Does the attendance of independent directors at shareholder meetings matter? The case of risk taking

ABSTRACT Information is the basis for independent directors to make correct decisions. The study found that the participation of independent directors in the shareholder meetings would help them to obtain more real information and curb the excessive risk taking phenomenon of firms caused by agency problems. The mechanism is that independent directors make more stable decisions after obtaining more real information, have higher probability and frequency of dissenting opinions, and reduce the risk tolerance of the board. The diligent behavior of the independent directors who actively participated in the shareholder meetings has been rewarded by the labor market, and they will get more allowances and directorships in the future. The research of this paper has reference significance for the regulators to broaden the access to information of independent directors and improve the performance of independent directors.


Introduction
Risk taking is a classic topic in the field of corporate governance and an important embodiment of the agency problem between shareholders and management (Jensen & Meckling, 1976).Risk taking, which reflects the willingness of a firm to invest in profitmaking projects, is conducive to corporate growth and macroeconomic development (Acharya et al., 2011;Boubakri et al., 2013;Lumpkin & Dess, 1996).In a perfect market, a firm and its managers should actively pursue investment projects with positive net present value (NPV), as these not only increase enterprise value but also benefit managers themselves (who can derive private benefits).However, risk taking may also be the embodiment of management agency problems.The separation of ownership and control leads to the management being in an information superiority position, and the management mastering the information superiority has the impulse to build an enterprise empire and take excessive risks for investment (Jensen, 1986).Excessive risk taking by the management can earn a lot of personal interests, but the cost is to increase the enterprise CONTACT Xiangting Kong kongxt5@mail.sysu.edu.cnSchool of Business/Center for Accounting, Finance and Institutions, Sun Yat-sen University, Guangzhou, China Paper accepted by Guliang Tang.risk and damage the wealth of investors (Zhan & Wang, 2013).Therefore, how to restrain the agency behavior of the management in the risk taking process and protect the wealth of investors is a classic topic in the field of corporate governance.
The independent director system is an important mechanism to solve the agency problem of enterprises and reduce the risk of enterprises, which helps to protect the wealth of investors and maintain the stability of the capital market.Studies have shown that many factors influence the risk choices of firms, including the external institutional environment (Deng et al., 2021;Hilary & Hui, 2009;John et al., 2008), corporate governance mechanisms (Coles et al., 2006;Faccio et al., 2011;Kini & Williams, 2012;Low, 2009), and manager characteristics (Faccio et al., 2016;Kong et al., 2020;Serfling, 2014).The independent director system is an important mechanism to solve the agency problem of enterprises and reduce the risks of enterprises.It helps to protect the wealth of investors and maintain the stability of the capital market (Fama & Jensen, 1983).Independent directors' ability to fulfill their governance role depends on their access to firm-specific information (Adams & Ferreira, 2007).However, the information on which the independent directors depends to make decisions mainly comes from managers, who often hide key information to protect the private benefits of management.When this happens, the governance effect of independent directors is weakened (Adams & Ferreira, 2007).The real governance effect of independent directors lies in the balance between independence and information acquisition.If independent directors can acquire more firmspecific information through their own means, how might this strengthen their governance effect and control the risk-taking behaviors related to managers' agency problems?
This paper holds that independent directors can independently obtain more real information when they participate in the shareholder meetings, which helps to suppress the agency problem of the management and reduce the risk of the enterprise.Firstly, shareholder meetings generally yield a large amount of real information that can improve the governance effect of independent directors.Compared with board meetings, shareholder meetings involve more diverse and geographically dispersed participants, who tend to have different knowledge backgrounds and interests and approach proposals from different perspectives.At such meetings, shareholders can discuss proposals at length, and independent directors in attendance can obtain real information that may be hidden by managers.Although the information of independent directors mainly comes from the board meeting, while attending the shareholder meetings can obtain additional information.The information which comes from shareholder meetings and the board meetings can confirm each other and even form new information, which is helpful for independent directors to perform their duties better.Second, limited by the form of convening the shareholder meetings, independent directors can only participate in the shareholder meetings through on-site participation, which improves the efficiency of information transmission.From the perspective of communication, on-site face-to-face communication makes the way of information transmission more diverse.People can transmit information through audio language and non-verbal symbols, which is conducive to the transmission and communication of information.The advantage of language communication lies in the symmetry, situational and timeliness of communication, and both sides can transmit and supplement information in time through interaction.The nonverbal symbols in the scene can affect the thinking ability and depth of the information receiver, and have a suggestive function.For example, those who contact the smiling speaker are more likely to have a positive attitude, while those who contact the frowning speaker are more likely to have a negative attitude (Burgoon et al., 1990;Ottati et al., 1997).Therefore, independent directors can obtain a lot of new information by attending the general meeting of shareholders on site.Finally, independent directors attending the shareholder meetings can observe the enterprise on the spot, communicate with shareholders face to face, and then obtain a large amount of new information.When independent directors go to the headquarters of a listed company to attend the shareholder meetings, it means that the geographical distance between them is shortened.Independent directors' on-the-spot observation of the production and operation of enterprises can generate new information, which can help to modify or form their views on corporate governance and improve the governance effect (Luo et al., 2018).
In our primary analyses, we test the relationship between independent directors' attendance at shareholder meetings and firm risk taking.We use two variables to measure risk taking, the first being the volatility of industry-adjusted profitability (σ(ROA)).Profitability is measured by a firm's operating return on assets (ROA), which is defined as the ratio of earnings before interest and taxes to total assets.We measure performance volatility in three-year overlapping periods.The second variable is the difference between the minimum and maximum ROA in a three-year period (Faccio et al., 2011).Holding all control variables constant, we find a statistically significant negative relationship between independent directors' attendance at shareholder meetings and the level of firm risk taking.This is consistent with the hypothesis that after obtaining more new information, independent directors can effectively identify the high-risk behaviors of the management, suppress their agency impulses, and thus reduce the degree of risk-taking of the firm.However, establishing causality is particularly challenging due to the endogenous nature of attendance rates.We have to address both the omitted variables concern (i.e.potential bias arising from unobservable, heterogeneous firm factors that correlate with both attendance rates and risk taking) and the reverse causality concern (i.e. the possibility that firms with a lower level of risk taking attract independent directors with a higher tendency to attend shareholder meetings).To establish causality, we use three efficient methods and perform a range of tests.
First, we used the high-speed rail opening as an exogenous event shock to test the impact of high-speed rail opening on the information content of the shareholder meetings, and the changes in the relationship between independent directors' participation in the shareholder meetings and firm risk taking before and after the high-speed rail opening.The opening of high-speed rail will help small and medium-sized shareholders to attend the shareholder meetings (Wang & Wang, 2021), which improves the information content of the shareholder meetings.Furthermore, after the high-speed railway is opened, independent directors can obtain more real information by participating in the shareholder meetings, which is conducive to better curb the problem of excessive risk taking by management.As an exogenous impact widely used, the opening of high-speed rail is conducive to establishing a causal relationship between independent directors' participation in the shareholder meetings and enterprise risk taking.
Second, we propose a novel instrumental variable (IV) approach that builds on the work of Han et al. (2018).We consider adverse weather, defining a day as an adverse weather day if the lowest or highest temperature passes a certain threshold (below 0°C or above 37°C) or if rainfall exceeds 10 mm.Consistent with the relevance condition of our instrument, we show that actual shareholder meeting attendance is indeed negatively related to adverse weather.Using the weather as an IV in two-stage least squares (2SLS) regression, we find strong and consistent support for the notion that a greater attendance rate causes a lower level of firm risk taking.
Third, we estimate the model using propensity score matching (PSM), which involves comparing risk taking in firms that have high attendance rates (the treatments) and those that have comparable propensity scores for attendance based on firm fundamentals but actually have lower attendance rates (the controls) (Hoi et al., 2019).Our results again remain unchanged.
We also address two important questions.First, we attempt to identify possible underlying mechanisms through which independent directors' shareholder meeting attendance reduce risk taking.Second, we investigate whether shareholder meeting attendance affects firm policy choices that determine risk taking.We find that after the independent directors obtain more information through attending the shareholder meetings, they will have lower risk tolerance to the management, and are more likely to express dissenting opinions subsequently.Additionally, we find that firms with higher meeting attendance rates for independent directors adopt less risky policies, including lower frequency of mergers and acquisitions, lower level of over investment and the will rely less on debt capital, hold more cash.After segmenting the risks, it was found that the independent directors' participation in the shareholder meetings and obtaining more information reduced the probability of violation and litigation risk of the enterprise, and reduced the probability of receiving inquiry letters.Finally, for independent directors, the more active they are in attending the shareholder meetings, the more allowance and directorships they will receive.
We further provide corroborative evidence by identifying cross-sectional factors that strengthen the baseline result.Specifically, we investigate whether the relation between independent directors' meeting attendance and firm risk taking varies according to the type of shareholder meetings, corporate information asymmetry, and management agency cost.We find that the negative relation is more pronounced when the shareholder meeting involves more proposals, when information asymmetry is greater and when the agency cost of management is serious.
This study makes three contributions to the literature.Firstly, through the unique perspective of independent directors participating in the shareholder meetings, it is demonstrated that mastering real information will help improve the monitor effect of independent directors.Compared with previous studies that indirectly demonstrated the relationship between information and independent directors' performance through model derivation and questionnaire surveys (Brooks et al., 2009;Masulis & Zhang, 2019), the evidence in this paper is more direct.As for the relationship between independent directors' information acquisition and their performance, the existing research mainly focuses on independent directors' participation in board meetings, but in fact, the governance role of the directors may have some limitations, such as the possibility of meeting information being deleted or even deliberately distorted by the management.In addition, China Securities Regulatory Commission forces independent directors to participate in board meetings, which leads to small differences between independent directors' participation behavior.However, the US enterprises only discloses whether independent directors are absent from more than 25% of board meetings, and outsiders cannot observe the details of independent directors' attendance at board meetings.By focusing on board meeting attendance, previous studies offer indirect evidence and mixed results (e.g.Adams & Ferreira, 2008;Chou et al., 2013;H. Liu et al., 2016;Vafeas, 1999).In fact, many behaviors of independent directors are worthy of further discussion, such as independent directors' participation in shareholder meetings (Zhou & Wang, 2021) and on-site office.These non-traditional work scenes have not attracted the attention of researchers.From the perspective of independent directors' participation in shareholder meeting, this paper examines whether independent directors can obtain more real information to improve their governance efforts.
Second, our findings enrich understanding of how independent directors affect corporate policies.Consistent with the arguments of Bernile et al. (2018) and Brick and Chidambaran (2008), we show that monitoring by independent directors is an important determinant of corporate policies and risk taking.Our evidence shows that independent directors' participation in the shareholder meetings obtains more real information, it is helpful for independent directors to identify and suppress high-risk decisions of the management, so as to reduce the unreasonable risk taking level of the management.It includes a series of policies such as merger and acquisition decision, investment decision, debt decision and cash holding decision.
Third, our novel IV approach provides a framework for making causal inferences under more general circumstances.Our method is similar in spirit to that of Han et al. (2018), who use the percentage of days in a given year with extreme weather conditions in the city where a company's headquarters is located as a proxy for shareholder meeting attendance.However, this method seems to be too rough, as it does not tell us whether the weather was extreme on a specific day.Our approach allows us to accurately identify the weather on the day of the shareholder meeting and use it as an IV.This is particularly important in our research context, as it alleviates concerns over the interaction between independent directors' choice to attend meetings and firms' inherent risk profiles.

