Introduction

Existing research on tax information primarily focused on taxation and corporate decision-making, notably in areas such as entrepreneurship (Edwards and Todtenhaupt, 2020; Djankov et al., 2010), innovation (Akcigit et al., 2022; Li et al., 2021a; Mukherjee et al., 2017), investment activities (Ohrn, 2018; Becker et al., 2013; Poterba and Summers, 1983) and corporate risk-taking (Armstrong et al., 2019; Langenmayr and Lester, 2018; Ljungqvist et al., 2017). However, there needs to be more research investigating the influence of tax information on financial reporting quality, particularly about the financial restatements.

According to Karpoff et al. (2017), financial restatements may stem from the negligence of accounting personnel or, in more severe cases, arise from the misconduct of company executives in achieving analysts’ expectations through earnings management. These financial restatements can result in adverse economic consequences, including negative reactions in the capital markets (Bauer et al., 2021; Kim et al., 2016), increased cost of capital (Kravet and Shevlin, 2010; Hribar and Jenkins, 2004), and diminished investor trust (Elliott et al., 2018; Garrett et al., 2014). Given the significant impact of financial restatements on capital markets, investigating how tax information influences financial restatements has become a crucial topic (Amel-Zadeh and Zhang, 2015; Garrett et al., 2014).

The theory proposed by Bartov and Bodnar (1996) suggests that increasing the accessibility of tax information may improve a company’s financial information quality. According to the theory, managers consider the costs of preparing and disclosing information against the potential benefits. Adopting new information technology can disrupt this balance, leading managers to enhance information disclosure if they expect the benefits to outweigh the implementation costs. Implementing a tax information system can make it harder for firms to undertake tax evasion through fraudulent activities or fund transfers (Overesch and Wolff, 2021; Bennedsen and Zeume, 2018). As a result, companies may put more effort into ensuring the accuracy and completeness of their financial reports to avoid higher legal and reputation costs (Gallemore et al., 2014).

However, the literature lacks conclusive evidence on the relationship between tax information and financial restatements. From one perspective, enhancing the transparency of corporate tax information may reduce financial restatements. Previous research indicates that tax information allows the public to gain deeper insights into the operational status of corporations (Hoopes et al., 2018; El Ghoul et al., 2011). For example, tax information can help stakeholders identify the aggressive nature of a company’s tax position (Overesch and Wolff, 2021; Lisowsky et al., 2013; Frank et al., 2009), curb profit shifting (Overesch and Wolff, 2021; Hope et al., 2013; Dyck and Zingales, 2004), and provide incremental information on corporate governance (Dhaliwal et al., 2013; Guedhami and Pittman, 2008; Kumar and Visvanathan, 2003). Because tax information often reflects the economic realities underlying a company’s operations, it provides a vital means to corroborate the financial statements delineated in annual reports. This assertion is supported by De Simone et al. (2015), Dhaliwal et al. (2013), and Hanlon and Heitzman (2010), who highlight the integral role tax data plays in the verification of financial documentation. Through enhanced information transparency, tax information may mandate firms to adopt a more discerning approach to disclosing financial information and push financial reporting quality.

From another perspective, some research indicates that increasing the accessibility of tax information may raise the likelihood of financial restatements. When tax information becomes more transparent, companies may face increased tax scrutiny and tax burdens (Overesch and Wolff, 2021; Joshi et al., 2020; Kerr, 2019), which may worsen their financial conditions (Waseem, 2018; Goh et al., 2016). In such cases, companies might adopt more aggressive earnings management measures to meet analysts’ expectations (Cazier et al., 2015; Erickson et al., 2013; Monem, 2003), diminishing financial reporting quality. Moreover, although the tax information system increases the transparency of tax information, it does not guarantee that all instances of tax-related financial manipulation will be exposed or detected. For example, managers can inflate profits on tax returns to prevent raising suspicions from stakeholders (Lennox et al., 2013; Dhaliwal et al., 2004). Erickson et al. (2004) report an analysis of 27 fraudulent firms that overstate pre-tax profits, but none of these financial frauds were discovered by the IRS (Lennox et al., 2013). Joshi et al. (2020) also show that because of the balancing effects of different tax avoidance practices that escape scrutiny, certain tax information does not significantly affect corporate tax-related behaviorFootnote 1.

This study attempts to reconcile the competing relationship between tax information and financial restatements. The greatest empirical challenge is identifying a causal relationship without reverse causality and omitted variables. First, sharing tax information may enhance a company’s financial information quality, and companies may adjust financial information to alter their effective tax rate. Second, the company’s endogenous factors may determine tax information and financial restatements. The financial condition, governance structure, management decisions, and other company characteristics may collectively affect the dynamics of tax information and financial restatements. Finally, tax information and corporate financial restatements may be related to unobserved economic and financial factors.

