INTRODUCTION
Pay gaps between top executives and subordinate employees have grown steadily over recent decades (Connelly, Tihanyi, Crook, & Gangloff, Reference Connelly, Tihanyi, Crook and Gangloff2014) not only in the United States and other developed economies (Sabadish & Mishel, Reference Sabadish and Mishel2012; Seager & Finch, Reference Seager and Finch2009) but also in emerging economies such as China (Firth, Leung, Rui, & Na, Reference Firth, Leung, Rui and Na2015). Widening pay gaps within firms usually mirror income disparity within society as a whole (Piketty, Reference Piketty2014). Consequently, media attention and political scrutiny are increasingly focusing on the effects of growing pay gaps. In the United States, the Dodd-Frank Wall Street Reform and Consumer Protection Act was established to require public firms to disclose pay gaps between CEOs and employees. In China, compensation policies in state-owned enterprises (SOEs) are strictly supervised and executive-employee pay gaps are compressed in alignment with governmental preferences for social equality (Firth et al., Reference Firth, Leung, Rui and Na2015). Lacking, however, is a clear understanding of the performance implications. Social comparison theories (e.g., relative deprivation theory and equity theory) suggest that wide pay gaps damage morale and cause employees to feel relatively underpaid and deprived (Adams, Reference Adams1963; Cowherd & Levine, Reference Cowherd and Levine1992; Crosby, Reference Crosby1984; Levine, Reference Levine1991; Martin, Reference Martin1981). Thus, social comparison theories suggest that firms should reduce pay gaps to create harmonious and cohesive environments. In contrast, tournament theory suggests that a competitive environment is formed as pay gaps grow between hierarchical levels: low-level employees strive for promotion, winners stay and losers leave, and firm performance increases (Becker & Huselid, Reference Becker and Huselid1992; Connelly et al., Reference Connelly, Tihanyi, Crook and Gangloff2014; Lazear & Rosen, Reference Lazear and Rosen1981).
China's unique cultural and economic environment adds new considerations to the theoretical debate. For example, traditional Confucianism, which deeply influences China, emphasizes equal wealth distribution and social harmony (Pan, Rowney, & Peterson, Reference Pan, Rowney and Peterson2012). A leading Confucian teaching says, ‘A nation or a family does not worry that it has little but that little is unevenly apportioned’ (Watson, Reference Watson2007: 115). Daoism (Taoism), another major dimension of Chinese culture, stresses that harmony is the ultimate way that nature functions and should be followed by human beings in their relationships with each other (Chen, Ünal, Leung, & Xin, Reference Chen, Ünal, Leung and Xin2016; Pan et al., Reference Pan, Rowney and Peterson2012). Similarly, Zhong Yong, another indigenous Chinese philosophy that emphasizes weighing and balancing, also seeks harmony of social relations (Pan & Sun, Reference Pan and Sun2018). However, increasing disparity within organizations inevitably violates the harmony principle embedded in the aforementioned philosophies. China's more recent communist ideologies have argued that the rich accumulated wealth by exploiting the poor, and thus wealth disparity should be eliminated (Wang & Qian, Reference Wang and Qian2011). As a result, in the Chinese context, wide pay gaps between top executives and employees may undermine the harmony principle of existing cultural values and ideologies, thereby lowering firm performance.
In China's transitional economic environment, however, the labor market comprises a large pool of homogenous unskilled labor with low marginal product of employees (World Bank, 2009). Managerial talent is more relevant for improving firm performance, but demand for qualified managers outstrips supply. Indeed, one of the biggest challenges facing China is the inadequate supply of managerial talent (Giannetti, Liao, & Yu, Reference Giannetti, Liao and Yu2015). This typical characteristic of the Chinese labor market suggests that attracting and motivating managerial talent are key factors in improving firm performance. By increasing pay gaps, firms will not only attract and retain talented executives but also provide incentives for outstanding employees who have the potential to be part of the management team in the future. As a result, firm performance is improved.
In this study, we take a contingent perspective to argue that firms that have predominantly egalitarian cultures will prefer smaller pay gaps and equal allocations; those that have competitive cultures will prefer larger pay gaps and unequal allocations. When firms shift from low pay gap environments to high pay gap environments, some employees initially feel relatively deprived as egalitarian norms break and internal harmony is disrupted. However, more productive employees support the change because they hope to benefit. As pay gaps continue to increase, so do motivating effects, gradually creating a competitive environment that replaces the original egalitarian norms. The tournament effect grows and finally overcomes the relative deprivation effect. Thus, we predict that pay gaps will have a U-shaped curvilinear relationship with firm performance.
Our fixed-effects regression results, based on panel data of 6,616 firm-year observations for 2,237 Chinese-listed firms from 2008 to 2012, reveal that pay gaps indeed have an attenuated U-shape relationship with firm performance. Most observations are on the positive slope of the curve, implying that only at the very low end of absolute levels of pay gap is there a negative effect of increasing pay gaps; fairly quickly, the impact of pay gaps becomes positive. We further investigate how ownership structure affects the weak U-shaped relationship between pay gap and firm performance, considering that SOEs and non-SOEs fundamentally differ in objectives and organizational culture. Our results reveal that the attenuated U-curve steepens in non-SOEs but further flattens and becomes approximately monotonic in SOEs. In short, different ownership structures cause the executive-employee pay gap to have a varying curvilinear effect on firm performance.
