According to the CSRC Accounting  No. 67 “Question and Answer of the Information Disclosure of Companies that Offer Securities to the Public No. 6―Payment of Accounting Firms and Their Disclosures”, China’s listed companies have been regulated to disclose their payments to public accountants in their annual reports since 2001. On April 5, 2010, the Ministry of Finance, the Audit Office, the China Securities Regulatory Commission, the China Banking Regulatory Commission and the China Insurance Regulatory Commission jointly issued the “Guidelines for Auditing the Internal Control of Enterprises”, requiring companies listed both domestically and overseas to start implementing internal control audit in 2011, while A-share main board listing companies in 2012. At the same time, in the “Contents and Formats for Information Disclosure of Companies that Offer Securities to the Public No. 2―Contents and Formats of Annual Reports”, which has been revised since 2012, the provisions are also added in the fifth section, stating that companies shall disclose the appointment of the internal control auditors and the remuneration paid to them during the reporting period. However, in practice, most companies disclose only the total audit fees paid to the accounting firms, but not the costs of internal control audits and financial reporting audits separately.
The available literature indicates that the disclosure of audit fees is considered to help reduce the “low-balling” behavior  , improve auditors’ independence   , enhance audit quality and auditing environment  , and the investors’ perceived audit quality of such companies is relatively high   . The implementation of internal control audit makes the internal control audit fee become a new economic bond between the client company and the accounting firm, which is largely the result of the compromise between management and auditors   . Therefore, this paper extends the disclosure of audit fees to the internal control audit situation, exploring whether the internal control audit fees disclosed by enterprises also have an impact on the quality of internal control audits.
The contributions of this paper to the existing research are as follows. First, it extends the research situation of audit fee information, demonstrating the impact of internal control audit fees on the internal control auditquality, and providing further empirical evidence for the economic consequences of audit fee disclosure. Second, it expands the research on the relationship between abnormal audit fees and auditor independence. In recent years, some studies have begun to pay attention to the impact of abnormal audit fees (that is, the residual items of the audit fee estimation model) on auditor independence or audit quality   , and non-disclosure of audit fees may be a manifestation of abnormal audit fees. Therefore, focusing on the impact of the internal control audit cost itself instead of disclosure can be viewed as supplement and verification to the previous studies. Third, it enriches the research related to internal control audit. Most of the relevant literatures in the past have focused on the economic consequences of internal control audits     , while few literatures regard the internal control audit fees except for Fang et al. and Tang et al. This paper finds that, from the public disclosure, the higher internal control audit fees may imply the purchase of internal control audit opinions, which means relatively low internal control audit quality.
2. Literature Review
2.1. Audit Fees and Audit Quality
A large amount of research has focused on the relationship between audit fees and non-audit service fees disclosed by companies and audit quality. Among them, some directly use the absolute amount of audit fee or non-audit expense, and its proportion to the total audit expenses as the key explanatory variables. Some further distinguish the nature of audit fees, using the audit pricing model to estimate the normal audit fees, and the difference between the actual audit fees and the normal expenses (the residual items or the abnormal audit fee). In addition, some literature studies the auditor’s economic dependence on specific clients on the audit quality. According to the dominant studies, the audit quality is mostly measured by the degree of discretionary accruals, the probability of being issued modified audit opinions, and the possibility of announcing financial restatement.
As to whether the level of audit fees will affect the quality of audits, the directly related research has not reached an agreed conclusion. Some scholars believe that excessively high audit fees can undermine audit independence and audit quality. When receiving higher audit fees, the auditor may be forced to indulge the opportunistic earnings management behavior of the management under the pressure of the client company. The management may purchase the audit opinion by giving the accounting firm an excessive audit fee, which supports the “audit collusion” hypothesis. The specific empirical results show that higher audit cost is accompanied by lower accrual quality and the probability of being issued a modified audit opinion      . At the same time, some scholars believe that auditors’ decision will be affected by the litigation and reputation risk, so a trade-off will be made in the decision-making process. The relevant research did not find a significant correlation between the audit fee and audit quality   , which leads to the opposite conclusion that supports “reputation risk” hypothesis   .