Literature review
Risk taking is an important force to enhance shareholders' wealth and promote social progress (John et al., 2008), but at the same time, the management also has the motivation to seek private interests through adopting high-risk policies.At present, there are rich research results on the influencing factors of enterprise risk taking.The most relevant literature can be roughly divided into the following two categories.
Firstly, the independent directors and management risk-taking.The purpose of introducing the independent director system is to solve the agency problems caused by the separation of ownership and control, which will affect the incentive of executives and ultimately affect their risk-taking behavior.W. Xie and Tang (2013) conducted a more comprehensive test on the relationship between the governance mechanism of listed companies and risk-taking, and found a positive correlation between the proportion of independent directors and enterprise risk-taking.Enterprises with high proportion of independent directors have more innovation output, but they reduce the quality of innovation (Wu & Zhang, 2019).Subsequent studies found that the proportion of independent directors alone could not improve their monitor role, nor significantly affect the risk-taking behavior of management, and the R&D investment of enterprises did not increase significantly (Xiao, 2016).The following research gradually shifted to the impact of independent directors on managements' risk-taking behavior.In view of the high internal uncertainty of R&D activities and serious information asymmetry, independent directors with technical background can effectively monitor managements' investment activities and improve it's efficiency (Hu, 2012).The main role of politically connected independent directors is to obtain political resources for enterprises, while the role of monitoring management is weak.In addition, politically connected independent directors can also increase the anti-risk ability of enterprises when they are punished for violations, and finally improve the risk-taking behavior of management (Zhou et al., 2018).However, foreign studies show that the system design with the majority of independent directors can increase the level of enterprise risk-taking.Taking the policy of Sarbanes-Oxley requiring the majority of independent directors on the board of listed companies as an exogenous impact, the study found that independent directors can improve the risktaking level of management, promote enterprises to invest in projects that are conducive to the long-term development of the enterprise (Balsmeier et al., 2016).
Second, information acquisition of independent directors and governance effect of independent directors.Information is the basic guarantee for independent directors to improve the governance effect (Adams & Ferreira, 2007).Independent directors' governance role needs to be based on sufficient information (Masulis & Zhang, 2019).Having a reliable information access channel is the premise for independent directors to obtain decision-making related information and improve governance effectiveness (Tang & Luo, 2006).Most of the information held by independent directors comes from the management, but the management will reduce the information supply to independent directors for the purpose of avoiding supervision, such as less prior communication, unreasonable meeting schedule and reduced meeting materials supply.56% of the independent directors interviewed believed that the enterprise had not fully negotiated with themselves on the time of the board meeting, which led to their absence from the board meeting (Y.Yang & Huang, 2015).Brooks et al. (2009) conducted a questionnaire survey on 684 independent directors of the largest 200 listed companies in Australia.Most of the independent directors participating in the survey believed that they lacked information access channels.At present, they are only limited to meeting with business executives, investigating business departments, conducting informal exchanges with board members and participating in other activities of enterprises.The lack of information led to independent directors even having to give up part of their monitor functions in exchange for information needed for governance (Adams & Ferreira, 2007).The optimal board size and structure of an enterprise is often a trade-off between encouraging insiders to disclose private information, reducing the coordination cost of outsiders and improving the professional ability of outsiders (Raheja, 2005).When the barriers to information access are too large, shareholders are more inclined to form a board controlled by insiders in exchange for insiders to provide more information (Harris & Raviv, 2008).Therefore, the more information is obtained, the better the performance of independent directors will be.Duchin et al. (2010) found through empirical research for the first time that simply increasing the number of independent directors can not significantly improve enterprise performance, and whether independent directors can play a governance role depends on their information acquisition cost.Huang and Hilary (2018) found that the longer the term of independent directors is, the more they understand the real business situation of the enterprise, and the more they can improve the performance of the enterprise.Masulis and Zhang (2019) found that when the independent directors are distracted by health problems, award-winning problems or major events of other enterprises which they holding directorships, its information access will decline, which will lead to less participation in board meetings, less trading in the company's shares, and easier sudden resignation from the company, which will lead to worse corporate performance.
From the literature review, it can be seen that although the existing research has found the importance of information for independent directors' decision-making, the existing information access channels are mostly limited to local information advantages, independent directors' networks or external information environment.There is a lack of necessary discussion on whether independent directors can open up channels to obtain real information independently.This paper finds that independent directors' participation in the shareholder meetings is a good channel to obtain information, which is also more direct and objective.
In addition, although some studies have found that obtaining information helps to improve the governance effect of independent directors and affects the risk-taking behavior of management.However, we still do not know how information is translated into the governance effect of independent directors, and what mechanism independent directors use to affect the level of risk-taking of enterprises.Based\on this, this paper takes independent directors' participation in the shareholder meetings as a special scenario to explore whether independent directors can obtain more information when they participate in the shareholder meetings and what mechanism can be adopted to improve the governance effect, and affect the level of risk taking of management.