To mitigate endogeneity concerns, we utilize a natural experiment, namely the third phase of the China Tax Administration Information System (CTAIS-3), which leads to an exogenous surge in the availability of tax information (Xiao and Shao, 2020). This setup is appealing for several reasons. First, CTAIS-3 is a tax policy gradually implemented in various provinces in China since 2013. Specifically, CTAIS-3 is a tax management system through which companies must submit taxes, apply for digital invoices in standardized formats, and pay taxes through banks. Therefore, the CTAIS-3 facilitates exchanges of tax information between tax authorities and third-party institutions such as financial institutions, industrial and commercial bureaus, and social insurance departments. The motivation for using CTAIS-3 is to ensure the effective collection and analysis of tax information. Therefore, the implementation of CTAIS-3 serves as a predominantly exogenous event rather than being influenced by the quality of a company’s financial information. Secondly, due to CTAIS-3, the reduction in tax opacity is quite significant. For example, by the end of 2016, the State Administration of Taxation had established collaborative relationships with 32 government departments, sharing 190 types of information, including equity transfers, social insurance, and asset transactions. In addition, tax information is shared with the State Administration of Taxation to prevent cross-regional fund transfers and profit concealment. Thirdly, implementing CTAIS-3 allows us to examine the impact of tax information supply shocks while controlling for other unobservable firm features or economic factors. Given that the rollout of CTAIS-3 occurs gradually across various provinces, it is less likely to be simultaneously implemented with other potentially confounding projects in the same province. Therefore, by comparing the impact of financial restatements in regions where CTAIS-3 is implemented (treatment firms) and regions where CTAIS-3 is not implemented (control firms), we can discern the direct link between tax information and financial restatements.

Applying a difference-in-differences (DID) technique, we study the relationship between tax information and financial restatements using a panel of 3546 public Chinese firms from 2010 to 2019. In line with our hypothesis, the rise in the availability of tax information substantially diminishes financial restatements. Additionally, we observe that the inverse relationship between tax information supply and financial restatements is intensified for firms possessing higher levels of information asymmetry.

Our study makes the following contributions. First, it adds to the existing literature on the drivers behind accounting information quality. Existing literature explores the impact of audit expertise (Ahn et al., 2020; Gunn and Michas, 2018; Jayaraman and Milbourn, 2015), top management characteristics (Zhang, 2019; Feng et al., 2011; Efendi et al., 2007), institutional investors (Cahan et al., 2024; Samuels et al., 2021), and government enforcement (Blankespoor, 2019; DeHaan et al., 2015) on accounting information quality. Yet, more investigation is required to understand the influence of tax information on financial information quality. Compared with the most relevant studies (Dhaliwal et al., 2013; Kumar and Visvanathan, 2003), this paper addresses the endogeneity issues in corporate tax information reporting by establishing the CTAIS-3 information system in China. This CTAIS-3 reform is a government-led initiative that can be exogenous to firms.

Secondly, our research adds to the developing literature investigating the economic implications of tax information technology. Prior studies reveal that tax information exerts a significant influence on tax compliance (Hoopes et al., 2018; Lisowsky et al., 2013; Chan et al., 2010), internal control quality (Amberger et al., 2021; De Simone et al., 2015; Becker et al., 2013), and financial reporting (Balakrishnan et al., 2019; Hope et al., 2013; Frank et al., 2009). These studies collectively suggest that tax information technology is instrumental in shaping firms. However, further investigation is warranted to explore whether the public dissemination of tax information contributes to a reduction in financial restatements by firms. Therefore, we extend this line of research, demonstrating that the exogenous changes in tax information supply can reduce corporate financial restatements by enhancing the external information environment of firms.

Thirdly, our findings expand the literature regarding the spillover effects of government behavior. Scholarly consensus indicates that government behavior can significantly influence corporate conduct. Previous studies discuss how the government affects financial reporting and auditing by establishing information disclosure standardsFootnote 2, regulations, and market forces (e.g., Houston et al., 2019; Chen et al., 2010). Although discussions around the influence of government behavior on financial information quality are escalating, empirical evidence on the spillover effects of government behavior remains limited. We extend this line of research by studying the spillover effects of government behavior through a quasi-natural experiment. After the government establishes a large-scale tax data information system, the external information environment of businesses will improve, motivating them to enhance the quality of financial information proactively. Therefore, we provide additional evidence indicating that government behavior is essential in improving corporate financial disclosure.

Finally, our study provides pivotal insights to policymakers and stakeholders. A fair, competitive, and sustainable tax system is vital for a country’s future prosperity. The ongoing debate on tax information transparency is crucial to this discussion (Lisowsky et al., 2013). Supporters advocate for transparency, believing the increased transparency of tax information can effectively combat corporate tax evasion. However, opponents argue that such transparency may encroach upon corporate privacy rights (Bradbury, 2013). This article provides evidence that tax information transparency enhances corporate financial reporting quality. This finding carries important policy and regulatory implications, suggesting that digital tax enforcement mechanisms can yield beneficial outcomes. Moreover, our research indicates that other market participants, including investors, creditors, and suppliers, stand to gain from understanding the broader implications of tax information systems. By integrating the effects of tax transparency into their evaluation frameworks, these stakeholders can better comprehend a company’s financial information.

