Can the rotation of chief accountants improve the accounting information quality of state-owned enterprises?

ABSTRACT This paper attempts to study the effect of the rotation of chief accountants of central SOEs. It is found that the rotation of chief accountants can improve the accounting information quality of SOEs. Mechanism analyses indicate that the above result is weakened when the tenure of chief accountants is long or after central SOEs carry out the pilot of establishing a standardised board or are audited by the National Audit Office. When further considering the types of rotation, the above result is only reflected when chief accountants come from SOEs in different industries or different cities or chief accountants are promoted. Economic consequence analyses indicate that the rotation of chief accountants alleviates the problem of over-investment and excess perks in SOEs by improving accounting information quality. In summary, the rotation of chief accountants can enhance their independence to improve the governance of SOEs and prevent the loss of state-owned assets.


Introduction
Since the reform and opening up, China's state-owned economy has kept growing.A series of effective reform policies have not only improved the efficiency of stateowned enterprises (SOEs) at the micro level, but also boosted the growth of national economy at the macro level (Huang, 2018;Z. Xu & Zhang, 2015).For example, existing literature indicates that mixed-ownership reform and dividend right incentive have significantly enhanced the market vitality of state-owned economy and improved the market competitiveness of SOEs (C.Cao & Zhang, 2020;Hao & Gong, 2017).The Guiding Opinions on Deepening the Reform of State-Owned Enterprises issued by the Central Committee of the Communist Party of China (Central Committee of CPC) and the State Council in 2015 clearly defines 'the combination of enhancing vitality and strengthening supervision' as one of the basic principles of deepening the reform of SOEs.It also points out that 'enhancing vitality is the essential requirement for improving the quality of SOEs, strengthening supervision is the important guarantee for improving the quality of SOEs, and it is necessary to effectively achieve the integration of the above two', and 'further improve the supervision and administration administration, this document specifically states that 'chief accountants should generally be rotated'.
As the leaders and financial executives of SOEs, chief accountants play an important role in the financial management and supervision of SOEs. 1 The Interim Rules on the Administration of the Responsibilities of Chief Accountants in Central SOEs issued by SASAC in 2006 requires that central SOEs should set up the position of chief accountants and chief accountant candidates can be from external rotation or internal transfer.I perform the statistics analysis on the sources of chief accountants of central SOEs.It can be found in Figure 1 that the proportion of chief accountants of central SOEs coming from the financial leadership position of other SOEs, that is, rotated chief accountants has remained at about 20% before 2017, but has increased significantly since 2018, and has exceeded 50% by 2020. 2 Therefore, it is an important and interesting research question that whether and how the rotation of chief accountants has the governance effect on SOEs.
Based on the data of listed firms controlled by central SOEs from 2007 to 2019, I examine the effect of the rotation of chief accountants of central SOEs on the accounting information quality of listed firms controlled by central SOEs.It is found that the rotation of chief accountants can improve the accounting information quality of SOEs.This result is robust to various endogeneity and robustness tests.Mechanism analyses indicate that the 1 For the convenience of literal expression, the main responsibilities of chief accountants are broadly summarised as financial management and supervision.Of course, chief accountants also participate in other aspects of management of SOEs.Compared with the overall management of chairmen or CEOs, the professional field of chief accountants helps me better identify the individual effect. 2The central SOEs in this paper mean the central SOE groups.I mainly focus on the rotation of financial leaders of SOEs in this paper.According to my statistics, besides the financial leadership position of other SOEs, the external sources of chief accountants include the non-financial position of other SOEs and government, both of which are in low and stable ratios.
above result is weakened when the tenure of chief accountants is long or after central SOEs carry out the pilot of establishing a standardised board or are audited by NAO, which suggests that independence is an important mechanism between the rotation of chief accountants and the accounting information quality of SOEs.When considering the types of rotation, the above result is only reflected when chief accountants come from SOEs in different industries or different cities or chief accountants are promoted, which further supports the above influence mechanism.Economic consequence analyses indicate that the rotation of chief accountants alleviates the problem of over-investment and excess perks in SOEs by improving accounting information quality.In summary, the rotation of chief accountants can enhance their independence and strengthen financial management and supervision to improve the governance of SOEs and prevent the loss of stateowned assets.
The theoretical contributions of this paper are mainly reflected in the following aspects.Firstly, state-owned assets supervision and administration system and the governance of SOEs are important research directions in the field of economics and management and are also important topics for deepening the reform of SOEs.Existing literature mainly focuses on the external and internal state-owned assets supervision and administration mechanisms (S.Chen & Lu, 2014;Chu & Fang, 2016;W. Li et al., 2017;Sheng & Liu, 2016;Yu et al., 2016), but pays little attention to the supervision and control mechanisms based on personnel appointment arrangements.It is the first time for this paper to explore the individual governance effect of state-owned assets supervision and administration system on SOEs from the perspective of the rotation of the leaders of SOEs.I also make an extended discussion on the substitutional relation among different supervision mechanisms in the governance effect on SOEs.
Secondly, the rotation of leading cadres and public performance are important topics in the field of public economics and management.Existing researches show that the rotation of local party and government leaders has a positive impact on local economic growth (X.Xu et al., 2007;Zhang & Gao, 2007), and the rotation of local leaders in judicial, tax, and audit departments has a positive impact on local judicial efficiency, tax collection, and audit supervision (G.Chen, 2012;Chu & Chen, 2021;Fan & Tian, 2016).Different from the above literature mainly focusing on macroeconomic development or some specific areas, I broaden research vision by studying state-owned assets supervision and administration and the rotation of chief accountants of enterprises at the micro level.I also identify the heterogeneity effect of the rotation of chief accountants from the perspective of the types of rotation.
Thirdly, financial executives have a significant individual effect in corporate information disclosure and financial decisions (Ge et al., 2011).Competence and independence are two important factors related to the individual effect.The literature on competence shows that financial executives from outside (Geiger & North, 2006;Mian, 2001), with professional qualification or experience (Hoitash et al., 2016;C. Li et al., 2010), and concurrently holding the position of directors (Bedard et al., 2014;Xiang, 2015) are perceived to be more competent.The literature on independence shows that the decisions and responsibilities of financial executives may be affected by CEOs (Feng et al., 2011;Leone & Liu, 2010;Mian, 2001) while staggered tenure and high rank of position can enhance the influence of financial executives (D.Jiang et al., 2018;F. Jiang et al., 2013).Based on these researches, my paper suggests that the rotation system of chief accountants in SOEs is an effective administrative appointment mechanism to enhance the independence of financial executives while maintaining professional competence.Moreover, political promotion incentive can strengthen the effect of rotation in the context of constrained outside labour market for chief accountants (D.Chen et al., 2018).All these obviously promote the researches in this field.

