Job-hopping executives and corporate social responsibility

Abstract

Purpose

The purpose of this study is to examine the relationship between the corporate social responsibility (CSR) performance of job-hopping executives at their former and subsequent firms.

Design/methodology/approach

We conduct regression analyses using a sample of firms listed on the Shanghai and Shenzhen Stock Exchanges from 2010 to 2020 to examine whether CSR performance is similar from one firm to the next as executives switch jobs.

Findings

We find a positive relationship between the CSR performance of former and subsequent firms under job-hopping executives. This relationship is the strongest in the year of the job switch; it weakens in the second year and eventually disappears in the third year. In addition, we show that this relationship benefits different CSR stakeholder groups and is contingent on executive and subsequent firm attributes and job-hopping characteristics. Furthermore, we demonstrate that firms that hire a new chief executive officer from a firm with a strong track record in CSR, the new firm experiences a significant surge in CSR performance compared with firms that do not experience such a shock.

Practical implications

This study has implications for executive hiring decisions.

Originality/value

This study extends the understanding of CSR determinants through the lens of inter-organisational ties associated with job-hopping executives.

Citation

Fang, J., Tian, Y. and Hu, Y. (2024), "Job-hopping executives and corporate social responsibility", China Accounting and Finance Review, Vol. 26 No. 2, pp. 196-228. https://doi.org/10.1108/CAFR-03-2023-0036

1. Introduction

Job-hopping by executives, which refers to the movement of executives across organisations, is a subject of constant debate in both the corporate world and academia. Some people regard it as a valuable practice that injects fresh perspectives and skills into organisations. This viewpoint finds substantial support in literature, which indicates that job-hopping by executives has a positive effect on various corporate outcomes in their successor firms, including driving strategic changes (Boeker, 1997), fostering innovation (Kaiser, Kongsted, & Rønde, 2015) and facilitating international mergers and acquisitions (Wang, 2019). Conversely, it is claimed that frequent executive mobility, particularly by executives who switch between rival firms, can lead to the loss of team and firm-specific tacit knowledge, disrupt established corporate culture and have adverse effects on employee morale. From this standpoint, firms with high rates of job-hopping by executives may deplete their reserves of developed human and social capital, which has the potential to undermine the viability of the firm the executive is leaving (Wezel, Cattani, & Pennings, 2006; Somaya, Williamson, & Lorinkova, 2008; Messersmith, Lee, Guthrie, & Ji, 2014).

Nonetheless, despite increasing interest in examining the impact of job-hopping by executives on corporate outcomes, evidence of the effects of job-hopping by executives on corporate social responsibility (CSR) remains sparse. Related research primarily examines how executive turnover affects the CSR of individual firms. This line of research includes studies that explore the influence of executive turnover on the CSR of the (same) firm under different leadership, and the role played in CSR by returning directors (e.g., Bernard, Godard, & Zouaoui, 2018; Luo, Chen, & Chen, 2021). What is missing from this research landscape is knowing whether there is a correlation between the CSR performance of the former and subsequent firms of the same executive. Our study aims to fill the gap by examining whether the CSR performance of firms under the same executive is similar and, if so, which stakeholders are most likely to benefit, how long the similarity persists and whether it is contingent on the executive, the firm and job-hopping characteristics. Furthermore, we assess the significance of this effect by comparing firms subjected to job-hopping with firms that do not experience a CEO (chief executive officer) job-hopping shock.

Anecdotal evidence strongly supports the notion that CSR experience gained by top executives in a previous firm significantly influences CSR practices in their subsequent firms. A striking example of this is Paul Polman, who is renowned for his decades-long service as the CEO of Unilever. Prior to joining Unilever in 2009, Paul had served as the chief financial officer of Nestlé, a firm widely recognised for its commitment to environmental and social responsibility. During his leadership at Unilever, Polman had initiated the Unilever Sustainable Living Plan to boost the company’s revenue while reducing its environmental footprint and bolstering its positive social impact. Under Polman’s stewardship, Unilever firmly established its reputation as a leader in CSR. Therefore, research into the influence of job-hopping by executives on the connection between CSR performance in an executive’s prior and subsequent firms holds significant merit, for several reasons [1].

First, comparing the CSR performance of different firms under the same executive unveils the extent to which CSR practices are transferable. This, in turn, helps to predict corporate outcomes and to foster a deeper understanding of best CSR practices that can be transmitted across firms. According to behavioural theorists, decision-makers often rely on their most available and recent memories (Tversky & Kahneman, 1973). Consequently, it is increasingly likely that corporate executives base corporate policies and decisions on their prior experiences (Enkhtaivan & Davaadorj, 2021). Second, unlike some relatively stable and objective determinants of CSR, such as firm industry classification, job-hopping by executives is a dynamic behavioural and subjective factor that facilitates the transfer of human capital, knowledge and expertise. Owing to the pivotal role top executives often play in shaping a company’s strategy, culture and decision-making (Hambrick & Mason, 1984), a new executive’s arrival can exert a substantial and immediate influence on the direction and priorities of CSR initiatives. Third, our results have practical implications for executive hiring decisions, especially for stakeholders who prioritise CSR. Over the past few decades, CSR has witnessed a significant increase in attention. As noted by Kitzmueller and Shimshack (2012), the widely accepted importance of CSR has led to a shift in research focus, from questioning its necessity to understanding its underlying drivers. Our study contributes new evidence to this research agenda.

We conduct our study in China because, first, China is a significant global economic power and its rapid development has caused a range of social and environmental challenges (Zou, Xie, Qi, & Yang, 2018). Investigating CSR determinants in an emerging economy is particularly timely. Second, job-hopping by executives is relatively common in China and this affects the way firms operate and make strategic decisions (Wang & Guo, 2022). These findings warrant solid ground to carry out our research. Third, owing to China’s unique institutional context, examining the correlation between CSR and job-hopping by executives in Chinese firms can offer valuable insights into CSR performance that is characterised by weak law enforcement and diverse stakeholder expectations (Hu & Fang, 2022a). Hence, understanding the influence of job-hopping by executives – a non-institutional channel – on CSR can aid firms in developing effective CSR strategies. Our study has broad implications for policy and practice concerning job-hopping by executives and CSR, not only in China, but also in other countries.

We posit that, when an executive switches firms, the CSR performance of the subsequent firm is correlated with that of the former firm. This draws on the concept of microfoundations and upper echelon theory, which argues that executives who switch jobs apply the accumulated experience and knowledge they gained at former firms to the strategic decisions of subsequent firms (Felin, Foss, & Ployhart, 2015). Job-hopping by executives provides a high-capacity channel for the transfer of organisational practices (Kraatz & Moore, 2002). Using data from listed Chinese firms from 2010 to 2020, we find a significant positive correlation between the CSR performance of former and subsequent firms under job-hopping executives. More interestingly, our further investigation shows this relationship is most prominent in the year of the job switch, weakens in the second year and eventually disappears in the third year. We also find that the strength of the job-hopping effect diminishes monotonically, which indicates that the job-hopping effect lasts only for a short time. This observation highlights the pressing question of how firms can leverage the transient nature of this impact and integrate its benefits into their long-term CSR strategies and practices. Additionally, our findings reveal that only the CSR practice of the former firm, up to one year prior to the job-hopping by the executive, has predictive power for the CSR practice of the subsequent firm. The evidence sheds light on what exact prior experience is relevant and transmitted to the subsequent firm, and results in the job-hopping effect. This finding is of interest to all firm stakeholders, as it can aid in predicting the potential effect of job-hopping by executives on subsequent firms.

To deal with endogeneity concerns, we also investigate the main research question by using a different setting, namely, propensity score matching difference-in-differences (PSM-DiD). We use CEO job-hopping (instead of job-hopping by all top executives) as a shock to analyse how CSR changes for firms that experience CEO job-hopping, compared with firms that do not experience such a shock. We find that, when a firm hires a new executive from a firm with greater CSR, the CSR of the focal firm subsequently increases relative to that of firms that do not experience such a change. The results provide significant evidence to support our main hypothesis.

Moreover, we test the job-hopping effect on separate components of CSR, to determine which stakeholder(s) are most likely to benefit from CSR after job-hopping by an executive. We find that the relationship is strong for CSR dimensions related to certain and specific stakeholder groups, such as shareholders, employees and consumers, because these dimensions share commonalities across firms. However, we observe no significant effect for the more general and broader social dimension. This may be the result of variations in the extent to which firms prioritise different social activities under this dimension; therefore, an executive’s CSR experience at their previous firm is less relevant for their subsequent firm.

Furthermore, we find that inter-firm CSR similarities arising from job-hopping by an executive depend on various specific job-hopping factors, such as whether the executive takes a career break, receives a pay rise, switches to a firm with higher or lower CSR performance, or changes their physical job location. The job-hopping effect is also highly sensitive to the demographics of executives, such as age, gender and overseas experience, and firm attributes, including size, leverage, ownership structure and quality of governance. Lastly, our results remain consistent through a battery of robustness tests, including a PSM balance test, a placebo test and a DiD method based on an entropy balancing approach.

Our study offers contributions that relate to two strands of literature. First, it complements the literature on the executive migration effect from the perspective of CSR. Our results highlight the significant effect of relevant and similar past experiences on corporate outcomes and we provide evidence that human behaviours tend to persist under various circumstances. Second, the study extends the understanding of CSR determinants through the lens of inter-organisational ties associated with executives who do job-hopping. The experience gained at former firms determines the executives’ knowledge and resources, which are then passed on to subsequent firms. The closest study similar to ours is that of Kim, Moon, and Kim (2022). They analyse CSR profile differentials between two firms after job-hopping by executives in the United States. Our evidence focuses on direct CSR correlations in China. We supply finer evidence on several crucial aspects, including how the correlation between CSR performance at a former and subsequent firm of an executive changes over time, which specific stakeholders benefit, and whether the job-hopping effect is contingent on the executive, successor firm and job-hopping characteristics. This level of insight is rare in the examination of the relationship between job-hopping and CSR and has practical implications beyond merely detecting the job-hopping effect. Specifically, our results can guide the design of effective executive compensation packages and inform the selection of executives with particular backgrounds to align hiring decisions with desired CSR outcomes.