All of the above studies directly examine the relationship between the level of audit fees (or the proportion of audit fees in total costs) and audit quality proxies. In recent years, scholars have further examined the relationship between them from the perspective of abnormal audit fees, and its impact on audit quality or earnings quality, according to its symbol. For positive abnormal audit fees, most empirical findings show that the higher the abnormal audit cost is, the higher the absolute value of the discretionary accruals and the likelihood of meeting or beating analyst forecasts will be    , indicating that excessive audit fees reflect the auditor’s economic dependence on clients and damage audit independence. However, some studies have found that positive abnormal audit costs are significantly negatively correlated with the absolute values of discretionary accruals or the probability of financial restatements   , and positively correlated with earnings response coefficients  , implying that positive abnormal audit fees reflect greater audit input rather than economic dependence and do not necessarily impair audit independence. For the negative abnormal audit fees, one view is that the abnormally low audit fees indicate that the client’s bargaining power is strong, which will possibly reduce the audit quality, and the negative abnormal audit fees and the company’s earnings management are positively relevant; another view is that auditors generally have less incentive to compromise because of the low audit fees, so negative abnormal audit costs do not significantly affect audit quality.
Most of the domestic research results support that the abnormal audit cost has a negative impact on audit quality. The literature based on the results of the audit opinion shows that the abnormal audit fees are significantly positively correlated with the improvement of the adverse audit opinions of listed companies, and significantly reduce the value relevance of accounting earnings, that is, the listed company successfully purchased the audit opinions by raising the audit fees. The increase in fees has jeopardized the quality of auditing     . When measuring audit quality with discretionary accruals, Duan et al.  found that both positive and negative abnormal audit fees would undermine audit quality. Guo  found that in the early stages of the audit tenure, abnormal audit fees and audit quality have a significantly positive correlation, but this effect is weakened with the extension of the audit term. Li and Shen  found that abnormally high audit fees may impair audit quality and the correlation is only established in areas with poor rule of law, but abnormally low audit fee does not. Qi and Han  found that the audit quality of private enterprises can be undermined under abnormally high and abnormally low audit fees, but the state-owned enterprises have not been affected. In recent years, a few scholars have attempted to measure the quality of audits indirectly for research. Wan  used the stock price crash risk as an alternative to measure audit quality for the first time. It found that the abnormal higher the audit fee was, the lower the stock price crash risk would be; but the negative abnormal audit cost had an insignificant impact on the stock price crash risk. Gao et al.  found that abnormal audit fees are negatively correlated with the persistence of earnings and the accuracy of analysts’ earnings forecasts, indicating that the low-quality financial reporting environment represented by abnormal audit fees reduces the sustainability of corporate earnings and the effectiveness of the earnings forecasts from security analysts. For the first time, using the reverse audit report as the measure of audit quality, Xu  considered rent factors from three aspects―the “audit collusion” hypothesis, the audit firm’s market competition intensity and its bargaining power over clients, which deepened the research in this area.
In addition, when the proportion of audit fees to the total income of the audit firm is used to measure the economic dependence of auditors on clients, the majority of studies based in China indicate that the independence of auditors is reduced in the face of large clients, which mainly use the discretionary accruals and probability of modified opinions as proxies for audit quality  -  . Some studies have reached the opposite conclusion  , while other literatures have not found a significant relationship between them   .
In general, the literature in this area reflects the economic bond between the auditor and the client through audit fees, non-audit service fees, abnormal audit fees or client importance, to explore how this bond affects the audit independence or audit quality. Among them, some scholars directly study the relationship between them by using the size of the discretionary accruals and the probability of issuing modified opinions as the proxy variables of audit quality, while the other part pays attention to the perception of audit quality or audit independence from information users (investors, creditors, analysts, etc.).
2.2. Internal Control Audit
Ghosh and Pawlewicz  found that during the post-SOX Act period, with the addition of internal control audit procedures, the total audit fees of U.S. listed companies increased. Domestic research on internal control audits appeared later and was less in number, mainly focusing on the economic consequences of implementing internal control audits and the market reaction of internal control audit opinions. Very few people paid attention to internal control audit fees.