Our study makes several contributions to the literature. First, we integrate relative deprivation theory and tournament theory to argue that the relationship between pay gap and firm performance is not simply linear but is contingent on the pay gap level. Our findings explain why the literature may have produced mixed findings (Becker & Huselid, Reference Becker and Huselid1992; Connelly, Haynes, Tihanyi, Gamache, & Devers, Reference Connelly, Haynes, Tihanyi, Gamache and Devers2016; Cowherd & Levine, Reference Cowherd and Levine1992; Ding, Akhtar, & Ge, Reference Ding, Akhtar and Ge2009). Second, most of the literature has focused on developed economies (e.g., Bloom, Reference Bloom1999; Bloom & Michel, Reference Bloom and Michel2002; Cowherd & Levine, Reference Cowherd and Levine1992; Shaw, Gupta, & Delery, Reference Shaw, Gupta and Delery2002; Siegel & Hambrick, Reference Siegel and Hambrick2005), but we extend the research to China, a major transitional economy, which has a special cultural and economic environment suitable for testing our ideas and enhancing the global relevance of this research field. Finally, we reveal that ownership structure is an important contingent factor affecting the curvilinear effect of pay gap.
THEORETICAL BACKGROUND AND HYPOTHESES
Background Theory
Many studies have analyzed outcomes associated with pay variations (dispersion/ disparity/range/gap) within organizations (Adams, Reference Adams1963; Brown, Sturman, & Simmering, Reference Brown, Sturman and Simmering2003; Cowherd & Levine, Reference Cowherd and Levine1992; Downes & Choi, Reference Downes and Choi2014; Ridge, Aime, & White, Reference Ridge, Aime and White2015; Shaw et al., Reference Shaw, Gupta and Delery2002; Trevor, Reilly, & Gerhart, Reference Trevor, Reilly and Gerhart2012; Wageman, Reference Wageman1995; Wageman & Baker, Reference Wageman and Baker1997). One stream of literature focuses on vertical pay variations, and two major viewpoints have emerged (Becker & Huselid, Reference Becker and Huselid1992; Cowherd & Levine, Reference Cowherd and Levine1992; Lazear & Rosen, Reference Lazear and Rosen1981).
The first view is based on relative deprivation theory, proposing that when employees find that they receive less compensation in comparison with referent others in the same organization, the result is damaged harmony, reduced cooperation, and negative employee attitudes and behaviors (Crosby, Reference Crosby1984; Martin, Reference Martin1981). That is, employees make upward comparisons in assessing the fairness and justice of the compensation system (Cowherd & Levine, Reference Cowherd and Levine1992). If they perceive unfair treatment, they will feel relatively deprived and react with shirking behaviors, excess turnover, strikes, or even vandalism (Crosby, Reference Crosby1984; Martin, Reference Martin1981). Thus, relative deprivation theory predicts that when pay variations increase across organizational hierarchies, employees are demotivated and show negative behavioral reactions.
The second view, based on tournament theory, argues that vertical pay variations provide incentives. Tournament theory suggests that wages based on rank orders are an efficient compensation scheme to motivate individuals (Lazear & Rosen, Reference Lazear and Rosen1981). In a rank-order tournament, the best relative performer is promoted to a higher level, while the others are passed over. The incentive effect of such a tournament is related to tournament spread (prize differential) (Becker & Huselid, Reference Becker and Huselid1992). The greater the pay variations across the hierarchy, the stronger the motivating effects (Connelly et al., Reference Connelly, Tihanyi, Crook and Gangloff2014). In a firm, top management team members serve as salient referents to employees (Wade, O'Reilly, & Pollock, Reference Wade, O'Reilly and Pollock2006), and promotion to higher levels carries with it higher pay and offers employees an incentive to increase their efforts, which in turn improves firm performance (Connelly et al., Reference Connelly, Haynes, Tihanyi, Gamache and Devers2016).
Empirical evidence supports the essentially opposite associations between vertical pay variation and organizational outcomes (Becker & Huselid, Reference Becker and Huselid1992; Connelly et al., Reference Connelly, Haynes, Tihanyi, Gamache and Devers2016; Cowherd & Levine, Reference Cowherd and Levine1992; Ding et al., 2009). In this study, we propose a more nuanced perspective to reconcile this conflict.
Hypotheses Development
Corporate compensation policies reflect cross-level differences in work content and contributions, organizational allocation preferences, and cultural norms. Companies that prefer egalitarian allocation rules, egalitarian organizational cultures, and the creation and maintenance of interpersonal harmony will design low pay gap compensation policies, as relative deprivation theory suggests (Bloom, Reference Bloom1999; Cowherd & Levine, Reference Cowherd and Levine1992; Williams, McDaniel, & Nguyen, Reference Williams, McDaniel and Nguyen2006). In contrast, companies that emphasize a competitive organizational culture to motivate and retain talent will prefer a high pay gap compensation policy, as tournament theory suggests (Becker & Huselid, Reference Becker and Huselid1992; Connelly et al., Reference Connelly, Tihanyi, Crook and Gangloff2014). When pay gaps increase from low to high levels, we expect to observe a curvilinear relationship with firm performance, especially in China's unique cultural and economic environment.
