
Province	Province	Location of company (province, municipality or autonomous region)
Year	Year	Year of the reporting period
Corporate governance characteristics
Board size	Boardsz	the natural logarithm of the number of directors in the board
Independent director ratio	Indpt	The proportion of independent directors among total directors
Whether the general manager has dual positions	Dual	1 if the chairman and general manager are served by the same person, otherwise 0
Top1 Shareholding ratio	Top1	Shareholding ratio of the shareholders holding the most shares

.

3.4. Model Settings

We present the following regression model, to test the influence of political connections on executive and employee compensations, and their gaps.

$$DV={\beta }_0+{\beta }_{1 }PC+{\beta }_{c}Controls+\epsilon$$

(5)

The dependent variables (DV) include $Exepay$, $Ordpay$ and $Paygap$ (in relative and absolute formats). PC is the political connection variable. “Controls” refers to a series of control variables, including firm size (Firmsize), operating performance (ROA), market performance (Return), operating risk (Volatility), company growth (MB), corporate financial risk (Lev), board size (Boardsz), independent director ratio (Indpt), whether the general manager has dual positions (Dual), the shareholding ratio of the largest shareholder (Top1) and the year and industry and region in which the company is located.

4. Data Sources and Descriptive Statistics

In this chapter, we will explain our research objects and method of samples selection, and then we will describe our data structure and characteristics. The related results are presented in

Table 2. Descriptive statistics of main variables.

Variables	N	Min	Mean	Max	P50	Sd
PC	10,562	0	0.194	1	0	0.395
Paygap_rel	10,562	1.032	7.742	119.3	5.786	7.361
Paygap_abs	10,562	7.828	12.74	16.13	12.76	0.825
Exepay	10,562	9.925	12.98	16.14	12.97	0.694
Ordpay	10,562	7.784	11.19	13.78	11.16	0.525
Firmsize	10,562	4.605	7.628	13.22	7.525	1.239
ROA	10,562	−9.486	0.0728	15.81	0.0724	0.397
Return	10,562	−1.187	0.0503	11.84	−0.0316	0.428
Volatility	10,562	0	0.0208	34.14	0.011	0.36
MB	10,562	−0.399	0.004	2.789	0.003	0.0285
Lev	10,562	0.007	0.436	8.612	0.426	0.261
Top1	10,562	2.197	36.07	89.41	34.28	15.44
Dual	10,562	0	0.253	1	0	0.435
Indpt	10,562	0.125	0.37	0.8	0.333	0.0547
Boardsz	10,562	1.386	2.164	2.89	2.197	0.196

,

Table 3. Politically connected firms VS non-politically connected firms.

Vairables	Samples of Political Connections		Samples of Non- Political Connections		Difference Test
	Mean Value	Median	Mean Value	Median	t-Value	z-Value
SOEs (samples of non-political connections = 3739; samples of political connections = 604)
Paygap_rel	8.581	5.894	7.229	5.274	4.128 ***	4.217 ***
Exepay	13.17	13.08	13.07	13.08	3.252 ***	2.376 **
Ordpay	11.31	11.31	11.35	11.34	−1.693 *	−2.035 **
Non-SOEs (samples of non-political connections = 4776; samples of political connections = 1443)
Paygap_rel	8.737	6.604	7.737	5.936	4.589 ***	5.866 ***
Exepay	12.94	12.91	12.91	12.89	1.363	1.319
Ordpay	11.01	10.97	11.1	11.07	−6.169 ***	−6.910 ***

Notes: The t-value in column (3) represents the mean test statistic of politically and non-politically connected samples, and the z-value represents the median Wilcoxon rank and test statistic of the two samples, * p < 0.1, ** p < 0.05, *** p < 0.01.

and

Table 4. Pearson correlation coefficient.