Institutional background
China's chief accountant system originated in the planned economy period.At that time, China learned from the experience of the Soviet Union and began to set up the position of chief accountants in some large and medium SOEs.As the person in charge of economic accounting and financial accounting, chief accountant was responsible for both the country and factory leader.The Draft of Several Provisions on the Establishment of Chief Accountants in State-Owned Industrial and Transportation Enterprises raised by the State Economic Commission and the Ministry of Finance and approved by the State Council in 1963 stipulates that state-owned industrial and transportation enterprises should set up the position of chief accountants and makes specific provisions on the rank, qualifications, appointment and dismissal, duties and authorities of chief accountants.The Regulations on the Duties and Authorities of Accountants issued by the State Council in 1978 also sets up a chapter to discuss the relevant provisions of chief accountant system.When state-owned industrial enterprises were overall rectified in 1982, various departments began to define chief accountants as leaders of enterprises and included them in the list of enterprise leaders.The Decision of the Central Committee of the Communist Party of China on Economic System Reform approved at the third Plenary Session of the twelfth Central Committee of the CPC in 1984 reaffirmed chief accountant system and listed chief accountants as members of the leadership team of enterprises.
Furthermore, it is the first time for the Accounting Law promulgated by the country in 1985 to clarify the requirement for the establishment of chief accountants in legal form, which states that 'large and medium enterprises, public institutions and administrative departments in charge can set up the position of chief accountants'.The Chief Accountant Ordinance issued by the State Council at the end of 1990 clarifies the authorities and rank of chief accountants and stipulates that 'large and medium enterprises owned by the whole people shall set up the position of chief accountants while public institutions and administrative departments in charge can have chief accountants as required and approved'.As a result, China's chief accountant system has entered a new stage of development.The Accounting Law revised in 1999 makes new provisions on the scope of establishing chief accountant system that 'large and medium SOEs and enterprises with state-owned assets predominate must set up the position of chief accountants'.In 2006, SASAC issued the Interim Rules on the Administration of the Responsibilities of Chief Accountants in Central SOEs, which clearly requires that 'central SOEs should set up the position of chief accountants in accordance with the provisions and allocate qualified chief accountants to effectively perform their responsibilities', and stipulates that 'the main responsibilities of chief accountants include basic management of enterprise accounting, financial management and supervision, construction of internal control system on financial accounting, and supervision of major financial issues' and 'central SOEs should give chief accountants corresponding authorities to effectively perform their duties, including the right to participate in major issues of enterprise, the right to supervise the implementation of major decisions and rules and regulations, the right to make personnel recommendations on the allocation of accountants, and the right to cosign the agreements for large capital expenditures'.It indicates that chief accountants are not only directly responsible for the financial management and supervision of SOEs, but also deeply involved in all aspects of the operation and management of SOEs.Therefore, chief accountants are the core members of the leadership team of SOEs.
In practice, Anshan Iron and Steel Corporation, the predecessor of Anshan Iron and Steel Group Corporation, set up the position of chief accountants at the headquarter as early as 1962.Subsequently, SOEs in different industries, such as metallurgy, machinery, chemistry and transportation, successively established chief accountant system.With the promulgation and revision of laws and regulations, such as the Accounting Law, the Chief Accountant Ordinance, and the Interim Rules on the Administration of the Responsibilities of Chief Accountants in Central SOEs, chief accountant system has been fixed in SOEs and has been continuously improved with economic and social development.In the meantime, the rotation system of chief accountants has also been implemented.Here I need to discuss the rotation system of leading cadres in China.After the reform and opening up, especially since the 1990s, the rotation of leading cadres has gradually become institutionalised, standardised and normalised.In July 1990, the Central Committee of CPC made the Decision on Implementing the Rotation System of Leading Cadres of the Party and Government.The General Office of the Central Committee of CPC issued the Interim Provisions on the Rotation of Leading Cadres of the Party and Government in 1999 and the Provisions on the Rotation of Leading Cadres of the Party and Government in 2006.Although the aforementioned documents do not directly target the leaders of SOEs, their characteristics of 'quasi officials' have contributed to the implementation of the rotation of chief accountants to a certain extent.The Interim Rules on the Administration of the Responsibilities of Chief Accountants in Central SOEs has already clarified that the source of chief accountants includes external rotation.The Interim Provisions on the Administration of the Leaders of Central SOEs issued by the General Office of the Central Committee of CPC and the General Office of the State Council in 2009 proposes to 'improve the rotation system of the leaders of central SOEs'.The leaders in this document include chief accountants.The Opinions on the Pilot of Delegating Chief Accountants in Central SOEs issued by SASAC in 2015 also emphasises that the delegated chief accountants are generally from rotation.SASAC also promotes the rotation of non-delegated chief accountants.In 2018, the General Office of the Central Committee of CPC and the General Office of the State Council further issued the Provisions on the Administration of the Leaders of Central SOEs, which explicitly states that 'chief accountants should generally be rotated'.As a result, the rotation system of chief accountants has been gradually specified through continuous practice.

