Introduction

There is a consensus that good governance leads to higher firm performance. This is the reason why corporate governance has long piqued the interest of academics, authorities, decision-makers, and investors. Particularly, researchers have shown interest in investigating how a corporate board’s structure, functions, and board of directors’ attributes affect the team and firm performance. For instance, scholars have looked into how structural and functional factors like board size, board independence, CEO duality, ownership structure, board committees, and board meetings, among others, affect the effectiveness of the board and, ultimately, the performance of the firm (i.e., Bhagat and Bolton 2008; Paniagua, Rivelles, & Sapena 2018). Likewise, studies have looked into how directors’ demographic or personal characteristics, such as gender, age, education level, nationality, and ethnicity, among others, impact firm performance (i.e., Livnat et al. 2021; Mohsni et al. 2021). Authorities are also making board-related regulations from time to time to safeguard stakeholders and boost performance. This demonstrates how crucial the corporate board is to the success of the firm. Owing to this significance, corporations work to build boards that perform better monitoring duties, make wise judgments, and ultimately improve firm performance. Nonetheless, it is clear that performance still lags even when firms make every effort to constitute an efficient board. What causes this? What other characteristics of the board affect the firm’s performance?

In this study, we attempt to look for the answers to these concerns in the board’s human component. Corporate boards are made up of humans, and as we know, human behavior and attitude change over time and in response to certain circumstances. For instance, the career stage theory of Super (1957) states that human behavior and attitude change with each career stage. This means that people exhibit different behavior and attitudes at the start, middle, and end of their careers. In the first two stages, the attitudes towards work and organization may be positive because people have to build their careers in these stages. However, in the last career stage or when people have the intention to quit their jobs, their attitude and behavior may be negative. This notion is confirmed by Gardner, Van Iddekinge, and Hom (2018) through their identification of several unfavorable pre-quitting behaviors among research participants. Moreover, Nguyen (2021) found that in Vietnam, directors of SOEs hired more employees the year before they were to retire in order to obtain bribes. This suggests that people exhibit negative work behavior and show less commitment toward their work and organization when they have to leave their job. We attempt to test this idea in our study empirically. We specifically look into whether corporate boards with more directors who plan to leave in a year will impact firm performance in the context of China. Kato and Long (2006) stated that in China, the capital market is still in the infant stage; therefore, both types of agency concerns are acute because of weak investor protection and poorly defined property rights, which in turn develop divergent interests between shareholders and top management that lead to poor firm performance. Unlike other developed nations such as the United States, ownership is concentrated in China, where major shareholders are the state or family (Jiang and Kim 2015; Ma and Khanna 2016; Zhu et al. 2016). Even if ownership is highly concentrated in China, studies reveal that corporate boards still play a meaningful role in shaping firm outcomes. For instance, studies reveal that various board characteristics, such as board size (Liang, Xu, & Jiraporn 2013), board independence (Liang et al. 2013; Liu et al. 2015), CEO duality (Zahid et al. 2024), board gender diversity (Li and Chen 2018; Liu et al. 2014), affect firm performance in China. Looking at the importance of corporate board, we attempt to investigate whether directors leaving their post in one year affect firm performance. Therefore, by employing a sample of Chinese non-financial firms for the period of 2003–2018, we examine whether directors in the last working year (hereinafter DLWY) influence firm performance.

Additionally, studies suggest that women’s behavior and attitudes are different from those of men (Dawson 1995; Ornstein and Isabella 1990). For this reason, we further investigate the moderating role of gender diversity. We find that the more gender diverse a corporate board, the less severe the negative influence of DLWY on firm performance. Finally, we address the potential endogeneity concerns and employ several robustness checks to validate our findings.

