Corporate Fraud and Accounting Firm Involvement: Evidence from China

: In some cases, accounting ﬁrms and individual auditors will be punished by the China Securities Regulatory Commission (CSRC) for involvement in the violations of their client companies. Taking the enforcement actions against listed companies and accounting ﬁrms of the CSRC from 2006 to 2019 as a research sample, this paper manually sorted out the speciﬁc characteristics of corporate fraud and empirically examined the regulatory authorities’ supervision tendency to auditors. The results show that accounting ﬁrms are more likely to be involved when their client companies’ fraudulent practices affect ﬁnancial statements, occur during the IPO process, and continue for a longer period of time. Income statement manipulation and higher fraud amounts also increase the probability of accounting ﬁrms being sanctioned. Further analyses show that regulators’ supervision intensity is increasing over time, and they impose penalties on auditors based on the severity of corporate fraud; however, the intensity and differentiation of the sanctions are still insufﬁcient. This study expands relevant research on accounting ﬁrm sanctions and provides empirical evidence for further improvement of audit industry supervision in an emerging market.


Introduction
In recent years, a wave of corporate fraud in the worldwide securities market has occurred, for which the accounting firms often bear significant responsibilities. For example, in the beginning of the twenty-first century, Arthur Andersen as an auditor of Enron had a great responsibility for misleading numbers presented in Enron's financial reports, and Enron's collapse directly led to the demise of Arthur Andersen, which was one of the big five accounting firms (Krishnamurthy et al. 2006;Nelson et al. 2008;Markoff 2012). Auditors are responsible for judging accounting information and ensuring the truthfulness and reliability of accounting information. Failure to detect ir-regularities during the audit process not only affects public trust in auditors but also causes serious financial losses to investors and hinders the normal functioning of the capital market (Chaney and Philipich 2002;Giannetti and Wang 2016;He et al. 2016).
According to current auditing standards, auditors are required to provide reasonable assurance as to whether the financial statements as a whole do not contain material misstatements. However, due to the inherent limitation of auditing, auditors are inevitably exposed to the risk that some certain material mis-statements are undetected. In other words, the fulfillment of audit responsibilities requires strong oversight by external enforcement agencies which will deter auditors' misconduct and protect investors' interests (Shleifer and Wolfenzon 2002;Defond and Hung 2004;Hilary and Lennox 2005).
However, we noticed that only 21% of the audit failure cases in China's securities market from 2006 to 2019 were punished by the CSRC. As a result, this paper aims to investigate an interesting and important question, which is whether the regulatory authorities' supervision tendency to auditors and the intensity of sanctions are affected by the characteristics of corporate fraud. In other words, what factors make auditors more or less likely to be involved in corporate fraud cases and to be subject to regulatory penalties. This question deserves great attention from not only regulators and accounting firms, but also investors in the capital market.
Taking the enforcement actions against listed companies and accounting firms of the CSRC from 2006 to 2019 as a research sample, this paper manually sorted out the specific characteristics of corporate fraud and empirically examined the regulatory authorities' supervision tendency to auditors. The results show that accounting firms are more likely to be involved when their client companies' fraudulent practices affect financial statements, occur during the IPO process, and continue for a longer period of time. Income statement manipulation and higher fraud amounts also increase the probability of accounting firms being sanctioned. Further analyses show that regulators' supervision intensity is increasing over time, and they impose penalties on auditors based on the severity of corporate fraud; however, the intensity and differentiation of the sanctions are still insufficient.
This study contributes to the literature in several ways. First, prior research on the administrative supervision on accounting firms generally has an early observation year and a limited sample period Wu 2007;Chen et al. 2011). In recent years, there have been significant changes in the structure and practice risk of the CPA profession, especially in an emerging market such as China's (Simunic and Wu 2009;Lisic et al. 2015;Wang et al. 2020). Therefore, our study uses a more adequate and recent sample to examine the audit responsibility determination and supervision tendency to auditors by the regulatory authorities in the case of corporate fraud, which has strong application and practical significance.
Second, most prior studies only analyzed the tendency of accounting firms to be punished from the perspective of the major violation types of listed companies and do not delve into other specific violation characteristics Wu 2007;Chen et al. 2011). Our study further details the specific characteristics of corporate fraud by reading the enforcement action announcements one by one. We add some quantitative variables, such as the time length and the amount of financial statements affected by fraudulent practices, to a previous model. Therefore, this paper provides further empirical evidence for the regulators' audit responsibility determination and supervision tendency to auditors.
Third, according to the New Securities Law of China promulgated in 2020, a filing management system (replacing the previous administrative licensing system) is implemented for accounting firms that want to engage in securities audit services, which means that many small accounting firms without a previous securities qualification are eligible to provide audit services for listed companies (Wang et al. 2020). Our results have strong reference values for the securities auditing market supervision under the new system. For example, regulators should impose differentiated penalties for audit failures of different severity and increase penalties for auditors who seriously lack audit independence and do not abide by professional ethics.
The remainder of the paper is organized as follows. Section 2 introduces the institutional background in China and related literature review. Section 3 states theoretical analysis and develops hypotheses. Section 4 introduces study design. Section 5 addresses empirical tests and performs several additional analyses and robustness tests. Section 6 concludes the paper.

