Does the standardisation of tax enforcement improve corporate financial reporting quality?

ABSTRACT Previous studies of tax enforcement neglect the effect of tax-penalty discretionary benchmarks on corporate financial reporting quality. In China, province-level variations in implementing tax-penalty discretionary benchmarks provide a quasi-natural experiment to explore how tax enforcement standardisation affects corporate financial reporting quality. We show that when tax-penalty benchmarks are implemented in every province, there is a more pronounced positive relationship between tax enforcement standardisation and corporate financial reporting quality compared with the ex-ante period. Moreover, this positive relationship is primarily driven by firms with higher degrees of tax avoidance and collusion with tax collectors. Consequently, the implementation of tax-penalty benchmarks means that the standardisation of tax enforcement can improve corporate financial reporting quality. In turn, tax enforcement standardisation can optimise the external governance environment for listed firms and improve financial information disclosure in capital markets.


Introduction
Theoretically, the marginal cost of corporate tax avoidance depends on the intensity and effectiveness of tax enforcement (Allingham & Sandmo, 1972), which are, in turn, affected mainly by tax-system personnel (Li, 2012;Shevlin et al., 2012). Under an imperfect taxenforcement system, it is easy for such personnel to collude over tax collection with managers, thereby weakening the intensity and effectiveness of tax enforcement and further encouraging enterprises to avoid tax (Tian & Fan, 2018). In practice, managers may attempt to avoid tax through earnings management activities in tax-related accounting projects (Dhaliwal et al., 2004;Gupta et al., 2016). It follows that corporate tax avoidance will have spillover effects on corporate financial reporting quality. Therefore, a strict tax enforcement system can strengthen tax collection and management, restrain corporate earnings management as a tax-avoidance mechanism, and have a positive spillover effect on corporate financial reporting quality. Specifically, strict tax enforcement substantially reduces a company's incentives to underestimate revenue for tax purposes (Hoopes et al., 2012), in turn reducing the degree of downward earnings management (Li et al., 2019). management, tax declaration management, tax collection management, tax inspection, invoice management, and tax guarantees. For each category, the discretionary benchmarks for tax violations and penalties are clarified. In general, the discretionary benchmarks for tax penalties are closely related to corporate tax-related accounting behaviour.
The discretionary benchmarks for local tax administrative penalties fully consider exante tax collection and administration factors, and differ remarkably from the unified tax administrative penalties specified in the Tax Enforcement Law. Moreover, they also introduce a greater legal deterrent than otherwise using discretionary benchmarks for local tax administrative penalties. For example, as shown in Table 1, the STA's Beijing, Jiangxi, and Guizhou branches have issued specific penalty benchmarks for enterprises that illegally print or forge tax payment vouchers. This activity is a form of earnings management that reduces corporate financial reporting quality. Beijing, located in the east, has set the highest penalties for this tax violation; Guizhou province, in the west, has the lowest penalties; and Jiangxi province, in the centre, falls between the other two in terms of penalty severity. Likewise, the three taxation bureaus vary in the quantities of illegally printed or forged tax payment vouchers that trigger each penalty. These examples demonstrate that with the implementation of discretionary benchmarks for tax penalties, local tax departments can impose benchmark penalties relevant to their respective regions. Deterring tax violations will inhibit corporate earnings management, thereby improving corporate financial reporting quality. Therefore, from the perspective of discretionary benchmarks for tax penalties, we focus mainly on how standardising tax enforcement impacts corporate financial reporting quality. This paper makes several contributions. First, we expand the research perspective and enrich the literature on tax enforcement. Studies of tax enforcement mainly measure the degree of tax collection and management using a results-oriented indicator of tax income (e.g. Xu et al., 2011;Zeng & Zhang, 2009). Furthermore, prior studies have also used the quasi-natural experimental scenario of Phase III tax reform in China (Sun et al., 2021;Zhang, 2020) to explore the economic consequences of tax enforcement. In general, these studies focus on the ex-post tax collection and administration factors, while neglecting pre-collection factors. In this study, we especially examine pre-collection factors by considering the discretionary benchmarks for tax penalties. Considering pre-collection factors enriches the literature on tax enforcement. Second, we advance research on corporate financial reporting quality, which mainly examines the impact of corporategovernance mechanisms (e.g. Aifuwa & Embele, 2019) and macro-institutional factors (e.g. Garrett et al., 2014) on corporate financial reporting quality. Few studies have examined the impact of tax enforcement on financial reporting quality from the perspective of Table 1. Discretionary penalty benchmarks for illegally printing or forging tax payment vouchers in Beijing, Jiangxi, and Guizhou.