This research makes a number of contributions. It provides fresh perspectives on corporate governance from a theoretical standpoint. Prior studies have investigated the employees’ behavior during the career stage (Aryee et al. 1994; Super 1957; Weeks et al. 1999); however, there is a lack of research on the board of directors’ behavior at career end. Our study extends this literature by investigating how directors’ last working year affects firm performance. Previous studies also linked executive turnover to firm performance. For instance, Chulkov and Barron (2023) investigate the sensitivity of CEO compensation to firm performance and find that this relationship is enhanced when CEO turnover is anticipated. In their study, Bates et al. (2018) investigate the directors’ turnover-performance sensitivity. They argue that weaker firm performance leads to higher director turnover. Most of the previous studies either use executive turnover as an outcome variable (i.e., Abe 1997; Bates et al. 2018; Rachpradit et al. 2012) or use executives’ actual turnover as a predictor of performance (i.e., Messersmith et al. 2014). These studies do not provide a direct link between directors or executives planning to leave the firm within one year and performance. Our study fills this gap by answering the question of how directors in their last working year affect firm performance. Directors play an imperative role in corporate governance, specifically in supervising top management. In other words, they are supposed to monitor the management’s actions, provide insight, and veto poor strategic decisions (Pearce and Zahra 1991). Weisbach (1988) states that directors act as a first defense line for stakeholders against management opportunistic behavior, and sometimes they even replace an errant CEO. Directors enhance the internal control framework of the enterprise (Jensen 1986), and poor performance of the firm is linked to CEO turnover (Warner et al. 1988). Thus, to develop the best practices for board composition, it is crucial for board nominations and governance committees, corporate executives, and shareholders to comprehend the impact of corporate boards on firm performance. Regulators and policymakers must also understand the ideal board makeup in order to protect the interests of investors and other stakeholders. Discovering if boards with directors who are leaving their jobs shortly affect firm performance would help inform such board debates.

Literature review and hypotheses

Directors’ last working year and performance

Prior literature suggests that certain structural, functional, and personal attributes of corporate boards and board members affect firm performance. On the structural side, researchers have identified various governance variables that affect board and firm performance. For instance, board size, board independence, CEO duality, and ownership structure (Bhagat & Bolton 2008; Ciftci et al. 2019; Demsetz and Villalonga 2001; Duru et al. 2016; Paniagua et al. 2018). Additionally, the evidence reveals that certain personal features of directors, such as gender, age, education, tenure, and ethnicity, influence firm performance (Ararat et al. 2015; Li and Chen 2018; Livnat et al. 2021; Miller and del Carmen Triana 2009; Mohsni et al., 2021). However, most of these attributes remain constant over time. This indicates that it would be simple for firms to fix the structural problems and keep diversity on the board for better performance if these were the only board features that affected the board and firm performance. The challenging part is that corporate boards work as a team, and it is not only the board’s structural and its members’ demographic attributes that affect performance, but also the members’ behavior may also have an influence on the team and corporate performance.

Moreover, literature suggests that individuals’ behavior changes over time. For instance, career stage theory (Super 1957) suggests that human attitudes and behavior change with each career stage, starting from early career to retirement, because of psychological needs and career concerns (Cohen 1991). The theory elaborates that as each stage is characterized by unique challenges, developmental tasks, and career-related needs, individuals’ behavior and job performance change accordingly. Earlier studies (i.e., Super 1957) classified career concerns into four stages of career, namely, exploration, establishment, maintenance, and disengagement. In the exploration stage, individuals are in search of better opportunities; in the establishment stage, they are more concerned about salary increment, promotion, job security, etc.; in the maintenance stage, they put all their efforts into maintaining their current position, job status, and performance level, and in disengagement stage, they show low interest in the job which lowers performance (Flaherty and Pappas 2002). This notion is supported by extant literature (i.e., Lee 2020; Ornstein et al. 1989; Slocum, Cron (1985)). In the given study, we focus on the disengagement stage of career stage theory. Cron et al. (1988) provide evidence that at the disengagement (pre-quitting) stage, individuals are less motivated to do their jobs efficiently since they experience a “psychological separation” from the enterprise, resulting in a decline in firm performance. Further, Gardner et al. (2018) find several pre-quitting behaviors (PQB), including loss of interest in the organizational mission, less focus on the job, leaving early from work, putting less effort and motivation, acting less as a team player, decreased productivity, and negative attitude, among others. They argue that employees’ behaviors change when they have intentions to quit the job by showing less interest in their duties and the organization. We build our argument on the basis of this notion. Following Gardner et al. (2018), we believe that board directors who have to leave their jobs in one year may show less interest in their responsibilities and the firm. What is more intriguing is that they might not care as much about the firm’s performance.

In sum, as per Super’s (1957) career stage theory, directors at the disengagement stage have less motivation for their jobs, and thus, in turn, firm performance declines. Based on Super’s (1957) career stage theory, we propose the following hypothesis:

Hypothesis 1: There is a negative relationship between corporate boards with higher proportions of directors having last working year and firm performance.