Theoretical analysis and hypothesis development
The research question I am interested in is the effect of the rotation of chief accountants on the accounting information quality of SOEs.Because the real owner of SOEs is the country and does not refer to a specific person, it creates a virtual 'owner vacancy'.Under this circumstance, the state-owned shareholders who exercise ownership right are in fact agents.SASAC delegates the management right of SOEs to managers.In order to control and supervise the behaviour of managers, SASAC designs a set of performance appraisal and evaluation system for leaders based on accounting information indicators (Du et al., 2018;Yu et al., 2016).But due to the multi-layer principal-agent relationship, there is severe information asymmetry in SOEs.Managers have motivation and ability to manipulate accounting information to confuse the public, thus leading to low accounting information quality of SOEs (Gul et al., 2010).Chief accountants are mainly responsible for financial management and supervision in SOEs.Their motivation and ability have direct effect on the accounting information quality of SOEs.The rotation of chief accountants can affect the accounting information quality of SOEs by changing their independence and competence.
On the one hand, the rotation of chief accountants can increase their motivation and strengthen their independence in financial management and supervision.As the members of the leading team of SOEs, chief accountants are mainly responsible for financial management and supervision and are also deeply involved in major decisions on the operation and management of SOEs.The financial and operating performance of SOEs are directly related to the performance appraisal and evaluation results of leaders including chief accountants and thus affect their compensation and promotion (Yang et al., 2013).As a result, chief accountants and other leaders have common interests.Meanwhile, chief accountants are often ranked at the back of some vice general managers in the list of leaders (Ding & Wen, 2012) and may be influenced by other leaders when performing financial management and supervision.All above factors have significantly weakened the independence of chief accountants in financial management and supervision.The rotation of chief accountants can reduce the possibility of insider collusion through staggered tenure with other leaders (F.Jiang et al., 2013).It can also alleviate the problem of insider control caused by working in an enterprise for a long time.Therefore, the rotation of chief accountants can strengthen their independence and then improve the accounting information quality of SOEs.
On the other hand, the rotation of chief accountants may increase their difficulty in performing duties and reduce their competence in financial management and supervision.Although 'learning by doing' can help chief accountants strengthen the general knowledge, experience and skills of financial management and supervision, rotated chief accountants still need to spend a lot of time and resources to get familiar with and understand the problems and underlying causes in financial management and supervision of a new enterprise and also need to establish the connection with and fit into a new leadership team.This is due to different historical background, development status and personnel composition in different enterprises.All above factors may destroy the professional stability and specialisation of chief accountants and increase their working difficulties and costs.Therefore, the rotation of chief accountants can reduce their competence and then decrease the accounting information quality of SOEs.
In summary, the effect of the rotation of chief accountants on the accounting information quality of SOEs is an empirical question.Therefore, I propose the following alternative hypotheses: H1a: When the independence mechanism dominates, the rotation of chief accountants can improve the accounting information quality of SOEs.

H1b: When the competence mechanism dominates, the rotation of chief accountants can reduce the accounting information quality of SOEs.
Then, I analyse the impact of relevant factors on the relationship between the rotation of chief accountants and the accounting information quality of SOEs based on the aforementioned theoretical mechanisms.Firstly, the tenure of chief accountants can affect their independence and competence and thus affect the relationship between the rotation of chief accountants and the accounting information quality of SOEs.If independence is the dominant mechanism through which the rotation of chief accountants improves the accounting information quality of SOEs, the possibility of collusion and the size of common interests between chief accountants and other leaders become greater when the tenure of chief accountants becomes longer, which impairs the independence of chief accountants.Therefore, the improving effect of the rotation of chief accountants on the accounting information quality of SOEs weakens as the tenure of chief accountants increases.
If competence is the dominant mechanism through which the rotation of chief accountants reduces the accounting information quality of SOEs, chief accountants have more time to get familiar with and understand the current situation of financial management and supervision of the new enterprise and are more likely to coordinate operation and management with other leaders when the tenure of chief accountants becomes longer.Afterwards they are more likely to carry out financial management and supervision according to inside and outside conditions, making them more competent to perform their duties.Therefore, the deteriorating effect of the rotation of chief accountants on the accounting information quality of SOEs weakens as the tenure of chief accountants increases.Therefore, I propose the following alternative hypotheses: H2a: When the independence mechanism dominates, the improving effect of the rotation of chief accountants on the accounting information quality of SOEs weakens as the tenure of chief accountants increases.

