Birds of a Feather Flocking Together: Sustainability of Tax Aggressiveness of Shared Directors from Coercive Isomorphism

: The purpose of the study is to examine the sustainability of the tax aggressiveness of shared directors from coercive isomorphism and whether social networks of directors have an impact on their tax aggressiveness. Speciﬁcally, the study intends to examine how tax knowledge diffuses across ﬁrms and how this knowledge diffusion affects connected ﬁrms. To test the constructed hypothesis, the panel logistic regression model is estimated using a ﬁrm-level panel dataset for the US and Pakistan to analyze cross-country differences, as the USA holds more legislation and effective governance mechanisms. The study covers the period of 2007–2019. The data required for the empirical analysis was collected from the Thompson Reuters database. The results of panel logistic regression show a signiﬁcant relationship between tax aggressiveness and director’s connections, suggesting that information diffuses by board interlocks. Speciﬁcally, the estimates suggest that there is a positive and signiﬁcant inﬂuence of connected directors on the probability that the tax aggressiveness spreads through coercive isomorphism, inferring that the sustainability of the tax aggressiveness of shared directors from coercive isomorphism is strong. Findings reveal that Pakistani ﬁrms, when compared to the USA, are more likely involved in tax aggression because of fewer legislations and tax reforms. The results also reveal that coercive isomorphism signiﬁcantly mediates the relationship between board interlocks and tax aggressiveness. These ﬁndings provide valuable insights into detecting the tax aggressiveness of ﬁrms and the channels through which this spread. The study contributes to the scarce research on the impact of board interlocks on tax aggressiveness and the inﬂuence of coercive isomorphism on these impacts. This study can help tax authorities in identifying tax-saving strategies through connected directors. Secondly, this study provides empirical evidence to support the diffusion of information regarding tax aggression and provides mechanisms with which to detect tax aggression. Third, our choice of empirical context also helps us contribute to the management practice of ﬁrms. CEOs and boards should be wary of interlocks with organizations, lest they inadvertently become reticent and hence prove to be of no good.


