A Study on Audit Regulation Engagement Interview and Audit Quality

Abstract

Purpose: This paper aims to investigate (1) whether the interviewed auditors conduct higher quality audit than the non-interviewed auditors and (2) whether the frequency of audit engagement interviews has an impact on audit quality. Research design, data, and methodology: Using a sample of Chinese A-share listed firms between 2011 and 2019, this paper empirically tests the effect of audit engagement interviews on auditor's behavior. We collect the data of audit engagement interviews on the CICPA's website. We use OLS regression, fixed-effect model and random-effect model to examine the association between audit engagement interviews and audit quality. Results: Findings indicate that the audit quality of the interviewed auditors is significantly greater than that of the non-interviewed auditors. The frequency of the audit engagement interviews is positively associated with audit quality. The interviewed auditors spend significantly more time on the audit. Furthermore, the positive association between audit engagement interviews and audit quality only exists in non-Big 4 auditors. Conclusions: Our findings provide evidence for the effectiveness of audit regulation enforcement. The results suggest that in an emerging market with weak legal systems, preventive regulations such as audit interviews have a deterrent effect and are necessary in alleviating information asymmetry and improving information environment.

1. Introduction

The China Institute of Certified Public Accountants (CICPA), affiliated to the Ministry of Finance, is a regulatory body that makes and enforces regulations upon auditors and audit firms, and punishes those who violate auditing regulations. Due to the increasing uncertainty of environment and fierce competition, Chinese firms are faced with increasing business risk. That will inevitably lead to an increase in auditor’s litigation risks (Pratt & Stice, 1994). In order to prevent audit risks and improve audit quality, CICPA has strengthened monitoring and supervision. In 2011, CICPA started an audit engagement interview program. Before the disclosure of annual reports, CICPA summons audit partners whose clients have potential high audit risks and warns them of the potential risks. After the audit interview, auditors may modify their risk assessment and add substantive procedures to prevent material misstatements in annual reports. Different from previous regulatory measures, the audit regulation engagement interview program initiated by CICPA is precautionary. Its purpose is to caution the interviewed auditors against the potential risks and urge the interviewed auditors to effectively control the audit risks.

Audit engagement interviews can provide investors with useful information. The interviewed auditors are likely to attract the attention of the media and the public. They may improve their audit quality after the audit interview. On the other hand, the interviewed auditors may have been aware of the relevant risks, and taken the risks into consideration in audit plan and audit procedures before the interview. Thus, there might be no significant improvement in audit quality for the interviewed auditors. Besides, the interviewed auditors may spend more time only on the audit of the clients mentioned by CICPA in the interview regardless of other clients with similar risks. Thus, there might be no significant improvement in the overall audit quality of the interviewed auditors or the audit firms.

This paper aims to investigate whether and how auditors respond to the audit interview at the audit firm level. We specifically examine (1) whether the interviewed auditors provide higher quality audit than the non interviewed auditors and (2) whether the frequency of audit engagement interviews has an impact on audit quality. We find strong evidence on auditor responses to audit engagement interviews. Compared with the non interviewed auditors, the interviewed auditors have significantly higher audit quality. The frequency of audit engagement interviews is positively associated with audit quality. In further research, we separate the full sample into Big 4 and non-Big 4 firms. Since previous studies show that significant differences exist in audit quality between Big 4 and non-Big 4 firms (Teoh & Wong, 1993). We test if significant differences exist in the impact of audit engagement interviews between Big 4 and non-Big 4 auditors. Big 4 auditors show no significant reaction to audit engagement interviews. The positive association between audit engagement interviews and audit quality only exists in non-Big 4 auditors. Finally, we examine the effect of audit engagement interviews on audit delay. We find that the interviewed auditors have significantly more audit delays than the non-interviewed auditors.

Our research makes several contributions. First, our research contributes to audit quality literature in attempting to investigate the impact of precautionary regulations on audit quality from the perspective of audit engagement interviews. Previous studies mainly focus on the credibility of peer reviews (Hilary & Lennox, 2005) and ex post sanctions (Chen, Firth, Gao, & Rui, 2005; Cahan & Wei, 2006; Ferris, Kumar, & Wolfe, 1992; Ghoshal, Bengtzen, & Roberts, 2020). Some studies investigate the economic consequences of SEC comment letters (Bing & Liu, 2017; Wang, 2016). Few studies examine the effect of audit engagement interviews. Our study explores the impact of preventive regulation from the perspective of CICPA audit interviews, enriching existing literature. Second, our research helps regulators, auditors and investors gain more knowledge about the effect of audit engagement interviews on audit quality at audit firm level. Our findings suggest that the interviewed audit firm improved the overall audit quality in spite of the fact that CICPA cautions audit firms against the audit risks of specific firms.