Institutional Background and Literature Review
China's legal environment has been changing greatly as the country's transition from a planned economy to a market-oriented one (Jiang and Kim 2015). In 1990 and 1991, the Shanghai and Shenzhen Stock Exchanges were established, respectively. Then, based on the need for an effective regulatory authority to manage capital market operations and protect investor interests, the CSRC was launched in 1992. The Securities Law of China promulgated in 1998 stipulates that the CSRC is the main supervision organization of Chinese securities market. Both the Shanghai and Shenzhen Stock Exchanges are governed by the ultimate authority of the CSRC and are empowered to regulate companies within their jurisdiction (Tondkar et al. 2003;Chen et al. 2005). In addition, the CSRC also has the right to supervise accounting firms and CPAs with a securities qualification.
Since the implementation of the Law of China on Certified Public Accountants and the Chinese Independent Audit Standards at the end of the twentieth century, relevant laws and regulations have been revised many times, which has significantly improved auditor independence and audit quality (Xiao et al. 2000;Simunic and Wu 2009). After the demise of Arthur Andersen, regulators paid more attention to the identification of auditors' responsibility in listed companies' violations. From 2007, China began to apply the New Accounting Standards, keeping pace with international and domestic economic development and enriching the system of enterprise accounting (Wang et al. 2012). With the formal implementation of the New Securities Law of China in 2020, the access system of accounting firms to engage in securities audit services has changed from the administrative licensing system to the filing management system, which means that a greater number of small accounting firms that did not have the securities qualification can obtain an opportunity to provide audit services for listed companies and will greatly change the audit market structure (Wang et al. 2020).
In summary, after three decades of development, China's securities market has gradually formed a sound supervision and self-discipline management system and continued to carry out institutional innovation. The constantly improving institutional system of accounting information disclosure of listed companies has played a key role in the development of China's securities market (Wang et al. 2008;Firth et al. 2009;Sun et al. 2016).
However, it is undeniable that in recent years there has still been a number of severe financial-fraud and audit-failure cases occurred in Chinese-listed companies, which has caused extremely heavy losses to the public. The reasons for audit failures have been discussed in the literature from two aspects of listed companies and accounting firms. Some characteristics of listed companies such as operating conditions, corporate governance mechanisms, and fraudulent financial reports are directly related to audit failures (Palmrose 1987;Bonner et al. 1998;Firth et al. 2005). Recent studies have also shown that there is a relationship between corporate social responsibility (CSR) and corporate fraud. Better corporate governance could improve CSR disclosure practices, which further alleviate the information asymmetry problem and reduce the probability of committing fraud in the future (Miras-Rodríguez et al. 2018;Hu et al. 2019). For accounting firms, imperfect internal quality control system, poor organizational structure, and intense external competition would reduce audit quality (Francis and Michas 2013;Gerakos and Syverson 2015;Moser 2021). In addition, CPA individuals also have a certain relationship with audit failures. For example, the CPA's failure to fully comply with professional ethics, lack of independence, and ineffective implementation of audit procedures could lead to audit failures (Dye 1993;Geiger and Raghunandan 2002;Blay 2005;Lennox and Li 2020). Audit failures are also related to CPA's educational level and prior audit experience (Ye et al. 2014).
The impact of audit failures on auditors includes administrative penalty, market debarment, revocation of professional qualification, and imprisonment. According to the previous studies on the consequences of audit failures, the main focus is on the impact of administrative penalty on auditors, as it is more common and can seriously affect auditors' reputations. An earlier study of accounting firm sanctions was conducted by Palmrose (1987) who used a sample of more than four hundred cases involving fifteen larger accounting firms in the US from 1960 to 1985 and found that nearly half of the audit lawsuits are related to corporate fraud. Wilson and Grimlund (1990) found that accounting firms subject to administrative penalties have a significant reduction in the number of audit clients. Administrative penalties also influence audit fees, and the market will also doubt their practical competence and professional ability, and then produce a negative market reaction to their clients (Chaney and Philipich 2002;Krishnamurthy et al. 2006). There are some controversial results regarding the relationship between administrative penalties and audit quality. Some researchers suggested that audit quality significantly improves after accounting firms are regulated (Lisic et al. 2015;Aobdia and Shroff 2017). However, Wang et al. (2011) found that audit quality of accounting firms or CPAs does not significantly improve after being disciplined by the CSRC. These results indicate that regulators are very concerned about corporate fraud, and listed companies and accounting firms that commit such violations are more likely to be regulated.
With the rapid growth of China's economy and the continuous improvement of relevant laws and regulations, China's securities market provides us a unique research context for studying the regulation of audit failures. Firth et al. (2005) analyzed the audit failures in China and found that compared with audit failures caused by other reasons, audit failures with material mis-statements in the company's financial statements are more likely to cause the CSRC to take enforcement actions against the accounting firms. Wu (2007) documented a significantly relieving trend of regulatory propensity toward auditor liability from 1999 to 2006. Chen et al. (2011) analyzed the CSRC's enforcement actions against accounting firms from 1993 to 2009 and found that the lower the profitability of audit clients and the higher the level of accounts receivable, the greater the probability of an accounting firm being penalized by the CSRC.
In summary, prior literature on the administrative supervision on accounting firms generally has an early observation year and a limited sample period Wu 2007;Chen et al. 2011). With the ever-changing economic environment and increasingly stringent audit regulations in China in recent years, the audit responsibility determination and supervision tendency to auditors by the regulatory authorities might have changed greatly (Simunic and Wu 2009;Lisic et al. 2015;Wang et al. 2020). Moreover, most studies only analyzed the tendency of accounting firms to be sanctioned from the perspective of the major violation types of listed companies (such as whether it is a financial statement fraud) and do not delve into other specific violation characteristics (such as the time length and the amount of financial statements affected by fraudulent practices). This paper will focus on expanding the research sample period and the characteristics of listed companies' violations in order to explore recent audit responsibility identification and supervision tendency to auditors in the case of corporate fraud.