Beijing
Jiangxi Guizhou 1. For fewer than 10 copies, a fine of at least RMB 10,000 but not more than RMB 20,000 shall be imposed; 1. For fewer than 10 copies, a fine of at least RMB 2,000 but not more than RMB 10,000 shall be imposed; 1. For fewer than 10 copies, a fine of at least RMB 2,000 but not more than RMB 5,000 shall be imposed; 2. For more than 10 copies but fewer than 50 copies, a fine of at least RMB 20,000 but not more than RMB 50,000 shall be imposed.
2. For more than 10 copies but fewer than 100 copies, a fine of at least RMB 10,000 but not more than RMB 20,000 shall be imposed.
2. For more than 11 copies but fewer than 50 copies, a fine of at least RMB 5,000 but not more than RMB 10,000 shall be imposed.
discretionary benchmarks for tax penalties. Third, our in-depth exploration makes an important practical contribution by revealing that with the implementation of discretionary tax-penalty benchmarks, tax enforcement standardisation significantly improves corporate financial reporting quality, helps to optimise the external governance environment of listed companies, and improves the information disclosure environment in capital markets.

Institutional background
As early as 2012, the STA issued its Guiding Opinions to standardise tax enforcement and build a harmonious relationship between tax collection and payment. The Guiding Opinions point out that when granting tax authorities discretionary power on penalising tax avoidance, the penalties must be regulated and controlled to improve the system of discretionary powers for tax administrative penalties. Moreover, according to the 'Implementation Outline for the Construction of a Rule of Law Government (2015-2020)', adhering to strict, standardised, impartial, and civilised law enforcement is a principal task in constructing China's rule of law government. In the field of taxation, the main need was to improve the benchmark system for tax penalties to promote the standardisation of tax enforcement. Consequently, in 2016 the STA issued the Exercise Rules, which standardised the management of discretionary powers for tax-avoidance penalties, refined and quantified the standards for discretion in tax administration. Subsequently, the provincial taxation bureaus issued their respective discretionary taxpenalty benchmarks in accordance with the Exercise Rules, further standardising the scope, types, and extent of discretionary powers over local tax penalties, as well as refining and quantifying them. They establish the discretionary benchmarks for local tax penalties, ensures the fairness and rationality of the tax-enforcement system, and assists China's efforts to achieve a discretionary tax-penalty benchmarking system. Additionally, in the 'Opinions on Further Deepening the Reform of Tax Collection and Administration' (issued on 24 March 2021), the General Office of the CPC Central Committee and the General Office of the State Council pointed out that strict regulation would bring continuing improvement in tax-law enforcement and tax services. The Opinions also reaffirmed commitment to standardising and continuously improving the discretionary benchmark system for tax penalties.
Improving the tax-penalty benchmark system is evidently an important measure in reforming the tax collection, administration, and governance of local tax authorities, in the process of building a government ruled by law. Indeed, the tax authorities do not have unfettered discretion to impose whatever tax penalties they choose, and they must establish reasonable legal benchmarks, strictly abide by legal procedures, and exercise their discretionary power in a fair and open manner. By formulating strict discretionary rules for the exercise of tax administration power, it is possible to effectively reduce the payment of 'personal' and 'relationship' taxes, maximise regulation of taxpayers, promote fair law enforcement and tax-law compliance, and establish a well-founded tax system that enables healthy economic, social, and environmental development. In general, the benchmark system regulates the exercise of discretionary powers over tax penalties, clarifies the specific standards for handling tax violations in various localities, effectively limits collusive rent-seeking behaviour between tax authorities and taxpayers, and strengthens the legal basis for tax penalties. Improved deterrence of tax evasion will help to promote the standardisation of tax enforcement.