The following section briefly reviews previous literature and develops research hypotheses. In Section 3, we discuss our sample data and research methodology. In Section 4, we provide the empirical findings and additional tests. The final section of the paper concludes.

2. Literature Review and Hypotheses

2.1. Literature Review

Accounting and auditing regulation aims to enhance public confidence in the effectiveness of the audit. The effect of regulatory enforcement has been studied extensively. Previous studies mainly discuss the characteristics of regulatory enforcement, investors’ responses and its impact on audit quality (Feroz, Park, & Pastena, 1991; Firth, Rui, & Wu, 2009; Nicholls, 2016; Smith, Stettler, & Beedles, 1984). Audit regulatory enforcement include self-regulated peer reviews, independent inspections, comment letters, audit engagement interviews and so on.

After a series of accounting scandals were exposed in the 1970s, AICPA initiated the mandatory self-regulated peer review program. Audit firms with SEC clients began to be monitored by other audit firms once every three years, which is called self-regulated peer review. A lot of research were conducted on the credibility of self-regulation. Fogarty (1996) holds that it was unlikely for reviewers to detect material deficiencies at reviewed firms. Hilary and Lennox (2005) posit that peer review opinions could provide credible information about quality differences between auditors. Casterella, Jensen and Knechel (2009) investigate the association between self regulated peer review and audit quality. They find that peer-review reports are useful in predicting audit failure and weak quality control.

After Sarbanes-Oxley Act, the self-regulated peer review program was gradually replaced by independent inspections by the Public Company Accounting and Oversight Board (PCAOB). Gramling, Krishnan and Zhang (2011) find that after the PCAOB inspection, auditors became more conservative and were more likely to issue a going-concern opinion for financially distressed clients. Lamoreaux (2016) find a positive and significant association between PCAOB inspection access and audit quality. DeFond and Lennox (2017) investigate the association between PCAOB inspections and the quality of internal control audits. They find that higher inspection deficiency rates result in an increase in the issuance of adverse internal control opinions and audit fees. Johnson, Reichelt, and Soileauthe (2020) explore the coinciding effects of the changes in auditing standards and PCAOB inspection regime as well as the Great Recession on audit fees and audit quality. They find a significant decrease in audit fees and a significant increase audit quality subsequent to the three events. Christensen, Lundstrom, and Newton (2021) find evidence that PCAOB inspection reports with audit deficiencies increase auditors’ litigation risk.

Some studies examine the determinants and effects of receiving comment letters (Ettredge, Johnstone, Stone, & Wang, 2011; Brown, Tian, & Tucker, 2018). Gunny and Hermis (2020) examine whether busyness influences SEC compliance activities. They find that when busy SEC is likely to issue fewer comment letters, focusing its limited resources on the most serious compliance issue. Dechow, Lawrence, and Ryans (2016) examine the effects of SEC comment letters on insider sales. They find evidence that insider trading is significantly higher prior to the public disclosure of SEC comment letters. They also find a negative return at the comment letter release date and a negative drift over the next 50 days subsequent to the release.

To sum up, with the improvement of audit regulation, numerous studies emerged in the field of audit regulation. However, prior literature mainly focusses on the effects of self-regulated peer reviews, PCAOB inspections and comment letters (Eutsler, 2020; Khurana, Lundstrom, & Raman, 2021). Few research investigates the impact of audit engagement interviews on audit behavior. Besides, extant literation mainly investigates the direct impact of audit regulation from individual CPA level or audit firm level. Few studies explore the spillover effects of audit engagement interviews. Last but not the least, until now there is not yet a consensus among scholars about the effects of audit regulation on audit quality. Thus, it is a question of interest and importance to investigate the spillover effects of audit engagement interviews on audit quality at audit firm level.