The following section provides an overview of the institutional background of the CTAIS-3 Reform in China and presents hypotheses. Section “Sample and descriptive statistics” contains a detailed account of the sample and data sources. Section “Empirical results” outlines empirical results. Section “Conclusions” concludes the study.

Institutional background and hypothesis development

Institutional backgrounds

The Chinese State Administration of Taxation (SAT) started the CTAIS-3 project in 2013 to fully promote tax collection and management information. This project steadily built up a national taxation information system across all provinces and cities in China and reached national coverage at the end of 2017. The CTAIS-3 does three measures: (1) establishing a unified technical platform covering all types of taxes; (2) Consolidating tax information at the national and provincial-level tax bureaus to facilitate data-sharing among geographically dispersed tax authorities; (3) sharing information with third parties such as financial institutions and other government departments to facilitate the utilization of tax credits.

The CTAIS-3 system can alleviate information asymmetry in two ways. First, CTAIS-3 reduces the prevalence of false invoices and enhances the reliability of firms’ operational information. After implementing CTAIS-3, firms can issue invoices in a standardized format through the system, with each invoice linked to a unique taxpayer. Therefore, tax authorities can quickly identify invoices without genuine transactions, thereby curbing income or profit manipulation. Additionally, through cooperation with banks, tax authorities can access information on abnormal fund transfers. As a result, CTAIS-3 makes it more difficult for businesses to manipulate income or profits by reducing the feasibility of false invoices and facilitating information exchange between tax authorities and society.

Secondly, CTAIS-3 lays the technical foundation for the social tax credit system. Aligned with the “Social Credit System Construction Plan” released by the Chinese State Council in 2014, the Chinese government encourages the establishment of a cross-departmental credit information-sharing mechanism. This mechanism includes sharing taxpayers’ basic information, transaction information, property retention and transfer information, and tax records. Tax authorities are also promoting cooperation with financial institutions such as banks. Companies can exchange “tax credits” for “bank credits” to facilitate standard debt contracts. In summary, with the establishment of CTAIS-3, firms’ business operations and transactions face heightened attention and oversight from tax authorities and society, effectively narrowing the information gap between companies and their stakeholders.

The implementation of CTAIS-3 commenced in 2013, starting from Chongqing, Shanxi, and Shandong. In 2014, CTAIS-3 extended to Guangdong, Henan, and Inner Mongolia. In 2015, the pilot program was expanded to include 14 additional provinces and regions, such as Jilin and Hainan. The project had expanded to achieve countrywide coverage by the close of 2016. The staged rollout of CTAIS-3 creates a quasi-natural experimental setting, allowing us to investigate the causal link between tax information accessibility and corporate financial restatements.

Hypothesis development

The previous literature suggests managers may engage in opportunistic financial restatements for corporate interests (e.g., Gao and Zhang, 2019; Pyzoha, 2015; Desai et al., 2006). Such opportunistic behaviors can trigger a range of detrimental economic outcomes, including adverse market reactions (Bauer et al., 2021; Kim et al., 2016), higher financing costs (Kravet and Shevlin, 2010; Hribar and Jenkins, 2004), and diminished investor trust (Elliott et al., 2018; Garrett et al., 2014). Therefore, the literature extensively explores how to curb opportunistic financial restatements (Cahan et al., 2024; Samuels et al., 2021; Cao and Pham, 2021). For example, previous research implies that external investors can exercise oversight and control over corporate managers, thereby reducing misreporting by companies (Elliott et al., 2020; Lin et al., 2018; Bird and Karolyi, 2016). Nonetheless, this perspective remains controversial. Specifically, other research suggests that the existence of external investors is linked to a decline in financial disclosure quality (Arif and De George, 2020; Bushee et al., 2019; Miller, 2002). This phenomenon is partially attributed to external investors’ challenges in obtaining comprehensive and timely company information. Companies may exploit asymmetric information to deliberately mislead investors, diminishing the quality of investor’s assessments regarding their conditions.

The issue of external investor monitoring’s ineffectiveness due to information asymmetry sparks researchers’ interest in how the information environment affects financial information quality. Chen (2016) finds that banks’ exceptional information acquisition and processing capabilities enable them to react to financial misreporting during restatement periods. Ashraf et al. (2020) find that audit committees’ information technology skills significantly improve corporate accounting quality, and Xiong et al. (2021) uncover that high-speed railway coverage significantly reduces corporate financial fraud by increasing the accessibility of corporate information. Overall, existing literature indicates that the external information environment is pivotal to improving corporate financial reporting quality.