Financial Statement Fraud vs. Information Disclosure Breach
According to the classification in the Securities Law of China, the violations of listed companies can be divided into financial statement fraud and information disclosure breach. Financial statement fraud affects the amount of financial statement items such as revenue and profit, while information disclosure breach refers to the existence of false content in the notes of the financial statements (such as material omission or inaccuracy in disclosure) but does not directly affect the amount of financial statement items. Because investors mainly use financial information to evaluate and judge a company's operating results and future performance, the mis-statement of financial statement information is an important risk for investors (Eilifsen and Messier 2015). If a company's financial information is misreported, either inadvertently or intentionally, it can mislead investors into making wrong decisions. Therefore, as the gatekeeper of the capital market, regulators have a responsibility to avoid incidents of financial statement fraud as far as possible. Bonner et al. (1998) found that auditors are more likely to be sued by companies that recognize revenue early, overestimate assets, and underestimate expenses or liabilities. The main reason is that relative to other information disclosure breaches, financial statement fraud involving mis-statements of assets and income can lead to more serious misunderstandings by investors about a company's true value and profitability, which may result in greater losses.
In addition, regulatory costs will also affect the enforcement of administrative supervision by the CSRC. With limited costs, they tend to choose to take enforcement actions for more serious violations, thereby increasing the probability of supervision on more harmful cases such as financial statement fraud (Jiang et al. 2009).
Taken together, we argue that regulators will impose stronger supervision for financial statement fraud (rather than information disclosure breach) that is of greater concern to investors and may have more serious consequences, so as to alert other auditors to raise their awareness of professional and legal responsibilities. This leads to the following hypothesis: Hypothesis 1. Regarding information disclosure breach, accounting firms are more likely to be involved when their client companies commit financial statement fraud.