As to economic consequences, the existing research found that implementing internal control audit can improve the efficiency of financial statement audit and accounting earnings quality, and reduce the cost of equity capital  . Wang et al. found that only when listed companies fail to disclose internal control deficiencies, will the internal control audit significantly reduce the company’s debt costs. Zhong et al.  discovered that the quality of financial reports is significantly higher if the company had undergone internal control auditing process.
Regarding the market response of internal control audit opinions, Wu et al. found that for companies that received “non-clean” opinions only in the internal control audit, investors did not make a significant negative reaction near the information announcement date, indicating that the investors’ response to the “non-clean” internal control audit opinion is not sufficient in China’s stock market at present stage. From the perspective of the creditor’s, Han’s study shows that if the listed company received a “clean” financial statement audit opinion and a “non-clean” internal control audit opinion in the same year, it is often accompanied by a significantly higher risk of financial distress in the current year and the next year, but financial institutions such as banks have not significantly reduced the short-term credit scale of such enterprises.
The literature on internal control audit fees is more limited. For example, Fang et al. studied the influencing factors of internal control audit fees, and found that the size of the company, the complexity of the business, the nature of ownership, the reputation of the accounting firm and the assurance degree of the internal control audit service provided by it are the main influencing factors. Yang Lin’s empirical test found that internal control audit fees are negatively correlated with earnings quality, and board governance has a regulatory role. Tang found that the voluntary disclosure of internal control audit fees increased the independence of internal control audits.
Based on the above literature review, it is obvious that China’s current research on internal control audit fees is quite limited, and the attention on internal control audit fees is also low in practice. The research on internal control audit fees may be a unique area under China’s institutional background. Therefore, this paper intends to further study whether the level of internal control audit fees also affects the quality of internal control audits.
3. Research Hypothesis
In the current situation where listed companies are required to implement internal control audits, internal control audit fees naturally become a new economic link between auditors and clients. One of the objectives of internal control is to ensure the reliability of financial reporting, and the scope of its current audit is limited to internal controls related to financial reporting. The internal control audit fee is the result of the “bargaining” of the auditor and the audited entity, reflecting both the risk of the internal control system of the audited entity and the cost of the auditor’s internal control audit. Under normal circumstances, when the quality of the internal control of the audited entity is low, the risk of internal control audit is higher, and the auditor needs additional inputs, such as measures to expand the scope of control testing, increase audit procedures, and communicate with clients’ management; and the higher the overall risk level of the clients is, the higher the litigation risk faced by the auditor will be, soaprice premium is required, which leads to higher internal control audit fees. In this case, if the auditor can make accurate professional judgment and maintain its independence, theoretically a modified internal control audit opinion is more likely to be issued. However, there is literature indicating that management can purchase audit opinions by giving audit firms excess audit fees  so higher internal control audit fees may not necessarily be the result of the auditor’s hard work and prudence, but may also reflect the economic rent received by the auditor  . Thereby the auditor’s independence or the quality of internal control audits is compromised to some extent.
According to the above discussion, on the one hand, the high internal control audit fee reflects the auditor’s input to the internal control audit process and the high internal control audit quality. On the other, it may also manifest the economic rent collected by the auditor and the lower internal control audit quality. In view of the fact that most domestic research has found excessive audit fees or positive abnormal audit fees will undermine audit independence, this paper extends its conclusions to the internal control audit situation and proposes the hypothesis:
Hypothesis: Under the same conditions, companies that disclose higher internal control audit fee will have a smaller probability of being issued a modified internal control audit opinion, which means lower quality of internal control audit.
4. Research Design
4.1. Sample and Data
This paper selects all A-share listed companies that have implemented internal control audits from 2011 to 2016 and separately disclosed the internal control audit fees as the original sample, and then excludes the companies that are: 1) in the financial industry, 2) with missing data, 3) listing less than one year, 4) audited by two different firms regarding internal control and financial reporting. In this way 2130 companies remain in the sample. In order to reduce the influence of extreme values, this paper performs Winsorize processing on the continuous variables from 1% to 99% of the quantile level. Among them, the internal control audit fee data is manually collected by the author from the listed companies’ annual reports. The internal control audit opinion and the internal control quality index data are from the DIB Internal Control and Risk Management database1, and the financial and other data are from CSMAR database.