Introduction
For more than ten years, several empirical studies have documented a huge set of variables and their impact on tax aggressiveness. Specifically, researchers have studied board composition, CEO narcissism, auditor quality, political connections, director gender, employees' whistleblowing, information asymmetry, insider trading, and independent directors with tax aggressiveness [1][2][3][4][5][6][7][8][9]. The literature on tax aggressiveness provides a variety of definitions for tax aggression, such as [7], who have defined tax aggressiveness as the avoidance of taxes by reducing tax payments. Moreover, Ref. [8] observed that firms make manipulations to lower taxes, due to activity known in tax literature as tax management. In today's era, firms are largely engaged in tax aggressive strategies, and there exists a substantial discrepancy in the taxes being paid among corporations [10]. Whilst the companies benefit from tax avoidance by paying lower taxes, it comes with a risk of penalties and may also involve reputational expenses [1]. This variation in the taxes being paid motivates researchers to conduct empirical research on other mechanisms through which firms get involved in tax-aggressive strategies [8,11]. Thus, we propose that the existing research on corporate tax aggressiveness is incomplete, as it does not consider the social structure in which tax management activity is embedded. We suggest that information about tax aggressiveness strategies and knowledge about implementing those strategies is shared among firms through connected directors.
The existing literature suggests a negative relationship between tax aggressiveness and corporate governance variables as corporate governance mechanisms such as boards, board sizes, and ownership structure reduce tax aggressiveness [12]. Corporate boards have been largely studied in literature as a monitoring mechanism, however none of the studies have investigated it as a triggering mechanism leading to tax aggression. Though scholars have made remarkable developments in classifying the characteristics of individual firms that are associated with the tax planning of firms [10,13,14], unexpectedly little research examines how tax planning spreads across firms. The impact of director interlocks on research and development studies suggested that managers imitate R&D intensity when they are connected through interlocks [15,16]. The impacts of social networks have also been studied in the literature on earnings management. The authors recommended that accruals management contagion arises for organizations having interlocks [17]. Further, firms whose members have been part of backdating firms tend to backdate more as a result of network effects [18]. In line with this research, scholars uncovered that shared auditors play a pivotal role in this case [19]. However, they did not consider the effect of board interlocks. Building on this literature, we examine whether director ties help explain variation in tax aggressiveness and how the relationship between director ties and tax aggressiveness varies depending on the context of those ties.
As part of a growing body of literature on director ties, scholars have identified a set of firms that tend to mimic others with which they have an informational connection, and called for additional research on isomorphic forces in different fields where isomorphic pressures exist [16]. The research also calls for different practices that diffuse through network ties but lack the degree and depth of those practices. Uncertainty, according to [20], encourages problematic search, which leads organizations to isomorph other organizations with which they share information. Isomorphic forces and resulting pressures vary from field to field, but none of the studies have paid attention to isomorphic pressures faced by connected directors. We examine tax aggressiveness in the context of director networks where connected firms coercively adopt practices.
The main purpose of the study is to link the literature on tax aggression with vigorous literature on shared directors. The study posits that the knowledge regarding tax management and the skill of employing that knowledge is transferred among firms through shared directors. The existing literature alludes to those social networks that encourage the diffusion of new ideas and hence influence corporate behavior [17,[21][22][23][24]. Prior studies also suggest that auditor network ties enable information transfer among client portfolios due to information advantage [25][26][27]. On similar grounds, director ties can also facilitate information diffusion. As directors sit on multiple seats, they may intentionally or unintentionally transfer information regarding certain strategies of the company. Hence, this study proposes that the shared directors diffuse knowledge in board meetings and interactions, causing knowledge spillover, and, hence, affecting the tax strategies of connected firms. On the contrary to the plethora of research on interlocks and mimetic isomorphism, little knowledge exists on the connection between interlocks and coercive isomorphism. Following the previous literature, this study explores tax aggression that firms spread through interlocks and other channels through which firms are likely to be involved in tax aggression.
By using the tax shelter prediction score by [28], the study aims at examining the relationship among connected directors and tax aggressiveness of a firm where coercive isomorphism is the underlying mechanism that explains this relationship. In this study, we use theory of diffusion of innovation to examine the relationship among connected directors and tax aggressiveness, as shared directors act as a conduit of information, coercing connected firms to be tax aggressive.
This study contributes to the accounting literature in several ways. First, this study ties the corporate governance (board interlocks) and tax aggressiveness literature together and directly responds to the call in [16] for research on diffused strategies and the isomorphic forces that exist among firms. We find evidence that director ties are associated with tax aggressiveness and that firms coercively imitate strategies, consistent with director ties acting as a conduit of information on tax management strategies. Furthermore, the results are robust against an alternative measure of tax aggressiveness; that is, the book-tax difference (BTD) (see Appendix A). We find that firms that have positive BTD's are more involved in tax aggression as compared to firms with low BTD's. Secondly, this study extends the findings of [29], as we find evidence that a specific strategy other than COLI diffused through board interlocks. We also extend the findings of [30], as we consider coercive isomorphism and identify how interpersonal ties map to an inter-organizational network by coercive isomorphism. Third, this study provides mechanisms to detect tax aggression, as we find that firms under the influence of connected firms adopt tax-aggressive strategies coercively so that practitioners and regulators can detect tax aggression by looking into a firm's intangibles, as they might be coercively adopting practices.
The remainder of the study is organized as follows: Section 2 briefly discusses the literature review and Section 3 explains the data and methodology used in the study. Section 4 explains the empirical finding of the study, and Section 5 concludes the discussion.