Hypothesis development
Risk taking is a classic topic in the field of corporate governance, and also an important embodiment of the agency problem between shareholders and management (Jensen & Meckling, 1976).At the macro level, the management actively takes risks.The enterprise's choice of high-risk but high return projects will help the enterprise obtain a large amount of capital gains and drive economic development (John et al., 2008).At the micro level, the active assumption of risk by the management means that the enterprise has a large amount of capital expenditure (Bargeron et al., 2010), which is conducive to the upgrading of the enterprise's products and equipment, or the realize of scale advantage, and ultimately contributes to the long-term development of the enterprise.However, in the process of risk taking, the management also has serious agency problems.For example, the management pursues the scale advantage of the enterprise through a large number of high-risk investments to meet its selfish desire to build an enterprise empire or seek higher salaries (Jensen, 1986).Correspondingly, in order to avoid reputation loss and even career loss, the management may also be risk neutral or risk averse.They will reject those projects with positive NPV but possible risks, so as to obtain control power for private benefit steadily (Bertrand & Mullainathan, 2003;Chakraborty et al., 2006).In a word, there are a lot of agency problems in the process of management taking risks.
In China's capital market, a large number of insiders take risks excessively for personal gain (W.Yang et al., 2018;Zhang et al., 2010).The reason for this phenomenon is mainly related to the imperfect capital market system.The main reasons are as follows: First of all, listed companies in China's capital market have a large amount of free cash flow, which provides an economic basis for the management to take excessive risks.Wind data statistics show that the dividend yield of listed companies in China's capital market will be only 1.4% in 2021. 1 The low proportion of cash dividends causes enterprises to hold a large amount of free cash flow.According to the free cash flow theory, free cash flow is very easy to cause management to carry out high-risk behaviours such as excessive investment (Stulz, 1990), or to carry out high on-the-job consumption, which ultimately increases enterprise risk (Xu et al., 2014).Secondly, the management made high-risk investments for the purpose of reducing its holdings, driving up the share price and increasing the enterprise risk.Due to equity incentive or free purchase and other reasons, the management of listed companies usually hold a certain number of shares.At this time, the personal wealth of the management is closely related to the stock price of the enterprise, and the way to seek personal interests is to reduce and cash out, and the higher the stock price is, the more in line with their personal interests.Therefore, a very natural logical behavior is that the management will conduct a large number of investment mergers and acquisitions before reducing its stock holdings, which will affect investors' perception by taking high risks and promote the rise of enterprise stock prices.However, this kind of agency behavior of the management will translate into the risk of stock price crash risk in the future (W.Yang et al., 2018).Third, the major shareholders of listed companies also have the motivation to seek private interests, and often transfer risks to listed companies while obtaining private interests.In China's capital market, there is a prominent agency conflict between large and small shareholders.In order to obtain the private interests of control, large shareholders will occupy the resources of listed companies, manipulate listed companies to take excessive risks, and cooperate with their opportunistic behavior.For example, the controlling shareholders will manipulated listed companies to conduct earnings management under the situation of equity pledge (D.Xie & Liao, 2018), and conducted mergers and acquisitions to reduce their holdings, and signed high-level performance commitments (H.Ye & Chen, 2022).These agency behaviors aggravate enterprise risks (Zhou & Wang, 2021) and seriously damage shareholders' interests.
Independent directors are an important institutional arrangement to protect the interests of investors (Fama & Jensen, 1983).Having sufficient decision-making related information can help independent directors curb the problem of excessive risk taking by enterprises.Independent directors should make correct decisions on the basis of mastering enough real information.For this, existing research has provided a lot of evidence from multiple perspectives.The relationship network formed by independent directors among different enterprises through part-time director seats is a good channel for obtaining information, which helps independent directors to obtain resources and information to supervise the agency problems of insiders, provide better advice, thus reducing excessive investment (Y.Chen & Xie, 2011), and improving corporate M&A performance (Schonlau & Singh, 2009), In particular, cross-border mergers and acquisitions with higher degree of information asymmetry and greater risk (Guo & Lv, 2018).Local independent directors can obtain enterprise information from formal and informal channels, and have rich social capital.This will help independent directors to improve the performance of enterprises' M&A of local enterprises and reduce the risk of M&A failure (C.Liu et al., 2015).
Independent directors can obtain more true information by attending the shareholder meetings.First of all, the shareholder meetings has rich information content.The shareholder meetings is the highest authority of an enterprise.Shareholders have the legal right to attend the meeting no matter how much they hold.The shareholder meetings is the main way for small shareholders to participate in corporate governance and vote by hand.The participation of numerous small and medium-sized shareholders in the shareholder meetings is conducive to giving play to collective wisdom, discussing and voting on enterprise proposals, and providing incremental information.Shareholders have different knowledge backgrounds, interest positions and appeals, and discussions among different shareholders are conducive to exploring the information that the management deliberately hides or the independent directors have not yet mastered (Willoughby & O'Reilly, 1998).
Secondly, a large number of stakeholders participated in the shareholder meetings, which is conducive to the follow-up interpretation of relevant information of the enterprise.In addition to corporate directors, supervisors and managements, there are many individual investors and institutional investors participating in the shareholder meetings, which requires the witness of a law firm to ensure that the convening and holding procedures of the shareholder meetings comply with the Company Law and other laws and regulations.The discussion at the shareholder meetings and even the conflict between shareholders will cause the follow-up attention, discussion and interpretation of stakeholders (Dyck et al., 2008), which is also conducive to independent directors obtaining more information.
Finally, from the perspective of communication, independent directors can obtain a lot of real information through verbal and nonverbal signs when attending the shareholder meetings on site.Subject to the requirements of the CSRC and the limitation of the convening form of the shareholder meetings, independent directors can only participate in the shareholder meetings in the form of on-site participation.The advantage of attending the meeting on site is that the way of information transmission is more diverse, and independent directors can obtain information in the form of vocal language and nonverbal symbols.Language communication is characterized by interactivity, context and timeliness.Communicators can timely conduct supplementary questions, in-depth questions, etc., which is helpful to mining hidden information.More importantly, the information transmitted by nonverbal symbols is more flexible and rich.Non linguistic symbols include body language, facial expression, etc. of the information publisher.In the interactive communication, the proportion of information transmitted by nonverbal symbols can reach 65% (Birdwhistell, 1970), mainly because: First, nonverbal symbols can affect the thinking ability and depth of the information receiver, especially when the information receiver has a single access to information.Second, nonverbal signs can influence the way of thinking of the information receiver.For example, communication with smiling people is more likely to generate positive attitudes, while communication with frowners is more likely to generate negative attitudes (Burgoon et al., 1990;Ottati et al., 1997).Third, nonverbal signs have suggestive functions.If the information receiver has a negative view on the body, expression and dress of the information publisher, it will lead the information receiver to refuse to trust the information (Briñol et al., 2007).
The independent directors' participation in the shareholder meetings to obtain more real information will help them to constrain enterprise risk decision-making and reduce the phenomenon of excessive risk taking by enterprises due to agency problems.Independent directors in China's capital market are mostly composed of professors, enterprise executives, accountants and law firms, and usually have high professional reputation (J.Tan et al., 2003).If the independent directors are punished due to firm risks during their performance of duties, their reputation will be damaged.The director market will severely punish the independent directors with damaged reputation (Fama & Jensen, 1983), and their directorships will decline significantly in the future (Fich & Shivdasani, 2007).Therefore, independent directors are usually very sensitive to corporate risks and the personal reputation risks that may arise from them, and even voluntarily resign before the outbreak of corporate financial risks (Gao et al., 2017).
It can be seen from this that the independent directors can obtain the true information of the enterprise when they attend the shareholder meetings, which helps them to accurately evaluate the matching degree between the current business situation and the risk level of the enterprise, predict the future risk events, and then adjust their performance of duties and reasonably regulate the enterprise risks (Y.Tan et al., 2022).The more risk related information independent directors get from shareholder meetings, the more independent they can make decisions, which will lead them to reduce risk tolerance (Ahmed & Duellman, 2007) and turn it into actual performance of duties.Specifically, the enhancement of reputation pressure will encourage independent directors to express more dissenting opinions (Jiang et al., 2016), restrain the unreasonable risk behavior of the enterprise, and ultimately reduce the enterprise risk and personal performance risk.
Based on the above analysis, we believe that the independent directors participating in the shareholder meetings can learn about the concerns of small and medium-sized shareholders from shareholders' exchanges, discussions, on-site nonverbal signs and other channels, obtain more real information about the enterprise, and understand the risk issues and potential sources of risk concerned by the capital market.Finally, it will help the independent directors to better play their supervision and consulting functions and reduce the unreasonable risk bearing of the enterprise.This paper proposes the following assumptions: Hypothesis: Independent directors attending the shareholder meetings can obtain more real information and reduce the enterprise's excessive risk taking behavior accordingly.