The Chinese are deeply influenced by the value of harmony, which is a salient concept in both traditional Chinese philosophies and current Chinese society (Li, Leung, Chen, & Luo, Reference Li, Leung, Chen and Luo2012). Confucian philosophy calls for equal wealth distribution to maintain harmony (Barkema, Chen, George, Luo, & Tsui, Reference Barkema, Chen, George, Luo and Tsui2015; Pan et al., Reference Pan, Rowney and Peterson2012). Daoist (Taoist) philosophy stresses that harmony is the ultimate way that nature functions and should be followed by human beings in their relationships with each other (Chen et al., Reference Chen, Ünal, Leung and Xin2016; Pan et al., Reference Pan, Rowney and Peterson2012). Zhong Yong philosophy also emphasizes that harmony is an ideal state and extremity should be avoided (Pan & Sun, Reference Pan and Sun2018). Significant income disparity between executives and employees obviously violates the harmony value of these philosophies. More recent communist ideology and its collectivism culture, which also influence contemporary Chinese employees (Chen, Reference Chen1995; Earley, Reference Earley1989, Reference Earley1994; Liu, Friedman, & Chi, Reference Liu, Friedman and Chi2005; Xiao & Tsui, Reference Xiao and Tsui2007), emphasize egalitarianism and a relatively equal distribution of wealth (Wang & Qian, Reference Wang and Qian2011). When the pay gap increases, it inevitably increases internal inequality, violates egalitarianism, and undermines harmony within organizations. Most employees who are habituated to the original harmonious environment and egalitarian norms are likely to resist an increase in pay gap that introduces competition. The increasing pay gap would motivate only the few productive employees who can expect to prevail (Grund & Westergaard-Nielsen, Reference Grund and Westergaard-Nielsen2008; Mitra, Gupta, & Jenkins Jr., Reference Mitra, Gupta and Jenkins1997). If only a few are motivated while most are demotivated, firm performance will experience a negative net effect.
However, high vertical pay variation can have varying effects on firm performance. First, firms that use high pay gaps will no longer prioritize internal harmony. Instead, they will focus on maintaining, motivating, and retaining the most valuable employees. As organizational goals change, employees focus more on reward distributions, so that differential rules become the new legitimate norm (Chen, Reference Chen1995). China has a unique history of economic, political, educational, and institutional reform. Since 1978, Chinese firms have turned from socio-political orientations to economic orientations, and Chinese employees have begun to support differential compensation policies (Chen, Reference Chen1995).
Second, two mechanisms are at work that lead to gradually emerging positive effects. First, productive employees who make greater marginal contributions to firm value are the ones most likely to get raises or promotions and will be most motivated by the change. China's labor market comprises a great pool of homogenously unskilled laborers (World Bank, 2009), especially in the manufacturing industry dominating China's economy, but high-quality labor is scarce. As a result, Chinese firms must incentivize their most productive employees and attract and retain skilled labor through greater rewards. Second, poor performers are more likely to disdain wide pay gaps because they have comparatively less chance of winning higher pay or promotion (Shaw, Reference Shaw2014). However, competition will gradually filter out poor performers and retain good ones (Shaw, Reference Shaw2014), leaving a workforce of productive competitors and eliminating the unproductive employees who regard wage differentials as relative deprivation. As the relative deprivation effect declines, the tournament effect rises to exhibit positive effects.
Figure 1 depicts the integration of the two effects. In summary, when pay gaps increase from low to high levels, egalitarian norms break and competitive norms form. At the first stage, internal equality and harmony are disrupted and the relative deprivation effect dominates. However, as more productive employees are attracted and retained, unproductive employees are filtered out. Egalitarian norms weaken in favor of competitive norms and the tournament effect gradually dominates. Therefore, we outline the following testable hypothesis:
Hypothesis 1: Executive-employee pay gaps will have a curvilinear U-shaped relationship with firm performance.
Figure 1. The integration of tournament effect and relative deprivation effect on firm performance
State ownership is a distinguishing feature of China's economy. Although China's economic reform has led to the rise of non-SOEs over the past decades, SOEs still account for a significant proportion of the economy. SOEs exhibit fundamental differences in organizational objectives (Peng, Bruton, Stan, & Huang, Reference Peng, Bruton, Stan and Huang2016) and culture (Tsui, Wang, & Xin, Reference Tsui, Wang and Xin2006), which may moderate the association between executive-employee pay gaps and firm performance.
First, SOEs differ from non-SOEs by having multiple objectives that may attenuate the effect of compensation on firm performance. That is, in SOEs the government is the controlling shareholder and has multiple social objectives including increasing employment, maintaining social stability, and implementing industrial policies. However, social objectives largely conflict with economic goals, so SOEs often choose social objectives over economic objectives. In contrast, non-SOEs prioritize the maximization of economic interests and are thus more focused on financial goals. Particularly, SOEs shoulder greater responsibility for achieving employment stability than non-SOEs, thus employees in SOEs enjoy higher job security (Gong & Chang, Reference Gong and Chang2008). This benefit of secure employment from working in SOEs may attract employees who prefer lower employment risk and thus are more likely to accept lower pay. As a result, pay gap exerts weaker influences on employees working in SOEs than those working in non-SOEs. As a result, non-SOEs and SOEs may have similar pay structures, but non-SOEs are more likely to show the pay gap-firm performance effect.
Second, SOEs have very different organizational cultures and compensation policies. For many decades, China's SOEs were under state-controlled communist rule that emphasizes collective and equal allocation of wealth. They believed that performance-based reward systems promote self-interest ideologies and glorify individualism. Instead, they based pay on predictable, stable personal attributes, especially seniority, to avoid triggering competition (Wu, Chen, & Leung, Reference Wu, Chen and Leung2011). Thus, their compensation systems contained few performance management elements and competition mechanisms. Although social and economic reforms are gradually shifting SOE organizational cultures away from egalitarianism and collectivism, long-standing practices are slow to change.