Literature Review and Hypothesis Development
Recent corporate tax practices have been widely discussed, including severe allegations of misconduct. Fresh terminology is emerging in this scenario to characterize corporate tax-related behavior, such as 'fair share,' tax-dodging,' etc. The world is seeing unprecedented levels of scrutiny and attempted 'tax shaming' by media and non-government firms against multinationals. However, there is a disturbing degree of confusion, disinformation, and falsehood, amid all the vocal demonstrations and verbiage, and we also know very little about the firms' real tax practices. We are put to wonder: who is behind this scheming and how these are spreading?
Researchers have described tax aggression as "the reduction of explicit taxes per dollar of pre-tax accounting earnings or cash flows". They defined the continuum of tax planning as a legal tax avoidance on one hand and tax non-compliance, evasion, aggression, and shelter on the other [10]. Following [10,28], this study also does not differentiate between legal and illegal tax saving, because the tax transaction's legality is determined ex-post [31].
With respect to the opportunistic managerial behaviors and resource diversion consequences of aggressive tax planning, it is documented that tax evasion benefits after deductions are higher for firms that are monitored effectively and, hence, are coerced and constrain managerial opportunism [32]. The literature also studies the focal firms' connections to tax havens to study tax evasion of firms [33]. Scholars found that tax planning spreads across firms through supply chains and customers [34]. Researchers also found that tax saving is more prevalent among firms in customer-supplier relationships [35]. Another study investigated banks as tax-saving liaisons and found that tax planning can spread among firms with common banking relationships [36]. The study examined whether having the same auditor diffuses tax planning [37]. The research reveals that companies show lesser ETRs when they are tied through common audit partners. One study found evidence suggesting that the tax disclosure of a firm affects the tax disclosure of its connected firms [38]. These findings indicate that tax-planning ideas tend to diffuse among corporations over time. Non-public information is mainly leaked through board interlocks. Numerous studies have recommended that board interlock transfer value-relevant information between firms [39][40][41]. The latest cases between Google and Apple hint towards the potential risks of information transfer as a result of board interlocks. For a well-connected firm, the possibility of information leakage is more severe, as the shared directors have more chances to communicate with directors of other organizations.
Distinct diffusion patterns are followed for strategies to transfer across firms [42]. For instance, a study by [43] proposed that board interlocks act as conduits of information and transfer corporate governance practices related to poison pills. Connected boards are seen to exhibit an impact on many business planning strategies similar to tax shelter adoption. Next, the conclusions of the study by [29] may not be generalizable, as they concern the spread of corporate-owned life. Additionally, the study provides information regarding how business strategies are adopted under the influence of connected boards, but it does not explore the effects of connected directors deciphered into rent extraction benefits [29]. The study found that firms might be exposed to new tactics via connected directors, or that they might be coaxed into adopting certain strategies after discovering that the connected partner had success in adopting that particular strategy [44].
Director ties are widely displayed in influencing business policies. Moreover, the above studies suggest that the connected board is a visible proxy for an organization's wider set of network connections. Connections to low tax firms have been said to be more prominent when the firms are connected through the same auditor [30]. Hence, board interlocks have been proposed to proxy for firms tied through the same directors. Precisely, this study is certain to find the relation between board interlocks and its tax avoidance behavior. Consequently, under the diffusion theory, we expect that firms having shared directors transfer tax knowledge and induce firms to be tax aggressive. Firms may be subjected to brand-new, tax-avoiding ideas via their connected directors or may be influenced to implement a new tax-avoiding tactic after discovering that the connected firm has been successful in utilizing that approach.

Hypothesis 1 (H1).
Director ties are positively associated with the tax aggressiveness of a firm.
The dictionary defines isomorphism as anything which is similar or identical or has the same form or structure [45]. While coercive isomorphism refers to the organizational similarity that "results from formal and informal pressures exerted on organizations by other dependent organizations and cultural expectations in the society in which organizations operate" [16,20]. Interlocking directors are human beings, and are not dispassionate, impersonal conductors of information [46]. It seems reasonable that coercive pressures would affect interlocking directors such that they foster further isomorphism by their past connection with the field in which the direct coercive pressures are active. The research has proven that, when there is information symmetry, mimetic strategic practices are diffused by board interlocks, for instance, by foreign expansion and even the practice of imitating competitor strategies [47][48][49]. Nonetheless, it is still a question of how exactly the compliance to coercive pressures would affect directors and how the connections will affect firms indirectly exposed to coercive pressures from the outside environment. People tend to internalize the behavior or opinion of the individuals expressed, as a result of pressures, to choose a particular behavior or opinion [50,51]. We argue that directors exposed to coercive pressures in one organizational field adopt those pressures as their attitudes, and subsequently spread the coercively adopted practices to other organizations' boards on which they sit, even if those organizations are not directly exposed to the same coercive pressures. Even when facilitating the diffusion of governance reforms, connected directors are viewed as conduits with which to transmit information regarding the best way to deal with coercive pressures, not to transmit the actual pressures.