Institutional background
In 2001, China Securities Regulatory Commission issued the guiding opinions on the establishment of independent director system in listed companies (The opinions), since then, the independent director system has been established in China's capital market.The 'opinions' specifies the proportion of independent directors in the board, their qualifications, election procedures and responsibilities.If an independent director fails to attend the board meeting in person for three consecutive times, the board of directors shall request the shareholder meeting to replace him.It can be seen that the opinion attaches great importance to the participation of independent directors in the board meeting, but it does not make necessary explanation for the participation of independent directors in the shareholder meetings.In other company laws and regulations, there are some corresponding requirements for independent directors to attend the shareholder meetings.
For example, article 150 of the company law promulgated in 2005 stipulates that 'if the shareholder meetings require directors, supervisors and senior managers to attend the meeting as nonvoting delegates, directors, supervisors and senior managers shall attend the meeting as nonvoting delegates and accept the questions raised by shareholders'.According to the requirements of the company law, all directors, including independent directors, should attend the meeting in accordance with the requirements of the shareholder meetings.The 'rules of shareholder meetings of listed companies' promulgated in 2006 has made more detailed provisions on independent directors' participation in shareholder meetings.Article 26 stipulates that 'when a listed company holds a shareholder meeting, all directors, supervisors and secretary of the board shall attend the meeting, and managers and other senior managers shall attend the meeting as non voting delegates'.Article 28 stipulates that at the annual shareholder meeting, the board of directors and the board of supervisors shall make a report on their work in the past year to the shareholder meeting, and each independent director shall also make a report on their work.In addition to the mandatory laws and regulations issued by government departments, article 8 of the guidelines for the performance of duties of independent directors of listed companies issued by China Association of Listed Companies in 2014 stipulates that independent directors shall attend the shareholder meetings of Listed Companies in person and communicate with shareholders on the spot.
To sum up, many rules and regulations stipulate the necessity for independent directors to attend the shareholder meetings, but they do not specify the punishment measures for the absence of independent directors from the shareholder meetings.Therefore, the provision of independent directors' participation in the shareholder meetings can be regarded as a 'semi mandatory' policy.According to the data of the article, at the individual level, independent directors participated in about 63.38% of the shareholder meetings.At the enterprise level, the frequency of independent directors attending the shareholder meetings is about 62.7%.Although a considerable number of independent directors are absent from the shareholder meetings, no independent directors have been punished by the regulatory authorities.However, the absence of independent directors from the shareholder meetings may be reported by the media, causing concern in the capital market and reputation loss.There are only two forms for independent directors to attend the shareholder meetings, which are on-site attendance and absence, excluding entrusted attendance and online attendance.