The rest of the paper proceeds as follows. Section 2 reviews the related literature and develops the hypotheses. Section 3 describes the data and the method. Section 4 reports the results. Section 5 provides further analyses and robustness tests. Section 6 concludes the paper.

2. Literature review and hypothesis development

2.1 Theoretical foundation

Executives play a key role in making decisions for organisations and developing routines. Studies suggest that executives are highly influential in establishing the microfoundations of organisational capabilities (Felin et al., 2015). The concept of microfoundations centres on the impact of individual actions and interactions on organisational outcomes (Teece, 2007; Felin et al., 2015), and suggests that firms can acquire valuable knowledge, experience and expertise by hiring executives who have job-hopped from other firms, because they bring strategic insights and knowledge with them when they switch jobs, therefore, job-hopping by executives can serve as a highly effective channel for the transfer of organisational practices (Kraatz & Moore, 2002). In line with this notion, upper echelon theory suggests that organisational decisions are heavily influenced by the experiences, background, personalities and values of its top management team (Hambrick & Mason, 1984). This implies that organisations are largely shaped by their top managers, and that changes in top management succession and demographics can have a significant effect on various aspects of organisational strategy and operations (Finkelstein & Hambrick, 1996).

The literature supplies ample evidence for the theoretical conjecture that job-hopping by executives can have learning effects that exert a significant influence on various organisational outcomes. Boeker (1997) finds that job-hopping by executives generates parallels in product and market entry. Westphal and Fredrickson (2001) find that, when new CEOs from outside take over firms, they tend to implement strategic changes that align their new organisations with their previous firms. Still and Strang (2009) find that firms linked through executive migration to Global Financial (a leading financial services firm) are more likely to be subject to benchmarking and have a greater influence on benchmarking teams than firms without such links. Wang (2019) finds that job-hopping executives’ inter-organisational learning is associated with international acquisitions. Enkhtaivan and Davaadorj (2021) highlight a significantly positive association between the cash holdings of the prior and the subsequent firms when they are managed by the same executive.

2.2 Related CSR literature

The CSR literature has developed significantly over recent years and reflects the increasing public and regulatory attention being paid to this topic (Kitzmueller & Shimshack, 2012). One strand of CSR research examines its economic benefits. Studies report that CSR is positively related to various financial outcomes, such as financial performance (Orlitzky, Schmidt, & Rynes, 2003; Kim, Kim, & Qian, 2018), firm value (Tsang, Hu, & Li, 2021) and trade credit from suppliers (Zhang, Ma, Su, & Zhang, 2014). Greater CSR engagement by a firm can also lead to more accurate analyst forecasts (Dhaliwal, Radhakrishnan, Tsang, & Yang, 2012), reduced risk of stock price crash (Du, 2018) and lower cost of capital (Dhaliwal, Li, Tsang, & Yang, 2014). Another strand of literature is motivated by the benefits and devotes significant efforts to identifying factors that explain CSR performance. Certain firm-specific characteristics (e.g., size, leverage, profitability) and external factors (e.g., stakeholders, institutional environment) are significant determinants of CSR (Campbell, 2007; Roberts, 1992; Artiach, Lee, Nelson, & Walker, 2010; Perez-Batres, Doh, Miller, & Pisani, 2012).

In recent years, studies have developed into the area of the heterogeneity of corporate managers and its impact on CSR performance and show how individual traits can be associated with firm variables. A large body of evidence supports the critical role played by an executive’s personal characteristics, such as gender, age and tenure (Manner, 2010; Borghesi, Houston, & Naranjo, 2014; Chen, Zhou, & Zhu, 2019), and psychological traits, for example, materialism (Davidson, Dey, & Smith, 2019) and hubris (Tang, Qian, Chen, & Shen, 2015), in determining the level of CSR performance. Along this line, CSR studies have examined the relationship between executive turnover (e.g., CEO turnover) and CSR performance and found that changes in executive leadership can affect CSR performance of a firm. For instance, Bernard et al. (2018) document a positive relationship between the change of executives and CSR performance in French firms and report that this positive effect is stronger in firms where the new executive is recruited from outside. Meng, Zeng, Tam, and Xu (2013) report that CSR is negatively associated with involuntary departure by executives (e.g., dismissals and forced resignations), but not with normal departures by executives (e.g., retirement and contract expiration). Using data from listed Chinese firms, Luo et al. (2021) demonstrate that returnee directors with relevant experience are associated with an increased level of corporate donations by their firms. Using firm-level data from the United States, Kim et al. (2022) present evidence that migrated executives incorporate elements of their CSR practices at their previous firms into their subsequent firms, thereby reducing the disparity between the CSR profiles of the two firms. Our study seeks to expand this strand of literature by investigating novel aspects, such as the duration of CSR performance similarity between former and subsequent firms, the stakeholder groups most affected, and the impact of job-hopping characteristics on different outcomes. In this study, we aim to fully comprehend the interplay of these variables.

2.3 Hypothesis development

According to the microfoundations concept and upper echelon theory, executives rely on their previous experience in similar roles to make decisions (Hambrick & Mason, 1984). This approach aligns with the behavioural consistency hypothesis, which asserts a strong connection between past and future decisions; furthermore, the behavioural styles exhibited by individuals in different situations tend to have a certain degree of similarity and stability (Epstein, 1979). We therefore posit that when executives switch jobs, there exists alignment in CSR between the former (departing) firm and the subsequent (hiring) firm. This alignment arises, in particular, from the discretionary nature of CSR decision-making (Barnett, 2007), which allows job-hopping executives to harness the CSR-related information, experience, and knowledge they had acquired in their prior roles to exert an influence on CSR initiatives at the subsequent firm. Consequently, these migrating executives serve as conduits for inter-organisational learning (Kraatz & Moore, 2002), which leads to a positive correlation between the CSR performance of their former firm and that of their subsequent firm. In particular, when executives move from a firm with strong CSR performance, they bring with them the successful CSR practices and strategies from their previous firms. This transfer of knowledge is instrumental in elevating the CSR performance of the subsequent firm. For instance, these executives can champion the implementation of proven best CSR practices they acquired from their former firm and identify areas for enhancement in the subsequent firm. Their commitment to CSR can instil a heightened CSR-conscious environment in the subsequent firm.

In contrast, it is crucial to recognise that adverse influences can also disseminate from one firm to another via executive job-hopping. These effects tend to diffuse through peer influence within the executive network (Hu & Fang, 2022b; Huang, Du, & Wu, 2023). In fact, migrating executives carry with them not only knowledge, but also the CSR culture and values cultivated at their previous workplaces, which may not always foster CSR. This phenomenon is substantiated by anecdotal evidence. For instance, Travis Kalanick, the former CEO of Uber, faced criticism for fostering an unethical corporate culture at Uber during his leadership, such as imposing long working hours without additional compensation. Subsequently, in 2018, he assumed the role of CEO at City Storage Systems, a company operating CloudKitchens. In 2021, CloudKitchens faced legal action over allegations of labour law violations and deceptive business practices. This example illustrates that when an executive from a firm with a poor CSR record switches to another company, it is likely to result in a decline in the CSR performance of the subsequent firm. Moreover, there is a possibility that certain shareholders may not favour (over) investment in CSR (Barnea & Rubin, 2010; Gillan, Koch, & Starks, 2021). In such instances, shareholders might intentionally hire new executives from firms that underperform in CSR to curtail CSR initiatives at the hiring firm.

Beyond the spillover effect in corporate practices, we contend that executive mobility may also result in adverse corporate outcomes, such as leadership disruption, interference with established corporate culture, and lower employee morale (Wezel et al., 2006; Somaya et al., 2008; Messersmith et al., 2014). Following this line of reasoning, an alternative hypothesis is plausible: a negative association may exist between the CSR performance of the former firm and the subsequent firm of the job-hopping executive. Three explanations support this claim. First, executives often build careers across multiple organisations (Won & Bidwell, 2023). When they job-hop, their focus may lean towards personal branding and career advancement, instead of the hiring firm’s long-term reputation, even when they transition from a firm with a strong CSR track record. This inclination could lead to decisions that favour short-term gains at the expense of long-term CSR initiatives, and ultimately result in a negative association between the CSR performance of the former (departing) firm and that of the latter (hiring) firm.

Second, job-hopping by an executive can signal their dissatisfaction with the values and culture of the departing firm. For instance, conflicts may arise within firms between managers and shareholders over issues such as the effect of CSR on firm value (Barnea & Rubin, 2010; Gillan et al., 2021). This discord can give rise to a lack of alignment or, in some cases, a negative relationship regarding CSR alignment between the departing and hiring firms, particularly when the hiring firm shares the departing executive’s CSR values.

Third, job-hopping executives may require an adjustment period to comprehend the distinctive challenges and opportunities at their new (hiring) firms. This learning curve can introduce delays in developing and implementing CSR initiatives. Additionally, a major shift in CSR leadership and direction may require time to disseminate among employees and throughout the organisation, thereby causing a delay in reflecting changes in the CSR performance of the new firm. These circumstances could contribute to a misalignment of CSR performance between the former and the latter firms.

Taken together, we present the following competing hypotheses:

H1a.

For the same job-hopping executives, CSR performance at their subsequent firm is positively influenced by the CSR performance at their former firm.

H1b.

For the same job-hopping executives, CSR performance at their subsequent firm is negatively influenced by the CSR performance at their former firm.

3. Data and method

Our data mainly comes from two sources: CSR data are collected from Hexun, and all other data are sourced from the China Stock Market and Accounting Research Database (CSMAR). Hexun CSR data have emerged as the primary source of CSR data in the literature in recent years (e.g., Hu & Fang, 2022a, b; Yi, Zhang, & Yan, 2021; Zhao & Xiao, 2019; Zhang, Zhang, & Jiang, 2021). The Hexun CSR index is based on 17 secondary and 37 tertiary indicators and aligns with the social responsibility guidelines issued by the Shanghai and Shenzhen Stock Exchanges. It is comparable with international CSR databases such as KLD (Guo & Lu, 2021). Hexun CSR data draws on information from both standalone CSR reports and annual reports, which makes it a comprehensive and reliable source for analysing CSR performance. This is particularly important, as Clarkson et al. (2020) warn against relying solely on CSR data from corporate annual reports or standalone CSR reports of firms. Therefore, the use of Hexun CSR data helps mitigate sample selection bias and enhance the accuracy of the analysis (Tang, Fu, & Yang, 2019). In addition, the Hexun CSR rating process is automated and unbiased, and it is not dependent on subjective evaluations by rating agencies. Another advantage of Hexun CSR is its use of a comprehensive weighted composite index that assesses the social responsibility of firms across five dimensions related to various firm stakeholder groups, namely, shareholders (30%), employees (15%), customers and suppliers (15%), the environment (20%), and society (20%). Additionally, Hexun publishes data pertaining to these CSR dimensions, which enables us to perform finer and more insightful further analysis later in this study. The highest attainable CSR score stands at 100, however, a firm may receive a negative CSR score when its CSR concern points exceed the CSR strength points, according to the evaluation criteria of Hexun (Yan, Wang, Wang, & Chan, 2023). Consequently, negative scores serve as a call to action for companies to improve their CSR performance.