As a crucial component of the information environment, tax information can facilitate investors to obtain higher-quality corporate information for external supervision, prompting companies to enhance financial disclosure quality. Prior literature shows that enhancing the accessibility of tax information can reduce information acquisition costs, improve the credibility of obtained information, and consequently enhance external supervision efficiency regarding companies’ financial information disclosure in several ways (Chiu et al., 2013; Dyck et al., 2010). First, tax information enhances the transparency of corporate operational conditions. For example, Dhaliwal, Kaplan, Laux (2013) and Kumar and Visvanathan (2003) found that managerial decisions concerning the valuation allowance for deferred tax assets offer additional insights into the enduring nature of accounting losses. Second, tax information enhances investors’ discernment of corporate disclosure. Prior studies document that tax information is an essential indicator for investors in evaluating corporate stock prices (Bauckloh et al., 2021; Hoopes et al., 2018).

In summary, implementing CTAIS-3, characterized by providing incremental tax information, contributes to heightened efficiency in external supervision, which induces a more cautious approach from companies in disclosing financial information. Consequently, there is an enhancement in the overall quality of corporate financial disclosure, accompanied by a reduction in financial restatements. We put forward our first hypothesis as follows:

Hypothesis 1A: Financial restatement decreases after the enactment of CTAIS-3.

Tax information transparency may elevate the likelihood of financial restatements. Specifically, the heightened transparency associated with tax information significantly diminishes the potential for companies to engage in deceptive tax declarations to evade taxes, which increases the corporate tax burden to some extent. Research demonstrates that an increase in the tax burden may hurt the quality of financial information (Waseem, 2018; Romanov, 2006). Maydew (1997) states that an increase in the tax burden corresponds with earnings management behavior in corporate financial statements. Chen and Schoderbek (2000) suggest that changes in tax policies may lead companies to make “one-time adjustments” in their financial statements, which could reduce the consistency of financial disclosures. The complexity and variability of tax policies may increase companies’ accounting and tax-based proprietary costs, affecting the accuracy and reliability of financial information (Yost, 2023). CTAIS-3 enhances the tax supervision capabilities of the governments, which may increase the tax burden and financial restatements of firms. Accordingly, we put forth our second hypothesis, which states:

Hypothesis 1B: Financial restatement increases after the enactment of CTAIS-3.

Sample and descriptive statistics

Sample

Our data came from the CSMAR (China Stock Market and Accounting Research) database, comprising financial and stock market variables and corporate financial restatements. We collected tax information data based on the rollout of the “CTAIS-3” project in various provinces and cities in ChinaFootnote 3. Following Srinivasan et al. (2015), our outcome variable, financial restatements (Restate), is a binary measure. It equals 1 if a company conducts financial restatements in a given year and 0 otherwise. We started with annual observations for 29,319 companies listed on the Shenzhen and Shanghai stock exchanges between 2010 and 2019. Our sample period began in 2010 when the China Securities Regulatory Commission revised the “Regulations on Information Disclosure and Reporting for Companies Issuing Securities to the Public” in 2010. Besides, we limit our sample period to 2019, recognizing the dramatic impact the COVID-19 pandemic had on business functions beginning in 2020, which could affect our coefficients. We excluded 2886 firm-year data points with missing values for the control variables and 211 singleton observations, resulting in a final sample of 26,222 observations from 3546 distinct companies.

The company-level control variables used as factors in investment decisions also come from the CSMAR database. Specifically, following Chin and Chi (2009), Amel-Zadeh and Zhang (2015), and Jayaraman and Milbourn (2015), we include company size (Size), company performance (ROA), and financial health (Lev). In addition, we also include liquidity (CashHold) to control for cash constraints (Cheng and Farber, 2008) and Tobin’s Q (TobinQ) to account for investment opportunities (Durnev and Mangen, 2009; Bergstresser and Philippon, 2006) and the proportion of independent directors (Indep) to control for internal control mechanisms (Cheng et al., 2016; Srinivasan, 2005). We also collect data on three provincial-level variables from the “China Statistical Yearbook” to control for provincial characteristics, including per capita GDP (GDP), fiscal revenue (FiscInc), and average income at the province level (AvgSalary).

Table 1 reports descriptive statistics of the dependent, independent, and control variables. The statistical mean of the outcome variable “Restate” is 0.243, suggesting that approximately 24.3% of the observations involved financial report restatements, consistent with other research findings (Palmrose et al., 2004). The mean firm size in the sample is 22.20 (logarithm of total assets in RMB), the mean leverage ratio is 43.6%, the mean ratio of cash and cash equivalents to total assets is 16.7%, the mean ROA is 3.6%, the mean Tobin’s Q is 2.05, and the mean ratio of independent director is 29.2%, which is similar to the statistical data from other studies (Lennox et al., 2018).