The impact of gender

As we mentioned earlier, the behavior of individuals changes when they are at the disengagement stage. We draw the same inference for the board of directors. Nevertheless, the corporate board consists of both male and female directors. According to the Gender Socialization Theory, as adults, the sexes will bring distinct ethical beliefs to their work responsibilities, which will have a varied impact on how they make judgments about their jobs (Dawson 1995). However, from the perspective of token status theory (Kanterm 1977), men are dominant in the top position; thus, they are more likely to hire or appoint male applicants than female applicants because it is believed that women have fewer imperative attributes for such positions (Lee and James 2007; Powell and Butterfield 2002). In addition, the critical mass theory, as an extension of token status theory, provides evidence that one woman on board is a token, two are pressure, and three are a voice (Kristie 2011). Similarly, Liu et al. (2014) document that board gender diversity enhances firm performance when there are three or more female directors on the board.

Further, testing the career stage theory of Super (1957) for women, Ornstein and Isabella (1990) find that, unlike men, women’s behavior does not change throughout career stages. Gilligan (1993) argues that gender differences exist in how each respond to moral choices. In one study, Weeks et al. (1999) contended that women appear to have stronger ethical judgment when it comes to breaking environmental pollution regulations, corporate espionage, choosing less qualified male candidates, bribing foreign clients, fair promotion practices, ignoring product safety hazards, and giving CEO pay raises that are unjustified. This literature suggests that gender differences exist in behaviors and ethical choices. Corporate governance literature also supports this notion.

The distinctions between the behaviors and attributes of male and female directors are well-documented in the literature. Javed et al. (2023) argue that both men and women are different in their leadership styles, which in turn ameliorates strategic decision-making (Ahern and Dittmar 2012; Gilbert and Ivancevich 2000; Greene et al. 2020). Therefore, gender diversity is an imperative determinant of firm performance. On the one hand, females are more conservative and risk-averse than their counterparts (Huang and Kisgen 2013; Tang and Li 2024), which in turn substantially influences firm strategic decision-making. On the other hand, studies reveal that, compared to males, female directors are less risk-averse (Carter et al. 2017; Gulamhussen and Santa 2015; Powell and Ansic 1997), less overconfident (Chen et al. 2019; Lonkani 2019; Lundeberg et al. 1994), bring more transparency in information sharing (Abad et al. 2017; Gul et al. 2011; Jurkus et al. 2011), mitigates earnings management (Arun et al. 2015; Kyaw et al. 2015) and attend more board meetings (Adams and Ferreira 2009; Boutchkova et al. 2020; Chen et al. 2021). Overall, the literature indicates that having more female directors enhances the board’s role in oversight and, as a result, boosts firm’s performance (Erhardt et al. 2003; Tejedo-Romero et al. 2017).

Based on this discussion, we predict that women’s behavior and commitment to their jobs and firms may not alter even during their last working year, as they are more dedicated and ethically conscientious. Considering this, we propose the following:

Hypothesis 2: The negative association between firm performance and boards with directors in their last working year is weaker in more gender-diverse boards.

Methodology

Data and sample

Our sample comprises 2866 A-share Chinese headquartered firms (non-financial only) that were listed between 2003 and 2018 on Shanghai (SSE) and Shenzhen Stock Exchanges (SZSE). Database of the China Stock Market and Accounting Research (CSMAR) is used to retrieve data for each variable. Due to the availability of board diversity and corporate governance data on CSMAR from that year, 2003 was chosen as the starting year. Financial institutions are not included in the sample since they may have an impact on financing decisions (Alves and Ferreira 2011) and have different characteristics. All continuous variables are winsorized by 5% at both tails to eliminate the impact of any outliers. Since our sample includes companies that have been listed for a variety of years, our data is an unbalanced panel with 25,461 yearly observations.

Variables

Dependent variable

Firm performance can be measured through both accounting-based variables and market-based variables (Brinkhuis and Scholtens 2018; Post and Byron 2015). Following previous studies (Matsa and Miller 2013; Yang et al. 2019), we use an accounting-based measure of performance, Return on Assets (ROA), as the main dependent variable in this study. We measured ROA as the ratio of net income to total assets (Chen and Keefe 2020; Liu et al. 2014).