Our sample began in 2010, at the inception of Hexun CSR and finished in 2020. Our sample selection process is primarily driven by the availability of CSR performance data for both the former and the latter firms of job-hopping executives, along with the job-hopping data of executives. Firms are excluded from our sample when there is insufficient data to measure all firm-specific control variables. Initially, our dataset comprises 4,596 firms with available Hexun CSR data, and firm-level and executive-level data from CSMAR are available for 4,153 and 4,333 firms, respectively. After matching CSR, firm-level and executive-level data, our merged sample has 4,076 firms. We then identify a total of 3,756 latter (i.e., subsequent) firms that have executive turnover during our sample period, underscoring the frequency and thus the importance of studying the impact of executive job-hopping. Next, we restrict our sample to firms with at least one incoming executive from other publicly listed firms, or put differently, we exclude firms with incoming executives solely from unlisted firms and those who were promoted internally, as this ensures data availability for the CSR of the former firms. This results in a final sample of 3,419 latter firms. If a latter firm has multiple new executives in a given year, we randomly select only one job-hopping executive for the focal firm to ensure sample representativeness. This setting makes a comparison across firms fairer in the main analysis. In our robustness check section, we also employ data from all job-hopping executives to confirm our results.

After the sample construction procedure described above, we have 11,023 observations for the main analysis. This final sample covers 6,236 unique job-hopping executives from 3,021 former firms [2]. In terms of the number of unique firms (including both former and latter firms), we have 3,463 firms in the final sample. The following model examines whether executives bring their CSR preferences to their new jobs:

(1)CSRLatteri,t+1=αi+β1CSRFormerj,t−1+β2Controlsi,t+1+Firm FE+Year FE+ɛi,t+1

where CSRLatter_i,t+1 captures the CSR performance of the subsequent firm i in year t + 1; in other words, it signifies the second year of the job-hopping executive who joins the latter firm. CSRFormer_j,t−1 captures the CSR performance of the former firm j in year t – 1, the year before the executive changes firms. It is worth highlighting our conservative and rigorous approach to identifying the timing of the job-hopping effect. Given that the exact timing of executive job-hopping in year t is not directly available from the database, measuring the job-hopping effect based on CSR correlation between the previous firm at the year prior to the job-hopping and the subsequent firm at the year after the job-hopping permits a cleaner conclusion of causality, which reduces endogeneity concerns regarding causality and increases the validity of the results. In our later analysis, we demonstrate how the results evolve when we choose CSR at different times to measure the effect of job-hopping by executives.

Following the literature (e.g., Çolak & Korkeamäki, 2021; Gillan et al., 2021), Controls_i,t+1 includes a range of firm-level and executive-level variables that are found to explain CSR performance: firm size (Size), firm financial performance (ROA), firm leverage (Leverage), sales growth (Growth), firm Tobin’s Q (TobinQ), ownership concentration (OwnershipCon), state-owned enterprises (SOE), management ownership (MgmtHolding), CEO power (CEODuality), age of the executive (ExeAge), gender of the executive (ExeGender), education level of the executive (ExeEdu), executives with overseas experience (ExeOverseas), executives with academic experience (ExeAcademic), and executives with financial experience (ExeFinBack). In all analyses, we lag one-year explanatory variables to mitigate reversal causality and include firm and year fixed effects to address endogeneity arising from omitted variable bias. All continuous variables are winsorised at the 1% and 99% levels. Appendix 1 defines the variables in detail.

Summary statistics presented in Panel A of Table 1 show that the mean values of CSR for the subsequent and the prior firms are very close, with only small differences in their distribution and standard deviations. Panel B of Table 1 reports the correlation coefficients of our main variables; notably, there is a positive correlation between CSRFormer and CSRLatter, which preliminarily supports our argument. Furthermore, the correlations between CSRLatter and other control variables are largely significant and align with the findings observed in the literature, therefore confirming the validity of our model.

4. Main results

4.1 Job-hopping effect on CSR

Table 2 reports the job-hopping effect, calculated by using Equation (1) and adding different control variables, namely, firm-level controls, executive-level controls, and all controls in Columns (1) to (3). The results for the same model using different fixed effects, including year and firm fixed effects, and year and industry fixed effects, are reported in Columns (3) to (4) in order [3].

For all specifications, we find that the CSR of subsequent firms is positively related to that of prior firms, which indicates that executives show a persistent pattern in their CSR decisions as they move between firms. This consistent decision-making regarding CSR is imprinted by executives’ past experiences and employment history (Dittmar & Duchin, 2016), leading to a contagion effect in CSR between former and subsequent firms. Regarding economic significance, based on Column (3), which includes all controls and firm and year fixed effects, one standard deviation increase in CSRFormer at year t−1 is associated with a 2% [(0.013 × 18.24)/14.35] standard deviation increase in CSRLatter at year t+1. This economic significance is sizeable given that it only captures the effect of a single executive’s job-hopping on the entire CSR landscape of a firm, and we allow for a conservative estimation of the effect by skipping the job-hopping in year t. The results exhibit consistency across alternative specifications.

As for the results of the control variables, the findings are largely consistent with our expectations. We find higher-CSR performance in bigger, more profitable and less leveraged firms in Column (3). In addition, CSR is significantly lower in firms with weaker governance (i.e., if a CEO also serves as chair of the board). In terms of executive-level characteristics, the education of an executive shows a significant positive correlation with CSR performance.

4.2 Job-hopping effect over time

Our results above support our Hypothesis 1a and indicate that job-hopping by executives leads to a positive correlation between the CSR performance of the former and the latter firms. However, an important follow-up question pertains to the extent to which this job-hopping effect persists over time. This inquiry is of particular relevance to firms and their stakeholders who seek to understand the long-term trajectory of firm CSR performance after executive migration, in order to facilitate better governance and monitoring.

We examine three specific questions along this line. First, how far back can the CSR practice of the prior firm influence the CSR of the subsequent firm? Our analysis in Section 4.1 is based on CSRFormer_t−1, and in this section, we ask if earlier CSR performance (e.g., CSRFormer_t−2) is still relevant. Second, how long does the job-hopping effect last? Alternatively, does CSRFormer_t−1 predict CSRLatter beyond year t+1 (the main analysis is based on CSRLatter_t+1)? Third, if there is a causal relationship between CSRFormer and CSRLatter as a result of executive job-hopping, we should observe no relationship after the job-hopping occurs. For instance, CSRFormer_t+1 and CSRLatter_t+1 should exhibit no correlation in this scenario. This absence of a causal link would be because the job-hopping executives, who served as the connecting factor influencing the CSR performance of both firms, had departed from their former companies. Consequently, their influence disappears, particularly when executives have been away from the former firm for longer. Similarly, prior to the job-hopping that links the two firms (i.e., the former and latter firms), their CSR performance should not be correlated. In other words, CSRFormer_t−1 and CSRLatter_t−1 should demonstrate no correlation. Answers to these questions can provide valuable information to firm stakeholders for making hiring decisions.

Table 3 presents the relevant results. Panel A replicates the main analysis by using CSRFormer_t−2 (CSR of the former firm two years before job-hopping) to predict the CSR of the subsequent firm in the year of the job-hopping (CSRLatter_t), as well as two years before (e.g., CSRLatter_t−2 and CSRLatter_t−1) and after the job-hopping year (e.g., CSRLatter_t+1 and CSRLatter_t+2). Similarly, Panels B to E show the results of the same analysis using CSRFormer_t−1, CSRFormer_t, CSRFormer_t+1, CSRFomer_t+2 to predict CSRLatter at the five different time points (from year t−2 to year t+2) relative to job-hopping. We choose a five-year window for both CSRFormer and CSRLatter, because the average tenure of the executives in our sample is 2.87 years (that is, roughly from t to t+2) [4]

The results reveal several interesting findings. Panel A displays that CSRFormer_t−2 has no significant predictive ability, regardless of the year of CSRLatter. However, Panels B and C yield different results, showing that both CSRFormer_t−1 and CSRFormer_t predict CSRLatter_t and CSRLatter_t+1, but not CSRLatter_t+2. Taken together, the results reveal that the CSR performance of the former firm has a significant impact on the CSR performance of the subsequent firm only in the year before (year t − 1) and the year of (year t) job-hopping by an executive. Notably, CSRFormer_t exhibits a higher correlation with CSRLatter and CSRLatter_t+1 than CSRFormer_t−1 (e.g., 0.034 versus 0.018 in Column 3, and 0.017 versus 0.013 in Column 4 of Table 3), which demonstrates the robustness of our main conclusion reported in Section 4.1, even after removing the one-year gap. In fact, the results are even more pronounced when the gap is removed.

As anticipated, the results align with the complexity of CSR, which is influenced by a multitude of factors, including internal factors such as board composition and firm strategy (Rao & Tilt, 2016; Uyar, Koseoglu, Kuzey, & Karaman, 2022), and external factors – those related to government, business and society (Vashchenko, 2017). As these variables exhibit disparities over time, CSRFormer follows a dynamic pattern that is a joint outcome of these influencing factors. According to the availability heuristic hypothesis, people rely on immediate memories and examples when they make decisions. Consequently, CSRFormer_t and CSRFormer_t−1 hold greater relevance for CSRLatter for up to two years after job-hopping. In other words, when executives move to a new firm, only their most recent CSR practices and influence tend to persist and have a noticeable effect. This finding provides an answer to our first question posed above in this section.