Income Statement Fraud vs. Other Statement Fraud
For financial statement fraud, the mis-statements in different statements may have diverse effects on investors' decisions and regulators' supervision tendency to auditors. Users of the financial statements of listed companies often pay more attention to operating results, particularly income (Eilifsen and Messier 2015). Previous studies also found that income-related mis-statements or ir-regularities are most common among financial statement fraud (Loebbecke et al. 1989;Dechow et al. 1996;Rezaee 2004).
Moreover, Chinese listed companies usually have a strong incentive for earnings management given a number of earnings-based regulations (Chen et al. 2001;Chen and Yuan 2004;Jing et al. 2021). Aharony et al. (2000) stated that revenue-related fraud is more consequential than asset-related fraud as income performance is an important criterion for IPO, rights offerings, and maintaining exchange-trading status. For example, China has regulations on the continuous profitability of listed companies. If a company operates at a loss for two consecutive years, it will be issued a delisting risk warning, also called a special treatment (ST). As a result, companies in financial distress have a strong incentive to whitewash their accounting earnings to prevent them from being "capped" by the regulators or to try to remove their "caps" (Yang et al. 2012).
The above studies show that compared with the balance sheet and cash flow statement, listed companies have a stronger incentive to manipulate their income statement. Therefore, we argue that regulators will impose stronger supervision for income statement fraud (rather than other statement fraud) that are the key audit areas in the entire audit process. This leads to the following hypothesis: Hypothesis 2. Regarding other financial statements, accounting firms are more likely to be involved when their client companies manipulate the income statement.

IPO Process Fraud vs. Post-IPO Fraud
Compared with the audit of listed companies' annual reports (post-IPO audits), the IPO audits are more risky, time-consuming, complicated, and require the participation and cooperation of various parties such as underwriters, law firms, and asset valuation agencies (Venkataraman et al. 2008). Once a company meets the listing qualifications, it can obtain a large amount of funds needed for the company's development. For accounting firms, IPO audits can receive higher fees and larger profit margins. If the company is successfully listed, the accounting firm's reputation can also be increased, thereby obtaining more business opportunities in the audit market. For this reason, auditors tend to commit irregularities to accommodate IPO companies to meet listing conditions (Aharony et al. 2000;Chan et al. 2021).
In recent years, the CSRC has paid more attention to the listing supervision of IPO companies and the determination of the responsibilities of IPO intermediaries, and also raised the threshold for listed companies to enter the securities market (Yang 2013). As the economic police of the securities market, accounting firms should ensure the accounting information quality of IPO companies.
Taken together, we argue that regulators will impose stronger supervision on accounting firms that fail to practice prudent audits in the IPO audit process than post-IPO audit process. This leads to the following hypothesis: Hypothesis 3. Regarding post-IPO fraud, accounting firms are more likely to be involved when their client companies commit fraud during the IPO process.

The Time Length of Fraud
Corporate fraud often exhibits a certain degree of inertia (Wu and Wang 2018). An analysis of fraudulent financial reporting by US public companies from 1998 to 2007 published by the Committee of Sponsoring Organizations (COSO) showed that the average time length of listed companies' violations is 31.4 months. Since the supervision is not completely effective and often has a certain lag, listed companies will continue to whitewash their financial statements in subsequent years if their fraudulent behavior is not detected quickly (Committee of Sponsoring Organizations COSO 2010). In our research sample, we also find that more than 40% of the observations have fraudulent periods longer than five years.
Previous studies showed that listed companies with long-term violations have stronger incentives to engage in opinion shopping from accounting firms in order to cover up unsatisfactory operating performance or achieve new financing needs. Intense industry competition and inefficient industry supervision also make accounting firms receptive to those requests from client companies (Schwartz and Menon 1985).
Long-term violations of listed companies not only indicate that accounting information was seriously distorted, but also mean that auditors failed to effectively identify misstatements for many consecutive periods, and did not perform audit responsibilities in accordance with relevant laws and regulations. Taken together, we argue that regulators will impose stronger supervision on accounting firms whose client companies have a longer period of fraud. This leads to the following hypothesis: Hypothesis 4. Accounting firms are more likely to be involved when their client companies commit fraud for a longer period of time.