	PC		Paygap_rel	Paygap_abs	Exepay	Ordpay	Firmsize	ROA
PC	1.000
Paygap_rel	0.063 ***		1.000
Paygap_abs	0.027 ***		0.647 ***	1.000
Exepay	0.015		0.622 ***	0.977 ***	1.000
Ordpay	−0.084 ***		−0.237 ***	0.293 ***	0.441 ***	1.000
Firmsize	−0.038 ***		0.360 ***	0.343 ***	0.329 ***	−0.042 ***	1.000
ROA	0.004		0.051 ***	0.081 ***	0.084 ***	0.037 ***	0.013	1.000
Return	0.004		0.024 **	0.023 **	0.021 **	−0.005	−0.036 ***	0.036 ***
Volatility	0.016 *		−0.002	0.012	0.013	0.013	−0.005	0.005
MB	−0.007		−0.014	−0.060 ***	−0.046 ***	−0.024 **	−0.038 ***	−0.066 ***
Lev	−0.056 ***		0.063 ***	0.034 ***	0.049 ***	0.035 ***	0.299 ***	−0.001
Top1	−0.020 **		−0.045 ***	0.056 ***	0.073 ***	0.146 ***	0.207 ***	0.023 **
Dual	0.248 ***		0.021 **	−0.013	−0.029 ***	−0.091 ***	−0.155 ***	0.001
Indpt	0.027 ***		−0.005	0.006	0.006	0.016	0.019 *	−0.016 *
Boardsz	−0.022 **		0.080 ***	0.134 ***	0.147 ***	0.102 ***	0.267 ***	0.012
	Return	Volatility	MB	Lev	Top1	Dual	Indpt	Boardsz
Return	1.000
Volatility	0.352 ***	1.000
MB	0.084 ***	0.005	1.000
Lev	0.006	0.017 *	0.020 **	1.000
Top1	−0.003	0.020 **	−0.030 ***	0.018 *	1.000
Dual	0.018 *	0.017 *	0.021 **	−0.155 ***	−0.047 ***	1.000
Indpt	0.007	−0.01	0.011	−0.019 *	0.062 ***	0.094 ***	1.000
Boardsz	−0.017 *	0.013	−0.021 **	0.144 ***	0.003	−0.173 ***	−0.441 ***	1.000

Notes: t statistics in parentheses, * p < 0.1, ** p < 0.05, *** p < 0.01.

.

4.1. Data Sources and Processing Methods

We used A-share listed companies on the Shanghai and Shenzhen Stock Exchange as the target. According to the “Reform Plan for the Remuneration System of Heads of Enterprises Managed by Central Government”, we chose the observation period between 2010 and 2014. This reform, which was officially in effect since 1 January 2015, imposed an upper bound on the salary of senior executives of central enterprises. Existing studies have shown that reform had practical binding force, which reduced the monetary salary level of central enterprise executives, shortened the salary gap within enterprises [57] and limited the growth rate of executive salary of SOEs. An observation period before 2015 will be more reasonable to avoid the influence of the government reform plan. Therefore, we used the observation period between 2010 and 2014.

We investigated these companies’ different aspects, including salary, executive political connections, company basic characteristics, financial status and corporate governance. We collected and processed data about executive political connections from the annual reports and official homepage of companies. The remaining data are from the CSMAR database.

In the process of data cleaning, we filtered out companies from the financial service industry and companies whose political connections information was unavailable. We filtered out companies whose sizes are smaller than 100 employees according to Liang and Feng [58], and removed companies whose asset-liability ratio (Lev) is greater than 10 and ROA is less than −10, according to Liu et al. [12]. We filtered out companies whose average executive compensations are lower than the average employee compensations. As a result, we investigated 2479 companies in the A-share market from 2010 to 2014. During this period, there were 3836 CEOs in these companies, and 696 CEOs of them had political connections. After merging into cross-sectional observation data, we have 10,562 firm-year observations (1669 firm in 2010, 2000 firm in 2011, 2268 firm in 2012, 2277 firm in 2013, and 2348 firm in 2014).

4.2. Descriptive Statistics

Table 2 reports the descriptive statistics of each variable. Among the total samples, 19.4% of them have political connections.

Further statistics found that the SOEs have a relatively lower political connection, about 13.8%, and the non-SOEs have a relatively higher political connection, about 23%. This result is consistent with Tian and Zhang [27], which reveals the fact that seeking political connection is very common in China, especially for non-SOEs, while the effects of political connections for different natures of ownership are not the same.

From the perspective of the pay gap, the average salary of executives is 7.74 times that of employees, and the maximum is 119 times; the standard deviation is 7.36, indicating that there is a large pay gap between executives and ordinary employees, and there are differences in the pay gap between different companies. Further statistics on the time trend show that the proportion of companies with a larger salary gap is decreasing, while the proportion of companies with a smaller salary gap is increasing, which is manifested in the fact that the proportion of the sample with high salary multiples (the pay gap is more than 7 times) is declining, while the proportion of the sample with low salary multiples (the salary gap is below 7 times) is increasing. To further observe the difference between the SOEs and non-SOEs in the pay gap, we conducted a group test. The results in Table 3 show that, in terms of the average, whether it is SOEs group or non-SOEs group, the pay gap within-firm of politically connected enterprises (salary multiples) is significantly higher than non-politically connected, and this difference is significant at the level of 1%.