Next, for our second question, we find that this significant executive job-hopping effect on CSRLatter lasts for up to two years (i.e., t+1) after the executive’s migration to the subsequent firm, before waning and eventually vanishing three years (i.e., t+2) after the executive joins the subsequent firm, as indicated by the monotonically decreasing coefficients of CSRFormer over time (e.g., 0.018 for CSRLatter_t versus 0.013 for CSRLatter_t+1 in Panel B, and 0.034 for CSRLatter_t versus 0.017 for CSRLatter_t+1 in Panel C). The results point to the transient nature of the effect of job-hopping by executives on CSR performance. This observation again underscores the intricate dynamics of CSR, where the job-hopping effect is most discernible in the first two years after job-hopping by an executive, before being overshadowed by other dynamic CSR determinants. This pattern is also in line with the average tenure of the executives in our sample, as some executives may have departed, leading to a reduction in their influence.

With respect to the third question, we find that prior to job-hopping by executives, and irrespective of the specific year being examined, the CSR performance of the executive’s former firm does not have the power to predict CSRLatter_t−2 and CSRLatter_t−1, as indicated by the insignificant coefficients of the CSRFormer reported in Columns (1) and (2) of in Table 3. This corroborates our main results, that the link between CSR performance of the executive’s former and latter firms is only present when job-hopping happens. In addition, in Column 5, after job-hopping, CSRFormer_t+1 and CSRFormer_t+2 show no predictive ability for CSRLatter, as indicated in Panels D and E. As executives spend more time away from their former firm, their influence diminishes, resulting in the absence of the relationship between CSRFormer and CSRLatter. These results lend strong support to the causal effect of job-hopping by executives on firm CSR performance.

Our results in this temporal analysis have important implications. We highlight that CSR performance can be shaped by multiple factors, including those with short-lived effects, as demonstrated in this study, and others that trigger deeper, long-lasting effects, such as social norms, culture and legal and regulatory compliance (Cai, Pan, & Statman, 2016). It is important for firms to respond effectively to short-term CSR determinants, thereby ensuring sustained, long-term commitment to social and environmental responsibilities. This area remains inadequately explored, which contributes to a gap in our comprehension of how CSR practices adapt and evolve in response to both transient and enduring factors to achieve sustained and impactful CSR performance. This finding underscores the need to develop comprehensive and forward-thinking CSR strategies.

4.3 Job-hopping shock: PSM-DiD

Although we made various efforts to consider potential endogeneity concerns in our empirical setting, including using lagged independent variables, controlling for firm, year and industry fixed effects, and randomly selecting an executive for each firm, there is still a possibility that endogeneity may persist. To address the remaining endogeneity concerns further, we employ an alternative method to quantify the effect of job-hopping by executives on CSR using a PSM-DiD framework.

In this alternative setting, we take CEO job-hopping as a shock to the CSR performance of firms. We narrow our focus to CEOs (as opposed to all top executives, as described in Section 3) owing to their chief role in shaping firm practices. A change in CEO is more likely to impose a significant shock, making it an essential event to examine. Additionally, analysing CEO job-hopping serves as a validation of our results with a different sample. Further, we concentrate on CEOs because CEO job-hopping occurs much less frequently than executive job-hopping, which enables a cleaner identification. Specifically, we argue that when a CEO comes from a socially responsible firm, it poses an exogenous positive shock to the CSR of the subsequent firm. While the OLS (ordinary least squares) model in Section 4.1 compares the CSR performance of former and subsequent firms, the PSM-DiD analysis identifies the average CSR performance change for firms that experience a CEO job-hopping event compared with firms that do not experience such an event. Following Kubick and Lockhart (2017), we perform the following analysis:

(2)CSRi,t=α+β1POST+β2Controlsi,t+YearFE+FirmFE+εi,t,

where POST equals 1 if firm i has an incoming CEO from a firm with greater CSR (i.e., CSRFormer > CSRLatter) and 0 otherwise. In the analysis, firms that experience such a shock during our sample period are classified as the treatment group; all other firms that have data available for our analysis are allocated to the control group. Therefore, β₁ captures the effect of the incoming CEO on the CSR performance of the firm. Following Mishra (2014), we employ nearest neighbour matching for the control and treatment observations to enhance the validity of our analysis before conducting a DiD analysis. We report the PSM balance test results in Appendix 2, where the t-test results show that the differences between the treated and the control samples after matching are all insignificant, indicating that the matching effect is effective.

Next, the PSM-DiD results are reported in Table 4. As expected, the coefficient of POST is positively significant, indicating that an incoming CEO from a higher-CSR firm effects a positive shock to the subsequent firm’s CSR, thereby further supporting our key hypothesis.

5. Further analyses and robustness check

In Section 4, we document evidence that, as executives switch jobs, CSR performance is similar from one firm to the next. Building on these similarities, we argue that the effect of job-hopping by executives may depend on various factors, such as different CSR dimensions, the specific types of job-hopping, and the executive and characteristics of the subsequent firm.

5.1 CSR dimensions

The analyses above are based on a composite CSR score that includes five dimensions related to five different firm stakeholder groups: shareholder responsibility, employee responsibility, consumer responsibility, environmental responsibility and social responsibility. Tsang et al. (2021) report that the positive association between CSR and firm value is sensitive to the selection of CSR dimension [5] This leads to a question about whether our main results are also affected by the choice of CSR dimensions. Specifically, we seek to understand (1) which dimension (i.e., stakeholder group) of CSR benefits the most from the executive’s job-hopping; (2) in what dimension does the job-hopping effect have a greater influence on the overall CSR of the latter firm; and (3) can the job-hopping effect be transmitted on a dimension-to-dimension basis? To address these issues, we rerun Equation (1) for different CSR dimensions.

Table 5, Panel A shows results that examine the relationship between the CSR dimensions of the latter firm and the overall CSR performance of the former firm. First, the results demonstrate that the effects of CSR spillovers that arise from job-hopping by executives are significant for dimensions such as shareholders, employees, consumers and the environment, which are associated with specific stakeholder groups. However, the effect of job-hopping is not evident for the broader and more general social responsibility dimension. The social responsibility dimension involves more abstract values and beliefs that are more difficult to measure, such as human rights and community involvement, and it is not directly correlated to specific stakeholders, unlike other dimensions. In addition, the spillover effects of this dimension may require a broader cultural shift within a firm. As a result, the spillovers from the social responsibility dimension pose greater challenges for transfer through job-hopping. Therefore, for stakeholders seeking to influence CSR in the social dimension, job-hopping by executives may not be an effective channel. These results also suggest that firms could prioritise CSR dimensions differently across industries. Overall, they indicate that our main results are not susceptible to the choice of CSR dimension, which underscores the robustness of our results.

Second, it is worth noting that the degree of impact caused by job-hopping varies for different stakeholders, with the most notable effects being observed in the areas of consumer and environmental responsibility (with coefficients of 0.022 and 0.021, respectively). This observation is logical, as consumer and environmental responsibility metrics typically encompass common responsibilities such as product quality, customer service, environmental protection and investment in environmental initiatives. Consequently, managerial CSR expertise in these two dimensions can be readily transmitted. In turn, shareholder and employee responsibilities are more closely tied to the internal operations and management of a firm, for instance, financial management and employee benefits. The effects of job-hopping by executives on these dimensions may therefore depend on the specific organisational context and the relationships between the job-hopping executive and the firm’s shareholders and employees. As a result, the effects of job-hopping on these two dimensions may be less pronounced than its effects on consumer and environmental responsibilities, which are more outward facing.

In Panel B, we segment CSRFormer into its five dimensions and observe the continued presence of the job-hopping effect. We find again that the CSR dimensions of the former firm significantly affect the CSR of the latter firm (CSRLatter), with the exception of the social dimension. In addition, the employee dimension exerts a greater effect on CSRLatter as a consequence of job-hopping by executives (with a coefficient of 0.068). Executives from firms with a stronger track record of employee wellbeing tend to bring with them the ability to foster a culture centred on employee welfare at their subsequent firms. This focus on the employee dimension of CSR aligns with sustainable business practices that yield long-term benefits for the company, which ultimately enhances the overall CSR performance (Harter, Schmidt, & Keyes, 2003; Bučiūnienė & Kazlauskaitė, 2012) [6]

In line with our approach of utilising the five dimensions of CSR to provide the best insights, we delve into Panel C to scrutinise the executive job-hopping effect across each of the CSR dimensions individually to determine its presence. In a similar vein, we find a significant and positive influence on each dimension, except for the social dimension, thereby affirming the results presented in Panels A and B and reinforcing the confidence in our results.

5.2 Job-hopping attributes

5.2.1 CSR level of prior firm

Up to this point, our analyses have not indicated whether the relationship between the CSR performance of the former and subsequent firms is contingent on the CSR performance level of the former firm.

Intuitively, we assess the strength of the job-hopping effect based on the former firm’s relative CSR performance in its industry. If the former firm is an industry leader in terms of CSR, would the job-hopping effect be stronger? Specifically, we evaluate the quality of the former firm’s CSR performance by benchmarking it against the industry average CSR performance of the former firm (CSRFormerIndAvg). The results are provided in Columns (1) and (2) in Panel A of Table 6. We find that the significant positive relationship of the CSR performance of the former and latter firms holds when the CSR performance of the former firm is above the industry average for CSR performance (CSRFormer > CSRFormerIndAvg). This outcome is intuitive because executives from industry-leading CSR firms typically bring with them specialised skills and expertise in CSR practices. When they move to new roles, their elevated expectations for CSR may prompt them to champion and insist on the implementation of CSR initiatives at their subsequent firms. This, in turn, fosters a culture of heightened CSR commitment and performance.

5.2.2 Direct versus gap job-hopping

According to Marquis and Tilcsik (2013), an individual’s behaviour becomes less persistent over time. In other words, the influence of their previous experiences on their behaviour decreases gradually. In support of this idea, Edin and Gustavsson (2008) find that taking a break from the job market can lead to a decline in executives’ skills. Similarly, Chen, Zhu, and Yu (2017) report that the persistence of executives’ earnings management behaviours decays as the employment gap between two firms increases. Based on these findings, we argue that job-hopping executives who have a long employment gap are less likely to have their decision-making at new (i.e., subsequent) firms influenced by their prior experiences. In Panel A of Table 6, we report the results in Columns (3) and (4), where we compare the effects of direct job-hopping (i.e., no gap between two jobs) and gap job-hopping (i.e., up to 5 years gap between two jobs). Our findings reveal that the effect is only significant for direct job-hopping, which accounts for 98% of the sample, which suggests that the decisions of executives with long employment gaps are influenced less by their experience at their former firm. These results further support our finding that the effect is mostly influenced by experiences just prior to job-hopping.