2.2. Hypotheses

Effective regulation not only includes investigation into and punishment for financial frauds, but also includes precautionary measures before the occurrence of financial frauds (Tian, Udell, & Yu, 2016). A timely warning can effectively prevent the occurrence of financial frauds. However, it is difficult to observe the preventive endeavors made by regulators. Therefore, there is little empirical evidence on the effect of precautionary regulation. The audit engagement interview program initiated by CICPA offers us a good opportunity to investigate the impact of precautionary regulation on audit quality. CICPA summons audit partners and warns them of the audit risk of specific firms before the disclosure of annual reports. For example, in February 2021, CICPA interviewed Pan-China audit firm (PCCPA), warning that there may be significant uncertainty related to going concern in some client firms. CICPA reminded PCCPA auditors to comprehensively consider whether the assumption of going concern for these firms is appropriate, and whether major uncertainties have been fully disclosed in their financial statements. According to the theory of the Hawthorne effect, people who receive extra attention can work harder and perform better (Levitt & List, 2011). During the audit engagement interviews, CICPA tries to make out how auditors evaluate and deal with the risk of material misstatements in the audit report. CICPA also invites experts to help the audit firms solve accounting and auditing problems if necessary. The interviewed auditors receive extra attention through communication in the interview, which may inspire them to devote more time to the audit. They may implement additional audit procedures and obtain more audit evidence. While negotiating with the clients, they may propose or insist on material audit adjustments. Thus, the audit quality may be improved subsequent to the audit interview.

On the other hand, the audit engagement interview has a deterred effect. CICPA is a national and professional organization, affiliated to the Ministry of Finance. It is uncommon for CICPA to summon audit partners. If CICPA summons an audit partner and has an audit interview with him/her, it will attract the attention of the public. The audit firm involved may become a concern in the capital market. It may exert great pressure on the interviewed audit partner and his/her colleagues in the audit firm. Furthermore, it is stated that CICPA will follow up the annual report audit of the interviewed auditor after the interview. CICPA also requires the interviewed audit firm to submit a detailed report on the annual report audit of the firm mentioned in the interview within a certain period of time after the disclosure of the annual report. These follow-up supervision measures may drive the interviewed auditors to improve audit quality.

According to sociological theory, a person’s thoughts, feelings and behaviors can be influenced by others (Bikhchandani & Welch, 1998). A rational man in social network judges the desirability of behaviors by verbal communication, observation of other people’s behaviors and the consequences of the behaviors, and decides whether to follow or not later (Bandura & Mischel, 1965). An audit firm is a social network. The auditors who are not summoned by CICPA but work in the same audit firm can observe the audit interview and the consequences of the interview. They may get risk tips and adjust their behaviors. This may lead to an overall increase in the audit quality of the audit firm. We therefore present the following hypothesis:

H1: The audit quality of the interviewed auditor is significantly greater than that of the non-interviewed auditor.

Mental set theory holds that people tend to solve problems in a fixed way based on early solutions in similar circumstances (Hunter, 1956). The demands, emotions, attitudes and values of the perceiver also have a set effect. For example, when the perceiver is happy, he tends to have a good perception of the surroundings. The mental set has both positive and negative effects. It makes the perception process more effective but more rigid, hindering or even misleading the perception (Ollinger, Gary, & Günther, 2008).

The audit engagement interview aims to help auditors make more accurate assessment of the audit risks of their clients. Some auditors have been interviewed several times during the sample period. The auditors interviewed for more than once are faced with more pressure. The perceptual set effect for them may be greater. They may form different perception sets from those who are interviewed for the first time. Different perception sets may influence the decision-making process of the interviewed auditors and lead to different decisions (Clor-Proell, Proell, & Warfield, 2014). When auditors are interviewed for the second time or the third time, they are likely to devote more time to the audit and act more prudently. Thus, we expect a positive association between the frequency of the audit engagement interviews and audit quality. We propose and test the following hypothesis:

H2: The frequency of the audit engagement interviews is positively associated with audit quality.

Reputation theory argues that a firm’s reputation affects stakeholders’ predictions of its current behavior and their actions (Herbig & Milewicz, 1993). Maintaining a good reputation is vital for the survival of the firm as it can influence firm loyalty and the stock market performance of the firm via profitability and growth (Rose & Thomsen, 2004). As audit quality is difficult to observe, firms and investors tend to use the reputation of audit firms to differentiate between good and bad. Reputable auditors are always regarded as higher-quality auditors (Geiger & Rama, 2008). Maintaining a good reputation is of the uttermost importance for audit firms (Krishnamurthy, Zhou, & Zhou, 2006). Prior studies find that damage to audit firms’ reputation may result in a loss of clients (Skinner & Srinivasan, 2012; He, Pittman, & Rui, 2016) and a decrease in the credibility of the audited financial statements (Cahan & Wei, 2006; Weber, Willenborg, & Zhang, 2008).