Table 1 Descriptive statistics.
Full size table

Empirical results

Baseline regression

To examine the causal relationship between the launch of the CTAIS-3 system and corporate financial restatements, we adopt the staggered difference-in-differences approach to estimate the regression model specified below:

$${{Restate}}_{{it}}=\,{\alpha }_{i}+{a}_{t}+{\beta }_{1}{{CTAIS}}_{i,t}+{\gamma }^{{\prime} }{X}_{{it}}+{\tau }^{{\prime} }{X}_{{pt}}+{\varepsilon }_{i,t}$$
(1)

The indices i and t correspond to firms and years, respectively. \({CTAIS}\) is a binary variable that equals 1 if the firm is in a region (province or prefectural city) where the CTAIS-3 platform is launched, and 0 otherwise. \({X}_{{it}}\) is a vector of time-varying, firm-level control variables, and \({X}_{{pt}}\) is a vector of time-varying, province-level control variables. \({{Restate}}_{{it}}\) represents a measurement of corporate financial restatements, which equals 1 if the company i conducted financial restatements in year t. Otherwise, it equals 0. \({\varepsilon }_{i,t}\) is the error term. See Supplementary Table 1 for all variable definitions.

\(\,{\alpha }_{i}\) is the firm-fixed effect that accounts for time-invariant omitted firm characteristics and captures the treated \({{firm}}_{i}\) variable from the original DID model. \({a}_{t}\) is the year-fixed effect that controls for the impact of the specific year on the regression and captures the \({{Post}}_{t}\) variable from the original DID model. We cluster standard error at the firm level in all regressions.

The coefficient of focus is the one on CTAIS, as it measures the differential change in the financial restatements for treatment firms compared to control firms in the periods before versus after the launch of CTAIS-3. The DID technique helps eliminate biases introduced by persistent differences between the two groups or time-varying confounding factors (Imbens and Wooldridge, 2009). If the CTAIS-3 pilot significantly reduces financial restatements,\(\,{\beta }_{1}\) should be negative and statistically significant, and vice versa.

Table 2 displays the multivariate DID analysis results. The coefficient of \({CTAIS}\) is negative and statistically significant at the 1% level for Restate (−0.025, t = −2.51 and −0.031, t = −3.06 in Columns 1 and 2), suggesting that the exogenous increase in tax information results in a significant reduction in corporate financial restatements, which supports H1A. Specifically, the CTAIS-3 pilot results in a decrease of 3.1% in financial restatements, which is considerably and economically significant, considering the sample mean value of this ratio is only 24.3%.

Table 2 Effect of tax information on corporate financial restatements.
Full size table

Parallel assumption

We conjecture that the CTAIS-3 pilot represents a shock to firms’ information environment. A significant concern is that the pilot might be based on unobservable local economic and political characteristics. Therefore, to establish the credibility of the DID specification, we must fulfill the parallel trends assumption, which necessitates there being no statistically significant difference in corporate financial restatements between the treatment and control groups before the implementation of the CTAIS-3 system.

To tackle the above concerns, we employ the technique used by Bertrand and Mullainathan (2003) to analyze the time-varying patterns of corporate financial restatements surrounding the establishment of the CTAIS-3 system by estimating the following equation:

$${{Restate}}_{i,t}={\alpha }_{i}+{a}_{t}+\mathop{\sum }\limits_{k=-3}^{3}{\beta }_{k}d{[t+k]}_{t}+{\gamma }^{{\prime} }{X}_{{it}}+{\tau }^{{\prime} }{X}_{{pt}}+{\varepsilon }_{i,t}$$
(2)

Where d[t + k], –3 ≤ k ≤ 3, is a binary variable equals 1 for each of the three years preceding and three years following the launch of the CTAIS-3 system and zero otherwise. d[t + 3] represents the impact of CTAIS-3 system on and after t + 3. d[t − 4], omitted from the model to prevent collinearity, represents the impact of the CTAIS-3 system on and before t − 4. The remaining variables included in the analysis are consistent with those used in the baseline DID specification.

The key coefficients are \({\beta }_{-1}\), \({\beta }_{-2}\), and \({\beta }_{-3}.\) If there is any pre-existing trend in corporate financial restatements between the treated and control firms before the launch of CTAIS-3, we would see statistically significant estimates for \({\beta }_{-1}\), \({\beta }_{-2}\), and \({\beta }_{-3}\). We report the dynamic DID results in Table 2, Columns (3) and (4). The coefficients on \({\beta }_{-1}\), \({\beta }_{-2}\), and \({\beta }_{-3}\) are statistically insignificant in all regressions, implying that the parallel assumption underlying the DID approach is satisfied. The absence of any pre-trends helps alleviate concerns about reverse causality. Moreover, the coefficients on d[t], d[t + 2], and d[t + 3] are significantly positive, implying that the launch of the CTAIS-3 system can aggravate firms’ shift backward from restating their financial reports both in the short term and over the long term.