Sample selection
Our sample consists of firms listed on the Shanghai Stock Exchange during the 2012 to 2017 period.We manually collect information on the participation of each independent director in shareholder meetings from annual reports and independent director work reports.The sample period starts from 2012 because this is the earliest year for which we are able to obtain data about independent directors' shareholder meeting attendance.The sample focuses on firms listed on the Shanghai Stock Exchange because this exchange, unlike the Shenzhen Stock Exchange, discloses detailed information on each independent director's meeting attendance.All other data are obtained from the China Stock Market and Accounting Research (CSMAR) database, a leading financial data provider in China.
Following the literature, we exclude financial firms such as banks, insurers, and investment trusts (due to their different accounting standards), special treatment firms, and firms for which data needed to measure control variables are not available.To minimize the influence of outliers, we winsorize all continuous variables at the top and bottom one percentile.We are left with a final sample of 4,510 firm-year observations in our baseline regressions.

Dependent variable: risk taking
Consistent with prior studies (e.g.Faccio et al., 2011;John et al., 2008), we consider two measures of risk taking.The first measure is the volatility of industry-adjusted profitability (σ(ROA)).Profitability is measured by a firm's operating ROA, which is defined as the ratio of earnings before interest and taxes to total assets.We measure performance volatility in three-year overlapping periods.The second measure is the difference between the minimum and maximum ROA in a three-year period.In a robustness test, we use fiveyear overlapping periods to recalculate the risk-taking variables.
N represents the total number of companies in J industry in year t.

Independent variable: independent directors' shareholder meeting attendance
In China's capital market, firms that are publicly listed on the Shanghai Stock Exchange have since 2012 been required to disclose the shareholder meeting attendance of independent directors in their annual reports.Specifically, the corporate governance section of an annual report must include the date of the shareholder meetings, the name of the proposals, and the number of times each independent director attended the shareholder meeting.
We use the variable AttendRate as our measure of independent directors' shareholder meetings attendance.AttendRate is calculated as the ratio of the number of shareholder meetings actually attended by independent directors scaled by the number of shareholder meetings they are required to attend during a year.The number of shareholder meetings an independent director is required to attend during a year is the number of shareholder meetings their firm holds.

Control variables
Following Coles et al. (2006) and Faccio et al. (2011), we select the following control variables to capture other determinants related to risk taking: firm size (Size), leverage (Lev), ROA (ROA), sales growth (Growth), CEO duality (Dual), director number (Board), independent director ratio (Independent), state-owned enterprise(SOE), the percentage of shares owned by the largest shareholder (Share1), capital expenditure (Capital), the ratio of fixed assets (PPE), and firm age (Age).
Detailed definitions for all variables used in this study are reported in Table 1.

Baseline regression models
We use the following empirical specification to test the implications of our hypotheses: where i indexes firm and t indexes time.The dependent variable Risk1 is the volatility of industry-adjusted profitability in a given period, and Risk2 is the difference between the minimum and maximum ROA in a given period.The independent variable AttendRate is the ratio of the number of shareholder meetings actually attended by independent directors scaled by the number of shareholder meetings they are required to attend during a year.X is a vector of the control variables defined in Table 1.We estimate ordinary least squares regressions and use year and industry fixed effects with heteroscedasticity in all specifications.

Risk1
Volatility of industry adjusted profitability in a certain period.

Risk2
The difference between the minimum and maximum ROA in a certain period.

Independent variable
AttendRate Ratio of the number shareholder meetings the independent directors actually attend scaled by the number of shareholder meetings the independent directors required during a year.

Size
Logarithm of total asset for a firm at year end.

Lev
Ratio of total liabilities scaled by total assets.

ROA
Return on assets for a firm in a given year.

Growth
Sales growth.

Dual
Dual is defined as a dummy that is equal to 1 if the CEO serves as chairman and is zero otherwise.

Board
The number of directors on the board at year end.Independent The percent of independent directors in a board.

Share1
The percentage of shares owned by the largest shareholder.

SOE
If the enterprise is state-owned holding, the value is 1; otherwise, the value is 0.

Capital
Ratio of capital expenditure divided by total assets.

PPE
Net properties, plant and equipment scaled by total assets.

Age
The natural logarithm of (1 + the number of years since found).

Baseline regression results
We first examine whether AttendRate influences firm risk taking and report the baseline regression results in Table 3.The baseline models use ordinary least squares (OLS) regression, with T-values adjusted by the White heteroscedasticity test.In the regressions of the dependent variables Risk1 and Risk2, the coefficients on AttendRate are negative and significant at the 5% level, at −0.004 and −0.008, respectively.These findings are consistent with our hypothesis.With regard to control variables, firms with larger size, higher profitability, and higher capital expenditure are less likely to take risks, while firms that are older and have more debt are more likely to take risks.These findings are consistent with prior studies (e.g.Faccio et al., 2016;John et al., 2008).

Endogeneity
We need to address potential endogeneity concerns in our specification.For example, the association between AttendRate and risk taking could be spurious if our baseline models omit factors that correlate with both AttendRate and risk taking.It is also possible that independent director's AttendRate may be non-random, potentially leading to self-selection bias.Additionally, there is the concern that the level of risk taking may influence the tendency of independent directors to attend shareholder meetings (i.e.reverse causality).
To solve endogeneity problems, we use several econometric techniques specifically designed for this purpose: Natural experiment, 2SLS regression, PSM, alternate measures, and an alternate regression method.

Exogenous impact of high-speed railway opening
Previous studies have shown that the opening of high-speed railway (HSR) is an exogenous event, which can compress the spatial distance of information transmission, and will have an important impact on the information environment of enterprises.After the opening of the high-speed railway, shareholders which in other cities can more easily participate in the shareholder meetings, which can improve the information content of the shareholder meetings, and then help the independent directors to obtain more real information.Therefore, we expect that when the high-speed railway is opened in the city where the enterprise is located, the risk-taking behavior of the enterprise will be inhibited.
Referring to previous studies (S.Chen & Liu, 2019;Chu & Fang, 2019;Pan et al., 2019).Firstly, this paper takes the opening of high-speed railway as an exogenous impact to test the impact of the opening of high-speed railway on the degree of enterprise risk taking.In this part, only the newly opened high-speed rail enterprises in the cities where they are registered during the sample period are retained.Hsropen is a dummy variable, if the city where the enterprise is registered has opened high-speed railway at the end of the year, the value is 1, otherwise it is 0. We use model (4) to verify the impact of high-speed rail opening on the results of this paper.
The results of column (1) in Table 4 show that the coefficient of Hsropen is −0.006, which is significant at the level of 5%.This shows that after the opening of high-speed railway, the level of risk taking of enterprises in the region has decreased significantly.In order to further determine the effect of independent directors getting more information, column (2) and column (3) respectively verify the relationship between independent directors attending the shareholder meetings and risk taking before and after the opening of highspeed railway.The results show that the relationship between the participation of independent directors in the shareholder meetings and the risk taking of enterprises is not significant before the high-speed railway is opened.However, after the high-speed rail opened, the independent directors' participation in the shareholder meetings was significantly negatively correlated with the risk taking of the enterprise.The results of Table 4 show that the independent directors obtain more information is one of the reasons for the decrease of the risk taking degree of the enterprises after the highspeed railway is opened, which verifies the robustness of the conclusion in this paper.