Literature review
Corporate financial reporting quality is a central discussion topic in corporate finance and accounting research, and is related to the healthy development of the financial information disclosure environment in capital markets. Studies have generally focused on three main types of factors affecting corporate financial reporting quality. The first factor type is corporate-governance mechanisms. For example, internal audits (Abbott et al., 2016), internal controls (Qi et al., 2010), audit committee expertise (Krishnan et al., 2011), institutional investors (Yang et al., 2012), and the characteristics of the board of directors (Aifuwa & Embele, 2019;H. Zhang et al., 2019) have been found to impact financial reporting quality. The second factor type is the personal characteristics of executives. For example, CEO age (Huang et al., 2012), CFO narcissism (Ham et al., 2017), and the characteristics of the executive team (Zhang, 2019) have been found to impact financial reporting quality. The third factor type is macro-institutional, such as the implementation of International Financial Reporting Standards (IFRS; Iatridis, 2010), business ethics (Choi & Pae, 2011), and the level of trust (Garrett et al., 2014). Conversely, few prior studies have considered whether and how tax enforcement standardisation affects financial reporting quality.
Tax enforcement, as an effective external corporate-governance mechanism, can substantially affect firms' financial behaviour (Desai & Dharmapala, 2006;Dyck & Zingales, 2004). Prior studies show that with the improvement of tax collection and management, the tax authorities becomes more likely to detect tax avoidance behaviour by enterprises. Once the tax authorities begin to question such behaviour, companies and managers face much greater costs of tax avoidance (Rego & Wilson, 2012). Thus, tax enforcement can prevent companies from solely pursuing their private interests (Dyck & Zingales, 2004) and inhibit corporate tax avoidance (Hoopes et al., 2012). An effective tax-collection system can also reduce agency costs (Xu et al., 2011;Zeng & Zhang, 2009), suppress earnings management (Ye & Liu, 2011), reduce stock price crash risk (Kim et al., 2011), improve financial reporting quality (Hanlon et al., 2014), and reduce the cost of capital (El Ghoul et al., 2011;Guedhami & Pittman, 2008).
In summary, literature on the impact of tax enforcement on corporate financial behaviour mainly focuses on agency costs, corporate tax avoidance, financial reporting quality, and financing behaviour. The study by Hanlon et al. (2014) is one of the few to consider the impact of tax enforcement on financial reporting quality. Although they find that tax enforcement can improve corporate financial reporting quality, their study's main indicator is the number of company inspections by the tax authority during a fiscal year. This indicator has a significant endogeneity problem: while tax audits can inhibit corporate tax avoidance, companies with a higher degree of tax avoidance are inherently more likely to be audited by taxation bureaus. More generally, this ex-post indicator of tax enforcement intensity does not take into account the prior deterrent effect of the law, and there are no clear discretionary benchmarks of tax penalties for specific tax violations. The enterprises' collusive rent-seeking behaviour is poorly managed with tax authorities . To address the deficiencies and gaps in analyses of the impact of tax enforcement on corporate financial behaviour, studies should consider not only tax enforcement efforts but also tax enforcement standardisation from the perspective of discretionary tax-penalty benchmarks.