Hypothesis 2 (H2). Director ties are positively related to Coercive Isomorphism.
Coercive isomorphism results from the direct pressures of powerful organizations, individuals, or societies [16]. Board interlocks encourage the diffusion of several mimetic strategies, such as the practice of imitating competitors' strategies, foreign expansion, the adoption of poison pills [47][48][49]. Yet, organizations experiencing coercive isomorphism have board interlocks, and it remains an open question as to whether the coercive pressures experienced by one organization affect directors in such a way that they subsequently influence the behavior of other organizations on whose boards the directors also sit. Theorists typically conceptualize isomorphic processes at the organizational level, paying less attention to the individuals responsible for disseminating and adopting the practices that become isomorphic [52]. We extend the theory in this area by discussing how conformity to coercive pressures in one organizational field affects the board members responsible, and, in turn, how those board members will influence other boards on which they sit. Hence, we propose that firms facing coercive pressures will adopt the strategy, ideas, and behaviors diffused by directors' ties. This strategy in the said study is tax aggressiveness. Following the above line of reasoning, the proposed hypotheses are:

Hypothesis 3 (H3).
Coercive Isomorphism is positively associated with the tax aggressiveness of a firm.

Hypothesis 4 (H4).
Coercive isomorphism mediates the positive association between director ties and tax aggressiveness of a firm.

Data and Methodology
The data for Pakistani non-financial firms and US firms are used to analyze crosscountry differences in tax aggressiveness. The main reason behind a cross-country analysis is the effective corporate governance mechanisms of US firms compared to a developing country like Pakistan. The study uses data starting from 2007 to 2019. The post-2008 crisis data is used because it was a catalyst for progress in international tax standards and other reforms in transparency and corporate governance. Consistent with the preceding literature [5,30,37,53], we excluded firms with negative before-tax income, as those firms have alternate reasons for tax saving [30]. The summary of the sample selection process is shown in Panel A of Table 1. Multiple sources have been used to determine the sample selection. One source used was [28], who investigated popular press to identify firms between 1990 and 2007. Ref. [54] also used the same methodology to identify firms. With these, we also searched the press for the period 2007 to 2019, using keywords "tax avoidant" and "tax shelter".

Measures of Tax Aggressiveness
The profile of tax sheltering firms is needed to be developed before further discussion. The positive value of the BTDs means that the firm is involved in tax aggression [28]. As the positive BTDs hint at tax sheltering activity, firms are identified as tax shelter participants by calculating BTDs. The following formula is used to calculate BTD: where BTD refers to the book-tax difference. which is calculated as book income less taxable income divided by total assets; NOL refers to net operating loss at time t, while NOL t−1 refers to net operating loss carry-forward. The tax aggressiveness measure uses the tax shelter score by [28] and is widely used in the literature [4,11]. SHELTER is the probability of the firm being involved in tax sheltering. Logistic regression is used to estimate the equation. The estimation equation is as follows: ln P shelter 1 − P shelter refers to the natural log of probability of sheltering firm. SHELTER represents tax sheltering, BTD i,t represents book-tax differences, DAP i,t refers to discretionary accruals, LEV i,t represents leverage, SIZE i,t reflects the size of the firm, ROA i,t represents the return on assets at time t and is a control variable, GROWTH i,t reflects the growth of the company, and CASH i,t represents cash holdings of the company at time t, FI i,t represents foreign income of company at time t, while LEV i,t , SIZE i,t , ROA i,t , GROWTH i,t , FI i,t and CASH i,t are control variables. A predicted probability estimate is found by incorporating organizations' features to ascertain a forecasted value. The following equation is used to find the predictive probability of tax sheltering: The higher the value of SHELTER means that there is a higher likelihood of being engaged in tax aggression, and DAP is a measure of discretionary accruals measured using the Modified Jones Model (1995).