H2b: When the competence mechanism dominates, the deteriorating effect of the rotation of chief accountants on the accounting information quality of SOEs weakens as the tenure of chief accountants increases.
Secondly, the pilot of establishing a standardised board in SOEs can affect the independence of chief accountants and thus affect the relationship between the rotation of chief accountants and the accounting information quality of SOEs.Pilot SOEs are required to allocate outside directors with numbers more than half of the board.The improvement of independence of the board ensures the independence of chief accountants because chief accountants need to report to the board.Meanwhile, the pilot of establishing a standardised board can alleviate the problem of insider control in SOEs (W.Li et al., 2017), which also strengthens the independence of chief accountants.If independence is the dominant mechanism through which the rotation of chief accountants improves the accounting information quality of SOEs, the pilot of establishing a standardised board and the rotation of chief accountants can substitute with each other.Therefore, the improving effect of the rotation of chief accountants on the accounting information quality of SOEs weakens after SOEs carry out the pilot of establishing a standardised board.
However, the pilot of establishing a standardised board in SOEs can neither directly affect the competence of chief accountants nor affect the relationship between the rotation of chief accountants and the accounting information quality of SOEs through the competence mechanism.Therefore, I propose the following alternative hypotheses: H3a: When the independence mechanism dominates, the improving effect of the rotation of chief accountants on the accounting information quality of SOEs weakens after SOEs carry out the pilot of establishing a standardised board.

H3b: When the competence mechanism dominates, the deteriorating effect of the rotation of chief accountants on the accounting information quality of SOEs does not change after SOEs carry out the pilot of establishing a standardised board.
Thirdly, audit supervision on SOEs can affect the independence of chief accountants and thus affect the relationship between the rotation of chief accountants and the accounting information quality of SOEs.In order to ensure the authenticity, legitimacy and effectiveness of the operation and management of SOEs, audit institutions shall, in accordance with laws and regulations, conduct audit supervision over the financial revenues and expenditures of SOEs.Audit outcomes are important decision bases for the reward, punishment, appointment and dismissal of the leaders of SOEs.The primary concern of audit supervision on SOEs is the authenticity of accounting information. 3When SOEs are audited and face subsequent rectification of issues found suspicious, chief accountants, as financial executives of SOEs, must pay more attention to accounting information quality.They are also less likely to be influenced by other leaders and thus their independence is significantly enhanced.Meanwhile, audit supervision can alleviate the problem of insider control in SOEs (Chu & Fang, 2016), which also strengthens the independence of chief accountants.If independence is the dominant mechanism through which the rotation of chief accountants improves the accounting information quality of SOEs, audit supervision and the rotation of chief accountants can substitute with each other.Therefore, the improving effect of the rotation of chief accountants on the accounting information quality of SOEs weakens after SOEs are audited.
However, audit supervision on SOEs can neither directly affect the competence of chief accountants nor affect the relationship between the rotation of chief accountants and the accounting information quality of SOEs through the competence mechanism.Therefore, I propose the following alternative hypotheses: H4a: When the independence mechanism dominates, the improving effect of the rotation of chief accountants on the accounting information quality of SOEs weakens after SOEs are audited by audit institutions.
H4b: When the competence mechanism dominates, the deteriorating effect of the rotation of chief accountants on the accounting information quality of SOEs does not change after SOEs are audited by audit institutions.

Data sources and sample selection
I attempt to empirically test the above hypotheses by taking the rotation of chief accountants in central SOEs as the starting point.Central SOEs fully comply with the applicable object standards of laws and regulations such as the Accounting Law, the Chief Accountant Ordinance, and the Interim Rules on the Administration of the Responsibilities of Chief Accountants in Central SOEs because they are the most representative of stateowned enterprises and their scale is generally large or medium.Meanwhile, I take listed firms controlled by central SOEs as empirical research object.The reasons are as follows.Firstly, chief accountants of central SOEs can directly control and influence the financial management and supervision activities of subordinate listed firms.Hence, the governance changes of subordinate listed firms in financial management and supervision can reflect the individual effect of chief accountants.Secondly, it is feasible to collect data and make research design when focusing on subordinate listed firms, which is also consistent with the general practice of existing literature (C.Cao & Zhang, 2020;Chu & Fang, 2016;W. Li et al., 2017;Yu et al., 2016).
Due to the new Accounting Standard for Business Enterprises implemented in 2007 and the impact of COVID-19 on firms' operation and management since 2020, I select a sample of listed firms controlled by central SOEs from 2007 to 2019.After deleting the sample firms in finance industry, listed as B-share, in Sci-Tech innovation board, and with missing variables, I finally get 3,530 listed firm-year observations.
The data about the individual characteristics of chief accountants is manually collected from the announcements on appointment and dismissal of leaders on the website of SASAC, directory of leaders on the website of central SOEs and Baidu Baike.The financial data and corporate governance data of listed firms controlled by central SOEs are from CSMAR database.The marketisation index is from Wang et al. (2021).