Hypotheses development
In practice, companies use tax-related items for earnings management (Dhaliwal et al., 2004;Gupta et al., 2016) as a means to avoid corporate taxation (Hanlon & Heitzman, 2010). However, with respect to enterprises engaging in serious tax avoidance, the tax authorities have incentives to more rigorously inspect their accounting records (Li et al., 2019), thereby limiting earnings management activities and reducing corporate tax avoidance. Strict tax enforcement substantially reduces companies' incentives to underestimate their taxable revenue (Hoopes et al., 2012), in turn reducing the degree of downward earnings management (Li et al., 2019). However, companies that engage in serious tax avoidance (directly manifested in large accounting-tax differences) may use a combination of taxable and nontaxable items to increase their earnings and reduce the suspicion of tax authorities. Tax authorities rarely pay attention to the accounting practice of companies making larger profits to increase taxable revenue. However, upward earnings management will only increase the tax paid by a company if these activities focus on taxable items, and not on nontaxable items. As tax authorities strengthen their enforcement regime, loopholes in the tax system will be reduced. A strict tax enforcement system will significantly restrict upward earnings management for nontaxable items and force enterprises to focus such activities on taxable items. Furthermore, the strengthening of tax enforcement will significantly increase the income tax cost of upward earnings management, thereby reducing the degree to which enterprises engage in such behaviour (Ye & Liu, 2011). Therefore, tax enforcement can restrain both upward and downward earnings management, in turn improving corporate financial reporting quality. Correspondingly, embedding tax enforcement standardisation in what we call the exante legal deterrence factors (i.e. tax penalties) may also affect financial reporting quality. We thus predict that the local implementation of discretionary benchmarks for tax penalties may significantly improve corporate financial reporting quality.
Along with implementing discretionary tax-penalty benchmarks, tax enforcement standardisation clarifies the specific administrative penalty standards for tax violations, which strengthens the deterrent effect of the law and clearly signals the cost of tax violations to enterprises. Greater transparency of tax-penalty information could reduce the risk of corporate tax violations, lower corporate earnings management behaviours, and thereby improve corporate financial reporting quality. For example, with regard to handling tax violations, the STA's Jiangsu branch clarifies that: if the amount of tax evasion is less than RMB 500,000, a fine of more than 50% and less than 1 times the unpaid tax shall be imposed; if the amount of tax evasion is more than RMB 500,000 but less than RMB 1 million, a fine of not less than 1 times but not more than 2 times the amount of tax in arrears shall be imposed; if the amount of tax arrears evaded is more than RMB 1 million, a fine of more than 2 times the amount of tax owed shall be imposed.
China's Tax Enforcement Law only specifies the legal liability for unpaid tax; it does not cover the discretionary benchmarks for tax violations. Based on the discretionary benchmarks established by the STA's Jiangsu branch, any local enterprises that owe tax face penalties corresponding to the specified degree of tax violation. Specifying the penalties ex ante increases the degree and effectiveness of punishment for tax violations; accordingly, it will increase the incentives for enterprises to carefully inspect their own taxrelated items, thereby improving their financial reporting quality.
Furthermore, the standardisation and implementation of discretionary tax-penalty benchmarks could greatly improve the tax enforcement system, limit the scope of collusive rent-seeking between tax enforcement personnel and company managers, and increase the transparency of corporate tax-related projects. In turn, this change in the external environment encourages companies to more thoroughly inspect their taxrelated accounting projects to reduce the risk of tax violations, thus inhibiting corporate earnings management. Consequently, this will improve financial reporting quality. Before the discretionary benchmarks for tax penalties were clarified, tax enforcement personnel had greater discretion in tax enforcement, which allowed for collusive rent-seeking with enterprises. In effect, this condoned earnings management activities by enterprises seeking to avoid tax, in turn undermining corporate financial reporting quality. However, clarification of the discretionary benchmarks for tax penalties changes this situation. For example, in the tax enforcement category, the STA's Zhejiang branch specifies penalties to which the tax collectors refer if taxpayers (including enterprises) do not file tax returns, fail to pay taxes, or underpay taxes. Specifically, if the taxpayer fails to pay or underpays taxes of less than RMB 100,000, the unpaid or underpaid taxes and overdue fines shall be pursued, and a fine of more than 50% but not more than 3 times shall be imposed. If the taxpayer fails to pay or underpays taxes of more than RMB 100,000, the unpaid or underpaid taxes and overdue fines shall be pursued, and a fine of more than 60% but not more than 5 times shall be imposed. Consequently, the discretion of tax enforcement personnel has been reduced and the tax-collusion environment disrupted, leading to lower risk of tax violations and improved corporate financial reporting quality.
Based on the above discussion, we can expect that with the implementation of discretionary benchmarks for tax penalties, tax enforcement standardisation can significantly improve corporate financial reporting quality. We thus propose the following hypothesis: H1: The standardisation of tax enforcement following the implementation of discretionary benchmarks for local tax penalties significantly improves corporate financial reporting quality compared with the prior situation.
Corporate tax avoidance has been found to increase the complexity of financial reporting and lower the information transparency for tax-related items, which harms corporate financial reporting quality (Balakrishnan et al., 2019;Bennedsen & Zeume, 2018;Donohoe & Robert Knechel, 2014). It is evident that compared with low-taxavoidance companies, those with high tax avoidance are more likely to increase the complexity of their financial reporting, thus reducing their quality. By clarifying in advance the specific penalty benchmarks for tax violations, the standardisation of tax enforcement could enhance the legal deterrent effect of ex-ante tax collection and management, making it widely known that enterprises attempting to avoid tax face the risk of penalties. To reduce their risk of incurring penalties, high-tax-avoidance companies will be strongly incentivised to suppress such behaviour; accordingly, the transparency of the tax-related information they provide will increase substantially. Therefore, implementation of the discretionary benchmark system for tax penalties can significantly inhibit the tax avoidance behaviour of companies that previously engaged in a high degree of avoidance, thereby improving such companies' information transparency on tax-related accounting projects. We thus propose the following hypothesis: H2: The positive impact of tax enforcement standardisation on corporate financial reporting quality after the implementation of discretionary benchmarks for local tax penalties is particularly reflected in companies with a high degree of tax avoidance.
An imperfect tax enforcement system will encourage tax collectors and enterprises to conspire to avoid tax. For example, tax collectors with greater discretion on enforcement may accept bribes to condone corporate tax avoidance (Tian & Fan, 2018). Companies avoid tax by increasing their earnings management, thereby raising the risk of tax violations, which then damages corporate financial reporting quality (Donohoe & Robert Knechel, 2014). It is, thus, evident that collusion between tax enforcement officers and company managers will damage corporate financial reporting quality. Besides the above-mentioned benefits, tax enforcement standardisation could reduce the scope for collusive rent-seeking between enforcement officers and firm managers by publicising the risks of corporate tax avoidance. This effectively restrains earnings management by companies with a high degree of tax enforcement collusion, thereby improving the information transparency of their tax-related accounting projects. We thus propose the following hypothesis: H3: The positive impact of tax enforcement standardisation on corporate financial reporting quality after the implementation of discretionary benchmarks for local tax penalties is particularly reflected in companies with a high degree of tax enforcement collusion.