Underlying mechanisms

This section examines the mechanisms that drive the effect of a tax information system on corporate financial restatements.

The efficacy of CTAIS-3 in influencing corporate financial restatements may be contingent on the extent of information asymmetry between firms and stakeholders. Stock price synchronicity (Synchronic) is a standard measure to access information asymmetry as it reflects the degree to which firm-level information is impounded into share prices. A higher Synchronic suggests a less transparent information environment and higher levels of information asymmetry (Chan and Chan, 2014; Crawford et al., 2012; Piotroski and Roulstone, 2004). Therefore, we follow Piotroski and Roulstone (2004) and partition the sample based on stock price synchronicity (Synchronic). Panel A of Table 3, Columns 1 and 2 divide the sample by stock price synchronicity. The coefficient of CTAIS is insignificant in Column 2 but is −0.052 (t = −3.54) in Column 1, verifying information asymmetry as an essential mechanism.

Table 3 Underlying mechanism.
Full size table

Next, previous research suggests that higher analyst forecast error suggests less firm-specific information being shared in the capital markets (Jiang and Yuan, 2018; Lang et al., 2012; Flannery et al., 2004; Krishnaswami and Subramaniam, 1999). We then divide the sample based on analyst forecast error. Following Eames and Glover (2003), we calculate Analyst forecast error (Ferror) as the absolute difference between the means of the actual and predicted earnings, scaled by the equity market values. Columns 3 and 4 partition sample by analyst forecast error. The coefficient of CTAIS is insignificant in Column 4 but is −0.028 (t = −1.85) in Column 3.

Third, the link between tax information and corporate financial restatements may exhibit greater resilience in higher bid-ask spread. The reason is that when information asymmetry is high, informed investors can exploit their informational advantage to the disadvantage of less informed investors (Krinsky and Lee, 1996). In such scenarios, uninformed investors will reduce their bid prices and raise their ask prices to safeguard themselves against the anticipated losses from trading with better-informed counterparties, leading to a higher bid-ask spread (Amiram et al., 2016; Wittenberg-Moerman, 2008; Affleck‐Graves et al., 2002). We calculate the bid-ask spread (Spread) by Roll’s (1984) approach. Columns 5 and 6 partition the sample by bid-ask-spread. The coefficient of CTAIS is insignificant in Column 6 but is −0.044 (t = −2.20) in Column 5.

Fourth, existing literature suggests that the participation of institutional investors can help mitigate corporate misconduct by increasing firm transparency and information production (Wu et al., 2016; Cheng et al., 2010; Chung et al., 2002). We then segment the sample according to institutional ownership (Institution). Panel B Columns 1 and 2 split the sample by the percentage of shares institutional investors hold (Tong and Zhang, 2024). The coefficient of CTAIS is insignificant in Column 2 but is −0.034 (t = −2.22) in Column 1.

Fifth, Existing studies find that social media in China (e.g., online stock forums), which operate independently of government press control, can leverage the “wisdom of the crowd” to provide valuable information to society (Wong et al., 2023; Bartov et al., 2018; Blankespoor et al., 2014)—Panel B Columns 3 and 4 departure sample based on the number of online stock forums. The coefficient of CTAIS is insignificant in Column 4 but is −0.056 (t = −3.85) in Column 3.

Finally, recent literature suggests that information asymmetry is associated with geographic proximity. Nearby investments have an informational advantage because investors can monitor firms more efficiently due to reduced travel time (Baik et al., 2010; Ivković and Weisbenner, 2005; Coval and Moskowitz, 2001). We then separate the sample based on the minimum distance from the firm to a local bank. The results show that the coefficient of CTAIS is −0.025 (t = −1.72) when the firm is located near the local bank but is −0.035 (t = −2.50) when the firm is far away from the local bank.

The findings above indicate that the impact of a CTAIS-3 launch on firms’ financial restatements is significant only when firms are under information asymmetry. Therefore, the increased transparency between firms and their investors is one channel through which the launch of a tax information system affects firms’ financial restatements.

Additional analyses

State-owned Enterprises and Tax Avoidance

The strength of CTAIS-3’s information effect can depend on factors like government ownership and companies’ tax avoidance behavior. State-owned enterprises (SOEs) tend to be less susceptible to the influence of CTAIS-3 compared to non-state-owned enterprises (non-SOEs). SOEs have weaker financial motives than non-SOEs because they often prioritize social and political objectives over economic goals. Moreover, SOEs generally enjoy a more secure financial position due to government support (Liu et al., 2021; Du, Tang and Young, 2012; Cull and Xu, 2005). Additionally, SOEs are subject to more extensive oversight by the government, such as through periodic government audits and anti-corruption initiatives (Hou et al., 2022; Li et al., 2021b). Thus, the exogenous increase in tax information is more likely to influence Non-SOEs than SOEs.

