Does the supply of tax information affect financial restatements? Evidence from the launch of Taxation Administration Information System III in China

The study investigates whether and how increased supply of tax information affects the incidence of corporate financial restatements. Financial restatements carry substantial market and trust costs, yet prior work provides mixed expectations about the role of tax information transparency. On one hand, greater tax information accessibility may deter misreporting and improve financial statement quality by increasing scrutiny and verification against tax data. On the other, enhanced tax transparency may raise firms’ tax exposure and pressure, potentially inducing more aggressive earnings management and more restatements. The authors address endogeneity by using the staggered rollout of China’s Taxation Administration Information System III (CTAIS-3) beginning in 2013 as an exogenous shock to tax information availability. They test competing hypotheses that restatements decrease (H1A) or increase (H1B) following CTAIS-3. The research is important for understanding determinants of financial reporting quality, the effects of tax information technology, and the spillover of government digitalization on corporate disclosure.

Prior research links taxes to corporate decision-making in entrepreneurship, innovation, investment, and risk-taking, but provides limited evidence on tax information’s effect on financial reporting quality and restatements. Financial restatements are associated with negative market reactions, higher capital costs, and reduced investor trust; determinants span audit expertise, management characteristics, institutional ownership, and government enforcement. Theoretical and empirical work suggests information technology and disclosure choices reflect cost-benefit tradeoffs; enhanced tax information can reduce tax evasion and improve reporting credibility, allowing external stakeholders to verify financial statements with tax data. Conversely, increased tax transparency may heighten tax scrutiny and burdens, potentially worsening firms’ conditions and prompting earnings management. Evidence shows some tax-related manipulation can escape detection, and certain transparency measures may not significantly alter behavior due to offsetting practices. Literature on information environments indicates that improved access and monitoring reduce misreporting; measures such as analyst informativeness, stock price synchronicity, bid-ask spreads, and investor attention proxy information asymmetry. The paper contributes by leveraging an exogenous, government-led rollout of CTAIS-3 to assess causal effects on restatements and by documenting mechanisms via information asymmetry.

Design: A staggered difference-in-differences (DID) framework exploits the provincial/city-level rollout of CTAIS-3 (2013 initial pilots; nationwide coverage by end-2016) as an exogenous increase in tax information availability. Treatment firms are those located in regions after CTAIS-3 launch; control firms are in regions yet to implement it. Data: Panel of Chinese A-share listed firms from 2010–2019 drawn from CSMAR (financials, stock variables, restatements). CTAIS-3 rollout timing by province/city is compiled from official sources (specific dates in Supplementary Table 2). Provincial controls come from China Statistical Yearbook. The final sample comprises 26,222 firm-year observations across 3,546 firms after excluding 2,886 firm-years with missing controls and 211 singleton observations. Variables: Outcome Restate equals 1 if a firm restates in a given year, 0 otherwise. Key regressor CTAIS equals 1 in firm-years after local CTAIS-3 launch. Firm-level controls: Size (log assets), ROA, leverage (Lev), cash holdings (CashHold), Tobin’s Q, independent director ratio (Indep). Province-level controls: per capita GDP (GDP), fiscal revenue (FiscInc), average salary (AvgSalary). Baseline model: Restate_it = α_i + α_t + β1*CTAIS_it + X_it + X_pt + ε_it, with firm and year fixed effects; standard errors clustered at firm level. Dynamic event-study specification includes indicators for k years before/after launch to test parallel trends. Mechanism and heterogeneity analyses: Partition samples by proxies for information asymmetry and monitoring: stock price synchronicity, analyst forecast error, bid-ask spread, institutional ownership, online stock forum activity, and geographic proximity to banks. Additional splits by ownership (SOE vs. non-SOE), tax avoidance (five-year average statutory-actual tax rate gap, LRATE_diff; and book-tax differences), industry (high-tech per government classification), multinational status (overseas affiliates vs local), and firm size (above/below median). Robustness: (i) Alternative financial reporting quality proxies: earnings management via Modified Jones (EMJones) and Dechow-Dichev (EMDD), going-concern audit opinion, accounting conservatism (C-score; ACF). (ii) Alternative dependent variable: log count of restatements (Restate_count). (iii) Adding province and industry linear trends and industry-by-year fixed effects. (iv) Placebo with 5,000 random assignments of treatment to test against unobservables; distribution centered at zero.