5.2.3 Job-hopping with an increase in pay versus without an increase in pay

Job-hopping by executives is often accompanied by a substantial salary increase (Topel & Ward, 1992). Additionally, the literature shows that executive pay is associated with CSR performance (e.g., Deckop, Merriman, & Gupta, 2006). In light of these findings, we investigate the effect of pay rises of executives who had job-hopped on the relationship between the CSR performance of the executive’s former and subsequent firms. We argue that job-hopping executives who do not receive pay increases are more likely to pay attention to non-monetary firm performance metrics, such as a commitment to CSR values and practices, hence, reflecting a greater job-hopping effect on CSR. For example, executives who transition from a former high-CSR firm without receiving pay rises may be more inclined to seek intrinsic rewards based on personal satisfaction (e.g. doing the right thing) and recognition for their contribution to CSR. This can create a positive cycle that strengthens their commitment to CSR values and practices and results in improved CSR performance of their subsequent firms. Rekker, Benson, and Faff (2014) provide evidence that intrinsically motivated CEOs are prepared to trade off some of their income in exchange for the satisfaction of leading a firm that acts in a socially responsible manner. In contrast, job-hopping that is accompanied by a pay rise may be driven more by extrinsic rewards, such as financial incentives, because executive compensation packages are typically conditional on firm financial outcomes, which can lead to a focus on short-term financial gains rather than long-term CSR performance. In this case, the job-hopping effect on CSR can be weak. Therefore, we examine the job-hopping effect separately for those executives who receive pay rises, and those who do not. The results are presented in Columns (5) and (6) in Panel A of Table 6. As expected, we observe that the CSR performance of the subsequent firm is positively related to that of the former firm for job-hopping executives who do not receive pay rises.

5.2.4 Job-hopping within the same city versus a transfer to a different city

Tambe and Hitt (2014) show that co-locating with other IT-intensive firms boosts productivity, because it makes it easier for technical workers to move within the region. This discovery prompts an interesting inquiry: does the location of job-hopping by executives also affect inter-firm CSR performance? We collect geographic location data of job-hopping executives, specifically the location of firm headquarters, and use it to analyse two subsamples in Equation (1): job-hopping within the same city and between cities. Columns (7) and (8) in Panel A of Table 6 reveal a significant positive relationship between the CSR performance of former and subsequent firms when job-hopping takes place within the same city. This highlights that the effects of job-hopping are amplified over shorter distances, which aligns with the findings of the literature on the role of geography in knowledge transfer. When an executive moves to a new firm within the same city, they often bring with them their established network of contacts and relationships. This network can include important stakeholders, such as suppliers, customers and regulators that can significantly affect a firm’s CSR performance. By leveraging these existing connections, executives can quickly build relationships that can ultimately benefit the CSR efforts of their subsequent firms.

5.3 Attributes of executives

Upper echelon theory proposes that executive characteristics shape their decision-making (Hambrick & Mason, 1984). There is substantial evidence to confirm the significance of management background and personal traits for CSR performance (see Section 2.2). In addition, our study is based on the argument that executives’ prior experiences influence their decision-making processes, leading to some degree of similarity and stability in their decisions. However, the degree of persistent behaviour may also be influenced by the unique characteristics of individuals and organisations (Marquis & Tilcsik, 2013). This motivates us to investigate whether the characteristics of job-hopping executives play a role in the CSR performance of their former and subsequent firms. Panel B of Table 6 provides compelling evidence to support our conjecture. The results show that older executives (Columns (1)), and those who have worked overseas (Columns (3)), exert a significant effect on inter-firm CSR performance. These executives gained a wealth of experience and diverse knowledge throughout their careers, and it enables them to have a greater influence on their subsequent firms. Our results therefore underline the importance of considering the background and experience of executives when making hiring decisions in relation to CSR.

In addition, we partition our sample into two subsamples based on the gender of the executives, and rerun Equation (1). The results in Column (5) of Panel B in Table 6 indicate a stronger job-hopping effect in firms with female executives. This aligns with the literature, which suggests that female executives have a greater tendency to prioritise CSR than their male counterparts (e.g., Zou, Wu, Zhu, & Yang, 2018). Consequently, the job-hopping effect is more pronounced in firms led by female executives.

Furthermore, the existing literature indicates that CEOs, in contrast to other firm executives, wield dominant power and exert prominent influence on executive boards (Finkelstein, 1992). Consequently, they play a central role in shaping corporate policies, strategies and performance (Mackey, 2008). Therefore, we expect that the effect of job-hopping will be more pronounced for CEOs than for their non-CEO counterparts. To investigate this, we divide our sample into CEO and non-CEO subsamples and reevaluate Equation (1). Our results, in Columns (7) and (8) of Panel B in Table 6, reveal that the positive effect of executive job-hopping persists for both groups, while the coefficient in the CEO subsample is greater than that in the non-CEO subsample (0.043 versus 0.038), which suggests that the job-hopping effect extends beyond CEOs, though it is greater for CEOs. This observation is consistent with our expectations, particularly given the highly discretionary nature of CSR decisions.

5.4 Attributes of firms

We postulate that the influence of a prior firm on the CSR performance of a subsequent firm is not monotonic, as firm-level characteristics affect CSR performance (Adams, 2002), as well as the extent of the job-hopping effect. Panel C of Table 6 offers supporting evidence for our proposition. The results reveal that the job-hopping effect is contingent on several characteristics of the subsequent firm. First, the effect of job-hopping may increase with diverse firm ownership (Dam & Scholtens, 2013). The results in Columns (1) to (6) of Panel C are consistent with this view. Large firms, firms with dispersed ownership and non-SOEs usually have a greater number of owners and are more concerned about CSR (Chen & Metcalf, 1980; Liu, Bredin, Wang, & Yi, 2014; Marquis & Qian, 2014). In addition, in these firms, large shareholders are less likely to restrict managers from making changes and decisions. Consequently, when executives transition to these firms, they can transfer CSR knowledge and practices to the new firms, which leads to a greater job-hopping effect and more effective integration of CSR into the operations of the subsequent firms.

Second, Columns (7) through (10) of Panel C reveal that the job-hopping effect is significant for firms with poor financial performance (e.g., high leverage, low Tobin’s Q). Poorly performing firms are often more motivated to seek changes to improve their financial outcomes. On the one hand, the results may suggest that the insurance effect of CSR is more salient for these firms, because doing good can mitigate the negative effects of poor performance (Minor & Morgan, 2011). Poor financial performance can harm a firm’s reputation and cause stakeholders to withdraw their support. To mitigate the risk of financial distress, subsequent firms may support migrating executives who have experience in implementing CSR activities (e.g., those who move from a high-CSR firm). Such firms have a greater incentive to engage in CSR activities, which could potentially serve as a form of ‘window dressing’ to offset poor financial performance. On the other hand, poorly performing firms may cut back on CSR investments and redirect resources towards improving financial performance when they hire an executive from a poorly performing CSR firm. In either situation, a positive correlation between the CSR of former and subsequent firms can be identified.

Third, Columns (11) to (14) in Panel C show that executives in firms with managerial ownership or powerful CEOs (i.e., CEODuality = 1) have a greater ability to make discretionary decisions (e.g., CSR); thus, executives’ previous experiences, preferences, knowledge and resources have a greater chance of being carried over to subsequent firms. The results are in line with the research of Finkelstein and Boyd (1998), who find a positive relationship between the strength of executive discretion and its impact on the firm.

5.5 Additional robustness check

We also assess the robustness of our results through a series of tests. First, we adopt two alternative measures to proxy the CSR performance of the firms where job-hopping executives were previously employed, namely CSRFormerSingle and CSRFormerAvg. The main results reported in Table 2 are based on the fact that when a firm has multiple job-hopping executives in a given year, we randomly select one job-hopping executive to capture the CSR performance of the former firm. A natural question is, do the results persist when we narrow our focus to firms with only one new (i.e., incoming) executive in a given year? In this context, CSRFormerSingle measures the CSR performance of former firms under such conditions. Next, related to this, if we take into account the collective impact of all newly appointed executives in a given year, would our results change? To capture this average effect, we employ CSRFormerAvg, which calculates the average CSR performance of all former firms in such a scenario.

We then replace CSRFormer with these two measures while retaining all control variables (averaging the control variables when CSRFormerAvg is used in the regression), and rerun Equation (1). Table 7 presents the results of these two alternative measures. We document a consistent effect of job-hopping by executives (the coefficients of CSRFormerSingle and CSRFormerAvg are 0.059 and 0.026, p < 0.05 and 0.1 respectively).

Second, we conduct a placebo test to assess the robustness of our results. The test randomly assigns two firms from our sample as a pair, and then we rerun Equation (1) to examine whether their CSR scores are correlated. We repeat this process 1,000 times and plot the distribution of the placebo estimates in Figure 1. If executive job-hopping influences CSR, we should see no correlation between firm CSR in the randomly matched pair. Figure 1 shows that the placebo estimates are not statistically different from zero, which confirms that our main results are not driven by unobserved or endogenous factors; rather, they demonstrate the effect of job-hopping on the transmission of CSR.

Third, the literature documents the limitations of the PSM method. Cram, Karan, and Stuart (2009) point out three threats to matched-sample studies in accounting research: (1) use of unconditional analysis, when analysis conditional on effects of the matching variables is needed; (2) failure to control for effects of imperfectly matched variables; and (3) failure to reweight observations according to different sampling rates. Heckman, Ichimura, Smith, and Todd (1998) find that selection bias could arise from drawing conclusions based on nonexperimental comparison groups. Shipman, Swanquist, and Whited (2017) document a substantial increase in the use of PSM in accounting research. They also emphasise that important design choices and implementing PSM in a theoretically consistent manner could affect the power of the test. To address concerns in this regard further, we apply the entropy balancing method developed by Hainmueller (2012) to reevaluate our DiD analysis. This method processes data before estimating the treatment effect. It leverages a maximum entropy reweighting technique to ensure that the reweighted treatment and control groups meet predefined balance criteria. This approach offers a distinct advantage over continual balance checking and PSM, which may stochastically balance the covariate distributions. The results of this analysis are provided in Table 8. Overall, our DiD analysis using the entropy balancing method remains robust, providing an extra layer of confidence to our research findings.