For the robustness check, we use return on equity (ROE) and a market-based variable, Tobin’s Q, as a measure of firm performance. Following previous studies (i.e., Iqbal et al. 2020), we measure ROE as the ratio of net income (loss) to total equity. Finally, we measure Tobin’s Q as a ratio of the firm’s market value to its replacement value (Post and Byron 2015; Yang et al. 2019).

Independent variables

The variable of interest in this study is the directors’ last working year. We measure Directors’ Last Working Year (DLWY) as the number of directors who will leave the post in one year divided by the total number of directors. For the robustness check, we employ an alternate proxy for our independent variable. Instead of leaving the post, we use the directors’ age as an alternate proxy. Nguyen (2021) argues that directors’ age before retirement significantly increases the number of employees hired by a firm. We follow the same concept and use the directors’ age before retirement as an independent variable. Most of the employees in China retire at the age of 65 (Fan et al. 2007; Wang and Luo 2019), we use age 64 as the directors’ age one year before retirement. We measure Directors’ Age before Retirement (DABR) as the number of directors aged 64 or above divided by the total number of directors.

Gender diversity is another variable that this study is interested in. We use gender diversity as a moderator in this study. We measure gender diversity using the Blau (2000) index.

$${Blau}=1-{\sum }_{i=1}^{n}{P}_{i}^{2}$$

where n is the total number of directors on the board, and p is the proportion of male and female directors.

Control variables

We include a number of control variables in our models since research indicates they may have an impact on the firm’s profitability. Following prior literature (Bhagat and Bolton 2019; Ciftci et al. 2019; Mertzanis et al. 2019), we added numerous governance variables; for instance, board size (BS), board independence (BI), and CEO duality (CEO_Dual). As recommended by prior research (Chen and Keefe 2020; Ducassy and Guyot 2017; Falk 2012; Mertzanis et al. 2019; Vithessonthi and Racela 2016), we also incorporate a number of company-level control variables. We include leverage (LEV), firm age (Age), firm size (Size), R&D intensity (RDI), and asset tangibility (Tang) as control variables. We also controlled industry and year effects by including industry and year dummies. All the variables used in this study are defined in Appendix 1A.

Model specification

We estimate the subsequent regression model to determine the impact of directors’ last working year on firm performance.

$$\begin{array}{l}{{Profitability}}_{{it}}={\beta }_{0}+{\beta }_{1}{{DLWY}}_{{it}}+\sum {\beta }_{i}{{Controls}}_{{it}}+{{\beta }_{2}{IND}}_{i}\\\qquad\qquad\qquad\quad\;\;+\,{\beta }_{3}{{Year}}_{t}+{\varepsilon }_{{it}}\ldots\end{array}$$
(1)

where profitability is measured by ROA, DLWY is the directors’ last working year, Controls include all control variables, IND and Year, and the industry and year dummies respectively. \(\varepsilon\) is the random error term.

In addition, we estimate the subsequent model to look into the moderating effect of gender diversity.

$$\begin{array}{l}{{Profitability}}_{{it}}={\beta }_{0}+{\beta }_{1}{{DLWY}}_{{it}}+{\beta }_{2}{{DLWY}}_{{it}}* {{Gender}{{\_}}D}_{{it}}+\\\qquad\qquad\qquad\quad\;\,+\,{\beta }_{3}{{Gender}{{\_}}D}_{{it}}+\sum {\beta }_{i}{{Controls}}_{{it}}+{{\beta }_{4}{IND}}_{i}\\\qquad\qquad\qquad\quad\;\,+\,{\beta }_{5}{{Year}}_{t}+{\varepsilon }_{{it}}\ldots\end{array}$$
(2)

Where \({{DLWY}}_{{it}}* {{Gender\_D}}_{{it}}\) is the interaction term of the directors’ last working year and gender diversity. The rest of the variables are similar to Eq. (1).

Results

Descriptive statistics

Table 1 reports summary statistics for all variables included in this study. Particularly, this table reports the number of observations, mean, standard deviation, and the 25th, 50th (Median), and 75th and 95th percentiles for each variable. The mean value of the profitability (ROA) is 0.038, which is consistent with Liu et al. (2015). The variable of interest in this study is the directors’ last working year (DLWY). The mean value of DLWY is 0.228, which indicates that 22.8% of the directors in our sample must quit their positions within a year. This figure is so high that we think it might have an impact on the firm’s performance.