Research model and variable definition
I use the following regression model (1) to test the effect of the rotation of chief accountants on the accounting information quality of SOEs: The subscript i represents listed firm controlled by central SOE.The subscript trepresents year.The subscript g represents central SOE (group).In model (1), I control firm fixed effect (λ i ) and year fixed effect (μ t ).
The dependent variable |DA| is accounting information quality, defined as Kothari et al. (2005).I run the following regression model ( 2) by industry and year to get the residual.After making the performance adjustment to the residual, that is, subtracting another residual of the firm with the same industry and year and the closest ROA, I take the absolute value of the above difference as the measure of |DA|.The smaller the value of | DA| is, the higher the accounting information quality of SOEs is.
TA is total assets.AC is total accrual earnings that equals the difference between net income and operating cash flow.ΔSale is the difference between current-and prior-year sales.ΔAR is the difference between current-and prior-year accounts receivable.PPE is the original value of fixed assets.
The independent variable ROTATE is the rotation of chief accountants, defined as a dummy that equals 1 if the incumbent chief accountants of central SOEs are from the financial leadership position of other SOEs, and 0 otherwise.Similar to Fan and Tian (2016), the research design based on multiple entry and exit shocks formed by the turnover of chief accountants is more exogenous than the general difference-in-differences model and is conducive to better identifying the casual effects.According to the aforementioned H1a, if the regression coefficient β 1 in model ( 1) is significantly negative, it means that the rotation of chief accountants can improve the accounting information quality of SOEs.
Meanwhile, I control several control variables in model (1).SIZE is defined as the natural logarithm of total assets at the end of year.LEV is defined as the total liabilities divided by total assets at the end of year.ROA is defined as the net income divided by total assets at the end of year.MTB is defined as the market value divided by book value of total equities at the end of year.LARGEST is defined as the percentage of ownership held by the largest shareholder.BOARD is defined as the natural logarithm of 1 plus the number of directors in the board.INDEP is defined as the number of independent directors divided by the number of directors in the board.DUAL is defined as 1 if the chairman of the board also holds position of CEO, and 0 otherwise.MSHARE is defined as the percentage of ownership held by managers.BIG4 is defined as 1 if annual report is audited by the international Big 4 audit firms, and 0 otherwise.MINDEX is defined as the decile ranking (0-9, divided by 9) of marketisation index for the province in which the firm is located.

Descriptive statistics
Table 1 reports the descriptive statistics of variables in model (1).It can be found that the mean (standard deviation) of |DA| is 0.075 (0.079), which is consistent with existing literature.The mean of ROTATE is 0.221, indicating that about 22% of chief accountants of central SOEs are rotated in the sample.In the dimension of corporate finance and governance, the indicators are quite similar to existing literature on SOEs.For example, the average leverage is 0.529, the average ROA is 0.028, the average largest ownership is 0.415, the average managerial ownership is 0.004, the average ratio of independent directors is 0.371, the average duality of chairman and CEO is 0.061, and the average probability of annual report audited by Big 4 is 0.152 in sample listed firms controlled by central SOEs.

Basic empirical results and analysis
Table 2 reports the results of the effect of the rotation of chief accountants on the accounting information quality of SOEs.It can be found in column (1) that the coefficient of ROTATE is significantly negative when controlling no control variables in model (1).The coefficient of ROTATE keeps significantly negative when adding control variables in column (2).Both statistical significance and economic significance are quite stable.The above results indicate that the rotation of chief accountants can improve the accounting information quality of SOEs.In the dimension of economic significance of the above effect, the accounting information quality of SOEs increases about 0.114 standard deviation when chief accountants change from non-rotated ones to rotated ones, which means the above effect is economically significant.
In summary, the rotation of chief accountants can increase their independence, thus leading to an improvement of the accounting information quality of SOEs.Therefore, H1a is empirically supported and H1b is empirically denied.