Last, we perform additional tests to ensure the robustness of our results. These tests exclude special treatment firms that report losses for two consecutive years and firms in the financial industry. After these tests, our results remain consistent (results are not tabulated, for brevity).

6. Conclusion

This study finds that job-hopping executives play an important role in explaining the positive relationship between CSR performance at their prior and subsequent firms. However, our finding is that this positive relationship is of short duration and disappears at two years after an executive joins the subsequent firm. In addition, only the CSR performance of the prior firm up to one year before the job-hopping has predictive power for the CSR of the executive’s subsequent firm. Furthermore, this relationship benefits various CSR stakeholder groups and is contingent on the characteristics of job-hopping and the executive, and attributes of the subsequent firm. The results suggest that job-hopping is a conduit through which executives’ prior knowledge, experiences and skills facilitate the transfer of CSR profiles between firms, thereby supporting the upper echelon theory. Furthermore, we demonstrate that spillover effects of job-hopping by executives are robust when we use CEO job-hopping as a shock to compare its effect on those firms that experience no CEO turnover.

Our study has several important practical implications. First, the study sheds light on the importance of knowledge transfer and learning in the context of CSR. Specifically, our findings suggest that firms that wish to improve their CSR performance can benefit from hiring executives who have a track record of success in CSR at their previous firms. This knowledge transfer can help the subsequent firm to build on its CSR practices and achieve better outcomes. However, it is equally important that hiring an executive from a firm with poor CSR performance can result in a significant decline in the CSR performance of the hiring firm.

Second, our study provides insights into how top executives convey their earlier CSR practices and experiences gained at their former firms to a new institutional environment (i.e., subsequent firms). By examining how past experiences of job-hopping executives influence their CSR decision-making in their current role, we may be able to improve predictions of the likelihood of good/poor CSR practices in their subsequent firms. This understanding can help organisations make informed decisions about executive hiring and to anticipate the potential impact on CSR performance.

Third, investigating how job-hopping by executives affects inter-firm CSR performance can help investors and stakeholders comprehend the impact of executive migration on the commitment of a firm to social responsibility. Our results indicate the specific stakeholder group that is most susceptible to the job-hopping effect on a firm’s CSR performance. This understanding can inform investment decisions and assist stakeholders in evaluating the long-term sustainability of a firm.

Several questions would be interesting to pursue in further work. If data are available, it would be useful to develop a more comprehensive understanding of how the effect of job-hopping executives on inter-firm CSR performance is influenced by forced versus voluntary job switches. A further interesting area of research would be comparing the effect of job-hopping executives to that of managerial social networks in terms of their influence on CSR. By investigating the relative strengths of these factors, we can gain a deeper understanding of the dynamics at play in inter-firm relationships and how they impact CSR performance.

Figures

Figure 1

Placebo test results

Table 1

Summary statistics and correlation matrix

Variables		N	Mean	SD	Min	Median	Max
Panel A: Summary Statistics
CSR variables	CSRLatter	11,023	25.87	14.35	−6.72	22.98	90.87
	CSRFormer	11,023	26.46	18.24	−14.89	22.22	90.87
Firm variables	Size	11,023	9.673	0.636	7.613	9.573	13.25
	ROA	11,023	0.055	0.064	−0.014	0.043	4.489
	Leverage	11,023	0.438	0.208	0.022	0.433	1.009
	Growth	11,023	0.079	0.445	−9.55	0.057	25.83
	TobinQ	11,023	2.661	2.351	0.594	1.921	38.21
	OwnershipCon	11,023	60.77	15.34	8.975	61.83	98.59
	SOE	11,023	0.229	0.420	0	0	1
	MgmtHolding	11,023	0.564	0.496	0	1	1
	CEODuality	11,023	0.256	0.437	0	0	1
Executive variables	ExeAge	11,023	52.21	8.113	25	52	83
	ExeGender	11,023	0.840	0.367	0	1	1
	ExeEdu	11,023	0.707	0.455	0	1	1
	ExeOveseas	11,023	0.143	0.35	0	0	1
	ExeAcademic	11,023	0.482	0.50	0	0	1
	ExeFinBack	11,023	0.221	0.415	0	0	1

Variables	CSR Latter	CSR Former	Size	ROA	Leverage	Growth	TobinQ	OwnershipCon	SOE	Mgmt Holding	CEO Duality	Exe Age	Exe Gender	Exe Edu	Exe Overseas	Exe Academic	Exe FinBack
Panel B Correlation Matrix
CSRLatter	1
CSRFormer	0.148***	1
Size	0.297***	0.148***	1
ROA	0.160***	−0.015	−0.114***	1
Leverage	0.096***	0.079***	0.513***	−0.290***	1
Growth	0.000	0.000	−0.016*	0.032***	0.044***	1
TobinQ	−0.046***	−0.041***	−0.441***	0.276***	−0.338***	0.042***	1
OwnershipCon	0.127***	0.043***	0.206***	0.157***	−0.023**	−0.016*	0.031***	1
SOE	0.086***	0.055***	0.199***	−0.035***	−0.153***	−0.007	−0.055***	0.168***	1
MgmtHolding	−0.069***	−0.061***	−0.242***	0.121***	0.128***	−0.012	0.130***	−0.049***	−0.149***	1
CEODuality	−0.081***	−0.048***	−0.175***	0.093***	−0.247***	−0.002	0.148***	0.045***	−0.102***	0.208***	1
ExeAge	0.002	0.048***	0.095***	−0.015	0.042***	−0.012	−0.062***	0.033***	0.016*	−0.011	−0.032***	1
ExeGender	0.018*	0.032***	0.039***	−0.010	0.038***	0.001	−0.002	0.042***	0.035***	−0.024**	−0.012	0.061***	1
ExeEdu	−0.009	−0.025***	−0.056***	0.066***	−0.111***	−0.024**	0.079***	0.107***	−0.033***	0.328***	0.110***	−0.023**	−0.012	1
ExeOveseas	0.051***	0.057***	0.058***	0.003	0.011	−0.007	0.026***	0.019**	−0.018*	0.014	0.015*	0.029***	0.028***	0.060***	1
ExeAcademic	−0.011	−0.017*	−0.055***	0.026***	−0.057***	−0.010	0.027***	−0.017*	−0.058***	0.064***	0.036***	0.257***	−0.034***	0.063***	0.180***	1
ExeFinBack	0.086***	0.085***	0.122***	0.019**	0.066***	0.000	−0.005	0.034***	0.026***	−0.029***	0.012	−0.071***	0.045***	0.064***	0.064***	−0.065***	1

Note(s): Table 1 reports the summary statistics and correlations of the variables. ***p < 0.01, **p < 0.05, *p < 0.1

Source(s): Table 1 by authors

Table 2

Executive job-hopping effect on CSR performance

Variables	(1)		(2)		(3)		(4)
	Firm controls		Executive controls		All controls		All controls
	CSRLatter		CSRLatter		CSRLatter		CSRLatter
CSRFormer	0.012*	(0.01)	0.014**	(0.01)	0.013**	(0.01)	0.033***	(0.01)
Size	6.547***	(1.58)			6.481***	(1.57)	7.274***	(0.63)
ROA	24.911*	(14.42)			24.667*	(14.27)	38.004*	(20.19)
Leverage	−5.255**	(2.35)			−5.275**	(2.33)	−5.694***	(1.96)
Growth	0.229	(0.45)			0.233	(0.45)	0.030	(0.10)
TobinQ	−0.221	(0.15)			−0.233	(0.15)	0.018	(0.14)
OwnershipCon	−0.041	(0.03)			−0.045	(0.03)	0.041***	(0.02)
SOE	−0.519	(0.81)			−0.517	(0.81)	−0.055	(0.58)
MgmtHolding	1.089	(0.84)			0.982	(0.83)	1.042**	(0.48)
CEODuality	−1.485**	(0.73)			−1.479**	(0.73)	−0.778*	(0.45)
ExeAge			−0.011	(0.02)	−0.010	(0.01)	−0.026*	(0.02)
ExeGender			0.343	(0.29)	0.354	(0.28)	0.062	(0.33)
ExeEdu			1.699***	(0.49)	1.501***	(0.49)	−0.176	(0.38)
ExeOveseas			−0.121	(0.33)	−0.149	(0.32)	0.439	(0.43)
ExeAcademic			−0.142	(0.25)	−0.147	(0.25)	0.157	(0.25)
ExeFinBack			−0.231	(0.32)	−0.252	(0.31)	0.001	(0.38)
Constant	−33.877**	(14.50)	24.745***	(0.89)	−33.621**	(14.54)	−46.630***	(5.51)
Year FE	Yes		Yes		Yes		Yes
Firm FE	Yes		Yes		Yes		No
Industry FE	No		No		No		Yes
N	11,023		11,023		11,023		11,625
R2	0.706		0.699		0.706		0.320

Note(s): This table reports the results of regressing the CSR of the subsequent firm (CSRLatter) in year t + 1 on the lagged CSR of the former firm in year t − 1 (CSRFormer) and the control variables in year t + 1. Robust standard errors clustered by firm and year are in parentheses. ***p < 0.01, **p < 0.05, *p < 0.1. All variables are defined in Appendix 1