Table 1 Summary statistics.
Full size table

Furthermore, Table 1 reports summary statistics for all control variables. It shows that the average board size (BS) of the sample firms is 8.937. Table 1 shows that 36.8% of corporate boards’ directors are independent. Our sample reveals that 22.4% of the directors serve as both the board chairman and CEO. Furthermore, we include descriptive statistics for the firm-specific factors. The mean value of leverage (LEV) is 0.451, while that of the log of total assets (Size) is 22. The average age of the firms (Age) included in the sample is approximately 14 years. R&D intensity (RDI) has a mean value of 0.002, while asset tangibility (Tang) has a mean value of 0.934. These numbers are approximately similar to prior research (Chen & Keefe 2020; Liu et al. 2015; Shih et al. 2021).

Correlation analysis

Table 2 lists correlations between each variable and its variance inflation factor (VIF). Interestingly, there is a negative and significant association between DLWY and ROA, which validates our hypothesis at this stage. Overall, correlations between explanatory variables are not higher than 0.55. These correlation coefficients imply that multicollinearity for regression analysis is not a significant problem. Also, all of the variables’ VIF values are lower than the cutoff value of 5 (Ringle et al., 2015), which further suggests that multicollinearity is not a problem in this research.

Table 2 VIF and pairwise correlations.
Full size table

Regression results

Directors’ last working year and performance

We empirically examine the nexus between DLWY and firm performance in this section. Table 3 reports these empirical results. Model (1) in Table 3 estimates one-to-one relationship between DLWY and firm performance. We estimate Models (2), (3), (4), and (5) using four distinct regression settings to obtain robust results. Model (2) shows that the coefficient of DLWY (−0.013) is negatively significant at 1% level. This indicates that on average, one standard deviation increase in DLWY causes approximately a 6% decrease in ROA, ceteris paribus. According to these findings, firms that have a higher percentage of directors quitting their positions within a year do poorly.

Table 3 DLWY and firm performance.
Full size table

Models (3) through (5) further test the impact of DLWY on ROA using different regression settings. More specifically, Model (3) uses Newey-West standard errors to cope with a serial correlation of standard errors. It is observed that the results are unchanged as the reported coefficient of DLWY in Model (3) is still −0.013 and is significant at 1% level. Additionally, we use random effect and fixed effect regression in Models (4) and (5), respectively, to evaluate the association between DLWY and ROA. The reported results in Model (4) show that the coefficient of DLWY is still negative (−0.010) and is highly significant (at 1%). Similarly, results in Model (5) support the previous findings as the coefficient of DLWY is negative (−0.008) and is significant at 1% level.

The coefficients of all control variables are significant except board size (especially in Model 2). Results in Table 3 reveal that CEO duality, firm size, and asset tangibility are positively related to ROA. However, board independence, leverage, firm age, and R&D intensity are negatively related to ROA. These findings are in line with previous studies (i.e., Bhagat & Bolton 2019; Chen and Keefe 2020; Mertzanis et al. 2019; Vithessonthi and Racela 2016).

Overall, the results in Table 3 support our hypothesis that a higher proportion of directors who will soon quit their jobs is negatively related to firm performance. These outcomes are in line with Super’s (1957) career stage theory. The reason behind this negative relationship may be, as mentioned by Gardner et al. (2018), that people develop some negative behaviors when they have to quit their jobs. As mentioned earlier, these negative behaviors include loss of interest in the organizational mission, less focus on the job, leaving work early, putting in less effort and motivation, acting less as a team player, decreased productivity, and a negative attitude, which may negatively affect the board and ultimately firm performance. Finally, the career stage theory (Super 1957) also lends support to our findings, which state that people’s behavior changes when they prepare for retirement. In conclusion, both theoretical and empirical findings support our argument.

The role of gender differences

We extend our analysis to find the moderating role of gender diversity in the relationship between DLWY and ROA. Results of the regression analysis with gender diversity as a moderator are reported in Table 4. These results reveal that the interaction term (DLWY*Gender_D) has a positive coefficient (0.027) and is significant at 5% level. These findings support our argument that women directors are less likely to experience unfavorable behavior changes throughout the course of their careers. Prior literature supports these results. For instance, Ornstein and Isabella (1990) found that, unlike men, women’s behavior did not change as they progressed through their careers. Moreover, as mentioned earlier, women are more committed to the organization (Weeks et al. 1999), bring more transparency in information sharing (Abad et al. 2017), mitigate earnings management (Kyaw et al.s 2015), attend more board meetings (Chen et al. 2021), and enhance the board’s role in oversight (Erhardt et al. 2003; Tejedo-Romero et al. 2017). This discussion leads us to the conclusion that a firm’s performance will less likely suffer when there are more female directors on the board, since there will be fewer unfavorable pre-quitting behaviors and more organizational commitment.