4.2. Variable Definition
Referring to the previous research on audit opinion, this paper selects internal control audit opinion (ICOP) as the proxy variable for internal control audit independence, namely the dependent variable. The key explanatory variables are the natural logarithm of the amount of internal control audit fees (ICAF) and the proportion of it to total audit fees (ICAF_r). According to the existing literature on audit fee disclosure and internal control audit, this paper includes the following control variables: the internal control audit opinion of the previous year (PRE_ICOP), internal control quality index (ICQ), earnings quality (DA), company size (SIZE), listing years (AGE), asset-liability ratio (LEV), financial status (LOSS), return on assets ( ROA), sales growth rate (GRTH), operating net cash flow level (CFO), inventory level (INVTA), accounts receivable level (RECTA), nature of ownership (SOE), internal control audit firm (ICBIG4) and dummy variables of years and industries. The specific variable definitions are shown in Table 1.
4.3. Model Specification
1The DIB Internal Control and Risk Management database is developed by Shenzhen Dibo Enterprise Risk Management Technology Co., Ltd. It provides enterprises, researchers, regulators and investors with information on the internal control status of listed companies in China through structured data compilation of internal control evaluation status, internal control audit status, internal control defects, and internal control information disclosure index.
Using a logitstic regression, the model (1) below tests whether the internal control audit fee and its proportion of the total audit fee are correlated to the probability that the enterprise is issued the modified internal control audit opinion. If the assumption of this paper is true, β1 is expected to be significantly negative.
5. Empirical Results
5.1. Descriptive Statistics
After removing the sample of companies in the financial industry, with missing
Table 1. Variable definition.
data and listing for less than one year, a total of 6257 original observations were obtained. Among them, as shown in Table 2, an average of 63.94% of companies in 2011-2016 disclosed internal control audit fees separately, and the number showed an increasing trend year by year. According to Table 3, the mean value of the variable ICAF_r is 0.273, indicating that the internal control audit fees disclosed accounted for an average of 27.3% of its total audit fees. Since 590 listed companies in the sample did not implement internal control audits in 2010, the number of observations including the audit opinions of the previous period was reduced to 3422. Overall, only less than 3% of the companies in the sample had been issued modified internal control audit opinions before.
5.2. Major Results
As shown in Table 4, the size of internal control audit fee (ICAF) and its proportion to the total audit fee (ICAF_r) are both significantly negatively correlated to the probability of issuing modified audit opinion at the level of 5%. Controlling other related factors, the higher the internal control audit fee paid by the enterprise, the lower the probability of being issued a modified internal control audit opinion. This suggests that higher internal control audit fees may
Table 2. Disclosure of Internal control audit fees for each year.
Table 3. Descriptive statistics.
Table 4. Internal control audit fees and internal control audit opinion.
Notes: 1) Significance (two-tailed) at: *0.10, **0.05 and ***0.01 levels, respectively; 2) 149 observations are automatically omitted because the industry dummy variables fully predict the dependent variable.
reflect the auditor’s collection of economic rents to compromise with the management, and thus the quality of internal control audits is relatively low.
5.3. Further Analysis
5.3.1. Nature of Ownership
Based on China’s institutional background, most scholars further distinguish the nature of ownership and find the research results differ between state-owned and private enterprises. On the one hand, state-owned companies generally have special agency problems, with the main body of their owners being absent. The actual owner entrusts the management personnel to perform the relevant control functions, which increases the length of the company’s agent chain. In this case the complicated agency problem exacerbates the difficulty of internal control construction. Moreover, most of the delegated managers have a political background, which may trigger the risk that the management is above its internal control, inhibiting the normal function of internal control system. In order to alleviate agency conflicts and establish a good corporate image, state-owned companies are more inclined to choose highly reputed audit firms. On the other hand, state-owned companies generally face more stringent risk management requirements, preferring to hire large-scale and high-quality accounting firms to deliver a positive signal to the stakeholders. In addition, compared with private enterprises, state-owned enterprises have less incentive to cater for security regulatory policies, and the risk of material misstatement in financial statement audits is significantly lower than that of private enterprises. Therefore most state-owned enterprises do not need to pay high audit fees to auditors, who tend to be independent.