The Amount of Fraud
As the most direct evidence of the seriousness of fraud, the high amount of fraud involved in financial statements is often the focus of attention from regulators and the public. In our research sample, the average amount of fraud is RMB 39.8 million; the maximum amount of fraud is RMB 1318.8 million; more than 50% of the observations with the amount of fraud over RMB 50 million. These figures show that the corporate fraud of Chinese listed companies has a great impact on the amount of financial statement items.
The financial statement fraud involving a large amount of money is not only attributed to the low moral level of the executives of listed companies, but also reflects the poor professional ability and ethics of auditors, which weakens the value of audit reports. If the CSRC does not sanction related accounting firms and CPAs when investigating fraud cases involving larger amounts of money, it will be detrimental to the improvement of the audit quality in the audit industry. This leads to the following hypothesis: Hypothesis 5. Accounting firms are more likely to be involved when their client companies commit fraud with a larger amount of money.

Data Source and Sample Composition
The main data used in this paper were hand-collected from enforcement action announcements against Chinese listed companies and accounting firms of the CSRC from 2006 to 2019. Other data were obtained from the China Stock Market Accounting Research (CSMAR) database. Table 1 shows the enforcement action announcements' collecting process. During our sample period, there are a total of 1133 enforcement action announcements published on the CSRC website. Since the main focus of this paper is to explore the relationship between the administrative supervision toward fraudulent behavior of listed companies and accounting firms' involvement, we first excluded 187 enforcement action announcements whose targets are not listed companies or accounting firms (e.g., underwriters, financial institutions, law firms, etc.). Then, we excluded 653 enforcement action announcements that are imposed for securities market violations (e.g., insider trading, stock price manipulation, excessive reduction of holdings, etc.), which are not related to financial statement audit. Finally, we obtained 293 effective enforcement action announcements, of which 242 announcements are imposed on listed companies and 51 announcements are imposed on accounting firms, with 21% of accounting firms being implicated.  Table 2 shows the sample composition of this paper. As the enforcement action announcements against listed companies usually involve multiple kinds of violations, the data at the announcement level cannot meet the needs of analysis of the characteristics of corporate fraud, so we transformed the data structure to the "company-year-violation type" level and gained 953 observations. After excluding 33 observations with missing data, the final sample is 920. According to the classification in the Securities Law of China, we classified violations of listed companies into financial statement fraud and information disclosure breach.
Referring to Zhu and Gao (2011), financial statement fraud is further classified as (1) Income statement fraud: a fraud that directly leads to a material mis-statement of the income statement.
(2) Balance Sheet Fraud: a fraud that directly leads to a material mis-statement of the balance sheet, but does not affect the income statement. (3) Cash flow statement fraud: a fraud that directly leads to a material mis-statement of the cash flow statement, but does not affect other statements. Information disclosure breach is further classified as: (1) False disclosure: a breach that fails to specify all or part of the matters. (2) Untimely disclosure: a breach that does not publicly disclose information within the appropriate period. (3) False statement: a breach that records facts that do not exist. (4) Misleading statement: a breach that misrepresents information that causes investors to make a misjudgment in their investment decisions. Table 2 shows that the proportion of financial statement fraud and information disclosure breach is 39.89% and 60.11%, respectively. Among financial statement fraud, income statement fraud accounted for the highest proportion (22.61%), followed by balance sheet fraud (16.74%) and cash flow statement fraud (0.54%). Among information disclosure breach, false disclosure accounted for the highest proportion (30.11%), followed by untimely disclosure (21.30%), false statement (7.50%), and misleading statement (1.20%).
In addition, Table 2 shows that 5.76% of the observations had committed fraud during the IPO process; 7.83% of the observations with a fraudulent period of one year; 51.95% with a fraudulent period of 2-5 years; 28.48% with a fraudulent period of 6-10 years; 11.74% with a fraudulent period of more than 10 years. This reflects that the accounting information quality of Chinese listed companies still has a lot of room for improvement.