Table 4 The role of gender diversity.
Full size table

Robustness checks

Alternate proxies

To validate our findings, we conduct various robustness checks. Firstly, we use an alternative proxy for our independent variable. Instead of leaving the post, we use the directors’ age as an alternate proxy. Following Nguyen (2021), we use the directors’ age before retirement (DABR) as the independent variable. Table 5 reports the regression results with DABR as the independent variable. The coefficient of DABR is negative (−0.003) and is significant at 5% level, hence confirming the robustness of our findings.

Table 5 Alternate proxy for DLWY.
Full size table

To further ensure the validity of our findings, we also use different proxies for firm performance in addition to ROA. Both a market-based indicator, Tobin’s Q, and an accounting-based measure of firm performance, ROE, are used. The results of the regression analysis with these alternative proxies are reported in Table 6. The coefficients of both ROE (−0.030) and Tobin’s Q (−0.071) are negative and are significant at 1% and 5% levels, respectively. This reaffirms the reliability of the findings.

Table 6 Alternative proxy for firm performance.
Full size table

Omitted variables and endogeneity concerns

Our primary investigation reveals a negative correlation between DLWY and firm performance. However, reverse causality or missing variables could have an impact on our findings. These are the concerns we tackle in our study. To address the endogeneity problem caused by missing variables, we first apply fixed-effect regression, which partially eliminates endogeneity due to omitted variables (Wooldridge 2010). Table 7’s column (1) shows the fixed-effect regression results. These outcomes continue to support our primary findings.

Table 7 Omitted variables and endogeneity concerns.
Full size table

In order to address the issue of reverse causality, we examine the effect of DLWY on firm performance once again using the dependent variable (ROA) at forward lag (found in column 2 of Table 7). The empirical results in Table 7, which demonstrate that DLWY considerably and negatively affects firm performance at a 1% level of significance, further corroborate our first hypothesis.

Furthermore, we use the PSM technique to address potential endogeneity problems resulting from reverse causality and the self-selection effect. We use DLWY as a dummy variable (“1” if DLWY is higher than the industry average and “0” otherwise) in order to use PSM. We employ every control variable from our primary regression model in PSM regression. Column (3) of Table 7 reports the PSM regression’s findings. The coefficient of DLWY_dummy is still significant and negative, as Table 7 demonstrates. This demonstrates that our results remain robust even after considering endogeneity concerns.

We use a two-stage least squares (2SLS) approach in our study to better address the endogeneity issues. The main problem of endogeneity is the possible reciprocal relationship between director departures and company performance; directors may quit in anticipation of subpar performance or because of subpar results. In order to address this problem, we used the industry median as an instrumental variable (IV) in the 2SLS procedure’s initial step. Since it captures patterns in industry performance and is presumably independent of the performance of a particular firm in the second stage, we think the industry median is a valid tool. In order to minimize bias from potential reverse causation or unobserved confounders, this method helps guarantee that the variance in director departures utilized to quantify the influence on company performance is exogenous. To further address the issue of reverse causality, we use another instrument. Instead of the industry median, we use directors who leave the firm due to health issues or due to retirement age as an instrumental variable. We believe this variable is a good candidate because it is highly correlated with directors’ turnover and is independent of firm performance. Results of the 2SLS regressions are reported in Table 8, which are in line with our main findings.

Table 8 2SLS.
Full size table

Entropy balancing

To further reduce the possibility of endogeneity bias, we additionally employ the entropy balancing method. The sample is initially split into two groups: the treatment group and the control group. Firm-year observations with DLWY_dummy equal to 1 are included in the treatment group, while firm-year observations with DLWY_dummy equal to 0 are included in the control group. We followed (Hainmueller and Xu 2013) to converge the mean, variance, and skewness of all covariates in the treatment and control groups in Table 8, Panels A (before entropy balancing) and B (after entropy balancing). We re-estimate the regression results based on the treated balance once again in Table 9, Panel C. The empirical results indicate a significant and negative relationship between DLWY and firm performance. This reveals that our results are resistant to endogeneity bias.