Table 4 reports the Pearson correlation coefficient between variables. The correlation coefficient between PC and pay gap is significantly positive, which initially indicates that political correlation has a significant positive effect on the within-firm pay gap of executives. In addition, the correlation coefficients of the various variables involved in the model are not large, and the results of estimating the variance inflation factor (VIF) after regression in the empirical stage show that the average value of VIF is 1.15. The results of these tests indicate that there is no serious multicollinearity problem.

5. Political Connections and Income Distribution Effect

In this chapter, we will make regression analysis between the political connection and the within-firm salary gap and the salary level of executives and employees. The related results are presented in

Table 5. The impact of political connections on the executive–employee pay gap.

	Full Sample		SOEs		Non-SOEs
	Paygap_rel	Paygap_abs	Paygap_rel	Paygap_abs	Paygap_rel	Paygap_abs
PC	0.831 ***	0.056 ***	0.721 **	0.057 *	0.444 **	0.045 **
	(4.788)	(3.025)	(2.120)	(1.722)	(2.231)	(2.065)
Firmsize	2.781 ***	0.281 ***	2.670 ***	0.244 ***	3.386 ***	0.320 ***
	(24.831)	(37.729)	(18.036)	(20.978)	(19.865)	(31.876)
ROA	0.755 ***	0.129 ***	0.904 **	0.150 **	0.406 **	0.094 ***
	(4.059)	(3.667)	(2.398)	(2.140)	(2.429)	(2.605)
Return	0.608 ***	0.058 ***	0.504 **	0.045	0.553 **	0.053 **
	(3.463)	(3.287)	(2.036)	(1.604)	(2.412)	(2.389)
Volatility	−0.259 ***	−0.006	−0.586 **	0.085 *	−0.329 ***	−0.018 *
	(−3.231)	(−0.515)	(−2.031)	(1.819)	(−3.734)	(−1.739)
MB	0.625	−1.093 ***	12.697 *	−0.556	−0.564	−1.233 ***
	(0.252)	(−6.309)	(1.647)	(−0.742)	(−0.375)	(−10.400)
Lev	−2.123 ***	−0.341 ***	−1.765 ***	−0.412 ***	−1.176 ***	−0.290 ***
	(−5.755)	(−8.082)	(−3.473)	(−3.986)	(−2.606)	(−6.723)
Top1	−0.063 ***	−0.002 ***	−0.097 ***	−0.004 ***	−0.009 *	0.000
	(−12.281)	(−5.076)	(−11.012)	(−5.407)	(−1.646)	(0.548)
Dual	0.913 ***	0.050 ***	1.059 ***	0.097 **	0.503 ***	0.045 **
	(5.453)	(2.950)	(2.767)	(2.373)	(2.837)	(2.359)
Indpt	−3.317 ***	0.088	−5.878 ***	0.066	3.944 *	0.517 ***
	(−2.595)	(0.652)	(−3.848)	(0.336)	(1.894)	(2.670)
Boardsz	−0.600	0.307 ***	−1.051 *	0.215 ***	2.081 ***	0.467 ***
	(−1.360)	(7.466)	(−1.740)	(3.608)	(3.182)	(7.842)
Industry/Year /Province	Y	Y	Y	Y	Y	Y
_cons	−7.523 ***	9.555 ***	−4.648 ***	9.877 ***	−20.568 ***	8.897 ***
	(−5.671)	(70.148)	(−2.780)	(51.782)	(−8.878)	(43.196)
N	10,562	10,562	4343	4343	6219	6219
Adj-R2	0.233	0.293	0.294	0.338	0.281	0.300
F	18.020	64.230	11.410	37.090	16.690	42.840

Notes: t statistics in parentheses, * p < 0.1, ** p < 0.05, *** p < 0.01. Robust standard error is used to overcome the possible heteroscedasticity of the model.

and Table 6.

5.1. Corporate Political Connections and the Pay Gap between Executives and Employees

Following H1 and H2, we use the regression model in (5) to examine the impact of political connection on the within-firm pay gap. The results in Table 5 show that in the full sample test, the coefficient of PC is positive, indicating that the political connection significantly increases the pay gap between executives and employees. We further run the regression in both SOEs and non-SOEs groups. In both SOEs and non-SOEs groups, the existence of political connection will significantly increase the pay gap within firm. Therefore, H1 and H2 cannot be rejected. The result is consistent with Du et al. [59], indicating that the political connection of non-SOEs will significantly increase the within-firm salary gap. We extend the research object to SOEs and all listed companies, and further expand the research in the field of salary gap.