Regression results using weather as an IV for AttendRate
We first address the reverse causality concern that risk-taking decisions may affect the likelihood of independent directors' shareholder meeting attendance.In other words, independent directors may attend shareholder meetings more frequently in firms with lower risk taking.To mitigate this concern, we use a novel weather-based IV approach based on the work of Han et al. (2018), using the weather conditions where a firm's headquarters is located as the IV.First, we identify the exact time and location (date and city) of a shareholder meeting.Second, we download weather conditions from the China Surface Climate Dataset provided by the China Meteorological Data Sharing Service System.2Third, we match the weather conditions with the previously identified shareholder meeting date and city.Fourth, we define a day as an adverse weather day if the lowest or highest temperature passes a certain threshold (below 0°C or above 37°C) or if rainfall exceeds 10 mm.The variable Weather is set to one if weather conditions are adverse on any day of a shareholder meeting in the city where a firm's headquarters is located, and zero otherwise.Obviously, Weather is an exogenous variable, able to hinder meeting attendance but unable to affect firm risk-taking behavior.
Table 5 reports the first-and second-stage IV estimates.The first-stage estimates show that there is a negative relation between Weather and AttendRate, which is significant at the 1% level.Column (1) shows that the T-statistic on the first-stage instrument is −7.414, and the F-statistic is greater than 54, indicating that our IV passes the weak instrument test.Columns (2) and (3) report the second-stage regression results.The coefficients on AttendRate^ remain negative and retain their significance at the 10% level.These findings offer plausible causal evidence of the effect of independent directors' shareholder meetings attendance on firm risk taking.

Results from the PSM sample
However, if independent directors' attendance decisions are endogenous, the sample may suffer from a self-selection problem.To mitigate this concern, we use the PSM method, which allows us to identify a control sample of firms that have high meeting attendance rates and exhibit no observable differences in characteristics relative to firms with low meeting attendance rates.The matching of observable firm characteristics mitigates concerns related to non random selection.Follow prior research (Caliendo & Kopeinig, 2008;Hoi et al., 2019), we rank AttendRate based on the data for a given year and classify firm-years in the top tertile as the treatment group and those in the bottom tertile as the control group.We then generate a new variable, HighAttendRate, which is set to one if a firm-year belongs to the treatment sample and zero if it belongs to the control sample.Based on this sample, we generate propensity scores by running a logistic regression with HighAttendRate as the dependent variable; the independent variables include all variables as specified in the baseline model.Then, based on the one-to-one matching principle, we use matching without replacement with a caliper of 1% to find the closest match.6 show that the estimates on AttendRate remain negative and significant at the 1% levels.

Sensitivity to alternative measures of risk taking
In the baseline regression, we measure risk taking using the volatility of industry-adjusted profitability (σ(ROA)) and the difference between the minimum and maximum ROA in a three-year period as proxies.In the robustness test, we use a five-year period instead and rename the resulting variables Risk3 and Risk4.This reduces our sample size to 2,525.Table 7 presents the results, showing that the coefficients on AttendRate are −0.005 and −0.011, which are significant at the 10% level.Thus, our results are robust to using Risk3 and Risk4 as alternative measures.

Sensitivity to an alternative estimation method
To further overcome the influence of auto correlation and heteroscedasticity on statistical inference, we cluster T-values at the firm level and time level simultaneously (Petersen, 2009).The results in Table 8 show that the baseline results still hold after adopting this more robust model.As seen in columns ( 1) and ( 2), the coefficients on the relationship between AttendRate and risk taking are 0.006 and −0.011, which are significant at the 10% level.In order to further ensure the robustness of the model and further alleviate the problem of potential omitted variables, the paper also uses the fixed effect model and made clustering adjustment at the firm level for further verification.The results are shown in columns (3) and ( 4) of Table 8.The coefficient of AttendRate in column (3) is −0.004 and that in column (4) is −0.008, which are significant at the 10% level, which is consistent with the previous conclusions.
After a rich set of endogeneity and robustness tests, the significant negative relation between independent directors' shareholder meetings attendance and firm risk taking still exists.These results strongly support our hypothesis.

Possible mechanism
Based on the above tests, our findings indicate that independent directors' shareholder meeting attendance can reduce firms' risk-taking behaviors.However, the reason for this phenomenon remains unknown.Based on the above theoretical analysis, we think that independent directors who participate in shareholder meetings are in a better position to acquire the real information.This will help them to accurately evaluate the matching degree between the current business situation and the risk level of the enterprise, and then adjust the performance behavior and reasonably regulate the enterprise risk.After independent directors obtain more risk related information from shareholder meetings, they will be urged to reduce their risk tolerance (Ahmed & Duellman, 2007) and turn it into actual performance of duties.Specifically, the increase of reputation pressure will prompt independent directors to express more dissenting opinions (Jiang et al., 2016), restrain the unreasonable risk behavior of enterprises, and ultimately reduce enterprise risks.
In order to verify the relationship between information acquisition of independent directors and risk tolerance, referring to previous studies, we choose CEO pay-forperformance sensitivity as a measure of enterprise risk tolerance.This practice has been applied in a large number of documents at domestic and abroad.Its internal logic is that the business operation and investment activities of enterprises have a high degree of uncertainty, so there is a high risk of failure.As the salary of the management is linked to the enterprise performance, it fluctuates with the performance.Therefore, when the performance of the enterprise declines, the salary of the management will also be greatly reduced, and the mechanism of salary linked to performance can automatically adjust the high-risk behavior of the management.Excessive risk taking by management will increase the possibility of their future earnings decline, which will prompt management to reduce agency problems in business and investment activities (Francis et al., 2011) and reduce some projects with excessive risks (X.Chen et al., 2021;Ederer & Manso, 2013;Jiang & Yao, 2015;Manso, 2011).If the company maintains a high level of executive pay-for-performance sensitivity, it shows that the company has a low tolerance for future investment failures of executives.
In addition, we hope to prove that independent directors have changed their personal risk attitude by observing the changes in the performance of their duties.Since the performance of duties of independent directors is mainly demonstrated by voting at the board meeting, and the independent directors' dissenting opinions will indeed delay the implementation of the enterprise's proposals, or urge senior executives to modify the project plan subsequently, the independent directors' voting can better represent the individual's risk attitude and supervise the enterprise's agency issues (K.Ye et al., 2011).
In Table 9, we examine the relationship between AttendRate and firm risk tolerance.We use CEO pay-for-performance sensitivity as the measure of firm risk tolerance, and define CEO pay as the natural logarithm of total CEO cash compensation.As seen in column (1) of Table 9, the coefficient of ROA is 2.233 and the coefficient of AttendRate*ROA is 1.150, the interactive item is significant at the 5% level.The results in Table 9 show that independent directors have a better understanding of the business situation and risk information of the enterprise after attending the shareholder meetings and obtaining more information.On the basis of mastering risk information, the independent directors will reduce the risk tolerance of the board and improves the CEO pay-for-performance sensitivity to reduce the phenomenon of excessive risk taking by the management (Manso, 2011).
The article then discusses whether the possibility, frequency of independent directors' dissenting opinions are increased after they attend the shareholder meetings to obtain more information.Referring to previous studies (Jiang et al., 2016;K. Ye et al., 2011), the article first selected all independent directors of enterprises who had dissented during the sample period as the sample.This choice is based on the fact that in the same enterprise, the contents of the proposals reviewed by independent directors and the performance environment faced by them are the same, but not all independent directors express the same independent opinions.This method can avoid endogenous problems to the greatest extent.Secondly, the deadline is the date on which the enterprise raises dissented opinions, and the frequency of the sample independent directors to attend the shareholder meetings in the year before the deadline is determined by manual search.After matching, the article finally obtained a sample of 97 enterprise year independent directors.Among them, 44 independent directors have expressed dissenting opinions, and 53 independent directors have not expressed dissenting opinions.The control variables at the individual level of independent directors include: gender, age, subsidy, directorships, educational background, legal background and financial background.The regression results are shown in Table 10.
Column (1) in Table 10 uses Logit model regression to examine the relationship between the frequency of independent directors' individual participation in the shareholder meetings and the possibility of dissenting opinions in the current year.Column (2) uses OLS model regression to examine the relationship between the frequency of independent directors' attending the shareholder meetings and the number of dissenting opinions expressed in that year.The results of column (1) and column (2) show that the higher the frequency of independent directors' participation in the shareholder meetings, the greater the possibility of expressing dissent opinions and the more times they express dissent opinions.The results in Table 10 prove that the independent directors' performance of duties becomes more stable after they obtain more information, and the enterprise risk is suppressed through the behavior of dissenting opinions.