In contrast, non-SOEs generally emerged after China's economic reform and thus lack the egalitarianism and collectivism traditions; instead, their organizational culture is more market-oriented (Tsui et al., Reference Tsui, Wang and Xin2006). In addition, they receive far less governmental support and protection. Instead, the market determines their survival. Consequently, their compensation systems will usually contain more competition mechanisms and be more strongly related with firm performance.
We suggest that multiple objectives and less competitive compensation policies may attenuate the effect of compensation schemes on firm performance. That is, when SOEs and non-SOEs have the same level of executive-employee pay gap, SOEs will show a weaker influence on firm performance, suggesting the second testable hypothesis:
Hypothesis 2: Pay gaps in SOEs will have a weaker curvilinear U-shaped relationship with firm performance than pay gaps in non-SOEs.
METHODS
Sample and Data
Our original sample included all firms listed on China's Shanghai and Shenzhen Stock Exchanges from 2008 to 2012. To improve sample comparability, we followed general practices to exclude firm-year observations where a focal firm belonged to financial industries, issued debt exceeding asset value, was listed on multiple stock markets, or had variables with missing values. We also excluded firms that had a negative pay gap; that is, managers earned less than average employees. The final sample included 2,237 firms and 6,616 firm-year observations. We collected all accounting data and corporate governance information from the China Stock Market & Accounting Research database (CSMAR; www.gtarsc.com) provided by Shenzhen GTA Information Technology Company, a major provider of data pertaining to China.
Measures
Dependent variable
To capture economic outcomes of pay gaps, the dependent variable is firm performance. We take Tobin's Q to measure firm performance, calculated as the ratio of the market value of common stock plus the book value of total debt over the book value of total assets. Tobin's Q represents how well a company is doing on the stock market. This measure goes beyond looking at simple assets to ascertain how positively stock owners feel about the future of the company. We use industry-adjusted Tobin's Q by subtracting its yearly industry median to control for industry influence.[Footnote 1] For robustness checks, we also use an accounting-based measurement–return on assets (ROA)–as a proxy for firm performance.
Independent variable
Our independent variable is the pay gap between top executives and subordinate employees, denoted by PAYGAP. Following Kale, Reis, and Venkateswaran (Reference Kale, Reis and Venkateswaran2009), we calculate PAYGAP as the natural log of the difference between the average of the top three executives’ pay and the average of employees’ pay, adjusted for social security payments (Li & Hu, Reference Li and Hu2012). Our pay gap measure is based on the top three executives because executives with the most significant compensation are most likely to attract employee attention, which in turn evokes the strongest positive or negative reactions among employees.[Footnote 2] We consider only cash compensation, given that China prohibited executive stock options until 2006 and that Chinese-listed firms have a very small ratio of compensation in the form of stocks or stock options (Bryson, Forth, & Zhou, Reference Bryson, Forth and Zhou2014). We use industry-adjusted PAYGAP by subtracting the yearly industry median to control for industry differences.
Predicting a U-shaped relationship between pay gap and firm performance, we include the square term of PAYGAP, denoted by PAYGAP_sqr, in regression models. Alternatively, following He and Fang (Reference He and Fang2016), we also calculate PAYGAP as the difference between the logarithm of the average of the top three executives’ pay and the logarithm of the average employee's pay in robustness checks. This method is able to capture the relative level of pay gap. We follow both methods to fully capture pay differences.
Moderating variable
In examining Hypothesis 2, we include ownership structure as a moderating variable. We generate an indicator variable, denoted as STATE. STATE equals 1 if the ultimate controlling shareholder is the government or a state agency and 0 otherwise.
Control variables
We first control for a set of financial variables that are likely to affect firm performance. As firm performance may be correlated with firm size, we control for firm size (SIZE), measured as the natural log of the book value of total assets. High leverage may have a negative effect on firm performance because of higher financial constraint. Thus, we control for firm leverage (LEVERAGE), measured as the ratio of total debt to total assets. As tangible assets incur lower agency cost (Titman & Wessels, Reference Titman and Wessels1988) and intangible assets are important in creating competitive advantages to enhance firm performance (Carmeli & Tishler, Reference Carmeli and Tishler2004), we control for asset tangibility (TANGIBLE), which is measured as the ratio of tangible assets to total assets. Firms with high growth rates are likely to exhibit high performance (Brush, Bromiley, & Hendrickx, Reference Brush, Bromiley and Hendrickx2000). Thus, we control for sales growth (GROWTH), measured as the rate of annual change in sales.
We also control for a set of governance variables. As employees’ compensation levels affect organization performance through work efforts and turnover rate (Campbell, Ganco, Franco, & Agarwal, Reference Campbell, Ganco, Franco and Agarwal2012), and high compensation can make employees more tolerant to the pay differences (Brown et al., Reference Brown, Sturman and Simmering2003), we control for average employee compensation (EMCOMP), measured as the natural logarithm of average employee compensation. As large shareholders can alleviate agency problems that are detrimental to firm performance (Shleifer & Vishny, Reference Shleifer and Vishny1997), we control for TOP1, measured as the proportion of equity shares held by the largest shareholder. Managerial ownership can increase firm performance by aligning managers’ interests with those of shareholders (Jensen & Meckling, Reference Jensen and Meckling1976). We control for such managerial incentive by including managerial ownership (MSHARE), measured as the percentage of equity shares held by the management team. Board characteristics are likely to affect firm performance through corporate governance (Fama & Jensen Reference Fama and Jensen1983; Finkelstein, Reference Finkelstein1992; Goodstein, Gautam, & Boeker, Reference Goodstein, Gautam and Boeker1994). Thus, we control for board size (BOARDSIZE), measured as the number of directors on the board, independent directors (INDBOARD), measured as the ratio of independent directors on the board, and Chairman-CEO duality (DUALITY), a dummy variable that equals a value of 1 if a firm's CEO is also the (vice) chairman of the board and 0 otherwise. We also control for firms’ special treatment (ST) status because firms with ST status are in financial distress and are likely to have lower firm performance because of financial constraint. ST is a dummy variable that equals 1 if a focal firm is under special treatment in the current year and 0 otherwise.