Measures of Board Interlocking
Board interlock is used as an independent variable. An interlock is defined as a state when firms have a shared director. We measured interlocks as the total number of board interlocks held by all directors on each corporate board each year, divided by the total number of directors on the board that year [16].

Measures of Coercive Isomorphism
Coercive isomorphism refers to the similarity that occurs when firms under influence of other firms to adopt new practices, strategies, or ideas. It is used as a mediator in the study. The coercive isomorphism is reflected in the firm's Intangible Asset Investment [16]. We have followed some prior research in examining two forms of investment in intangible assets, for which data are publicly available: research and development intensity and advertising intensity. Research and development intensity reflects investments in the firm's knowledge assets and advertising intensity reflects investments in the firm's reputational and brand assets. We created a summative index for intangible investments incorporating both measures, as some firms may be only R&D-intensive and some may be only advertising-intensive. The two measures have been extensively used in the literature [16,[55][56][57].

Regression Model
The regression model is as follows: where SHELTER i,t represents tax sheltering, BLOCKS i,t represents director ties, CISOM i,t reflects coercive isomorphism, LEV i,t represents leverage, SIZE i,t reflects the size of the firm, ROA i,t represents the return on assets, GROWTH i,t reflects the growth of the company at time t, FI i,t represents foreign income of a company, CASH i,t represents cash holdings of the company, and ε i,t represents error term at time t while SHELTER i,t , LEV i,t , SIZE i,t , ROA i,t , GROWTH i,t , CASH i,t are the control variables. The presence of binary variable (SHELTER = 1 or 0) hints using logistic regression as a predictor to examine outcome relationship [58]. In generalized linear models, researchers use odd ratios to interpret results.   This table presents descriptive statistics for the key variables used in our regression analysis. The sample period is 2008-2019. The zero values refer to missing values of the specific years of data. All the continuous variables are winsorized at 1 and 99 percent. The variable of foreign income is removed from final estimation due to non-availability of data.

Descriptive Statistics and Correlations
SIZE has a mean of 14.94 and a standard deviation of 4.93. Foreign Income has a mean value of 0.35 and a standard deviation of 0.47. The mean value of R&D is 0.07 with a standard deviation of 014. CASH has a mean of 0.08 and a standard deviation of 0.11. GROWTH has a mean value of 0.06 and a standard deviation of 0.31. Blocks have a mean value of 0.83 and a standard deviation of 0.31. CISOM has a mean value of 0.07 and a standard deviation of 0.14. Table 3 reports the Pearson's correlation for tax shelter score variables and other variables of the study. The correlation coefficient for the relationship between SIZE and LEV is 0.35, which is statistically significant at a 1% level of significance. The relationship between SIZE and ROA is also statistically significant, with a correlation coefficient of 0.18 at the 1% level of significance. The correlation coefficient for the relationship between SIZE and FI is 0.29, which is statistically significant at the 1% level of significance. The relationship between SIZE and R&D is also statistically significant with a correlation coefficient of 0.21 at the 1% level of significance. The correlation coefficient for SIZE and BTD is 0.15 with a 1% level of significance, suggesting that larger firms are less involved in tax aggression. The correlation coefficient for the relationship between SHELTER and CISOM is 0.56, which is statistically significant at the 1% level of significance. The relationship between BTD and CISOM is also statistically significant with a correlation coefficient of 0.38. The correlation coefficient for the relationship between BTD, SHELTER, and DAP is significant, and suggests that firms involved in earnings management may also be involved in tax shelters. Similarly, the variables of GROWTH and CASHHOLDINGS also show insignificant coefficients and therefore indicate no relationship; this infers that growing firms are investing their cash in ventures, hence lesser cash holdings. Moreover, the correlation coefficient for two tax aggressiveness proxies, Shelter and BTD, are positive with a value of 0.71, which suggests that both measures capture tax aggressiveness. Similarly, the relationship between tax aggressiveness measures and accruals is positive and significant, which is consistent with the findings of [28,59].  Table 4 reports VIF and Tolerance for variables of the study. All of the VIFs are below five thresholds, which statistically infer that multicollinearity is not present. The VIF for the tax shelter score variables are below two, therefore there is no multicollinearity present. For the validity of our results, Ramsey reset test or else. We tested for the normality of residuals, heteroscedasticity and autocorrelation problems before panel data analysis.   In this section, the results for the test of Hypothesis 1 are presented, stating that there is a positive association between director ties and tax aggressiveness. As in Table 6, the measure of tax aggressiveness is positively and significantly associated with directors' ties (board interlocks). The effect of connected directors on the tax aggressiveness of a firm is significant in probability and also in terms of percentage. For one unit increase in director ties (board interlocks), the odds of a firm involved in tax sheltering (versus non-tax sheltering) increase by a factor of 1.43. The second column reports the results for an alternative measure of tax aggressiveness, which states that, for one unit increase in director ties (board interlocks), the odds of a firm involved in tax sheltering (versus non-tax sheltering) increase by a factor of 3.76. In other words, the odds of a firm involved in tax sheltering increase three-fold when those of the firms are connected through directors. This result suggests economically significant social network effects on tax aggressiveness. The results are hence consistent with Hypothesis 1. The country dummy value 1 for Pakistan and 0 for the US is also significant, suggesting that Pakistani firms are involved in tax sheltering, as compared to the US, where there are more corporate governance codes and regulations. It suggests that the odds of a firm relative to a specific country, to become involved in tax sheltering, increases by a factor of 92.31. This result suggests that Pakistan, when compared to the USA, is more involved in tax sheltering. The reason behind this skyrocketing effect may be that the firms in Pakistan are family-owned and because regulations are less stringent compared to the US. The strong corporate governance mechanisms and tax regulations play their part in controlling taxes, while fewer regulations and corruption make Pakistani firms more prone to tax sheltering. Hence, in-part institutional problems such as corruption in the tax department, bribery, and a lower quality of governance are the main reasons for Pakistani firms' increased tax aggression.