In order to clarify the factors affecting the internal salary gap, we control the factors of economy and corporate governance. From the perspective of economic influence factors, the larger the scale of the enterprise, the greater the pay gap within the enterprise. The coefficients of ROA and return are significantly positive, indicating that the pay gap between executives and employees is positively related to the company’s accounting performance and market performance. These results are consistent with Lin and Lu [60]. In addition, the growth of non-SOEs is significantly negatively correlated with the within-firm pay gap. This may be because the quality of employees in high-growth enterprises at the forefront of contemporary economy is generally high, and the human capital is not much different, so the pay gap is small. Corporate financial risk is significantly negatively correlated with the pay gap, indicating that the greater the financial risk of an enterprise, the smaller the gap between executives’ and employees’ salaries.

From the perspective of influencing factors of corporate governance, the company with the largest shareholding ratio of the largest shareholder has a smaller gap between executives and employee’s compensation, while in those companies where the roles of chairman and general manager are served by the same person and the board of directors, the gap is larger, the within-firm pay gap is bigger. These results are consistent with Xu et al. [61] in that the larger salary gap results from the governance environment where the management receives less supervision and they can raise their salary level more easily. The role of independent directors in effectively supervising executives and reducing the pay gap is only effective in SOEs. In non-SOEs, the higher proportion of independent directors has widened the pay gap between executives and employees. This difference indicates the necessity for us to discuss ownership.

5.2. Corporate Political Connections and the Executive and Employee Salary

In this section, we examine the impact of corporate political connections on the compensation of executives and employees at both ends of the gap, which corresponds to the sub-hypotheses. Table 6 presents the regression results. In the full sample, political connections increase executive compensation (Exepay, 0.037), while decreasing employee compensation (Ordpay, −0.058). This result of political connections significantly increasing executive compensation is consistent with the conclusion of Tang and Sun [9].

We further ran the regression in non-SOEs and SOEs groups. In the non-SOEs group, the political connections significantly increase executive compensation (Exepay, 0.032) and significantly reduced employee compensation (Ordpay, −0.034). Therefore, the sub-hypotheses (H1.1, H1.2) cannot be rejected. The result of political connections significantly decreasing employee compensation further explains the phenomenon that the political connections can reduce the labor cost of private enterprises, according to the study of Wei, Dong and Liu [51].

A different situation appears in the SOEs group. In Table 6, the SOEs group has similar results to the non-SOEs group in terms of the coefficient signs of PC. However, these coefficients (0.044 on Exepay, and −0.018 on Ordpay) are not statistically significant. These results can hardly provide strong evidence for the test of H2.1 and H2.2, though the main hypothesis H2 has been supported previously. A possible reason may be that, on the one hand, the influence of political connection on increasing executive compensation would be limited to the extent to which pay regulation restricts executive compensation in SOEs, and, on the other hand, politically connected executives in SOEs do not need to use their background to access government resources or engage in policy rent-seeking activities in the same way that politically connected executives in non-SOEs do, and this is what leads to higher pay premium. However, although the effect of political connections on both executive compensation and employee compensation in SOEs does not reach the significance level (the effect of political connections on higher executive compensation is already close to the significance level), the superimposed effect of both brings the effect of political connections on the pay gap to the significance level.

Table 6. The impact of political connections on executive and employee compensation.

	Full Sample		SOEs		Non-SOEs
	Exepay	Ordpay	Exepay	Ordpay	Exepay	Ordpay
PC	0.037 **	−0.058 ***	0.044	−0.018	0.032 *	−0.034 ***
	(2.381)	(−5.087)	(1.563)	(−0.797)	(1.773)	(−2.641)
Firmsize	0.219 ***	−0.050 ***	0.178 ***	−0.087 ***	0.252 ***	−0.068 ***
	(34.897)	(−9.437)	(18.766)	(−12.829)	(29.095)	(−9.789)
ROA	0.113 ***	0.038 ***	0.134 **	0.054 *	0.083 ***	0.035 **
	(3.810)	(2.591)	(2.129)	(1.707)	(2.937)	(2.051)
Return	0.042 ***	−0.016	0.029	−0.018	0.039 **	−0.014
	(2.764)	(−1.403)	(1.265)	(−0.960)	(2.045)	(−1.077)
Volatility	−0.002	0.015	0.092 **	0.138 ***	−0.012	0.013 **
	(−0.198)	(1.572)	(2.223)	(3.833)	(−1.384)	(2.371)
MB	−0.572 ***	−0.307	−0.724	−1.768 *	−0.654 ***	−0.237 **
	(−6.351)	(−1.610)	(−1.218)	(−1.897)	(−8.689)	(−2.285)
Lev	−0.256 ***	0.037	−0.312 ***	0.002	−0.243 ***	−0.080
	(−7.565)	(0.941)	(−3.942)	(0.084)	(−6.581)	(−1.543)
Top1	−0.001 ***	0.004 ***	−0.002 ***	0.007 ***	0.000	0.000
	(−3.081)	(14.134)	(−3.013)	(16.258)	(0.581)	(0.883)
Dual	0.026 *	−0.069 ***	0.079 **	−0.031	0.030 *	−0.017
	(1.835)	(−6.681)	(2.353)	(−1.343)	(1.869)	(−1.516)
Indpt	0.162	0.533 ***	0.186	0.642 ***	0.469 ***	0.306 ***
	(1.437)	(6.196)	(1.157)	(5.231)	(2.879)	(2.591)
Boardsz	0.329 ***	0.395 ***	0.260 ***	0.378 ***	0.419 ***	0.213 ***
	(9.668)	(14.646)	(5.420)	(10.211)	(8.372)	(5.832)
Industry/Year/Province	Y	Y	Y	Y	Y	Y
_cons	10.144 ***	9.802 ***	10.414 ***	9.877 ***	9.742 ***	10.577 ***
	(90.342)	(112.111)	(67.184)	(77.389)	(56.149)	(85.797)
N	10,562	10,562	4343	4343	6219	6219
Adj-R2	0.312	0.303	0.363	0.361	0.309	0.305
F	70.34	93.05	40.11	40.79	43.71	48.38