Finally, we also include one-year-lagged term of Tobin's Q to control for persistence in firm performance and year indicators to control for time fixed effects. To minimize outlier effects, we winsorize all continuous variables at the 1st and 99th percentiles.
The Model
Our sample is a classical panel data, so we conduct fixed-effects regressions to test the hypotheses. Fixed-effects regressions counter endogeneity problems resulting from omitted time-invariant variables. We used lagged values of independent and control variables to alleviate endogeneity from reverse causality. According to the results from the F-test for choosing pooled Ordinary Least Squares model vs. fixed-effects regression model and Hausman's test for choosing fixed-effects vs. random-effects model, the fixed-effects regression model was the best fit to analyze our panel data (Wooldridge, Reference Wooldridge2002). The fixed effects are controlled by including firm dummies in the regression models. We center the square term of PAYGAP and interaction variables to avoid multicollinearity. As the average variance inflation factor of each regression model is less than the cut-off point of 10, multicollinearity seems insignificant.
RESULTS
Descriptive Statistics
Panel A of Table 1 reports descriptive statistics for the main variables. The distribution of raw Tobin's Q shows that Tobin's Q is greater than 1 at the 25th percentile and close to 1 at the minimum, suggesting that most of our sample firms are valued by the markets to be greater than their asset values. Specifically, the mean value of raw Tobin's Q is 2.417, suggesting that the sample firms have a market value of 2.417 times their book value. The mean and median value of raw PAYGAP is 12.520 and 12.570, respectively, indicating that the mean and median pay gaps between top executives and employees reach about 273,758 (e12.518) RMB and 287,793 (e12.570) RMB, respectively. Considering that China's national urban disposable income was 28,844 RMB in 2014,[Footnote 3] the average pay gap between executives and employees is up to nine times the national urban disposable income, which reflects the income disparity. Of the total sample, 43.8% are SOEs with the government or state agencies as ultimate controlling shareholders. On average, the largest shareholder holds 36.4% proportion of equity stakes, indicating that China still has the one dominant controlling shareholder phenomenon (yigududa). In particular, the mean, median, and standard deviation values of managerial ownership (MSHARE) are 0.087, 0, and 0.183, respectively, suggesting that management teams on average hold only 8.7% of total ownership; more than half of the sample firms are not under manager ownership. In 20.8% of firms, one person serves as both CEO and (vice) chairman of the board of directors; 6.6% of these firms are under ST status. The boards have an average of 9 (9.040) directors; 36.7% are independent directors.
Table 1. Descriptive statistics and correlation matrix (N = 6,616)
Notes: *, **, and *** denote significance at the 10%, 5%, and 1% level (two sided), respectively.
Panel B of Table 1 presents the pair-wise correlation matrix of main variables. A significant and negative correlation occurs between PAYGAP and Tobin's Q (r = −0.096, p < 0.01), while a significant and positive correlation occurs between PAYGAP_sqr and Tobin's Q (r = 0.059, p <0.01), consistent with predictions in Hypothesis 1. The correlation coefficient between Tobin's Q and lagged Tobin's Q is up to 0.633 and significant at the 1% level, suggesting a trend in firm performance. STATE displays a significant and negative correlation with Tobin's Q (r = −0.128, p < 0.01), implying that SOEs perform more poorly than private firms (Boycko, Shleifer, & Vishny, Reference Boycko, Shleifer and Vishny1996; Kato & Long, Reference Kato and Long2011). In addition, correlation coefficients among other variables are no larger than 0.5, indicating insignificant multicollinearity.
Regression Results
Table 2 displays the regression results of the fixed-effects model. We conduct hierarchical multiple regression to test our hypotheses. Specifically, Model 1 is a baseline model with only control variables. Model 2 adds the single term, i.e., PAYGAP. Model 3 further adds the square term of PAYGAP (i.e., PAYGAP_sqr) to examine Hypothesis 1 predicting a U-shaped relationship between pay gap and firm performance. Model 4 adds the moderating variable, STATE. Model 5 adds the interaction between STATE and PAYGAP and the interaction between STATE and PAYGAP_sqr to examine the moderating effect of ownership structure on the curvilinear effect. As Table 2 shows, the adjusted R 2 for all regression models is above 0.7, which indicates that more than 70% variance of firm performance can be explained by our regression models.
Table 2. Regression results for testing hypotheses (N = 6,616)
Notes: Exact p-values in brackets. The square term variable and the interaction variables are mean centered. Year indicators are included but not reported here. *, **, and *** denote significance at the 10%, 5%, and 1% level (two sided), respectively.