In addition, the influence of introducing the CTAIS-3 might be more decisive in firms with a higher tax avoidance level. For example, firms that evade tax more can reduce tax evasion after the introduction of CTAIS-3 due to increased transparency in tax enforcement (Overesch and Wolff, 2021; Joshi et al., 2020; Kerr, 2019). Therefore, these companies may show a stronger correlation between the CTAIS-3 and financial information quality. In contrast, even if the introduction of the CTAIS-3 strengthens external supervision for companies with solid tax compliance, it is difficult to improve their tax compliance further.

We then compare the influence of CTAIS-3 on corporate financial restatements of SOEs to non-SOEs. Table 4 Panel A, Columns 1 and 2, show a sample breakdown dependent on whether the firm is an SOE. The coefficient of CTAIS is insignificant in Column 1 but is −0.033 (t = −2.57) in Column 2, confirming our expectation.

Table 4 Additional analyses.
Full size table

Columns 4 to 6 stratify the sample by the degree of tax avoidance. We calculate the tax avoidance measure by taking the five-year average of the gap between the statutory tax rate set by tax law and the actual tax rate paid by the firm (LRATE_diff) (Hanlon and Heitzman, 2010; Dyreng et al., 2008), as in Column 3 and 4, and book-tax-difference (Chen et al., 2022; Badertscher et al., 2019), as in Column 5 and 6. The coefficient of CTAIS is −0.036 (t = −2.77) in Column 3 and is −0.032 (t = −2.39) in Column 5, but it is statistically insignificant in Columns 4. The coefficient in Column 6 is −0.034 (t = −1.76). The findings suggest that the launch of CTAIS-3 had a more significant impact on financial restatements for non-SOEs and firms with higher tax avoidance levels.

High-technology, multinational, and large companies

Next, we investigated the variations in the effect of CTAIS-3 on financial reporting quality across different industries, company types, and firm sizes. Previous studies have shown that companies in the high-tech sector exhibit higher levels of information opacity (Mohd, 2005; Coff and Lee, 2003; Aboody and Lev, 2000). For example, Barron et al. (2002) reported that there is a lower level of consensus in analyst forecasts for high-tech firms due to the existence of a substantial amount of intangible assets. Palmon and Yezegel (2012) also contend that R&D investments create higher information asymmetry, potentially leading to lower informational content in the stock prices of high-tech companies. Therefore, the effects of CTAIS-3 may be more significant in the high-tech industry. Table 4 Panel B columns 1 and 2 display the heterogeneous influence of CTAIS-3 on restatements contingent on whether the company is in a high-tech industry. We obtained industry codes for the high-tech industry from the Chinese government’s websiteFootnote 4 and matched them with the company’s industry codes. We found that the coefficient of CTAIS is −0.050 in the first Column (t = −2.76) but is −0.023 (t = −1.85) in the second Column, confirming our expectations.

Besides, the impact of CTAIS-3 may be more significant in companies with overseas affiliates. Companies with overseas affiliates can exploit tax policy differences between countries or regions. They can minimize their tax liabilities to the greatest extent possible by shifting profits to tax havens (Overesch and Wolff, 2021; Bennedsen and Zeume, 2018; Dyreng et al., 2016). For example, Dyreng et al. (2019) found that companies establishing subsidiaries in tax havens have more significant tax uncertainty. The CTAIS-3 system establishes extensive cooperation and information-sharing relationships between the tax authorities and other departments such as customs and banks, thus reducing the possibility of companies manipulating profits to some extent. Panel B Columns 4 and 6 reflect the sample division based on whether the company has overseas affiliates. Based on data of the company’s overseas affiliates from the CSMAR database (including subsidiaries, joint ventures, and cooperative companies), we categorized companies with overseas affiliated companies as Multinational and those without overseas affiliated companies as Local, as shown in Columns 3 and 4. The coefficient of CTAIS is −0.033 (t = −2.07) in Column 3 but is statistically insignificant in Column 4.

Finally, CTAIS-3 may have a more significant impact on larger enterprises. Previous literature indicates that tax planning exhibits economies of scale (Rego, 2003). Large enterprises dedicate more resources toward tax evasion and are more adept at it (Brown and Drake, 2014; Dyreng et al., 2008; Hanlon et al., 2005). For example, Kotchen (2021) illustrates that the advantages provided by subsidies for oil, gas, and coal production are primarily accrued by a select group of large-scale enterprises. Dyreng, Hills, and Markle (2022) show that a small number of large enterprises drive most income-shifting activities in the United States. As a result, the impacts generated by the tax information system may be more pronounced in large enterprises. In Columns 5 and 6, we separate the sample by the median value of firm size. The coefficient of CTAIS is −0.055 in Column 5 (t = −3.88) but is statistically insignificant in Column 6.

These results show that CTAIS-3’s impact on corporate financial restatements is more significant for high-tech companies, companies with overseas affiliated companies, and larger companies.