Tests of Hypothesis
The coefficients for controls are also consistent with those in the literature. We find evidence that firms that are involved in tax sheltering have lower LEVERAGE, lower SIZE, and lower ROA. It is evident that less-levered firms are more prone to becoming involved in tax aggression. The probable reason for this may be because they are not already availing of tax shelter by using interest deductions due to debt. This may also suggest that tax sheltering itself removes liability from reports, hence decreasing leverage [60]. Furthermore, the study suggests that tax shelter organizations are not highly levered, and hence do not seek to remove debt from their balance sheet.
In addition, we also find that large-sized firms face deeper political and reputational pressures, resulting in less tax sheltering. Firms in Pakistan are smaller hence are more prone to tax aggressiveness because of the less political and reputational pressures faced. Results also suggest that low-performing firms are involved in tax aggressiveness. This is inconsistent with the literature, as it suggests that low profitability organizations have fewer motivations to be involved in tax aggression [30,53]. It suggests that, in a developing country like Pakistan, where there are smaller firms, there are more chances for increasing profits by using tax-saving strategies. Consistent with the literature [53], we find that growing firms become involved in tax sheltering. According to [61], financially constrained firms are more likely to be involved in tax aggressiveness. The result is inconsistent with the literature because our results show that firms having more cash holdings (CASH) are more involved in tax aggression. This infers that the firms in Pakistan have more incentive to increase cash through tax management because of its developing nature as compared to a country as developed as the USA. *, **, and *** refer to 10%, 5%, and 1% levels of significance, respectively. The results for logit regression with odds ratios are reported. The variable SHELTER (tax aggressiveness) is the probability of a firm involved in tax aggression. BTD is another measure for tax aggressiveness, which is a dummy variable for a firm going for tax saving while 0 for non-tax savers. Blocks is a ratio of board members interlocked to total board members.