Hypothesis 1 predicts a U-shaped association between pay gap and firm performance. As Model 2 shows, the coefficient for PAYGAP is positive and significant (Model 2: 0.062 with t = 2.157). In Model 3, the single term of PAYGAP still acquires a positive and significant coefficient (Model 3: 0.102 with t = 3.360), while the coefficient for PAYGAP squared is also positive and significant (Model 3: 0.063 with t = 3.829).[Footnote 4] Moreover, the adjusted R 2 is larger in Model 3 (0.716) than that in Model 2 (0.715); the difference is significant at the 1% level (F value = 22.223), suggesting that the square term of PAYGAP has significant incremental explanatory power for the variation of firm performance. The results indicate a curvilinear association between PAYGAP and Tobin's Q.
Hypothesis 2 pertains to the moderating effect of ownership structure. As Model 5 displays, the coefficient is not significant for the interaction between PAYGAP and STATE (PAYGAP * STATE) (Model 5: −0.004 with t = −0.091 and p = 0.927), but the coefficient is significant for the interaction between PAYGAP_sqr and STATE (PAYGAP_sqr * STATE) (Model 5: -0.057 with t = −2.050 and p = 0.040). The coefficient sign of PAYGAP_sqr * STATE is opposite to that of PAYGAP_sqr, suggesting that the U-shaped association between PAYGAP and Tobin's Q is less pronounced in SOEs than in non-SOEs. Therefore, Hypothesis 2 is also supported.
As to the control variables, Lagged Tobin's Q acquires significant and negative coefficients, suggesting that there may be a reversal trend of Tobin's Q. The coefficients on SIZE are also significant and negative, indicating that small firms on average have higher performance. Both TOP1 and MSHARE acquire significant and positive coefficients, suggesting that stronger shareholder monitoring and greater convergence of managerial interests can improve firm performance. The coefficients on ST are significantly positive, which could be because investors give ST firms a premium because of their shell value (Liu, Robert, & Yu, Reference Liu, Robert and Yu2019). However, we fail to find other control variables having consistent and significant effects on firm performance.
To further explore whether pay gap has a clear U-shaped relationship with firm performance as hypothesized, we follow Aiken and West (Reference Aiken and West1991) to transform the mean-centered variables to their initial values, constrain all control variables to their mean values, and then plot the relationship between PAYGAP and Tobin's Q for the full sample, the SOE subsample, and the non-SOE subsample, respectively. As Figure 2 shows, an attenuated curvilinear relationship occurs between PAYGAP and Tobin's Q in the full sample, with most of the observations on the right side of the curve, suggesting that when the pay gap increases from low to high, a negative effect emerges only in the short term, while the positive effect gradually increases and dominates. Furthermore, the curvature increases in the non-SOE subsample but decreases and approximates monotonic in the SOE subsample, consistent with our second hypothesis that pay gap has a curvilinear effect mainly in non-SOEs. We also follow Aiken and West (Reference Aiken and West1991) to conduct simple slope analyses. In the non-SOE subsample, the simple slope is -0.097 (t = −1.993, p = 0.046), 0.102 (t = 2.202, p = 0.028), and 0.502 (t = 4.495, p < 0.01) for the values -1SD, 0, +1SD of pay gap;[Footnote 5] in the SOE subsample, the simple slope is 0.071 (t = 1.490, p = 0.136), 0.116 (t = 2.471, p = 0.014), and 0.161 (t = 2.324, p = 0.020) for the values -1SD, 0, +1SD of pay gap. The results indicate that the curvilinear effect is significant in non-SOEs but insignificant in SOEs, further supporting our second hypothesis.
Figure 2. The moderating effect of ownership structure on the relationship between pay gap and firm performance
We further examine the economic significance of pay gap. For SOEs, a 25% increase in pay gap (68,439 RMB) from the sample mean (273,758 RMB) would cause Tobin's Q to rise from 2.082 to 2.100, which is about 1.8% (2.100–2.082) of book value (1.442 × 108 RMB). For non-SOEs, the same amount of increase in pay gap (68,439 RMB) from the sample mean (273,758 RMB) would cause Tobin's Q to rise from 2.403 to 2.428, which is about 2.5% (2.428–2.403) of book value (2.004×108 RMB). That is, when pay gap increases by 25% from the sample mean, one RMB increase of pay gap on average would bring about 2,107 RMB (1.442 × 108 / 68,439) for SOEs and 2,928 RMB (2.004 × 108 / 68,439) for non-SOEs, indicating that pay gap has significantly different economic effects for SOEs and non-SOEs.
Finally, the inflection point (i.e., the point where the effect of PAYGAP turns from negative to positive) is 65,131 RMB for SOEs and 160,735 RMB for non-SOEs, which implies that increasing the executive-employee pay gap has a negative (positive) economic effect on firm performance if the pay gap is lower (higher) than 65,131 RMB for non-SOEs and 160,735 RMB for non-SOEs. In the full sample, the transformed value of PAYGAP at the inflection point is 122,723 RMB. That is, expected firm performance is higher at either much lower pay gap or much higher pay gap compared to this intermediate level of pay gap (122,723 RMB) in the Chinese context.
Alternative Empirical Approach
Organizations may select pay structures to fulfill internal strategies and human resource policies or to meet external needs for economic development (Milkovich, Newman, & Gerhart, Reference Milkovich, Newman and Gerhart.2011). Employees generally accept some inequality if it appears reasonable (Kepes, Delery, & Gupta, Reference Kepes, Delery and Gupta2009; Shaw et al., Reference Shaw, Gupta and Delery2002; Trevor et al., Reference Trevor, Reilly and Gerhart2012), but they do not when normally accepted factors fail to account for pay inequality. In our samples, employees may expect and accept a certain degree of pay gap. To corroborate our hypotheses, we next examine effects occurring when observable factors fail to explain excess pay gaps.