Sample and data
Our study focuses on the 22 provinces whose local taxation bureaus issued discretionary benchmarks for tax penalties up to 31 December 2018 (listed in Appendix I). We obtain sample data for A-share listed companies from 2013 to 2018. After excluding (1) financial companies, (2) *ST and ST companies, and (3) companies with missing data, the final sample comprises 10,500 companies. To reduce the influence of outliers, we winsorise all continuous variables at the 1% and 99% levels. Tax revenue and other macroeconomic data for each province are sourced from the Chinese Research Data Services (CNRDS) and RESSET databases. Data on the legal environment are sourced from the 'China market index report by province ' , and financial data are taken from the China Stock Market & Accounting Research Database (CSMAR) database.

Corporate financial reporting quality
It will take time for the implementation of discretionary tax-penalty benchmarks to influence corporate financial reporting quality, and our focus is on the degree of earnings management, rather than the direction. Accordingly, we adopt the calculation method of Kothari et al. (2005). Specifically, we use the absolute value of the discretionary accrual value in period t + 1 as a surrogate variable for corporate financial reporting quality (FRQ i,t+1 ). The calculation method is as follows: calculate the normal accrued profit (NDA i ) at the company level according to formula (1), then subtract NDA i from the total accrued profit (TA i ) to obtain the controllable accrued profit (DA i ). DA i represents corporate financial reporting quality (FRQ i,t+1 ). The higher (lower) its value, the poorer (better) the quality of the company's financial reporting.
where TA i is the total accrued profit of company i for the current period, measured as the difference between current net profit and net cash flow from operating activities; A i is the assets of company i at the end of the previous period; ΔREV i is the difference in the main business income between the end of the current period and the end of the previous period for company; ΔREC i is the difference in accounts receivable between the end of the current period and the end of the previous period for company i; PPE i is the value of fixed assets at the end of the current period for company i; and ROA i is the return on assets of company i, calculated as the ratio of current net profit to total assets.

Standardisation of tax enforcement
Different provinces have issued and implemented the discretionary tax-penalty benchmarks at different times. This situation provides suitable conditions for establishing a variable to represent the standardisation of tax enforcement: specifically, Rov equals 1 if the year of analysis is in and after the first year in which the benchmark system is issued, and 0 otherwise.

Corporate tax avoidance
The literature largely focuses on corporate income tax avoidance (e.g. Hanlon & Heitzman, 2010;Tang et al., 2021;Zhang et al., 2021), for the following reasons. Corporate income tax is a direct tax that cannot be passed on and is directly borne by the enterprise. By contrast, value-added, consumption, and business taxes are indirect taxes whose burden can be passed on to other taxpayers. Therefore, corporate income tax is easier to manipulate than indirect tax and the implementation of discretionary benchmarks for tax penalties is more likely to impact on the avoidance of corporate income tax, rather than of indirect taxes. Therefore, we choose the effective tax rate of corporate income tax to measure the degree of corporate tax avoidance. Drawing on the calculation method of Dyreng et al. (2008) and Liu and Ye (2014), an enterprise's effective tax rate (ETR) is equal to the ratio of income tax expenses to pre-tax accounting profit. Following Liu and Zhao (2019), we shorten ETR to the interval [0, 1]. Because many Chinese enterprises enjoy preferential taxation, it may not be appropriate to measure the degree of tax avoidance using the actual tax rate. Therefore, we use the nominal corporate tax rate minus the actual tax rate to measure the degree of corporate tax avoidance (TAXAVOID i,t ). The values are categorised based on the median ETR adjusted by the nominal tax rate. If ETR adjusted by the nominal tax rate is greater than or equal to the median, TAXAVOID i,t equals 1, indicating that the degree of corporate tax avoidance is relatively high; otherwise, TAXAVOID i,t equals 0, indicating a low degree of corporate tax avoidance.