Big 4 auditors are more concerned about their firm reputation (Becker, DeFond, Jiambalvo, & Subramanyam, 1998). They spend many years in building a prestigious firm image. As they are global audit firms, audit failure in one country or area will be detrimental to their global operations. They are more concerned with the improvement of the audit quality. Previous studies find evident that the audit quality of Big 4 auditors is significantly higher than that of non-Big 4 auditors (Chae & Hwang, 2017; De Angelo, 1981; Pittman & Fortin, 2004; Fan & Wong, 2005). Firms audited by Big 4 auditors have significantly lower discretionary accruals (Becker et al., 1998) and higher earnings response coefficients (Ghosh & Moon, 2005). As Big 4 audit firms have long been devoted to quality control and internal governance. We expect no systematic improvement in audit quality after the audit interview. The association between audit engagement interviews and audit quality is expected to exist only in local audit firms. Accordingly, we put forward the following hypothesis:

H3: The association between audit engagement interviews and audit quality is more pronounced in local audit firms.

3. Data and Methodology

3.1. Sample and Data Source

Our sample period is from 2011 to 2019. The start date coincides with when CICPA initiated the audit engagement interview program. Our sample includes all the audit interviews conducted by CICPA during the period of 2012-2020. As audit engagement interviews are conducted between the balance sheet date and the audit report date, the sample period for our regression analyses is 2011-2019. During the sample period, CICPA conducts 154 audit engagement interviews with audit firms. 44 audit firms, including all the Big 4 audit firms, are involved in the interviews. In 2016, CICPA conducted 43 interviews, which is the most frequent of the sample period. While in 2020 CICPA conducted only 6 audit interviews. The interviews are either conducted face-to-face (43%), in writing (54%), or on the phone (3%).

We use Chinese A-share firms listed in the Shanghai Stock Exchange and the Shenzhen Stock Exchange as our sample. First, we exclude financial firms and special treatment firms. Financial firms differ from other industries in accounts and business models, so we exclude financial firms according to the industry classification standard of CSRC 2012. Special treatment firms have great business risks and audit risks. Then, we delete observations with missing data. The final sample is reduced to a total of 21330 firm-year observations. We collect the data of audit engagement interviews on the CICPA’s website. The financial data is collected from China Stock Market & Accounting Research Database (CSMAR).

3.2. Model Design

To examine the impact of audit engagement interviews on audit quality, we estimate the following models:

\(\begin{aligned} \text { ABSDA }_{i, t}=\beta_{0}+& \beta_{1} \text { Interview }_{i, t}+\beta_{2} \text { Size }_{i, t}+\beta_{3} \text { Lev }_{i, t} \\ &+\beta_{4} \text { Current }_{i, t}+\beta_{5} \operatorname{Rec}_{i, t}+\beta_{6} \operatorname{Inv}_{i, t} \\ &+\beta_{7} \text { CFFO }_{i, t}+\beta_{8} \operatorname{Loss}_{i, t}+\beta_{9} R O A_{i, t} \\ &+\beta_{10} \text { Growth }_{i, t}+\beta_{11} \text { Age }_{i, t}+\beta_{12} \text { SOE }_{i, t} \\ &+\beta_{13} \text { BIG } 4_{i, t}+\beta_{14} \text { Switch }_{i, t}+Y E A R \\ &+\text { INDUSTRY }+\varepsilon \end{aligned}\)       (1)

\(\begin{aligned} \text { ABSDA }_{i, t}=\beta_{0}+& \beta_{1} \text { Times }_{i, t}+\beta_{2} \text { Size }_{i, t}+\beta_{3} \text { Lev }_{i, t} \\ &+\beta_{4} \text { Current }_{i, t}+\beta_{5} \text { Rec }_{i, t}+\beta_{6} \text { Inv }_{i, t} \\ &+\beta_{7} \text { CFFO }_{i, t}+\beta_{8} \text { Loss }_{i, t}+\beta_{9} \text { ROA }_{i, t} \\ &+\beta_{10} \text { Growth }_{i, t}+\beta_{11} \text { Age }_{i, t}+\beta_{12} \text { SOE }_{i, t} \\ &+\beta_{13} \text { BIG4 }_{i, t}+\beta_{14} \text { Switch }_{i, t}+Y E A R \\ &+\text { INDUSTRY } \\ &+\varepsilon \end{aligned}\)       (2)

\(\begin{aligned} \text { ABSDA }_{i, t}=\beta_{0}+& \beta_{1} \text { Interview_1 }_{i, t}+\beta_{2} \text { Interview_2 }_{i, t} \\ &+\beta_{3} \text { Size }_{i, t}+\beta_{4} \text { Lev }_{i, t}+\beta_{5} \text { Current }_{i, t} \\ &+\beta_{6} \text { Rec }_{i, t}+\beta_{7} \text { Inv }_{i, t}+\beta_{8} \text { CFFO }_{i, t} \\ &+\beta_{9} \text { Loss }_{i, t}+\beta_{10} \text { ROA }_{i, t}+\beta_{11} \text { Growth }_{i, t} \\ &+\beta_{12} \text { Age }_{i, t}+\beta_{13} \text { SOE }_{i, t}+\beta_{14} \text { BIG } 4_{i, t} \\ &+\beta_{15} \text { Switch }_{i, t}+\text { YEAR }+\text { INDUSTRY } \\ &+\varepsilon \end{aligned}\)       (3)