Following the method of Fredrickson, Davis-Blake, and Sanders (Reference Fredrickson, Davis-Blake and Sanders2010) and He and Fang (Reference He and Fang2016), we first estimate the expected pay gap by regressing pay gap on determinants including SIZE, LEV, GROWTH, TANGIBLE, TOP1, MSHARE, DUALITY, STATE, PROFIT (the ratio of operation profit to total assets), RISK (the standard deviation of daily stock returns in the year), R&D (the ratio of R&D expense to sales income), MAGE (the average age of top management team), MTENURE (the average tenure of top management team), MKT (regional market development index from Fan, Wang, & Zhu, Reference Fan, Wang and Zhu2011), and year indicators (Connelly et al., Reference Connelly, Haynes, Tihanyi, Gamache and Devers2016; Fredrickson et al., Reference Fredrickson, Davis-Blake and Sanders2010).[Footnote 6] Next, we measure expected pay gap, denoted as EXPECTED PAYGAP, as the predicted value of the regression. The excess pay gap, denoted as EXCESS PAYGAP, is measured as the difference between the actual and expected pay gap (Fredrickson et al., Reference Fredrickson, Davis-Blake and Sanders2010). We then control for the expected pay gap and use excess pay gap to re-examine whether pay gap still has a curvilinear effect and ownership structure still has a moderating effect. Table 3 shows the results.
Table 3. Regression results of alternative empirical approach (N = 6,593)
Notes: Exact p-values in brackets. The square term variable and the interaction variables are mean centered. *, **, and *** denote significance at the 10%, 5%, and 1% level (two sided), respectively.
As Table 3 shows, the coefficients on EXPECTED PAYGAP are positive and significant at the 1% level in all models, implying that expected pay gap has a positive effect on firm performance. The coefficient on EXCESS PAYGAP is nonsignificant in Model 2. However, in Model 3, when the square term of excess pay gap (EXCESS PAYGAP_sqr) is included, both the single term and square term become positively significant at the 5% level (p = 0.015 / 0.006), suggesting that excess pay gap has a nonlinear relationship with firm performance. Moreover, the coefficient on the interaction EXCESS PAYGAP_sqr * STATE is significantly negative at the 5% level (p = 0.014), suggesting that the curvilinear relationship is weaker in SOEs than in non-SOEs. Thus, regression results in Table 3 are similar to those in Table 2, suggesting that pay gap continues to have a curvilinear effect on firm performance when expected pay gap is controlled in the model.
Robustness Checks
We further conducted two robustness checks. First, rather than market-based performance (i.e., Tobin's Q), we measure firm performance with an accounting-based measure of ROA. The results are shown in Model 1 and Model 2 of Table 4. Second, in addition to absolute pay gap, we follow He and Fang (Reference He and Fang2016) to calculate relative pay gap, defined as the difference between the logarithm of the average of top three executives’ pay and the logarithm of the average employee's pay. The results are presented in Model 3 and Model 4 of Table 4. As Table 4 shows, coefficients on both PAYGAP and PAYGAP_sqr are significantly positive at the 1% level in all models, implying that pay gap continues to have a curvilinear relationship with firm performance under both robustness tests. Moreover, coefficients on the interaction term PAYGAP_sqr * STATE are significantly negative at the 10% level in both checks (Model 2: p = 0.073; Model 3: p = 0.069), suggesting that ownership structure continues to have a moderating effect under both checks. Overall, the regression results of both robustness tests further support our hypotheses.
Table 4. Regression results of robustness Checks (N = 6,616)
Notes: The variable PAYGAP in Model 3 and Model 4 is relative pay gap, which is calculated as the difference between the logarithm of the average of top three executive pays and the logarithm of the average employee pay. Exact p-values in brackets. The square term variable and the interaction variables are mean centered. *, **, and *** denote significance at the 10%, 5%, and 1% level (two sided), respectively.
DISCUSSION
In this study, we suggest why mixed findings are common in studies of how executive-employee pay gaps affect organizational performance. Taking a contingent perspective, we argue that pay gaps have a curvilinear association with firm performance. Our empirical results, based on a panel data set of 2,237 Chinese listed firms from 2008 to 2012, support our predictions. Moreover, the curvilinear relationship varies according to ownership structures. That is, in non-SOEs, the U-shaped curve steepens and becomes more significantly U-shaped; in SOEs the curve flattens and approximates monotonic with a positive slope, indicating that pay gap has a more significant U-shaped relationship with firm performance in non-SOEs. Thus, ownership structure is an important contingent factor in shaping the effects of pay gaps.
Contributions
Our study makes several contributions to the literature. First, although extant studies have examined the outcome of vertical pay variation within organizations, evidence based on either tournament theory or relative deprivation theory is mixed (Becker & Huselid, Reference Becker and Huselid1992; Connelly et al., Reference Connelly, Haynes, Tihanyi, Gamache and Devers2016; Cowherd & Levine, Reference Cowherd and Levine1992; Ding et al., 2009). Responding to the call for a more balanced and integrated perspective (Downes & Choi, Reference Downes and Choi2014), we integrate tournament theory and relative deprivation theory to investigate the comprehensive effect of pay gap. We argue that the relationship between executive-employee pay gap and firm performance is not simply linear but contingent on the pay gap level. Consistent with our prediction, we find that the relationship between pay gap and firm performance exhibits an attenuated U-shape, potentially explaining the previous mixed findings and supplementing Shaw et al. (Reference Shaw, Gupta and Delery2002) by showing that pay gap level is an important contingency.