Model Setting and Variables Definition
The following logistic models were conducted to test our research hypotheses: AUDSAC = α 0 + α 1 MISFRA + α 2 IPO + α 3 VIOYEAR + α 4 BIG10 + α 5 MAO + α 6 CHANGE + α 7 SIZE + α 8 ROA + α 9 LEV + α 10 SOE + Year FE + Industry FE + ε (1) AUDSAC = β 0 + β 1 INCFRA + β 2 AMOUNT + β 3 IPO + β 4 VIOYEAR + β 5 BIG10 + β 6 MAO + β 7 CHANGE + β 8 SIZE + β 9 ROA + β 10 LEV + β 11 SOE + Year FE + Industry FE + ε (2) In the above models, the dependent variable AUDSAC indicates whether an accounting firm is involved in corporate fraud, which equals one if the company is subject to administrative supervision and its accounting firm is also sanctioned by the CSRC, and zero otherwise. The experimental variable MISFRA denotes the violation type of listed companies, which equals one if the company commits financial statement fraud, and equals zero if the company commits information disclosure breach; INCFRA denotes the type of financial statement fraud, which equals one if the company manipulates income statement, and equals zero if the company only manipulates balance sheet or cash flow statement; IPO equals one if the company commits violations during the IPO process, and zero otherwise; VIOYEAR proxies for the time length of fraud, which equals the natural logarithm of the consecutive years of corporate fraud; AMOUNT proxies for the amount of financial statement fraud, which equals the natural logarithm of the mis-statement amount of net profit or net asset. 1 We include several control variables in the above models, including BIG10 (indicator variable equal to one if the company is audited by Big 10 accounting firms, and zero otherwise; Big 10 accounting firms refer to international Big 4 accounting firms and domestic top six accounting firms ranked by auditing revenue in a given year); MAO (indicator variable equal to one if the accounting firm issues a modified audit opinion, and zero otherwise); CHANGE (indicator variable equal to one if the company changes its accounting firm, and zero otherwise); SIZE (=natural logarithm of total assets); ROA (=ratio of the net income to total assets); LEV (=ratio of the total liabilities to total assets). We also add SOE (indicator variable equal to one if the company's controlling shareholder is the government or a state-owned enterprise, and zero otherwise) to control for the political connection of a listed company which might influence the regulatory authorities' supervision tendency to the company as well as its auditors. In addition, we include year fixed effect (Year FE) to control for time-varying factors and industry fixed effect (Industry FE) to control for heterogeneity across industries. Table 3 shows the descriptive statistics for the variables. To eliminate the influence of outliers, 1% winsorizing was performed for all continuous variables. The results show that accounting firms in 30.9% of the observations are involved in the violations of their client companies. The proportion of financial statement fraud is 56.8%, among which the proportion of income statement fraud is 80.1%, indicating that income-related misstatements or ir-regularities are most common among financial statement fraud. In addition, 5.8% of the observations committed violations during the IPO process; the mean of fraud periods is 4.29 years (with the mean of natural logarithm value of 1.458); the mean of fraud amounts is RMB 39.8 million (with the mean of natural logarithm value of 17.500). The mean test and Wilcoxon ranksum test show that for the involved accounting firms, their client companies' fraudulent practices are more likely to affect financial statements, occur during the IPO process, continue for a longer period of time, and impact a larger amount of money. Moreover, the involved accounting firms are less likely to issue modified audit opinions. Table 4 presents the regression results of regulatory authorities' supervision tendency to auditors based on different characteristics of corporate fraud. In Column (1), the coefficient of MISFRA is significantly positive, indicating that accounting firms are more likely to be involved when their client companies commit financial statement fraud rather than information disclosure breach. In Column (2), the coefficient of INCFRA is significantly positive, indicating that accounting firms are more likely to be involved when their client companies manipulate the income statement rather than other financial statements.   (1) and (2), the coefficients of IPO are significantly positive, indicating that accounting firms are more likely to be involved when their client companies commit fraud during the IPO process rather than in the post-IPO period; the coefficients of VIOYEAR are significantly positive, indicating that the longer the time length of fraud of listed companies, the more likely the accounting firms are involved; the coefficient of AMOUNT is significantly positive, indicating that there is a significant impact of corporate fraud amounts on accounting firms' involvement.

Regression Analysis
The above results support all hypotheses of this paper, proving that the regulators' administrative supervision inclination toward auditors varies depending on the characteristics of listed companies' violations, and they pay attention to not only the major violation types, but also the affected statements and other quantitative indicators such as the time length and the amount of corporate fraud when determining the responsibility of audit failures.
We also find a significantly negative association between SOE and AUDSAC, indicating that the political connection of a listed company might influence the regulatory authorities' supervision tendency to its auditors, and the auditors of state-owned listed companies are less likely to be involved when their clients commit corporate fraud. 2