Table 9 Entropy balancing.
Full size table

Additional analysis: changed sample

We also discuss the implications of reforms pertaining to non-tradable shares (NTS). Chinese firms were once governed by the government. The establishment of the Shanghai Stock Exchange (SSE) and the Shenzhen Stock Exchange (SZSE) in 1990 and 1991, respectively, marked the beginning of the privatization movement. Nevertheless, the majority of Chinese businesses were still under state control even after the establishment of the SSE and SZSE (Sun et al. 2017). In April 2005, the China Securities Regulatory Commission (CSRC) announced the NTS changes with the goal of enhancing corporate governance and mitigating agency issues. These changes made the shares that were not tradable into tradable ones. Reducing ownership concentration and enhancing liquidity were the goals of these reforms (Khaw et al. 2016). According to certain academics, the corporate governance processes have improved as a result of the NTS reforms (Sun et al. 2017). Roughly 90% of the firms had completed these reforms by 2007. Given that the study’s sample was chosen between 2003 and 2018, it is probable that the 2005 NTS reforms had an impact on the findings. As a result, we adjust the sample by excluding the observation prior to 2008 and used the updated sample to examine the effect of DLWY on firm performance. Table 10 presents the findings. These findings corroborate with our primary conclusions and imply that our results are robust and unaffected by the NTS reforms.

Table 10 Sample change.
Full size table

Directors’ career concerns

Our main findings reveal that directors in their final working year are negatively related to firm performance. Nonetheless, research indicates that individuals at varying stages of their careers have distinct anxieties regarding their careers. For instance, Alfonso et al. (2019) state that CEOs in their early career stage are more concerned about their reputation and avoid inefficient strategies. They further argue that late-career CEOs are involved in expectations management. Following this concept, we expect that young directors might be more concerned about their careers because they must remain active in the job market. Conversely, directors who are near to retirement may not be too concerned about their careers. To address this issue, we employ regression analysis for different age windows. First, we use an independent variable as the number of directors leaving the post in one year, who are 40 years of age or younger. Second, we use an independent variable as the number of directors leaving the post in one year, who are 50 years of age or below. Finally, we use an independent variable as the number of directors leaving the post in one year, and they are 60 years of age or below. Results in Table 11 show that the findings do not change even after using different career stages. Even directors who intend to stay in the workforce may have the motivation to uphold their reputation, but if they place their future professional prospects ahead of their current position, they may even lessen their governance efforts. Considering this, our results show that directors in their final year have a detrimental impact on firm performance, regardless of their future career prospects.

Table 11 Reputation concerns at different career stages.
Full size table

Conclusions

Literature suggests that people show unfavorable behavior when they intend to quit their jobs (Gardner et al. 2018). Following this notion, we investigate the nexus between corporate directors’ last working year and firm performance. Analyzing a sample of Chinese listed firms, we find a negative relationship between DLWY and firm performance. Our findings confirm the idea that people exhibit negative work behavior and show less commitment to their work before quitting their jobs, which negatively affects their own and ultimately the firm’s performance. In addition, studies indicate that men and women behave and think differently (Dawson 1995; Gilligan 1993) and unlike males, women’s behavior does not change as they advance through the career stages (Ornstein and Isabella 1990). Owing to its significance, we look into how gender diversity on corporate boards moderates the association between DLWY and firm performance. We find that companies with greater gender diversity experience less negative impact of directors’ last working year (DLWY) on performance. To make our results more reliable, we use alternative proxies for DLWY and firm performance and address the endogeneity problem.

Our findings are important for corporations, regulators, and policymakers. First, to boost the motivation level of directors at the disengagement stage, corporations should employ tenure-related incentives, which in turn reduce opportunistic behavior and enhance firm performance. Second, while appointing a board of directors, firms must consider the critical mass theory perspective and appoint a greater proportion of female directors on the board because they inhibit agency concern and improve firm performance. Third, regulators and policymakers should formulate a framework and conduct training for individuals who are in the disengagement stage in order to shape their psychological bias, which is developed at the last career stage.

This study has some limitations. First, this study is conducted on a single country, China. To make the results more generalizable, this study can be replicated using a multi-country sample, as other countries might have different legal, cultural, structural, or institutional attributes. Second, even though we demonstrate that DLWY reduces firm performance after controlling for industry, year, and several firm-specific variables, it is possible that we have overlooked other mechanisms by which DLWY could affect firm performance.