Second, we extend the ongoing debate about the economic consequences of the pay gap in China, an underexplored but noteworthy context. Most research on pay inequality focuses on Western developed economies (e.g., Bloom, Reference Bloom1999; Bloom & Michel, Reference Bloom and Michel2002; Cowherd & Levine, Reference Cowherd and Levine1992; Shaw et al., Reference Shaw, Gupta and Delery2002; Siegel & Hambrick, Reference Siegel and Hambrick2005) and has paid relatively less attention to China, the world's largest emerging economy. However, China's unique cultural and economic environment is an intriguing context for examining the economic consequences of pay gaps. The Chinese are deeply influenced by traditional Confucianism and communist ideologies that emphasize equal wealth distribution and harmony. Other important dimensions of Chinese culture such as Daoism (Taoism) and Zhong Yong philosophies also share the principle of harmony. But as China transitions into the global economic environment, large pay differentials are needed to attract and retain talented employees and executives from China's underdeveloped labor market. As a result, compensation practices in China may be shaped by the unique cultural and market environment and thus differ from those in other contexts. Overall, our study globalizes pay gap research by revealing how specific cultural and economic environments may shape the effects of pay gap.
Third, we respond to calls for more in-depth work regarding distinctions between SOEs and non-SOEs (Bruton, Reference Bruton2015; Peng et al., Reference Peng, Bruton, Stan and Huang2016). In China and some other countries, SOEs play an important role in the economy and behave differently from non-SOEs in various aspects. For example, China's SOEs pursue multiple social objectives such as increasing employment, maintaining social stability, and implementing governmental industrial policies (Boycko et al., Reference Boycko, Shleifer and Vishny1996). However, social objectives largely conflict with profit maximization, so SOEs may choose social goals over economic objectives. Consequently, compensation would have a weaker association with firm performance. In addition, China's SOEs traditionally emphasized egalitarianism and stressed collective allocation of wealth (He & Fang, Reference He and Fang2016). Accordingly, SOE compensation systems contain fewer performance management elements and competition mechanisms in comparison with non-SOEs. Thus, although SOEs and non-SOEs may have similar compensation schemes, effects of these schemes are weaker in SOEs. By revealing ownership structure as an important contingency, we enrich the literature and provide relevant practical understandings for SOE managers and policymakers.
Limitations and Future Research Directions
Despite the above contributions, our study has several limitations that offer opportunities for future research. First, because of a lack of data, our pay gap measure does not include stock-based compensation, on-the-job consumption, and other implicit benefits, which may distinctly influence performance. Therefore, future studies will need access to more detailed compensation information to more comprehensively investigate the distinct effects of compensation dimensions.
Second, we consider only two important and relevant contingent factors: pay gap levels and ownership structure. Future investigations could expand to other significant contingent factors, such as subnational formal institutions that may shape the antecedents and consequences of pay inequality among top management teams (He & Fang, Reference He and Fang2016). Moreover, informal social norms and culture play important roles in emerging economies with underdeveloped formal systems other than China (Li & Hu, Reference Li and Hu2012). Although we considered those characteristics, we did not directly examine their effects. Thus, future research should explore how specific informal social institutions and cultural norms directly influence effects of pay gaps.
Finally, theory generalization may be another limitation. We develop and examine our theory in the context of China, where the cultural environment is deeply influenced by Confucian philosophies and socialist ideologies (He & Fang, Reference He and Fang2016). Although exploring the impact of cultural characteristics enriches the literature, the conclusions may be culturally specific, so caution is advised in generalizing our theory. Our theory may be applicable to countries that share China's cultural background, such as Korea, Japan, and some other East Asian countries. However, such conjecture has yet to be examined. Thus, future studies could test our theory in different cultural environments to explore whether the curvilinear relationship is culturally specific or universal.
Practical Implications
Our study may have two practical implications. First, we show that when pay gap increases from low to high levels, results are initially unfavorable but gradually promote firm performance. Therefore, ceteris paribus, firms might set pay gap levels meaningfully low or meaningfully high but avoid intermediate levels. Second, to motivate SOE managers more effectively, one possible way is to relax restrictions on SOE managers’ compensation, particularly for those without stock-based compensation. Another possible way is to partly privatize SOEs’ equity to make SOEs more market-oriented, because pay gap yields greater effect in non-SOEs. However, it should be noted that the actual results of these suggestions may vary greatly among firms with different characteristics; thus, firms should be cautious about their real effects.
CONCLUSION
Although pay gaps between top executives and subordinate employees are common worldwide and long regarded as a main incentive mechanism for all employees, debates are ongoing regarding the outcomes. Our results, based on the context of China, suggest that executive-employee pay gaps have an attenuated U-shaped relationship with firm performance. Moreover, the U-shaped relationship is weaker in SOEs than in non-SOEs. Thus, pay gaps have nonlinear rather than simply linear effects, while pay gap levels and ownership structures are important contingencies. Overall, drawing on a contingent perspective, we integrate tournament theory and relative deprivation theory to reconcile extant mixed findings and deepen our understanding of the performance implications of pay gaps.