Notes: t statistics in parentheses, * p < 0.1, ** p < 0.05, *** p < 0.01. Robust standard error is used to overcome the possible heteroscedasticity of the model.

Table 6. The impact of political connections on executive and employee compensation.

	Full Sample		SOEs		Non-SOEs
	Exepay	Ordpay	Exepay	Ordpay	Exepay	Ordpay
PC	0.037 **	−0.058 ***	0.044	−0.018	0.032 *	−0.034 ***
	(2.381)	(−5.087)	(1.563)	(−0.797)	(1.773)	(−2.641)
Firmsize	0.219 ***	−0.050 ***	0.178 ***	−0.087 ***	0.252 ***	−0.068 ***
	(34.897)	(−9.437)	(18.766)	(−12.829)	(29.095)	(−9.789)
ROA	0.113 ***	0.038 ***	0.134 **	0.054 *	0.083 ***	0.035 **
	(3.810)	(2.591)	(2.129)	(1.707)	(2.937)	(2.051)
Return	0.042 ***	−0.016	0.029	−0.018	0.039 **	−0.014
	(2.764)	(−1.403)	(1.265)	(−0.960)	(2.045)	(−1.077)
Volatility	−0.002	0.015	0.092 **	0.138 ***	−0.012	0.013 **
	(−0.198)	(1.572)	(2.223)	(3.833)	(−1.384)	(2.371)
MB	−0.572 ***	−0.307	−0.724	−1.768 *	−0.654 ***	−0.237 **
	(−6.351)	(−1.610)	(−1.218)	(−1.897)	(−8.689)	(−2.285)
Lev	−0.256 ***	0.037	−0.312 ***	0.002	−0.243 ***	−0.080
	(−7.565)	(0.941)	(−3.942)	(0.084)	(−6.581)	(−1.543)
Top1	−0.001 ***	0.004 ***	−0.002 ***	0.007 ***	0.000	0.000
	(−3.081)	(14.134)	(−3.013)	(16.258)	(0.581)	(0.883)
Dual	0.026 *	−0.069 ***	0.079 **	−0.031	0.030 *	−0.017
	(1.835)	(−6.681)	(2.353)	(−1.343)	(1.869)	(−1.516)
Indpt	0.162	0.533 ***	0.186	0.642 ***	0.469 ***	0.306 ***
	(1.437)	(6.196)	(1.157)	(5.231)	(2.879)	(2.591)
Boardsz	0.329 ***	0.395 ***	0.260 ***	0.378 ***	0.419 ***	0.213 ***
	(9.668)	(14.646)	(5.420)	(10.211)	(8.372)	(5.832)
Industry/Year/Province	Y	Y	Y	Y	Y	Y
_cons	10.144 ***	9.802 ***	10.414 ***	9.877 ***	9.742 ***	10.577 ***
	(90.342)	(112.111)	(67.184)	(77.389)	(56.149)	(85.797)
N	10,562	10,562	4343	4343	6219	6219
Adj-R2	0.312	0.303	0.363	0.361	0.309	0.305
F	70.34	93.05	40.11	40.79	43.71	48.38

Notes: t statistics in parentheses, * p < 0.1, ** p < 0.05, *** p < 0.01. Robust standard error is used to overcome the possible heteroscedasticity of the model.