- Baseline DID: CTAIS is associated with a significant reduction in restatements. Coefficients: -0.025 (t=-2.51) and -0.031 (t=-3.06), implying about a 3.1 percentage-point decrease relative to a mean restatement rate of 24.3%. - Parallel trends: Pre-treatment coefficients (pre3, pre2, pre1) are insignificant; post-treatment indicators are negative and increasingly significant, supporting causal interpretation. - Mechanisms via information asymmetry (Table 3): - High stock price synchronicity subsample: CTAIS -0.052 (t=-3.54); low synchronicity: insignificant. - High analyst forecast error: -0.028 (t=-1.85); low error: insignificant. - High bid-ask spread: -0.044 (t=-2.20); low spread: insignificant. - Lower institutional holdings: -0.034 (t=-2.22); higher holdings: insignificant. - Fewer online forums (lower external retail attention): -0.056 (t=-3.85); higher forums: insignificant. - Greater distance to nearest bank: -0.035 (t=-2.50); nearer: -0.025 (t=-1.72). These patterns indicate effects concentrate where information asymmetry is higher and external monitoring is weaker. - Additional heterogeneity (Table 4): - Ownership: SOEs insignificant; non-SOEs -0.033 (t=-2.57). - Tax avoidance: High LRATE_diff -0.036 (t=-2.77) vs. low insignificant; High BTD -0.032 (t=-2.39); Low BTD -0.034 (t=-1.76). - Industry/type/size: High-tech -0.050 (t=-2.76) vs. low-tech -0.023 (t=-1.85); Multinationals -0.033 (t=-2.07) vs. locals insignificant; Large firms -0.055 (t=-3.88) vs. small insignificant. - Robustness (Table 5): Negative effects persist across alternative quality proxies: EMJones -0.005*; EMDD -0.009**; going-concern opinion -0.012***; conservatism C-score -0.053*; ACF -0.060*. Alternative dependent variable Restate_count: -0.016* and -0.022**. Results hold with province/industry trends and industry-year FE. Placebo tests show randomized treatment yields coefficients centered near zero, supporting robustness.

The findings support the hypothesis that increasing the supply of tax information via CTAIS-3 reduces financial restatements by improving the external information environment and mitigating information asymmetry between firms and stakeholders. Concentrated effects in settings with higher information asymmetry (e.g., high synchronicity, higher forecast errors, wider spreads, weaker outside monitoring) corroborate the information channel. The lack of significant effects for SOEs and stronger effects for non-SOEs, high-avoidance, high-tech, multinational, and large firms suggest that firms with greater incentives and capacity for opaque reporting respond more to enhanced tax information transparency. The results underscore governmental digital tax infrastructure as a policy lever that indirectly elevates financial reporting quality with positive spillovers to investors and markets.

Using a quasi-natural experiment based on the staggered rollout of China’s CTAIS-3 and a DID framework on 26,222 firm-year observations (2010–2019), the study shows that exogenous increases in tax information availability significantly reduce the incidence of financial restatements. Effects are stronger where information asymmetry is higher and among non-SOEs, tax-avoidant firms, high-tech, multinational, and larger firms. Robustness checks across alternative proxies and specifications, as well as placebo tests, support the validity of the results. The study contributes to understanding determinants of accounting information quality, the economic consequences of tax information technology, and the spillover effects of government digitalization. Policy implications include supporting digital tax enforcement mechanisms to improve firms’ information environments, with guidance for investors, creditors, and suppliers to incorporate tax transparency effects into assessments.