Control variables
Referring to Ye and Liu (2011) and Hanlon et al. (2014), we include the following control variables: (1) return on equity (ROE i,t ), which equals the ratio of net profit to the owner's total equity; (2) book-to-market ratio (BM i,t ), which is the book value of the company divided by its market value; (3) degree of collusion in tax enforcement (EC i,t ), which is the ratio of business entertainment expenses to operating income (Tian & Fan, 2018): when testing H3, if EC i,t is greater than or equal to (less than) its median among the sample companies, it takes the value 1 (0), indicating companies with a higher (lower) degree of tax enforcement collusion; (4) assetliability ratio (LEV i,t ), which equals the ratio of the company's total liabilities to total assets; (5) inventory ratio (INV i,t ), which equals the ratio of total inventory to total assets; (6)cash flow ratio (CASH i,t ), which equals the ratio of cash flow from operating activities to total assets; (7) fluctuations in cash flow (CASH_SD i,t ), which equal the standard deviation of the company's net cash flow from operating activities in the first three periods; (8) company size (SIZE i,t−1 ), which equals the value of the logarithm of all assets at the beginning of the current period; (9) board size (BDSIZE i,t ), which equals the logarithm of the number of board directors; (10) proportion of independent directors (IDR i,t ), which equals the ratio of the number of independent directors to the total number of directors. (11) Big Four auditing (BIG4 i,t ), which takes the value 1 if the company is audited by a Big Four accounting firm, and 0 otherwise; (12) audit opinion of the previous year (OPINION i,t−1 ), which takes the value 1 if the company was issued a non-standard audit opinion in the previous year, and 0 otherwise; (13) nature of the enterprise (SOE i,t ), which takes the value 1 if the company is a state-owned enterprise, and 0 if a private enterprise; (14) legal environment (LAW i,t ), which is the logarithm of the 'development of market intermediaries and the score for the legal system environment' in the 'China market index report by province ' ; (15) tax enforcement intensity (TAX_EN i,t ), which we calculate using the method of Tang et al. (2021): if the tax enforcement intensity index is greater than or equal to (less than) the median calculated by year and province, TAX_EN i,t takes the value 1 (0), indicating an area with strong (weak) tax collection and management. Considering the potential influence on the regression results of time-invariant factors at the firm level, we also control for firm fixed effects (γ i ). Finally, we control for industry and year fixed effects with industry and year dummy variables.

Hypotheses testing approach
Following Ye and Liu (2011) and Hanlon et al. (2014), among others, we test H1 by constructing the following difference-in-differences (DID) fixed-effects model with multiple time points: where a negative and statistically significant value of α 1 means that tax enforcement standardisation significantly improves corporate financial reporting quality after the implementation of discretionary benchmarks for local tax penalties. Second, to test H2 and H3, we group sample companies according to their degrees of corporate tax avoidance and tax enforcement collusion, respectively, and further perform the regression analysis using formula (2). Table 2 lists the descriptive statistics for the main variables. The mean of FRQ i,t+1 is 0.048, while the mean of Rov is 0.304. Furthermore, the mean of TAX_EN i,t is 0.608, indicating that 60.8% (39.2%) of sample companies are located in areas with high (low) tax enforcement intensity. Table 3 presents the results for examining H1. As shown in column (1) and column (2), the negative correlation between Rov and FRQ i,t+1 becomes statistically significant . To summarise, tax enforcement standardisation significantly improves corporate financial reporting quality, thus supporting H1. An important assumption underlying the DID method is that the treatment and control groups have parallel trends before the policy shock. As shown in column (3) of Table 5, none of Rov(−3), Rov(−2), and Rov(−1) have a statistically significant relationship with FRQ i,t+1 , thus supporting the parallel trend assumption. In addition, there is a negative correlation between Rov(1) and FRQ i,t+1 significantly at the 5% level, indicating there is a significant difference between the treatment and control groups after the implementation of the discretionary penalty benchmarking system, further supporting H1. Table 4 presents the results for examining H2 and H3. The results in column (1) show that in the high-tax-avoidance group, Rov and FRQ i,t+1 have a negative correlation significantly at the 10% level; conversely, column (2) shows that this negative correlation is not statistically significant in the low-tax-avoidance group, thus supporting H2. Furthermore,  * p < 0.1; ** p < 0.05; *** p < 0.01; the t values are provided in parentheses in columns (1) to (4), adjusted for heteroscedasticity; robust standard errors are clustered by province.