6. Robust Test

In this chapter, we use a variety of methods to test the robustness of the conclusions of this paper. Specifically, we use PSM method (propensity score matching) to control endogenous problems. In addition, we also carried out the other five robustness tests, and the specific analysis process and results are as follows.

6.1. Endogenous Test of Political Connection

Referring to the research of Erkens et al. [62], Sakawa and Watanabel [63], we use the PSM (propensity score matching) method to overcome the bias caused by sample selection and eliminate possible endogenous problems. In this paper, 2047 company-annual samples with political connections are taken as the treatment group, and 8515 company-annual samples without political connection are taken as the control group. The idea of PSM originates from the matching estimator, and its key idea is to find a certain company j in the control group where companies with no political connection, to make it as similar (matched) as possible to the company i with political connections in the treatment group, in order to ensure that the salary distribution and salary gap are explained by the feature of whether the company has a political connection. As the relationship between political connection and within-firm pay gap is most easily influenced by corporate governance variables, we take whether the company has political connection as the dependent variable and corporate governance variables such as Firmsize, ROA, Return, Volatility, MB, Lev, Boardsz, Indpt, Dual and Top1 as the explanatory variables. In

Table 7. Logit Models Predicting on PC.

	PC
	Coefficient	t-Value
Firmsize	0.056 **	(2.159)
ROA	0.032	(0.440)
Return	−0.015	(−0.221)
Volatility	0.050	(0.521)
MB	−1.689	(−0.609)
Lev	−0.662 ***	(−4.791)
Boardsz	0.363 **	(2.231)
Indpt	1.139 **	(2.128)
Dual	1.420 ***	(24.716)
Top1	−0.003 *	(−1.711)
Industry/Year	Yes
N	10,560
Adj-R2	0.098

Notes: t statistics in parentheses, * p < 0.1, ** p < 0.05, *** p < 0.01.

, we report the results estimated by logit model. We can find that the ratio of companies with political connections are larger, with larger directors and lower asset-liability ratio than those without political connection. In a word, the control variables are an important factor affecting the estimation results of this paper.

We use the kernel matching method to calculate the tendency score and match the samples. To illustrate the reliability of PSM, we need to prove that there is no significant difference in observable corporate governance variables between the matched treated group and the control group. If there is a significant difference between the post-matching treatment group and the control group, it means that the selection of matching method is inappropriate. Before reporting the PSM estimation results, this paper makes a balance test on the matching results. Compared with before matching, the absolute value of the standard deviation of each variable after matching is less than 10, which indicates that there is no significant difference between the matched treatment group and the control group on the observable corporate governance variables. Therefore, it can be considered that the observable variables selected in this paper are appropriate, and the selection of matching methods is also appropriate. According to the matching results, the relationship between corporate political connection and internal salary gap is tested by regression. The results are shown in

Table 8. Regression after PSM: Test of political connection and within-firm salary gap.

	Full Sample		SOEs		Non-SOEs
	Paygap_rel	Paygap_abs	Paygap_rel	Paygap_abs	Paygap_rel	Paygap_abs
PC	0.830 ***	0.056 ***	0.722 **	0.058 *	0.458 **	0.046 **
	(4.793)	(3.025)	(2.124)	(1.757)	(2.308)	(2.112)
Firmsize	2.868 ***	0.285 ***	2.711 ***	0.249 ***	3.488 ***	0.324 ***
	(25.566)	(38.308)	(18.273)	(22.504)	(20.350)	(31.413)
ROA	0.974 ***	0.141 ***	0.977 **	0.149 **	0.659 ***	0.103 ***
	(4.630)	(3.727)	(2.469)	(2.148)	(3.442)	(2.615)
Return	0.527 ***	0.050 ***	0.464 *	0.061 **	0.386	0.043 *
	(2.890)	(2.805)	(1.808)	(2.150)	(1.616)	(1.907)
Volatility	−0.473	0.072	−0.339	−0.986 **	−0.172	0.103
	(−1.273)	(1.214)	(−0.116)	(−2.428)	(−0.164)	(0.436)
MB	47.657 ***	1.092	28.482 **	0.096	74.357 ***	2.046
	(3.676)	(1.027)	(2.025)	(0.072)	(4.072)	(0.905)
Lev	−2.936 ***	−0.380 ***	−2.326 ***	−0.539 ***	−1.895 ***	−0.312 ***
	(−8.860)	(−9.605)	(−4.858)	(−8.595)	(−4.513)	(−6.213)
Boardsz	−0.568	0.309 ***	−1.050 *	0.218 ***	2.017 ***	0.465 ***
	(−1.285)	(7.513)	(−1.732)	(3.665)	(3.077)	(7.798)
Indpt	−3.259 **	0.094	−5.776 ***	0.079	3.730 *	0.511 ***
	(−2.550)	(0.697)	(−3.774)	(0.402)	(1.787)	(2.633)
Dual	0.864 ***	0.047 ***	1.026 ***	0.090 **	0.466 ***	0.044 **
	(5.167)	(2.772)	(2.678)	(2.199)	(2.643)	(2.297)
Top1	−0.063 ***	−0.002 ***	−0.097 ***	−0.004 ***	−0.010 *	0.000
	(−12.391)	(−5.190)	(−11.071)	(−5.507)	(−1.810)	(0.480)
Industry/Year/Province	Y	Y	Y	Y	Y	Y
_cons	−8.109 ***	9.527 ***	−4.861 ***	9.898 ***	−21.124 ***	8.872 ***
	(−6.082)	(69.562)	(−2.910)	(51.784)	(−9.109)	(42.845)
N	10,547	10,547	4335	4335	6210	6210
Adj-R2	0.236	0.290	0.295	0.341	0.285	0.295
F	19.21	63.08	11.90	37.90	17.12	39.84