Based on the above analysis, this paper further explores whether the relationship between internal control audit fees and internal control audit quality will vary among clients with different ownership. The results in Table 5 show that in the sample of non-state-owned enterprises, the internal control audit fee and its proportion are significantly negatively correlated with the probability of modified internal control audit opinions at the level of 5%, while this relationship is not significant in the state-owned counterparts. The results imply that the phenomenon of purchasing internal control audit opinions by increasing the fees paid to the audit firm is more obvious in the non-state-owned enterprises.
5.3.2. Client Size
Smaller-scale enterprises are mostly in the early stage of development and their internal control system construction started relatively late, which results in lower internal control quality weaker ability to withstand risk. In addition, smaller companies may not give adequate attention to internal control audit, so they are more inclined to purchase audit opinions. Thus, this paper speculates that companies with relatively small scale in the same industry are more likely to pay higher internal control audit fees to obtain internal control audit opinions favorable to them. Grouping the full sample by the industry mean of total assets, the test results listed in Table 6 support the speculation. That is, the negative relations between the internal control audit fees and their proportions and the probability of modified internal control audit opinions are only significant for smaller clients.
Table 5. Group by nature of ownership.
Notes: 1) Significance (two-tailed) at: *0.10, **0.05 and ***0.01 levels, respectively; 2) 313 observations are automatically omitted because the industry dummy variables fully predict the dependent variable.
Table 6. Group by client size.
Notes: 1) Significance (two-tailed) at: *0.10, **0.05 and ***0.01 levels, respectively; 2) 326 observations are automatically omitted because the industry dummy variables fully predict the dependent variable.
5.3.3. Total Audit Fee
It has been shown in the literature that integrated audit can significantly reduce total audit fees, which is mainly due to the enhanced efficiency of audit brought by scaling synergy  . At the same time, in the case of integrated audit, auditors face certain challenges in ensuring the audit quality of internal control and financial reporting under the constraints of limited auditing costs. Further, adopting integrated audit mode may lose the advantages of double auditing by different auditors, which has an adverse impact on audit quality  . In addition, China’s internal control auditing system is still in its infancy, and the information content of internal control audit opinions is low. Under the integrated audit, it is possible that the auditors tend to make the internal control audit opinions consistent with the audit opinions of the financial statements. In this case, the internal control audit opinion may be issued with the absence of necessary internal control audit procedures, resulting in low internal control audit quality. Therefore, if the higher internal control audit fee is related with lower internal control audit quality, due to the unobservability of internal control audit quality, the integrated audit model may be more conducive to cover up the client’s purchase of internal control audit opinions at a relatively low overall cost. In light of the above discussion, this paper groups the sample by the industry mean of total audit fees, and speculates that in the sample of companies with lower total audit fees, the negative correlation between internal control audit fees and internal control audit opinions will be more significant. The results in Table 7 also support the assumption.
Table 7. Group by total audit fee.
Notes: 1) Significance (two-tailed) at: *0.10, **0.05 and ***0.01 levels, respectively; 2) 325 observations are automatically omitted because the industry dummy variables fully predict the dependent variable.
5.4. Robustness Tests
The following robustness checks were conducted to test the relative stability of the main hypothetical results. Due to space limitations, the specific statistical results are not included in the text.
5.4.1. Using Abnormal Internal Control Audit Fees as Key Explanatory Variable
Since the higher internal control audit fees may include the normal part of the auditor’s increased input and effort, this paper further distinguishes the internal control audit fees with reference to the literature on abnormal audit fees. Referring to the internal control audit fee model established by Fang et al., this paper first used the pricing model to obtain the predicted value of internal control audit fee for each observation, and then compared it with the actual internal control audit fee disclosed. The difference between the two values is considered to be the abnormal audit fee (AB_ICAF), which cannot be accounted for by known explanatory variables. In the meantime, the sign of AB_ICAF is distinguished. Finally, this paper added the positive or negative AB_ICAF to model (1) to replace the IACF for regression. The results show that the positive abnormal internal control audit fee is negatively correlated with the probability of issuing modified internal control audit opinion. Although the result is not significant, it supports the main hypothesis to some extent.