Source(s): Table 2 by authors

Table 3

Timeline results

Variables	(1)	(2)	(3)	(4)	(5)
	CSRLattert−2	CSRLattert−1	CSRLattert	CSRLattert+1	CSRLattert+2
Panel A: CSRFormert−2
CSRFormert−2	0.005	0.010	0.004	0.009	0.006
	(0.007)	(0.007)	(0.007)	(0.006)	(0.006)
All controls	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes
Firm FE	Yes	Yes	Yes	Yes	Yes
N	7,404	8,139	9,173	10,143	7,627
R2	0.795	0.766	0.739	0.717	0.731
Panel B: CSRFormert−1
CSRFormert−1	0.011	0.008	0.018**	0.013**	0.000
	(0.007)	(0.007)	(0.007)	(0.006)	(0.007)
All controls	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes
Firm FE	Yes	Yes	Yes	Yes	Yes
N	7,782	9,063	9,945	11,023	8,463
R2	0.797	0.767	0.728	0.706	0.727
Panel C: CSRFormert
CSRFormert	0.009	0.015	0.034***	0.017**	−0.004
	(0.006)	(0.009)	(0.008)	(0.007)	(0.007)
All controls	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes
Firm FE	Yes	Yes	Yes	Yes	Yes
N	7,969	9,227	9,900	10,964	7,445
R2	0.797	0.770	0.726	0.702	0.739
Panel D: CSRFormert+1
CSRFormert+1	0.007	0.013	0.025	−0.002	0.007
	(0.006)	(0.009)	(0.014)	(0.008)	(0.005)
All controls	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes
Firm FE	Yes	Yes	Yes	Yes	Yes
N	7,895	9,251	9,492	10,049	7,797
R2	0.801	0.767	0.719	0.711	0.735
Panel E: CSRFormert+2
CSRFormert+2	0.008	0.006	0.009	0.003	0.005
	(0.008)	(0.007)	(0.008)	(0.006)	(0.007)
All controls	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes
Firm FE	Yes	Yes	Yes	Yes	Yes
N	7,505	8,962	9,365	10,022	7,780
R2	0.791	0.766	0.720	0.712	0.733

Note(s): This table reports the results of regressing the CSR of the subsequent firm (CSRLatter) on the CSR of the former firm (CSRFormer) and the control variables in different periods. Robust standard errors clustered by firm and year are in parentheses. ***p < 0.01, **p < 0.05, *p < 0.1. All variables are defined in Appendix 1

Source(s): Table 3 by authors

Table 4

Difference-in-differences analysis

Variables	CSRLatter
Post	2.398***	(0.840)
Size	5.771***	(0.929)
ROA	14.074*	(7.513)
Leverage	0.803	(0.597)
Growth	2.315*	(1.207)
TobinQ	−0.083	(0.223)
OwnershipCon	0.054**	(0.021)
SOE	1.154***	(0.447)
MgmtHolding	0.207	(0.472)
CEODuality	0.244	(0.433)
ExeAge	−0.014	(0.029)
ExeGender	−0.64	(0.688)
ExeEdu	1.035*	(0.572)
ExeOverseas	0.257	(0.731)
ExeAcademic	0.383	(0.540)
ExeFinBack	−0.11	(0.606)
Constant	−36.388***	(9.226)
Year FE	Yes
Firm FE	Yes
N	14,903
R2	0.600

Note(s): This table reports the DiD results using CEO job-hopping event as a shock. POST equals 1 if firm i has an incoming CEO from a firm with greater CSR (i.e., CSRFormer > CSRLatter) and 0 otherwise. Robust standard errors clustered by firm and year are in parentheses. All other variables are defined in Appendix 1. ***p < 0.01, **p < 0.05, *p < 0.1

Source(s): Table 4 by authors

Table 5

Further analyses by CSR dimensions

Variables	(1)		(2)		(3)		(4)		(5)
	ShareholderLatter		EmployeeLatter		ConsumerLatter		EnvironmentalLatter		SocialLatter
Panel A: CSRFormer vs CSRLatter dimensions
CSRFormer	0.011**	(0.005)	0.016**	(0.007)	0.022***	(0.008)	0.021**	(0.009)	−0.001	(0.005)
All controls	Yes		Yes		Yes		Yes		Yes
Year FE	Yes		Yes		Yes		Yes		Yes
Firm FE	Yes		Yes		Yes		Yes		Yes
N	11,023		11,023		11,023		11,023		11,023
R2	0.831		0.705		0.632		0.608		0.823

Variables	(1)		(2)		(3)		(4)		(5)
	CSRLatter		CSRLatter		CSRLatter		CSRLatter		CSRLatter
Panel B CSRFormer dimensions vs CSRLatter
ShareholderFormer	0.028*	(0.015)
EmployeeFormer			0.068**	(0.029)
ConsumerFormer					0.045**	(0.019)
EnvironmentFormer							0.042**	(0.019)
SocialFormer									0.015	(0.019)
All controls	Yes		Yes		Yes		Yes		Yes
Year FE	Yes		Yes		Yes		Yes		Yes
Firm FE	Yes		Yes		Yes		Yes		Yes
N	11,023		11,023		11,023		11,023		11,023
R2	0.711		0.711		0.711		0.711		0.711

Variables	(1)		(2)		(3)		(4)		(5)
	ShareholderLatter		EmployeeLatter		ConsumerLatter		EnvironmentalLatter		SocialLatter
Panel C CSRFormer dimensions vs CSRLatter dimensions
ShareholderFormer	0.010**	(0.004)
EmployeeFormer			0.017***	(0.006)
ConsumerFormer					0.015**	(0.007)
EnvironmentFormer							0.018**	(0.007)
SocialFormer									0.001	(0.005)
All controls	Yes		Yes		Yes		Yes		Yes
Year FE	Yes		Yes		Yes		Yes		Yes
Firm FE	Yes		Yes		Yes		Yes		Yes
N	11,023		11,023		11,023		11,023		11,023
R2	0.804		0.705		0.660		0.623		0.784

Note(s): This table reports analyses for the CSR dimensions. Panel A regresses five CSR dimensions of the subsequent firm in year t+1 on lagged overall CSR of the former firm in year t-1 (CSRFormer) and control variables in year t+1. Panel B regresses the overall CSR of the subsequent firm (CSRLatter) in year t+1 on lagged five CSR dimensions of the former firm in year t-1 and control variables in year t+1. Panel C regresses five CSR dimensions of the subsequent firm in year t+1 on lagged five CSR dimensions of the former firm in year t-1 and control variables in year t+1. Robust standard errors clustered by firm and year are in parentheses. All variables are defined in Appendix 1. ^***p < 0.01, ^**p < 0.05, ^*p < 0.1

Source(s): Table 5 by authors

Table 6

Further analyses by job-hopping, executive and firm attributes

Variables	(1)		(2)		(3)		(4)		(5)		(6)		(7)		(8)
	CSRFormer < CSRFormerIndAvg		CSRFormer > CSRFormerIndAvg		Direct job-hopping		Gap job-hopping		Pay increase		No pay increase		Same city		Different city
	CSRLatter		CSRLatter		CSRLatter		CSRLatter		CSRLatter		CSRLatter		CSRLatter		CSRLatter
Panel A: Job-hopping attributes
CSRFormer	−0.030	(0.030)	0.078*	(0.040)	0.014**	(0.01)	0.033	(0.05)	0.001	(0.011)	0.027**	(0.011)	0.036**	(0.02)	0.006	(0.01)
All controls	Yes		Yes		Yes		Yes		Yes		Yes		Yes		Yes
Year FE	Yes		Yes		Yes		Yes		Yes		Yes		Yes		Yes
Firm FE	Yes		Yes		Yes		Yes		Yes		Yes		Yes		Yes
N	6,534		4,489		10,181		198		3,597		3,393		2,364		6,599
R2	0.861		0.851		0.712		0.878		0.788		0.739		0.771		0.729

Variables	(1)		(2)		(3)		(4)		(5)		(6)		(7)		(8)
	Age > median		Age < median		Overseas experience		No overseas experience		Female executives		Male executives		CEO		Non−CEO
	CSRLatter		CSRLatter		CSRLatter		CSRLatter		CSRLatter		CSRLatter		CSRLatter		CSRLatter
Panel B: Executive Attributes
CSRFormer	0.023**	(0.01)	0.008	(0.01)	0.040***	(0.01)	0.005	(0.01)	0.034**	(0.014)	0.012*	(0.007)	0.043*	(0.023)	0.038***	(0.011)
All controls	Yes		Yes		Yes		Yes		Yes		Yes		Yes		Yes
Year FE	Yes		Yes		Yes		Yes		Yes		Yes		Yes		Yes
Firm FE	Yes		Yes		Yes		Yes		Yes		Yes		Yes		Yes
N	4,405		5,309		933		9,261		1,201		8,639		611		10,406
R2	0.772		0.757		0.857		0.706		0.880		0.708		0.343		0.256

Variables	(1)		(2)		(3)		(4)		(5)		(6)		(7)
	Size > median		Size < median		OwnershipCon > Median		OwnershipCon < Median		SOE		Non-SOE		Leverage > median
	CSRLatter		CSRLatter		CSRLatter		CSRLatter		CSRLatter		CSRLatter		CSRLatter
Panel C: Firm Attributes
CSRFormer	0.017*	(0.01)	0.001	(0.01)	0.004	(0.01)	0.017***	(0.01)	−0.001	(0.01)	0.016*	(0.01)	0.024**	(0.01)
All controls	Yes		Yes		Yes		Yes		Yes		Yes		Yes
Year FE	Yes		Yes		Yes		Yes		Yes		Yes		Yes
Firm FE	Yes		Yes		Yes		Yes		Yes		Yes		Yes
N	5,471		5,202		5,293		5,343		2,337		8,335		5,404
R2	0.737		0.761		0.734		0.753		0.812		0.740		0.734

Variables	(8)		(9)		(10)		(11)		(12)		(13)		(14)
	Leverage < median		TobinQ > median		TobinQ < median		MgmtHolding = 1		MgmtHolding = 0		CEODuality = 1		CEODuality = 0
	CSRLatter		CSRLatter		CSRLatter		CSRLatter		CSRLatter		CSRLatter		CSRLatter
CSRFormer	0.005	(0.01)	0.004	(0.01)	0.018**	(0.01)	0.014*	(0.01)	0.013	(0.01)	0.014*	(0.01)	0.004	(0.01)
All controls	Yes		Yes		Yes		Yes		Yes		Yes		Yes
Year FE	Yes		Yes		Yes		Yes		Yes		Yes		Yes
Firm FE	Yes		Yes		Yes		Yes		Yes		Yes		Yes
N	5,253		5,135		5,300		6,093		4,660		8,076		2,644
R2	0.746		0.813		0.734		0.749		0.720		0.727		0.787