Notes: t statistics in parentheses, * p < 0.1, ** p < 0.05, *** p < 0.01. Robust standard error is used to overcome the possible heteroscedasticity of the model.

and

Table 9. Regression after PSM: Test of political connections on executive and employee compensation.

	Full Sample		SOEs		Non-SOEs
	Exepay	Ordpay	Exepay	Ordpay	Exepay	Ordpay
PC	0.037 **	−0.058 ***	0.045	−0.017	0.033 *	−0.035 ***
	(2.383)	(−5.109)	(1.598)	(−0.780)	(1.805)	(−2.742)
Firmsize	0.220 ***	−0.060 ***	0.182 ***	−0.090 ***	0.253 ***	−0.078 ***
	(34.831)	(−12.331)	(19.807)	(−12.697)	(28.228)	(−12.033)
ROA	0.119 ***	0.022	0.131 **	0.044	0.085 ***	0.014
	(3.682)	(1.242)	(2.106)	(1.391)	(2.767)	(0.648)
Return	0.036 **	−0.013	0.046 *	0.002	0.033 *	−0.004
	(2.407)	(−1.104)	(1.925)	(0.082)	(1.687)	(−0.294)
Volatility	0.075	0.112 ***	−0.912 ***	−0.824 *	0.103	0.115
	(1.415)	(2.897)	(−2.636)	(−1.802)	(0.494)	(1.438)
MB	0.097	−4.732 ***	−0.550	−3.464 **	0.828	−6.340 ***
	(0.107)	(−3.581)	(−0.460)	(−2.147)	(0.402)	(−2.738)
Lev	−0.258 ***	0.146 ***	−0.405 ***	0.010	−0.238 ***	0.013
	(−7.953)	(5.409)	(−7.954)	(0.240)	(−5.345)	(0.333)
Boardsz	0.330 ***	0.391 ***	0.262 ***	0.378 ***	0.418 ***	0.219 ***
	(9.686)	(14.583)	(5.470)	(10.208)	(8.347)	(6.023)
Indpt	0.166	0.530 ***	0.195	0.637 ***	0.465 ***	0.325 ***
	(1.468)	(6.173)	(1.210)	(5.193)	(2.852)	(2.761)
Dual	0.025 *	−0.064 ***	0.073 **	−0.032	0.030 *	−0.013
	(1.771)	(−6.288)	(2.187)	(−1.382)	(1.870)	(−1.158)
Top1	−0.001 ***	0.004 ***	−0.002 ***	0.007 ***	0.000	0.000
	(−3.155)	(14.352)	(−3.076)	(16.336)	(0.563)	(1.153)
Industry/Year/Province	Y	Y	Y	Y	Y	Y
_cons	10.135 ***	9.858 ***	10.440 ***	9.926 ***	9.734 ***	10.624 ***
	(89.709)	(111.553)	(67.150)	(77.231)	(55.812)	(85.598)
N	10,547	10,547	4335	4335	6210	6210
Adj-R2	0.308	0.308	0.365	0.363	0.304	0.308
F	68.54	81.66	41.05	41.12	40.75	47.64

Notes: t statistics in parentheses, * p < 0.1, ** p < 0.05, *** p < 0.01. Robust standard error is used to overcome the possible heteroscedasticity of the model.