Endogeneity tests and robustness tests
I do the following tests to alleviate endogeneity concern and strengthen the robustness of my results.(1) Use change model based on the turnover of chief accountants.Non-R to R is defined as 0 before the turnover of chief accountants from non-rotated ones to rotated ones, and 1 after that.R to Non-R is defined as 0 before the turnover of chief accountants from rotated ones to non-rotated ones, and 1 after that.It can be found in column ( 1) and (2) of Table 3 that the coefficient of Non-R to R is significantly negative and the coefficient of R to Non-R is significantly positive, indicating that the accounting information quality of SOEs becomes better when chief accountants change from non-rotated ones to rotated ones and deteriorates when chief accountants change from rotated ones to non-rotated ones.This result further supports the above finding.(2) Parallel trend test.1YBefore is defined as 1 for the one year before rotated chief accountants assume office, and 0 otherwise.1YAfter is defined as 1 for the one year after rotated chief accountants leave office, and 0 otherwise.I add these two variables in model (1).It can be found in column (3) of Table 3 that both coefficients of 1YBefore and 1YAfter are insignificant, indicating that rotated chief accountants do not selectively join in SOEs with high accounting information quality.This result further supports the above finding.(3) Use PSM sample.To control the difference between SOEs with rotated chief accountants and SOEs with non-rotated chief accountants, I take propensity score matching method.In the first step, I run the logistic regression of ROTATE on control variables in model ( 1) and year fixed effect to calculate the propensity score.In the second step, I get the PSM sample based on 1 by 1 nearest matching by using the propensity score.Finally, I rerun the regression model (1) based on this PSM sample.It can be found in column (4) of  6) of Table 3 with inside chief accountants as baseline group that compared with these chief accountants, chief accountants rotated from external financial position can significantly improve the accounting information quality of SOEs while chief accountants from external non-financial position or government have no additional governance effect.It can be found in column ( 7) of Table 3 with inside chief accountants promoted from lower financial position as baseline group that compared with these chief accountants, chief accountants rotated from external financial position can significantly improve the accounting information quality of SOEs while chief accountants from internal or external nonfinancial position or government have no additional governance effect. 4The above result indicates that rotation from external financial position does not obviously impair the professional competence of chief accountants but improves the independence in performing their duties, which makes them carry out financial management and supervision activities effectively.(6) Change the measurement of accounting information quality.I define |DA_MJ| as the absolute value of discretionary accruals calculated on the modified Jones model (Dechow et al., 1995), AQ as the standard deviation of discretionary working capital accruals calculated on the modified DD model (McNichols, 2002), and AQ_DISC as the discretionary part of AQ (Francis et al., 2005).These are the alternative indicators of accounting information quality.It can be found in column (8) to (10) of Table 3 that the result is consistent with the above finding. 5(7) Rule out the alternative explanation.I mainly use accrual earnings management to measure accounting information quality.Due to the potentially substitutional relationship between accrual earnings management and real earnings management, the rotation of chief accountants may lead SOEs to adopt more obscure real earnings management when decreasing accrual earnings management, which will worsen accounting information quality.To rule out this alternative explanation, I construct comprehensive real earnings management indicator RM, defined as Cohen et al. (2008).It can be found in column (11) of Table 3 that the rotation of chief accountants does not significantly influence the real earnings management of SOEs.So, this alternative explanation is not established.

Mechanism tests
Based on the aforementioned theoretical analysis and corresponding empirical finding, the rotation of chief accountants can improve the accounting information quality of SOEs by increasing their independence.To verify the theoretical mechanism, I do the following mechanism tests.Firstly, according to the theoretical analysis, if the independence mechanism is valid, the independence of chief accountants weakens with the tenure and then the effect of the rotation of chief accountants on the accounting information quality of SOEs also weakens with the tenure.Therefore, I construct the interaction term of the rotation and tenure of chief accountants.It can be found in column (1) of Table 4 that the coefficient of ROTATE×TENURE is significantly positive, which indicates that the longer the tenure of chief accountants is, the weaker the improving effect of the rotation of chief accountants on the accounting information quality of SOEs is.Therefore, H2a is empirically supported.
Secondly, according to the theoretical analysis, if the independence mechanism is valid, the pilot of establishing a standardised board, as a substitutional inside governance measure, can improve the independence of chief accountants and then the effect of the rotation of chief accountants on the accounting information quality of SOEs weakens after the implementation of the pilot.Therefore, I define PILOT as 1 after central SOEs carry out the pilot of establishing a standardised board, and 0 otherwise.Then I construct the interaction term of the rotation of chief accountants and the pilot of establishing a standardised board.It can be found in column (2) of Table 4 that the coefficient of ROTATE×PILOT is significantly positive, which indicates that the improving effect of the rotation of chief accountants on the accounting information quality of SOEs weakens after SOEs carry out the pilot of establishing a standardised board.Therefore, H3a is empirically supported.
Thirdly, according to the theoretical analysis, if the independence mechanism is valid, audit supervision, as a substitutional outside governance measure, can improve the independence of chief accountants and then the effect of the rotation of chief accountants on the accounting information quality of SOEs weakens after the implementation of audit supervision on SOEs.Therefore, I define AUDIT as 1 after central SOEs are audited by NAO, and 0 otherwise.Then I construct the interaction term of the rotation of chief accountants and audit supervision.It can be found in column (3) of Table 4 that the coefficient of ROTATE×AUDIT is significantly positive, which indicates that the improving effect of the rotation of chief accountants on the accounting information quality of SOEs weakens after SOEs are audited by NAO.Therefore, H4a is empirically supported.
In summary, the rotation of chief accountants does improve the accounting information quality of SOEs by increasing their independence.