Robustness tests

To strengthen the validity of our baseline findings, we undertook a diverse set of robustness checks to eliminate potential alternative explanations.

First, the introduction to CTAIS-3 reduces corporate financial restatements by mitigating information asymmetry, and it might generate the same results on other proxies of financial reporting quality. As such, we proceed to compute the subsequent variables of financial reporting quality, including the modified Jones model (EMJones) from Dechow et al. (1995) and the Dechow and Dichev (2002) model (EMDD) that represent earnings management, the auditor’s opinion type (Opinion) which equals 1 if the auditor of the firm provides a going-concern opinion and otherwise 0, C-score (Khan and Watts, 2009), and ACF score (Ball and Shivakumar, 2006) that represents the timeliness of the corporate response to bad news. Table 5 Panel A displays the results. We find robust and negative coefficients across all measurements of financial report quality, indicating that the exogenous increase of tax information indeed increases financial reporting quality.

Table 5 Robustness tests.
Full size table

Second, we test the robustness of our findings by employing different measures. We incorporate the natural logarithm of the number of financial report restatements conducted by a firm in a fiscal year (Restate_count) as the dependent variable. Table 5 Panel B Columns 1 and 2 show similar results (−0.016, t = −1.87 and −0.022, t = −2.51, respectively).

Third, to address the possibility that our findings may be influenced by the temporal trends in economic growth, urbanization, and natural and ecological environment that vary by province or industries, we follow Barrios et al. (2022) to control the province linear trend and industry linear trend in the primary regression and re-estimate Eq. (1) in Column 1 and 2 of Table 5 Panel C, and we add industry*year-fixed effect in Column 3. We continue to find negative and statistically significant coefficients of CTAIS (−0.030 with t = −2.92, −0.031 with t = −3.06, and −0.030 with t = −2.88), suggesting that the increased financial reporting quality is not due to the temporal trends in different province or industries, or the effect of the industry in combination with the specific year.

Finally, after accounting for firm-level variables, industry × year-fixed effect, and a series of factors, there may still be some unobservable factors, resulting in deviations in the estimated results. Following Ferrara et al. (2012), we conduct a placebo test that turns whether each enterprise is subject to the influence of CTAIS-3 into a completely random process, repeats the random process 5000 times, and estimates whether the coefficient of our interest follows the standard normal distribution. Figure 1 shows the distribution of 5,000 random experiments. The coefficient is close to the standard normal distribution with a mean of 0, proving that our study has controlled the influence of unobservable factors well and that the estimation result is robust.

Fig. 1: Distribution curve of coefficient α after random treatment.
figure 1

The figure illustrates the distribution of regression coefficients obtained by replicating the regression process 5000 times, where each firm impacted by CTAIS-3 is transformed into a random event. The regression coefficient of CTAIS-3 on corporate financial restatements is estimated through this randomization process. The x-axis represents the coefficient of the random event. The y-axis on the left represents p-values ranging from 0 to 1. The y-axis on the right represents the kernel density, corresponding to the kernel density plot. All regressions include firm-level and province-level controls as defined in Supplementary Table 1, with firm and year-fixed effects included in the analysis. Figure 1 shows that the coefficient distribution is close to the standard normal distribution with a mean of 0. Source: Calculated based on data from the CSMAR database.

Full size image

Conclusions

Using a panel of Chinese-listed firms from 2010 to 2019, our analysis reveals that firms reduced their financial restatement after the launch of the third praise of the China Taxation Administration Information System (CTAIS-3). Moreover, we observe a more pronounced negative association between CTAIS-3 and financial restatements when firms face higher levels of information asymmetry. Our results are robust to different measurements of financial reporting quality, varied measures, and alternative model specifications. Additional analyses suggest that the main results are significant in Non-State-Owned Enterprises and firms with higher levels of tax avoidance. Collectively, we prove that the exogenous increase in tax information leads to a lower incidence of corporate misreporting by providing a better information environment.

According to our study, the exogenous increase in tax information can serve to provide the public with additional information and mitigate the information asymmetry between companies and their investors, which contributes to studying the determinants of accounting information quality (Ahn et al., 2020; Gunn and Michas, 2018; Jayaraman and Milbourn, 2015), the growing literature examining the economic consequences of tax information (e.g., Amberger et al., 2021; Balakrishnan et al., 2019; Simone, Ege and Stomberg, 2015), the spillover effects of government behavior (Houston et al., 2019; De George, Li and Shivakumar, 2016; Chen et al., 2010) and provide insights which indicate that government behavior plays a vital role in improving corporate financial disclosure.

Our findings support that tax information is essential in reducing information asymmetry and shed light on the positive spillover effect of government activities in establishing a tax information system. We infer that policymakers and regulators can further promote the digitalization of government procedures, promoting a better information environment. Other stakeholders, such as investors, creditors, and suppliers, should incorporate the information effects of government actions into their assessment and decision-making process to better understand companies’ financial information.