Regression results
the results in column (3) show in a high degree group of tax enforcement collusion, there is a negative correlation between Rov and FRQ i,t+1 significantly at the 1% level. Conversely, column (4) shows that this correlation becomes positive and statistically non-significant in the group of companies with a low degree of collusion, thus supporting H3.

Placebo tests
To confirm the robustness of the results of our quasi-natural experiment, we conduct two placebo tests. In the first placebo test, we shift the event period forward by two years. The basic principle is demonstrated by the following example. In Zhejiang province, the implementation date of the tax-penalty benchmarking system is 13 July 2017. For the placebo test, we set the date two years earlier (i.e. 13 July 2015) and reassign Rov to take the value 1 for observation years in Zhejiang province of 2015 or later, and 0 otherwise. We apply the same approach for all 22 provinces in our sample. Following Almeida et al. (2017), if Rov and FRQ i,t+1 are not significantly correlated in the placebo test regression results, this indicates that our quasi-natural experiment results are robust. As shown in Table 5, the results demonstrate that H1, H2, and H3 do not hold when the event period is moved forward by two years. This indicates that the results of our quasi-natural experiment are robust. Further following Almeida et al. (2017), our second placebo test shifts the event period forward two years and also removes the period in which the discretionary penalty benchmarking system was actually implemented. As shown in Table 6, the results demonstrate that H1, H2, and H3 do not hold when the event period is moved forward by two years and the original event period is excluded. These findings further indicate the robustness of the main results.

Considering the impact of the STA's Exercise Rules
At the provincial level, the system is based mainly on the Exercise Rules issued by the STA on 30 November 2016. Before the provincial tax authorities formally introduce discretionary benchmarks, local tax departments may deploy and implement regional tax- penalty benchmarks in line with the spirit of the Exercise Rules. This raises the concern that our main research findings may indicate the impact of the Exercise Rules, rather than of local tax-penalty benchmarks. To overcome this concern, we control for the impact of the Exercise Rules using the variable Post: if the year is 2016 or later, the variable Post takes the value 1, and 0 otherwise. We rerun formula (2) after controlling for Post and test H1, H2, and H3 separately. If the results still support the hypotheses, this will indicate that our main results reflect the impact of discretionary tax-penalty benchmarks imposed in various localities, rather than the influence of the Exercise Rules, thereby confirming that our findings are robust. As shown in Table 7, after considering the impact of the STA's Exercise Rules, our main findings remain valid.

Considering the impact of the Golden Tax Project Phase III
The impact of the Golden Tax Project Phase III on enterprises' financial behaviour is already becoming evident. For example, it has enhanced the intensity of big-data-based tax collection and management, reduced the degree of tax evasion by enterprises (Zhang  et al., 2020), and inhibited corporate earnings management (Sun et al., 2021). These findings suggest that the implementation of the Golden Tax Project may have influenced our main results. To test this, we establish the implementation variable DID1 for the Golden Tax Project Phase III. If the observation year for a given locality is the same as or later than the year in which the Golden Tax Project Phase III was locally implemented, DID1 takes the value 1, and 0 otherwise. After controlling for DID1, if we find that the main results hold, it will confirm they are robust. As shown in Table 8, after considering the impact of the Golden Tax Project Phase III, we find that our main results continue to hold.