The influence of independent directors' participation in the shareholder meetings on the enterprise and themselves
In this section, to better understand how independent directors' shareholder meeting attendance affects firm risk-taking behaviors, we shift our focus to corporate policies.We predict that firms with higher meetings attendance rates adopt less risky investment and financial policies.We test this prediction using M&A and over-investment as a proxy for investment policies, leverage and cash holdings as proxies for financial policies (Bernile  et al., 2018).The M&A variable is defined as M&A numbers in the next year, the overinvestment in the next year is calculated as the method of Richardson (2006), and the leverage variable is defined as total debt scaled by total assets in the next year, the cash holdings variable as cash and cash equivalents scaled by total assets in the next year.We examine the relations between AttendRate and M&A, over-investment, leverage and cash holdings and present the results in columns (1) to (4) of Table 11.The respective coefficients of AttendRate are −0.0598,−0.009, −0.021 and 0.007, which are all significant.These results indicate that, all else equal, firms with higher shareholder meetings attendance rates for independent directors have less M&A frequency and lower overinvestment level, and rely less on debt capital, hold more cash.This suggests that independent directors, based on their advisory and monitoring functions, can influence investment and financial policies by attending shareholder meetings.Furthermore, such policy changes can be interpreted as a change in firm risk-taking behaviors.
Next, we conducted a deconstruction analysis of the risk composition of the enterprise.Table 12 examines the relationship between the participation of independent directors in the shareholder meetings and whether the enterprise violates the rules, is sued and receives the inquiry letter in the next year.The results show that the higher the frequency of independent directors' participation in the shareholder meetings, the lower the probability of violations, litigation and receipt of inquiry letters in the next year.These results prove from the perspective of enterprise risk sources that independent directors' participation in the shareholder meetings can obtain more real information, which will help them reduce the future risks of the enterprise.
The independent directors' active participation in the shareholder meetings will help them to obtain more information, which will help them improve corporate governance and reduce their unreasonable risk bearing level.Then, will the active performance of independent directors contribute to personal career development in the future?To explore this issue, Table 13 examines the relationship between individual independent directors' participation in the shareholder meetings and their individual salary level and directorships in the next three years.The results show that if independent directors actively participate in the shareholder meetings, they will be rewarded by the directors' labor market, and will be able to obtain significantly higher salaries and directorships in the future.

What types of shareholder meetings matter more for firm risk taking?
In the next set of tests, we examine whether the baseline results reported in Table 3 reflect the dominant effects of some aspect of independent directors' shareholder meetings attendance.We first examine what types of shareholder meetings matter more for firm risk taking.We hypothesis that independent directors who participate in shareholder meetings are in a better position to acquire the real information necessary to mitigate agency problems, which then leads to a reduction in firm risk-taking.Therefore, the negative relationship between meeting attendance and risk taking should be more pronounced when the amount of real information obtainable by directors is larger.We examine the moderating effects of certain shareholder meeting features and independent directors features.This paper summarizes the number of proposals considered by the shareholder meetings held by the firm in that year, and then takes the median number of annual summarized proposals as the standard and divides the samples into 'more proposals' and 'less proposals' for grouping test.The calculation method of the number of shareholders and the shareholding ratio of shareholders participating in the meeting is consistent with the number of shareholders.
Panel A of Table 14 presents the results of the split-sample analysis.The dependent variables are Risk1 and Risk2, and the independent variable is AttendRate.Specifically, Risk1 is the dependent variable in columns (1) and ( 2).The coefficient estimate for AttendRate is 2.5 times larger in the more proposals subsample, and the coefficient is only significant in the more proposals group.The results in columns (3) and ( 4) are similar to those in columns (1) and (2).Panel B presents the result of the number of shareholders and the shareholding ratio of shareholders participating in the meetings, and the result is consist with Panel A.
Next, we test whether meeting attendance by independent directors who serve on compensation committees (denoted as IDCM) or chair a compensation committee (denoted as IDCCM) matters more in terms of firm risk.The above theoretical analysis shows that independent directors will carefully assess the enterprise risk and reduce

M&A-Num
the risk tolerance of the board after obtaining more real information.An important performance is to improve CEO pay-for-performance sensitivity.Therefore, we speculate that the participation of independent directors serving in the remuneration committee in the shareholder meetings can more effectively reduce the level of enterprise risk bearing.
The dependent variables in Table 15 are Risk1 and Risk2, and the independent variable is AttendRate.In column (1), the coefficient estimate for AttendRate is −0.005 and significant at the 1% level, but the coefficient in column (2) is 0.005 and insignificant.This indicates that the participation of IDCM in shareholder meetings has a more pronounced negative effect on firm risk-taking behaviors.Likewise, in column (5), the coefficient estimate for AttendRate is 0.006 and significant at the 1% level, but the coefficient in column (6) is 0.001 and insignificant.Again, this indicates that the attendance of IDCCM at shareholder meetings has a more pronounced negative effect on firm risk-taking behaviors.At the same time, the absolute values of the coefficients of AttendRate in columns ( 5) and ( 7) are bigger than those in columns (1) and (3), which indicates that IDCCM can acquire more real information than IDCM by attending the meetings.