5.4.2. Controlling the Impact of the Audit Opinion of the Previous Financial Report
In order to alleviate the endogenous problems caused by missing variables, this paper added the audit opinion of the previous year’s financial statements (pre_Auditop) as one of the control variables by referring to the study of Tang and Zhang. The results are still consistent.
5.4.3. Including the Corporate Governance Variables
Referring to the previous literature, the internal governance of the company also has an impact on the audit opinion. Therefore, this paper added a series of common corporate governance variables as control variables, such as “whether the chairman and CEO are the same person” (Dual), the proportion of independent directors to total directors (Indpdt), the size of the supervisory board (Supervise), the separation of two powers. The results did not change.
6. Conclusions and Implications
6.1. Research Conclusions
Based on the internal control auditing system, this paper studies the relationship between internal control audit fees and internal control audit quality. The research results are as follows: 1) under the control of other possible conditions, the higher the internal control audit fee and its proportion, the lower the probability of being issued a modified internal control audit opinion, which means that the relatively high internal control audit fee may be paid by companies to purchase more favorable internal control audit opinions; 2) the above result is found to be more significant in non-state-owned, relatively smaller companies, and clients whose total audit fees are higher. Overall, from the research conclusions of this paper, higher internal control audit fee damages the quality of internal control audit.
6.2. Policy Implications
The study also has certain enlightenments in practice. On the one hand, although there are documents that require listed companies to disclose the payment to the accounting firms, there is no exact requirement regarding how detailed the disclosure of audit fees should be. Furthermore, there is no corresponding penalty mechanism for companies that do not disclose audit fees. In this case, enterprises have considerable arbitrariness in the disclosure of audit fees. At the same time, no clear pricing mechanism for internal control audits has been implemented so far in China, so the rent-seeking behavior between client companies and auditors is difficult to avoid. On the other hand, the results of this paper indirectly reflect that when China’s internal control auditing system is still in its early stage, the information content of internal control auditing opinions is minimal, supporting the conclusions of Wu et al. and Han.
Given the above, this paper proposes the following policy recommendations. 1) Standardizing the information disclosure related to internal control audit. The regulatory body should strengthen the enforcement of the audit fee disclosure system, and force the listed company to separately disclose the internal control audit fees and the financial report audit fees, standardizing the relevant formats for the disclosure of audit fees to facilitate the use of accounting information and protect the relevant interests of investors. Moreover, certain penalties could be imposed on companies which do not disclose audit fees. 2) Establishing specific standards for internal control audit fees. The regulatory body should also promote the establishment of a pricing model for internal control audits and establish reasonable charging standards, such as the reasonable scope and gradient of internal control audit fees based on the characteristics of listed companies such as asset size and business complexity. It helps control the internal control audit fees within a reasonable proportion to reduce the possibility of listed companies using internal control audit fees for audit opinion purchases. 3) Improving the internal control audit system. Relevant departments should strengthen the enforcement of internal control audits, update and improve relevant laws and regulations to further specify and standardize the audit procedures. It can also help guide and constrain the behavior of certified public accountants, making internal control audits fully exert its due value and escort the development of the enterprise and the order of the market.
6.3. Limitations and Future Prospects
This paper is subject to some possible limitations. 1) The data on internal control audit fees in this paper are collected manually, so errors may exist, which could affect the test results. 2) Before 2014, non-state-owned companies listed in the main board had not been required to implement internal control audits; and even until now SME board and GEM companies are still in the stage of voluntary internal control audit. This inevitably leads to sample self-selection bias. For companies that do not separately disclose internal control audit fees, we cannot observe the relationship between their internal control audit fees and internal control audit quality. Although certain measures have been taken to mitigate the endogeneity problem, the bias cannot be eliminated. 3) At present, there is no specific proxy variable used to measure the quality of internal control audit. The measurement of explanatory variable in this paper is relatively simple.
This paper only examines the impact of internal control audit fees on internal control audit opinions from the results of the disclosure, yet the incentives for listed companies to disclose internal control audit fees and the influencing factors of internal control audit fees may themselves affect the internal control audit results. These can be further explored in the future studies.