Mediation
This section explores the mediation between tax aggression and director ties. For Hypothesis 2, the results show that director ties are positively associated with coercive isomorphism. Table 7 reports the findings of mediation using coercive isomorphism as a mediator. The findings also suggests that there is a positive association between connected directors and coercive isomorphism and that it is also significant. This result is consistent with our prediction in Hypothesis 2 that direct pressures of powerful organizations and executives are diffused through ties, and that shared directors are the conduits of information, as they transfer strategies within firms. A tax-saving strategy is, therefore, one of the strategies or practices that are diffused through board interlocking. In Hypothesis 3, we contend that coercive isomorphism is positively related to tax aggression. Our findings suggest that there is a positive relationship between coercive isomorphism and tax aggressiveness and that is also significant. The coefficient is 1.46 with a p-value of 0.001. The result is also consistent with our hypothesis. In Hypothesis 4, we assert that coercive isomorphism mediates the relationship between director ties and tax aggressiveness.
Direct, indirect, and total effects are reported in Table 7. The direct effects of director ties to the shelter are also significant, which are estimated using paths. The indirect effects are reported for the path of director ties and tax aggressiveness, which are significant and positive. The total effects are reported, which shows that all the paths are statistically significant, which is consistent with Hypothesis 4. Coercive isomorphism fully mediates the relationship between director ties and tax aggressiveness, as there are significant direct effects [62]. Therefore, it has been proved that firms having shared directors tend to deliver company-specific information to connected firms and that, hence, firms adopt the tax-saving strategy under their influence. We argue that connected directors transmit strategies, practices, and tacit knowledge, adopted in response to coercive pressures from other connected firms, and that the web of networks continues this diffusion. Overall, our findings recommend that connected directors sitting on multiple boards exert pressure on organizations, coercing firms to isomorph the strategies, and that tax knowledge is diffused, leading to tax sheltering, which is consistent with our hypothesis.

Number of Observations 2297
Goodness of Fit R 2 0.029 0.029 *, **, *** refer to 10%, 5%, and 1% levels of significance, respectively. Path analysis is reported with odd ratios and SEM results. The dependent variable SHELTER (tax aggressiveness) is the probability of a firm involved in tax aggression. BTD, another dependent variable, is a dummy variable equal to 1 for a firm having a positive BTD value and zero otherwise. CISOM is a coercive isomorphism calculated as advertising and R&D intensity scales by lagged assets. All variables are winsorized at 1 and 99 percent.
There were an extensive set of controls in our key analysis that are known for their correlation with tax aggressiveness. To mitigate concerns, we tested our model with an alternative measure which suggests that the results are robust. Table 7 shows the results for controls. We find that the alternative tax measure of BTD is also significant and consistent with our hypothesis. The control variable ROA is positively opposite to the tax sheltering measure. It suggests that high-performing firms are involved in tax aggressiveness. The consistent results with the literature suggest that low profitability firms have fewer motivations to be involved in aggressive taxation [30,53]. Furthermore, unlike the result for cash holdings, which is positive for a tax shelter prediction score, it comes out to be negative for BTD, suggesting that the less available cash, the higher the probability of becoming involved in tax aggressiveness. This result is consistent with the findings of [61].