Note(s): This table reports subsample analyses for the job-hopping, executive and subsequent firm characteristics. We regress the CSR of the subsequent firm (CSRLatter) in year t + 1 on the lagged CSR of the former firm in year t−1 (CSRFormer) and the control variables in year t + 1. Robust standard errors clustered by firm and year are in parentheses. All variables are defined in Appendix 1. ***p < 0.01, **p < 0.05, *p < 0.1

Source(s): Table 6 by authors

Table 7

Additional robustness tests

Variables	(1)		(2)
	CSRLatter		CSRLatter
CSRFormerSingle	0.059**	(0.023)
CSRFormerAvg			0.026*	(0.014)
All Controls	Yes		Yes
Year FE	Yes		Yes
Firm FE	Yes		Yes
N	6,652		11,023
R2	0.890		0.796

Note(s): This table reports additional robustness test using alternative measures for CSRFormer. We regress the CSR of the subsequent firm (CSRLatter) in year t + 1 on the lagged CSR of the former firm in year t−1 (CSRFormerSingle and CSRFormerAvg) and the control variables in year t + 1. The average of the control variables is used for reporting the results in Column (2). Robust standard errors clustered by firm and year are in parentheses. All variables are defined in Appendix 1. ***p < 0.01, **p < 0.05, *p < 0.1

Source(s): Table 7 by authors

Table 8

DiD test with entropy balanced matching

Variables	CSRLatter
POST	2.520**	(0.975)
Size	6.594***	(1.507)
ROA	5.914	(3.603)
Leverage	−0.293	(0.182)
Growth	1.802*	(0.900)
TobinQ	0.416*	(0.186)
OwnershipCon	0.095*	(0.041)
SOE	0.767	(0.613)
MgmtHolding	1.319*	(0.634)
CEODuality	−1.363	(1.006)
ExeAge	0.083	(0.048)
ExeGender	0.008	(1.175)
ExeEdu	2.058*	(1.076)
ExeOverseas	−0.263	(0.920)
ExeAcademic	0.694	(0.890)
ExeFinBack	−0.016	(0.886)
Constant	−54.809***	(13.835)
Year FE	Yes
Firm FE	Yes
N	14,817
R2	0.654

Note(s): This table reports the DiD results with entropy balanced matching using CEO job-hopping event as a shock. POST equals 1 if firm i has an incoming CEO from a firm with greater CSR (i.e., CSRFormer > CSRLatter) and 0 otherwise. Robust standard errors clustered by firm and year are in parentheses. All other variables are defined in Appendix 1. ***p < 0.01, **p < 0.05, *p < 0.1

Source(s): Table 8 by authors

Table A1

Variable Definitions

Variables		Definition
CSR variables	CSRLatter	Dependent variable: overcall CSR score of the subsequent firm. The maximum CSR score is 100. A firm can have a negative CSR score if its CSR performance is poor
	CSRFormer	Independent variable: overall CSR score of the former firm. The maximum CSR score is 100. A firm can have a negative CSR score if its CSR performance is poor
	CSRFormerIndAvg	Industry average CSR performance of the former firm
	CSRFormerSingle	CSR performance of the former firm with only one job-hopping executive
	CSRFormerAvg	Average CSR performance of all former firms with multiple job-hopping executives
Firm variables	Size	Firm size, measured as the natural logarithm of total assets
	ROA	Firm performance, measured as net income divided by total assets
	Leverage	Leverage is measured as total liabilities divided by total assets
	Growth	Sales growth, calculated as return on net assets × earnings retention rate divided by (1 − return on net assets × earnings retention rate)
	TobinQ	Tobin’s Q, measured as firm market value divided by total assets
	OwnershipCon	Ownership concentration, measured as the sum of squares shareholding by the top 10 shareholders
	SOE	State-owned enterprise = 1; otherwise = 0
	MgmtHolding	Management shareholdings; executive owns shares = 1, otherwise = 0
	CEODuality	CEO is chair of the board = 1, otherwise = 0
Executive variables	ExeAge	Age of executive
	ExeGender	Gender of executive: male = 1, female = 0
	ExeEdu	Education of executive: executive with Bachelor’s degree or higher = 1, otherwise = 0
	ExeOverseas	Executive with overseas experience = 1, otherwise = 0
	ExeAcademic	Executive with academic experience = 1, otherwise = 0
	ExeFinBack	Executive with financial background = 1, otherwise = 0

Source(s): Appendix 1 by authors

Table A2

PSM Balance Test Results

Variables	Unmatched	Mean		t-test
	Matched	Treated	Control	t	p>|t|
Size	U	9.594	9.523	1.84	0.065
	M	9.599	9.613	−0.24	0.810
ROA	U	0.048	0.035	1.32	0.185
	M	0.048	0.040	1.59	0.113
Leverage	U	0.454	0.436	0.20	0.838
	M	0.453	0.424	1.52	0.130
Growth	U	0.086	−0.030	0.43	0.665
	M	0.085	0.136	−0.55	0.586
TobinQ	U	3.201	2.793	1.70	0.090
	M	2.875	2.642	0.98	0.329
OwnershipCon	U	58.872	59.128	−0.26	0.793
	M	59.032	58.894	0.10	0.923
SOE	U	0.230	0.160	2.99	0.003
	M	0.230	0.242	−0.31	0.756
MgmtHolding	U	0.510	0.667	−5.30	0.000
	M	0.512	0.488	0.53	0.597
CEODuality	U	0.284	0.319	−1.18	0.238
	M	0.281	0.289	−0.20	0.845
ExeAge	U	46.984	49.869	−6.88	0.000
	M	47.016	47.188	−0.27	0.786
ExeGender	U	0.825	0.930	−6.45	0.000
	M	0.824	0.828	−0.12	0.907
ExeEdu	U	0.700	0.869	−7.86	0.000
	M	0.703	0.699	0.10	0.923
ExeOverseas	U	0.117	0.097	1.07	0.284
	M	0.117	0.160	−1.41	0.160
ExeAcademic	U	0.070	0.217	−5.69	0.000
	M	0.070	0.094	−0.97	0.335
ExeFinBack	U	0.160	0.083	4.39	0.000
	M	0.160	0.141	0.62	0.537

Note(s): This table reports the results of the PSM test. The t-test results show that the differences between the treated and the control samples after matching are all insignificant, which indicates that the matching effect is effective

Source(s): Appendix 2 by authors

Table A3

Table: Hexun CSR Framework

CSR dimensions	Indicators
Shareholders (30%)	Profitability (10%)
	Debt paying (3%)
	Return (8%)
	Credit (5%)
	Innovation (4%)
Employees (15%) (10% in consumption sector)	Performance (5%) Safety (5%) Caring for employees (5%)
Customers and suppliers (15%) (20% in consumption sector)	Product quality (7%) Customer service (3%) Mutual good faith (5%)
Environment (20%) (30% in manufacturing sector, 10% in service sector)	Environmental governance (20%)
Society (20%) (30% in service sector, 10% in manufacturing sector)	Contribution (20%)

Source(s): Appendix 3 Table by authors

Notes

1. https://marketingclubimi.wordpress.com/2018/08/27/paul-polman-hard-work-has-no-substitute/ In addition, Section 2.3 provides further anecdotal evidence illustrating how job-hopping executives who leave firms with poor CSR can diminish the CSR of the hiring firms.

2. To break down our data further by year, 642 (in 2011), 637 (in 2012), 929 (in 2013), 1,499 (in 2014), 1,501 (in 2015), 1,832 (in 2016), 1,465 (in 2017), 1,328 (in 2018) and 1,190 (in 2019) job-hopping events by executives are identified. As our study examines how job-hopping by executives in year t causes CSR in year t-1 of the former firm to correlate with CSR in year t+1 of the latter firm, we establish the sample period and corresponding control variables from 2010 to 2020. Additionally, given that some executives change jobs multiple times during the sample period and may hold board positions in multiple connected companies in the same year (e.g., subsidiaries and parent companies), our sample includes a total of 6,236 unique executives.

3. Note that Table 2 Column (4) results are based on 11,625 observations, because only industry and year fixed effects are taken into account. In Columns (1) to (3), where firm and year fixed effects are considered, firms with only one observation in the sample are omitted from the analysis, resulting in a reduced sample size of 11,023.

4. According to all available executive-level data from CSMAR before and after matching firm-level characteristics and CSR data, the average tenures of executives are 3.46 years and 2.22 years, respectively. When there are multiple job-hopping executives for a firm, we randomly select one executive for each firm. In this final sample, the average tenure of executives is 2.87 years.

5. The CSR dimensions are described in Appendix 3.

6. We also regress all five CSR dimensions of the former firm against the CSR of the latter firm (CSRLatter) simultaneously. The untabulted results reveal that only the shareholder and consumer dimensions exhibit a significant impact on CSRLatter, and the coefficients for the employee, environment and social dimensions switch signs and become insignificant. The results are typical, given the high multicollinearity among the CSR dimensions by construct. To substantiate this, we further conduct correlation tests (untabulated) between the dimensions and observe high levels of correlation, confirming the existence of multicollinearity.

Appendix 1

Appendix 2

Appendix 3 CSR dimensions

The Hexun CSR score evaluates the social responsibility performance of firms across five different stakeholder groups (weights in parentheses): shareholders (30%), employees (15%), customers and suppliers (15%), the environment (20%) and society (20%). For each dimension, weight allocation is adjusted according to industry, to recognise the variations in emphasis that different industries place on these dimensions.

Specifically, the shareholder dimension gauges a firm’s financial performance in terms of earnings, solvency, shareholder returns, creditworthiness and innovation. The employee dimension focuses on evaluating a firm’s commitment to employee development, safety, wellbeing and per capita income for its workforce. The customer and supplier dimension examines aspects such as product quality, after-sale service, promotion of fair competition and efforts to combat bribery in supplies and customer relationships. The environmental dimension assesses eco-conscious practices of the firm, including environmental certification, investment in sustainability, pollutant discharge management and energy-saving initiatives. Last, the social dimension measures the societal impact of a firm, by examining areas such as public donations and its tax contributions. Appendix 3 table presents a detailed breakdown of the five dimensions, to offer a comprehensive insight into the evaluation of a firm’s CSR performance across various critical areas.