. The conclusion shows that after overcoming the deviation of sample self-selection, political connection still significantly affects the salary distribution and salary gap of enterprises. Specifically, in the whole sample, SOEs and non-SOEs, the coefficients of political connection and pay gap are negative and significant, which supports H1 and H2 (in Table 8). Further, in Table 9, for the non-SOEs group, the coefficients of PC in the regressions of Exepay and Ordpay are statistically significant and thus support H1.1 and H1.2. However, H2.1 and H2.2 are still less supportive due to the statistical insignificance. In summary, our estimation results are acceptable after using PSM method.

6.2. Other Robust Test

In order to verify the reliability of conclusions, we further provide the following robustness test:

First, referring to the studies of Yu and Pan [32] and Tian and Zhang [27], the chairman or CEO’s political connection is used as a substitute variable for the CEO’s political connection, and the results of repeating the above tests are consistent with the previous test.

Second, among Chinese companies, the chairman has an absolute influence. The chairman of some companies is responsible for the company’s specific operation and management “in person”, becoming the company’s “leader” and acting as the actual decision-maker [3,27]. Therefore, in the regression test of political connections to executive compensation, we re-regressed the average salary of the top three directors with the highest salary and the average salary of the top three directors, supervisors, and senior executives with the highest salary on behalf of the company’s executive compensation. The regression results remain stable.

Third, drawing on the studies of Liu and Sun [42] and Tang and Sun [9], the average of “total compensation of the top three directors and supervisors with highest remuneration” and the average of “total salary of the top three directors with highest compensation” are used as alternative indicators of executive compensation; and the total amount of “compensation paid to and for employees” and net compensation after removing social basic security expenses such as pension insurance are used to calculated the average salary of ordinary employees, respectively. From this, the relative and absolute salary gaps between executives and ordinary employees are obtained and subsequent regression analysis is performed. The results are similar.

Fourth, in order to prevent the results of this study from being affected by regional and annual factors, all regression analyses related to salary and salary gap control the fixed effects of province and year, and the results are similar.

Fifth, in order to eliminate the effects of extreme values, this article performs winsorize processing on the 1% and 99% percentiles of continuous variables and re-examines them, and the results remain stable.

7. Conclusions and Limitation

The economic consequences of political connections and the determinants of income distribution gaps within enterprises are two current research topics that have received much attention in academia, but there are few studies that combine the two topics. This study uses empirical data from 2010 to 2014 of China’s non-financial listed companies to empirically test the impact of political connections on the corporate executive–employee salary gap. The study shows that, overall, the political connections of senior executives in Chinese listed companies increased the pay gap between executives and employees within the enterprise. It was further found that the political connection of non-SOEs significantly increased the salary of executives, but also significantly reduced the salary of employees, thereby increasing the salary gap of executives and employees within the enterprise. This phenomenon does not exist in SOEs. It is shown that in non-SOEs, profits over-paid for executives with political connections will partially erode employee compensation, resulting in a reduction in their average compensation.

This paper organically links the two areas of research: the economic consequences of political connections and the determinants of income distribution gaps. At the same time, the research also has strengthened practical significance. On the one hand, the research explains that political factors are one of the determinants of the salary gap between senior executives and average employees among listed companies in China. That is, in SOEs, executives improve performance through evaluation and pursuing political promotion, while in non-SOEs, executives seek political relationship rent. In enterprises with different ownerships, these two types of behaviors have affected the internal income distribution mechanism of the enterprise, which has led to the widening of the salary gap between executives and employees. The empirical evidence shows that, in order to achieve the goal of “reducing the income distribution gap” proposed by the 18th National Congress of the Communist Party of China in deepening the reform of the income distribution system, it is necessary to pay attention to the problems affecting the income distribution at the enterprise level caused by political and economic factors. Fundamentally, the root cause of non-market factors such as political connections affecting income distribution is that China’s legal system is not yet sound, and corporate governance and market mechanisms are imperfect. Only the legal system is further improved to regulate market operation mechanisms and minimize political forces in economic life. Intervention can create a good business and political-business system environment for the enterprise. Only by improving the internal governance of the company is it possible to improve the internal income distribution pattern of the enterprise and promote the income distribution of the entire society to a healthy development track.

Although this research provides valuable insights, it has some limitations, which should help the further research. First, in the selection of research objects, we selected the samples of listed companies from 2010 to 2014. This is because China has implemented salary control measures since 2014, which may affect the research conclusion. Secondly, in the mechanism of the influence of political connections on the salary distribution of enterprises, considering China’s unique political system, some politically connected executives may have served as deputies to the National People’s Congress, while others may have served in government departments. The political rights and influence of these two departments are different, so it is valuable to consider the influence of different types of political connections on the salary gap of enterprises. In further research, we will try our best to expand the research scope, research the influence of salary control on the relationship between political connection and enterprise salary distribution, and take different types of political connection into account to improve the above problems.