When does independent directors' meeting attendance matter more for firm risk taking?
The evidence so far consistently shows that independent directors' shareholder meeting attendance leads to lower firm risk.However, it is plausible that this relationship may vary along dimensions that determine the degree of information asymmetry.Therefore, we examine whether the effect of meeting attendance depends on firm characteristics.Furthermore, to the extent that the effect arises due to access to real information, we expect it to be more pronounced when the (3)   degree of information asymmetry is higher or the agency cost of executive is higher.
We use analyst coverage and big four auditors to proxy for information asymmetry.We split the sample along the median of the number of analysts covering a firm in a given year and industry, and the results are shown in columns (1) to (4) of Table 16.We also split the sample according to whether a firm appoints one of the big four auditors, with the results shown in columns (5) to (8).
Table 16 reports the moderating effects of analyst coverage and auditor size on the relationship between independent directors' shareholder meeting attendance and risk taking.As seen in columns (1) to (4), independent directors' meeting attendance has a larger effect on firm risk taking when there are fewer analysts covering the firm.In addition, as seen in columns ( 5) to (8), independent directors' meeting attendance has a larger effect on firm risk taking when the appointed auditor is not one of the big four.The evidence suggests that information asymmetry plays a key role in terms of the strength of the link between independent directors' meeting attendance and firm risk taking.
The previous analysis shows that management will take excessive risks for personal gain.It is reasonable to speculate that when the executive's agency problem becomes more serious, independent directors will be able to hear more discussions on executive agency at the shareholders' meeting, obtain this part of information, and better supervise executives in the subsequent performance of their duties.Finally, it can reduce the enterprise risk more effectively.Table 17 is divided into high agency cost and low agency cost groups according to the median annual industry of executive agency cost.The results show that when the executive's agency problem is serious, independent directors can obtain more real information about management by participating in the shareholders meetings, and then play a more active supervisory role in the follow-up performance of their duties and reduce the level of enterprise risk-taking.

Conclusion
This study provides a convincing analysis of the impact of independent directors' shareholder meetings attendance on firm risk taking.As a large number of heterogeneous shareholders attend the shareholder meetings, their discussions and questions will overflow a large number of undisclosed true information of enterprises.Therefore, independent directors' on-site participation in the shareholder meetings is an effective way to obtain the true information of the enterprise.Independent directors who participate in shareholder meetings are in a better position to acquire more real information and thus improve their governance effect.The participation of independent directors in shareholder meetings can reduce the risk tolerance of enterprises and thus reduce the risktaking impulse of managers.More specifically, the participation of independent directors in the shareholder meetings can promote the steady operation of the enterprise, reduce its future M&A frequency, excessive investment level and debt level, and improve its cash holdings.In addition, the probability of violation of rules, the probability of being sued and the probability of receiving inquiry letters of the enterprise in the future decreased significantly.More importantly, the director labour market can effectively identify the diligent behavior of independent directors who actively participate in the shareholder meetings.The more active the independent directors participate in the shareholder meetings, the more their salary levels and directorships will be in the future.Finally, this paper also finds that the performance effect of independent directors' participation in the shareholder meetings is closely related to the information content of the shareholder meetings, the enterprise information environment and the information access ability of independent directors.Taken together, these findings prove that the role of independent directors in the shareholder meetings is a fruitful area for further exploration.
This paper holds that the shareholder meetings is rich in information, which can provide new information for the independent directors attending the shareholder meetings and improve their performance.In order to verify the logic, first of all, based on the theory of corporate governance and communication, it demonstrates the theoretical logic that independent directors can obtain real information when they participate in the shareholder meetings.In the next, we carries out strict endogeneity test and robustness test to verify the reliability of results.Then a large number of heterogeneity tests are carried out around the information content of the shareholder meetings.The above results support the role of information content of the shareholder meetings.This study provides a new perspective for understanding the information access channels and performance behaviors of independent directors, and provides a new topic for subsequent theoretical research.At the same time, it has reference significance for the regulatory authorities to improve the guidance on the performance of independent directors, the rules for convening shareholder meetings, and the way of enterprise information disclosure.

Disclosure statement
No potential conflict of interest was reported by the author(s).

Funding
regressions control for year and industry fixed effects.The baseline models using ordinary least square (OLS) regressions with t-statistics are adjusted by White Heteroscedasticity.***, **, and * indicate significance at the 1%, 5%, and 10% level, respectively.
Faccio et al. (2011)iptive statistics for all variables used in the baseline regressions.The mean values of Risk1 and Risk2 are 0.038 and 0.071, results that are consistent withJohn et al. (2008)andFaccio et al. (2011).As for AttendRate, there is significant variation in our data; the standard deviation is 0.331, with a mean value of 0.627 and corresponding maximum and minimum values of 1 and 0. Regarding other variables, the average firm has a Size of 7 billion yuan, Lev of 48.9%, ROA of 3.4%, sales growth of 16.4%, capital expenditure ratio of 4.2%, and PPE-to-assets ratio of 24.3%.In our sample, the CEO and chairman are the same person in 16.1% of boards, and the average board has 8.949 members, an independent director ratio of 37.2%, and an age of 18.37 years.

Table 3 .
Independent directors' shareholder meeting attendance and risk-taking.

Table 4 .
Regression results of exogenous impact of high-speed rail opening.
All regressions control for year and industry fixed effects.The baseline models using ordinary least square (OLS) regressions with t-statistics are adjusted by White Heteroscedasticity.***, **, and * indicate significance at the 1%, 5%, and 10% level, respectively.

Table 5 .
Regressions with instruments for AttendRate based on weather.Using the 1281 treatment observations and 1281 control observations, we are able to capture average treatment effects by comparing observations in firms with high AttendRate values against observations in firms with comparable propensity scores but low AttendRate values, holding other factors constant.The results in Table

Table 6 .
Regressions based on a propensity score matched sample.

Table 7 .
Regressions based on alternate measures of risk-taking.

Table 8 .
Sensitivity to an alternative estimation method.

Table 9 .
The relationship between information acquisition of independent directors and risk tolerance of the board.

Table 10 .
Relationship between independent director information acquisition and independent director voting.
All regressions control for year and industry fixed effects.The baseline models using ordinary least square (OLS) regressions and logit regressions with t-statistics are adjusted by White Heteroscedasticity.***, **, and * indicate significance at the 1%, 5%, and 10% level, respectively.

Table 11 .
Independent directors' shareholder meetings attendance and corporate policies.

Table 12 .
Independent directors' shareholder meetings attendance and corporate special risk.

Table 13 .
Independent directors' shareholder meetings attendance and career development of independent directors.

Table 14 .
Information content of the shareholder meetings, participation of independent directors in shareholder meetings and firm risk taking.

Table 15 .
Independent directors of remuneration committee, participation of independent directors in shareholder meetings and firm risk taking.All regressions control for year and industry fixed effects.The baseline models using ordinary least square (OLS) regressions with t-statistics are adjusted by White Heteroscedasticity.***, **, and * indicate significance at the 1%, 5%, and 10% level, respectively.

Table 16 .
Analyst and big4 auditor moderating effect on the relation between AttendRate and Risktaking.All regressions control for year and industry fixed effects.The baseline models using ordinary least square (OLS) regressions with t-statistics are adjusted by White Heteroscedasticity.***, **, and * indicate significance at the 1%, 5%, and 10% level, respectively.

Table 17 .
Impact of executives' agency problem on the relationship between independent directors' participation in the shareholders' meeting and risk-taking.