Heterogeneity tests on the types of rotation of chief accountants
I further do heterogeneity tests on the types of rotation of chief accountants.Firstly, I divide ROTATE into ROTATE_SAMEIND and ROTATE_DIFFIND according to the industries which former SOEs of chief accountants belong to.ROTATE_SAMEIND is defined as 1 if the incumbent chief accountants of central SOEs are from the financial leadership position of other SOEs and the two SOEs share the same industry, and 0 otherwise.ROTATE_DIFFIND is defined as ROTATE minus ROTATE_SAMEIND.It can be found in column (1) of Table 5 that the coefficient of ROTATE_DIFFIND is significantly negative while that of ROTATE_SAMEIND is not significant, and there is significant difference between the two coefficients, indicating that chief accountants rotated from different industries can improve the accounting information quality of SOEs while those from same industry can't do so.This finding further supports the independence mechanism and rejects the competence mechanism, such as industry experience.
Secondly, I divide ROTATE into ROTATE_SAMECITY and ROTATE_DIFFCITY according to the cities where former SOEs of chief accountants are located.ROTATE_SAMECITY is defined as 1 if the incumbent chief accountants of central SOEs are from the financial leadership position of other SOEs and the two SOEs share the same city, and 0 otherwise.ROTATE_DIFFCITY is defined as ROTATE minus ROTATE_SAMECITY.It can be found in column (2) of Table 5 that the coefficient of ROTATE_DIFFCITY is significantly negative while that of ROTATE_SAMECITY is not significant, and there is significant difference between the two coefficients, indicating that chief accountants rotated from different cities can improve the accounting information quality of SOEs while those from same city can't do so.This finding further supports the independence mechanism and rejects the competence mechanism, such as local information.
Thirdly, I divide ROTATE into ROTATE_PROMOTE and ROTATE_TRANSFER according to the changes in administrative levels from former SOEs to current SOEs of chief accountants.ROTATE_PROMOTE is defined as 1 if the incumbent chief accountants of central SOEs are from the financial leadership position of other SOEs and also get promotion in administrative levels, and 0 otherwise.ROTATE_TRANSFER is defined as ROTATE minus ROTATE_PROMOTE.It can be found in column (3) of Table 5 that the coefficient of ROTATE_PROMOTE is significantly negative while that of ROTATE_TRANSFER is not significant, and there is significant difference between the two coefficients, indicating that chief accountants rotated from lower administrative levels can improve the accounting information quality of SOEs while those from same administrative level can't do so.This finding underlines that political promotion can further strengthen the motivation of chief accountants and their independence.
Fourthly, I divide ROTATE into ROTATE_DELEGATE and ROTATE_NODELEGATE according to whether chief accountants are delegated by SASAC.ROTATE_DELEGATE is defined as 1 if the incumbent chief accountants of central SOEs are from the financial leadership position of other SOEs and are also delegated by SASAC, and 0 otherwise.ROTATE_NODELEGATE is defined as ROTATE minus ROTATE_DELEGATE.It can be found in column (4) of Table 5 that the coefficient of ROTATE_NODELEGATE is significantly negative while that of ROTATE_DELEGATE is not significant, and there is insignificant difference between the two coefficients, indicating that delegation does not influence the relationship between the rotation of chief accountants and the accounting information quality of SOEs. 6n summary, the rotation of chief accountants improves the accounting information quality of SOEs only when chief accountants come from SOEs in the different industries or different cities or they are promoted.

Rotation of chief accountants and corporate governance of SOEs
Based on the aforementioned findings, I further test the effect of the rotation of chief accountants on corporate governance in SOEs.Accounting information is the basis for firms' operation and management decisions.Accounting information quality is related to the efficiency and effect of firms' regular management and major decisions (Biddle et al., 2009;Gul et al., 2011).According to the Interim Rules on the Administration of the Responsibilities of Chief Accountants in Central SOEs, the responsibilities of chief accountants cover not only accounting information but also regular management and major decisions in SOEs.Therefore, I further study the governance effect of the rotation of chief accountants on the investment decision and perks of SOEs.
The investment decision is one of major decisions in SOEs.Due to the agency problem caused by insider control, investment efficiency is quite low and over-investment problem is quite severe in SOEs (Zhong et al., 2010).The investment plan review, financial feasibility demonstration, investment capital guarantee, investment implementation and other related processes are under the control and influence of chief accountants during the investment decision-making of SOEs.Chief accountants also have the right to participate in and supervise major investment decision and the right to co-sign the agreement for large capital expenditures.Therefore, the rotation of chief accountants can improve the investment efficiency of SOEs and especially curb serious overinvestment.Moreover, accounting information is the basis for firms' investment decision.The higher the accounting information quality is, the higher the investment efficiency is (Biddle et al., 2009).Therefore, the rotation of chief accountants can curb over-investment of SOEs by improving accounting information quality.I construct the over-investment indicator OVERINV, defined as Richardson (2006).It can be found in Table 6 that the coefficient of ROTATE is significantly negative and that of |DA| is significantly positive, which indicates that the rotation of chief accountants can curb the over-investment of SOEs and low accounting information quality leads to serious over-investment.Combined with the above finding that the rotation of chief accountants can improve the accounting information quality of SOEs, the rotation of chief accountants can curb the overinvestment of SOEs by improving accounting information quality.
In the regular management of SOEs, due to compensation regulation, insider control has led to serious excessive perks (Quan et al., 2010).Perks, as regular expenditures, involve relevant rules and regulations in SOEs.Chief accountants are responsible for the construction of internal control system on finance and accounting, cost control coordination, and overall budget management advancement in SOEs.They also have the right to supervise the implementation of rules and regulations and the right to co-sign the agreement for large regular expenditures.Therefore, the rotation of chief accountants can curb the excessive perks of SOEs.Moreover, accounting information is the basis for firms' regular management.The higher the accounting information quality is, the higher the management efficiency is (Gul et al., 2011).Therefore, the rotation of chief accountants can curb the excessive perks of SOEs by improving accounting information quality.I construct the indicator of excessive perks UNPERK, defined as Quan et al. (2010).It can be found in Table 7 that the coefficient of ROTATE is significantly negative and that of | DA| is significantly positive, which indicates that the rotation of chief accountants can curb the excessive perks of SOEs and low accounting information quality leads to serious excessive perks.Combined with the above finding that the rotation of chief accountants can improve the accounting information quality of SOEs, the  1), OCF is defined as the operating cash flow divided by total assets, and MPAY is defined as the natural logarithm of managerial compensation.
rotation of chief accountants can curb the excessive perks of SOEs by improving accounting information quality.In summary, the rotation of chief accountants can alleviate the problem of overinvestment and excessive perks in SOEs, in which the improvement of accounting information quality plays a mediating role.