Conclusions and Policy Implications
The purpose of the study is to examine the sustainability of tax aggressiveness of shared directors from coercive isomorphism and whether social networks of directors have impacts on their tax aggressiveness. The results reveal that there is a significant relationship between tax aggressiveness and directors' connections, suggesting that information diffuses by board interlocks. Specifically, the estimates suggest that there is a positive and significant influence of connected directors on the probability that the tax aggressiveness spreads through coercive isomorphism, inferring that the sustainability of the tax aggressiveness of shared directors from coercive isomorphism is strong. The results also reveal that coercive isomorphism significantly mediates the relationship between board interlocks and tax aggressiveness. These findings provide valuable insights into detecting the tax aggressiveness of firms and channels through which this spreads. Board directors are taken to be blue-eyed while they make the most out of it. Coercive isomorphism has been experienced by firms having board interlocks, but it is still questionable whether these pressures experienced by a firm's board members affect other connected firms and their boards as it does for the first.
It is evident that shared directors transfer knowledge and practices, and that connected members adopt those practices under coercive pressures. Our findings suggest that boards are not always monitoring, but that they trigger other compeers to adopt certain strategies and practices. Corporate governance has always been in debate and has been studied widely [12,32]. Thus, this study offers a new understanding of the tax aggressiveness, coercive isomorphism, and board interlocks literature, as tax sheltering is proved to be one of the strategies that are diffused through isomorphism and interlocks. The earlier strategies diffused are poison pill adoption [47], private equity offers (Yim and Stuart, 2010), and earnings management [17].
The study provides better insights for firms to have better expertise of their executives so that they can have help in tax planning. Furthermore, these results can be used by regulatory authorities, the Securities and Exchange Commission (SEC), the Federal Board of Revenue (FBR), and other agencies to identify firms doing tax aggression. They can generate more revenues by identifying firms involved in tax aggression. In addition, the firms who have competitive positions need to check the balance on their executives regarding information dissemination, as they might prove mole in the company. Tax aggressiveness is mainly accomplished via the information dissemination of connected directors. Thus, the regulatory authorities need to make certain regulations to stop information and knowledge dissemination. There is also a need to develop certain guidelines to be followed by directors when they are interacting in director meetings. In developing countries such as Pakistan, the difficult access to capital markets and the lack of investor protection laws leads companies to assume more tax aggressive positions as compared to the USA. An important question that might arise in readers' minds is why directors would share information. The legitimate answer to this thought lies in the idea of a firm's wealth maximization drive. Tax aggression is not a criminal offence, but rather a moral or ethical offence. The major objective of directors is to maximize company wealth without damaging the interests of others.
The limitations of the study are as follows: First, the study is not generalizable, as the laws, regulations, governance practices and particularities of the countries, such as ease of access to market, legal issues, and tax enforcement perceptions, must be considered. Second, there is a lack of publicly available data for tax measures and discretionary accruals measures, as unregistered firms are not liable to disclose financial data. Hence, unlisted companies have not been included in this research. Third, our study uses data from 2007 to 2019, without taking into account the effects of the COVID-19 pandemic. Fourth, the Pakistani stock exchange has a limited number of listed companies, and, hence, has lesser corporate culture diversity than US firms, as they are enrichment oriented. Therefore, future studies are advised to check for cultural diversity and its impact on tax aggression. Another limitation of the study is that the sample does not include any law firm in the sample of 192 firms. Moreover, Pakistani-listed firms are approximately 600, out of which very few have publicly available data. Hence, future studies can check for auditor and lawyer effects depending on the availability of data.
The use of post-financial crisis data, as an exogenous shock, implies that COVID-19 has also impacted the tax aggressiveness of firms; hence, future studies are advised to include the effects of the COVID-19 pandemic. Further research is recommended to identify other strategies and practices that might be diffused through corporate boards. Our paper examines the sustainability of the tax aggressiveness of shared directors from coercive isomorphism. Extensions of our paper could apply the model we used in our paper to other areas, for example, in economic policy and stock market returns [63], the sustainability of mergers and acquisitions [64], the sustainability of Energy Substitution [65], or corporate culture and cultural diversity [66]. Readers may read [67] and others for other areas in which one could apply the model we used in our paper.

Conflicts of Interest:
The authors declare that they have no competing interest.

Variables Description
Tax aggressiveness (Wilson, 2009) Tax aggressiveness is measured by the tax shelter score.
ln P shelter 1 − P shelter = β 0 + βX + εβX = β 1 BTD it + β 2 DAP it + β 3 LEV it +β 4 SIZE it + β 5 ROA it +β 6 FOREIGN I NCOME it +β 7 R&D it BTD A proxy for tax aggressiveness is the BTD factor which is a dummy variable and takes value 1 when there is a positive BTD value and zero otherwise

Board interlocks
It is measured by the total number of board interlocks held by all directors on each corporate board each year, divided by the total number of directors on the board that year Coercive isomorphism (Krause et al., 2018). It is the sum of R&D and Advertising divided by total assets Size It is measured as the Natural log of total assets at the end of the year Return on assets It is measured as operating income before depreciation (OIBDP) in the current fiscal year scaled by lagged total assets (AT)

Leverage
It is measured as long-term debt (DLTT) at the end of the current fiscal year scaled by lagged total assets (AT)

Growth
It is measured as a difference between sales and lagged sales divided by the lagged sales Cash holdings It is measured as cash and cash equivalents scaled by total assets