Additional Analyses
As we mentioned above, in recent years, regulators have introduced a number of laws and regulations to strengthen auditor's accountability. To examine whether the intensity of administrative supervision strengthens over time, we control for the time variable T and its interactions with MISFRA, IPO, and VIOYEAR. 3 T equals one for the first year of the sample period (2006), equals two for the second year of the sample period (2007), and so on. Table 5 shows the time trend regression results of supervision intensity. In Column (1)-(3), the coefficients of the interaction terms between T and specific characteristics of corporate fraud are all significantly positive, indicating that the intensity of administrative supervision on accounting firms is increasing over time in our sample period. When an accounting firm is subject to administrative supervision, it will face a fine both at the level of accounting firms and CPAs. We manually collected penalty information in the enforcement action announcements in an attempt to further examine the differential impacts of specific characteristics of corporate fraud on the intensity of sanctions imposed on accounting firms and CPAs.
We replaced the dependent variable in Model (1) with FINE_AUDFIRM (=natural logarithm of the fines imposed on the accounting firms) and FINE_AVGCPA (=natural logarithm of the average fines imposed on the CPAs). Due to data availability, the sample size is limited to 284 observations. The unreported descriptive statistics show that the mean of fines imposed on accounting firms is RMB 0.43 million and the maximum value is RMB 9.69 million; the mean of fines imposed on CPAs is RMB 0.06 million and the maximum value is RMB 0.14 million. Overall, the fines imposed on both accounting firms and CPAs are relatively lower (compared to their business income) and the variation is not significant. Table 6 presents the regression results of the factors influencing the intensity of sanction. In Column (1) and (2), the coefficients of MISFRA are significantly positive for both dependent variables, indicating that both accounting firms and CPAs are subject to significantly higher fines when financial statement fraud occurs in listed companies. IPO is significantly positively associated with FINE_AVGCPA, indicating that for corporate fraud during the IPO process, the CSRC imposes greater fines mainly on the CPAs who are directly responsible for the prospectus' reliability, while no higher fines are imposed on the accounting firms. The regression results of other control variables show that the CSRC's sanction intensity for both accounting firms and CPAs is significantly weaker when the auditors have higher audit quality as well as perhaps stronger political ties (BIG10) and issue a modified audit opinion (MAO).

Robustness Tests
To overcome a potential endogeneity problem in this study, we check the robustness of our findings with the propensity score matching (PSM) method to match listed companies that committed financial statement fraud with those that did not commit financial statement fraud. Specifically, we use SIZE and ROA as the matching variables by the nearest neighbor 1:2 method. For Model (1), the sample of the treatment group is 491 and the matching group is 1964. For Model (2), the sample of the treatment group is 277, and the matching group is 679. Table 7 shows that both the coefficients of MISFRA and INCFRA are significantly positive after using the PSM method, demonstrating that our conclusions are robust after controlling the sample selection bias.

Conclusions
Using the enforcement actions against listed companies and accounting firms of the CSRC from 2006 to 2019 as a research sample, this paper manually sorted out the specific characteristics of corporate fraud and empirically examined the regulatory authorities' supervision tendency to auditors. The results show that accounting firms are more likely to be involved when their client companies commit financial statement fraud (rather than information disclosure breach), income statement fraud (rather than other statements fraud), fraud during the IPO process, for a longer period of time, and with a larger amount of money. Additional analyses show that regulators' supervision intensity is increasing over time, and they impose penalties on auditors based on the severity of their client companies' violation practices.
The findings of this paper have certain practice significance for both regulators and auditors. From the perspective of regulators, the foremost issue is to increase the penalties for audit failures. On the one hand, more effective investigations should be conducted into violations of listed companies and auditors who were not acting with due diligence in the process of auditing. On the other hand, differentiated penalties should be imposed for audit failures of different degrees. For auditors who seriously lack audit independence and do not abide by professional ethics, regulators should increase fines, implement market prohibition, and even impose criminal penalties.
From the perspective of auditors, the quality control of accounting firms needs to be further strengthened. From the CSRC's enforcement action announcements, it can be found that most accounting firms have deficiencies in standard setting, personnel management, and information processing. As a result, auditors need to effectively fulfill their audit responsibilities, gradually improve their professional capabilities, and raise their awareness of audit independence.
Although a more adequate and recent sample is used to examine the relationship between corporate fraud and accounting firm involvement in this paper, our study is still limited due to fewer enforcement action announcements of accounting firms. Future research could explore whether there are signs before accounting firms being sanctioned and whether this kind of differentiated regulation is effective.

Conflicts of Interest:
The authors declare no conflict of interest.

1
We constructed two logistic models because the sample of Model (2) was limited in the observations that commit financial statement fraud (MISFRA = 1), and only for these observations we could calculate the amount of financial statements affected by fraudulent practices (AMOUNT).