Research conclusions and policy implications
How to improve the governance of SOEs and maintain the safety of state-owned assets through effective institutional arrangements for state-owned assets supervision and administration is an important theoretical and practical question.This paper attempts to answer this question by taking the rotation of chief accountants in central SOEs as the starting point.It is found that the rotation of chief accountants can improve the accounting information quality of SOEs.Mechanism analyses indicate that the above result is weakened when the tenure of chief accountants is long or  1), OCF is defined as the operating cash flow divided by total assets, and MPAY is defined as the natural logarithm of managerial compensation.
after central SOEs carry out the pilot of establishing a standardised board or are audited by NAO.When further considering the types of rotation, the above result is only reflected when chief accountants come from SOEs in the different industries or different cities or chief accountants are promoted.Economic consequence analyses indicate that the rotation of chief accountants alleviates the problem of overinvestment and excess perks in SOEs by improving accounting information quality.In summary, the rotation of chief accountants can enhance their independence and strengthen financial management and supervision to improve the governance of SOEs and prevent the loss of state-owned assets.
The research conclusions have important policy implications.Firstly, the rotation of leaders is an important way to cultivate and exercise the comprehensive quality and ability of managers of SOEs, and also an important institutional arrangement to strengthen state-owned assets supervision and administration.From the perspective of the regulator, it is necessary to comprehensively consider the scope, object, and frequency of rotation to balance the independence and competence of managers when performing their duties in SOEs.Secondly, chief accountant system is an important guarantee for SOEs to achieve effective financial management and supervision, which is closely related to chief accountants' status and authority.Therefore, it is necessary to improve the rank of chief accountants among the leadership team of SOEs and ensure their independence in performing financial management and supervision during the process of rotation.Last but not least, the personnel arrangement for leaders to rotate positions, the establishment of a modern enterprise system with a standardised board, and external audit supervision may all have the governance effect on SOEs in a substitutional way.Therefore, it is necessary to achieve the coordination of various reform policies of SOEs, so as to maintain and increase the value of state-owned capital.

Figure 1 .
Figure 1.Ratio of rotated chief accountants of central SOEs.
based on robust standard errors that are clustered by central SOEs are displayed in parentheses.***, ** and * represent 1%, 5% and 10% significance levels, respectively, based on a two-tailed test.Control variables are same as those in model (1).

Table 2 .
Rotation of chief accountants and accounting information quality of SOEs.

Table 3 .
Endogeneity tests and robustness tests.
Table 3 that the result is consistent with the above finding.(4) Control other individual characteristics of chief accountants.I further control tenure (TENURE), age (AGE), gender (FEMALE), education level (EDULEVEL) of chief accountants in model (1).It can be found in column (5) of Table 3 that the result is consistent with the above finding.(5) Change the baseline group in model (1).To further compare the governance effect of rotated chief accountants with that of chief accountants from other sources, I add INOTH and EXOTH in model (1).INOTH is defined as 1 if the incumbent chief accountants of central SOEs are from the non-financial position of the same SOEs, and 0 otherwise.EXOTH is defined as 1 if the incumbent chief accountants of central SOEs are from the non-financial position of other SOEs or government, and 0 otherwise.It can be found in column (

Table 5 .
Heterogeneity tests on types of rotation.based on robust standard errors that are clustered by central SOEs are displayed in parentheses.***, ** and * represent 1%, 5% and 10% significance levels, respectively, based on a two-tailed test.Control variables are same as those in model (1).

Table 6 .
Rotation of chief accountants and over-investment of SOEs.based on robust standard errors that are clustered by central SOEs are displayed in parentheses.***, ** and * represent 1%, 5% and 10% significance levels, respectively, based on a two-tailed test.Besides control variables in model (

Table 7 .
Rotation of chief accountants and excess perks of SOEs.based on robust standard errors that are clustered by central SOEs are displayed in parentheses.***, ** and * represent 1%, 5% and 10% significance levels, respectively, based on a two-tailed test.Besides control variables in model (