Considering the impact of the value-added tax (VAT) reform
Since 2012, China has been gradually implementing VAT reform, aiming to eliminate double taxation of enterprises, implement tax reductions for enterprises, and stimulate their innovative vitality. With the pilot programme having reached full coverage, the impact of the VAT reform on enterprises' financial behaviour is gradually emerging. For example, the reform has improved the corporate technical level of enterprises (Fan & Peng, 2017). To test whether the VAT reform influenced our main findings, we establish the implementation variable DID2 to control for this reform. If the observation year for a pilot-industry enterprise is the same as or later than the year in which the VAT reform was locally implemented, DID2 takes the value 1, and 0 otherwise. Although the VAT reform primarily targets the service industry, it also reduces the tax burden of all manufacturing enterprises (Fan & Peng, 2017) by enabling them to use the many productive services they purchase as input deductions, which was not possible pre-reform. We thus place all manufacturing enterprises in the treatment group (DID2 = 1). After controlling for DID2, if the main findings hold, it will confirm they are robust. As shown in Table 9, after considering the impact of the VAT reform, our main results remain valid.

Further analysis focused on real earnings management activities
As mentioned earlier, tax enforcement standardisation can improve corporate financial reporting quality by inhibiting earnings management behaviour. Besides accrued earnings management, which our main analysis considers, companies may also engage in real earnings management activities. Some studies find that accrued earnings management and real earnings management are substitutes (Zang, 2012). For example, after the passage of the Sarbanes-Oxley Act in the United States in 2002, strict accounting regulations significantly reduced the level of accrued earnings management, but the level of real earnings management increased (Cohen et al., 2008). This suggests that when corporategovernance mechanisms constrain the management of a firm's accrued earnings, its real earnings management activities substantially increase (Zang, 2012). The question thus arises of whether, with the implementation of discretionary tax-penalty benchmarks, tax enforcement standardisation can effectively constrain real earnings management to improve corporate financial reporting quality. To further examine this issue, we follow Roychowdhury (2006) by calculating the real earnings management level. We first calculate three indicators: abnormal cash flow from operating activities (AbCFO), abnormal production costs (Ab_PROD), and abnormal discretionary fees (Ab_DISEXP). We then use the sum of the negative values of these three indicators to proxy for the level of real earnings management (REM). In addition, because we focus on the degree of earnings management rather than the direction , we take the value of REM after taking the absolute value as an alternative variable for FRQ i,t+1 : the higher the value, the poorer the company's financial reporting quality. As shown in Table 10, column (1) indicates that Rov has a negative correlation with FRQ i,t+1 significantly at the 5% level. This shows that tax enforcement standardisation significantly inhibits enterprises' real earnings management behaviour, thereby improving the quality of their financial reporting. Furthermore, column (2) indicates the results confirms the parallel trend assumption. Overall, tax enforcement standardisation can improve corporate financial reporting quality by inhibiting real earnings management behaviour. * p < 0.1; ** p < 0.05; *** p < 0.01; the t values are shown in parentheses in columns (1) to (5), adjusted for heteroscedasticity; robust standard errors are clustered by province.

Conclusion
This study used the staggered local implementation of tax-penalty discretionary benchmarks as a quasi-natural experiment to examine the relationship between tax enforcement standardisation and corporate financial reporting quality. Taking A-share listed companies from 2013 to 2018 as the research sample, we explored whether the implementation of benchmarks influenced corporate financial reporting quality. We find that tax enforcement standardisation significantly improves corporate financial reporting quality compared with the pre-implementation period. Furthermore, considering the degrees of corporate tax avoidance and collusion between tax collectors and corporations, we find that the positive effect on corporate financial reporting quality is primarily driven by companies with high tax avoidance and those with high tax enforcement collusion.
In general, our findings demonstrate the necessity of focusing on not only the consequences of tax collection and management but also the pre-reform environment and how discretionary tax-penalty benchmarks influence corporate financial reporting quality. With the implementation of this system, the external governance environment of listed companies can be optimised by standardising the tax enforcement system, thereby improving the information disclosure environment in capital markets. Therefore, at the provincial level and according to the local benchmarks for tax penalties, each local taxation authority should fully implement these tax enforcement regulations as part of the process of standardising tax collection and administration. The discretionary powers of local taxation authorities regarding tax penalties and standardising tax enforcement  (1) and (2), adjusted for heteroscedasticity; robust standard errors are clustered by province.
should be exercised fairly and reasonably to effectively encourage improved corporate financial reporting. In addition, to prevent abuse of this discretion, it is necessary to strengthen the supervision of tax enforcement personnel and thereby maintain the independence of the tax enforcement system.