Our dependent variable is audit quality, which is measured by accrual-based earnings management, ABSDA_i,t. Drawing on prior literature (Dechow, Sloan, & Hutton, 1995; Guay, Kothari, & Watts, 1996), we define ABSDA_i,t as the absolute value of discretionary accruals estimated by industry and year using the Modified Jones Model. The independent variable Interview_i,t is an indicator variable which takes the value of 1 if the sample firm is audited by an interviewed audit firm before the issuance of the annual report, and 0 otherwise. To explore the impact of the frequency of audit engagement interviews on audit quality, we set several indicator variables, i.e., Times_i,t, Interview_1_i,t and Interview_2_i,t. Times_i,t denotes the frequency of audit engagement interviews for an audit firm during the sample period. Times_i,t takes the value of 1 if the audit firm was interviewed by CICPA for the first time, takes the value of 2 if the audit firm was interviewed for the second time, etc. Interview_1_i,t takes the value of 1 if the audit firm was interviewed for the first time, and 0 otherwise. Interview_2_i,t takes the value of 1 if the audit firm was interviewed more than once, and 0 otherwise. Following prior research (Al-Absy, Ismail, Chandren, & Al-Dubai, 2020; Roychowdhury, 2006; Cohen, Dey, & Lys, 2008; Dechow, Ge, & Schrand, 2010), we control for firm level attributes and auditor-specific attributes that may have an effect on earnings quality. We control time and industry fixed effects with YEAR and INDUSTRY. The main variables are as defined in Table 1.

Table 1: Definition of Variables

4. Results and Discussion

4.1. Descriptive Statistics

Descriptive statistics for main variables are reported in Table 2. The mean and median of ABSDA in the sample are 0.083 and 0.051 respectively. The mean Auditlag in the sample is 4.538. The mean Interview is 0.552, suggesting that 55.2% of the observations are audited by auditors who have been interviewed by CICPA. The minimum and maximum Size are 19.785 and 26.109 respectively, suggesting that a big difference exists in firm size for the sample. The mean Loss is 0.096, suggesting that almost 10% of the firm-years make a loss during the sample period. Only 5.6% of the firm-years are audited by Big Four auditors, indicating a highly competitive audit market in China. More than one-third of the firm-years are state owned enterprises and 11.1% of the firm-years change their auditors.

Table 2: Descriptive Statistics

4.2. Baseline Regression Results

Table 3 shows the regression results of estimating Eq. (1) using Interview measure of audit engagement interviews. Results are shown in columns 1, 2 and 3 for OLS regression, panel data regression of fixed effect and random effect respectively. The results are similar across all the regression models. Each of the three regressions is highly significant (p<0.000) and explains about 5 percent of audit quality of our sample firms. The explanatory ability of the model needs to be improved. More search is needed to investigate the determinants of audit quality in China. As can be seen in column 1 of Table 3, the coefficient of Interview is negative and significant (coefficient =-0.0051; t value=-2.04). This provides support for hypothesis H1, that the interviewed auditors provide significantly higher audit quality than the non interviewed auditors. Column 2 and 3 of Table 3 draw similar inferences. Column 2 of Table 3 reports the result of the fixed effect model. The coefficient of Interview is negative (coefficient =-0.0066) and significant at the 5% level. Column 3 of Table 3 reports the result of the random effect model. The coefficient of Interview is negative (coefficient =-0.0051) and significant at the 5% level.

Table 3: Audit Engagement Interviews and Audit Quality

Note: ^*, ^*** and ^*** denote significance level of 10%, 5% and 1% respectively.

Additionally, the results indicate positive and significant coefficients on Lev, Loss, ROA, Growth, Age, and Switch, and negative coefficients on Size, CFFO, and SOE. This suggests that firms with higher level of leverage, loss-firms, firms with higher growth, established firms and firms who change auditors are likely to have lower earnings quality. While firms with a larger size, firms with more cash flows and stated-owned firms are likely to have higher earnings quality.

4.3. The Frequency of Audit Engagement Interviews and Audit Quality

Table 4 reports the regression results of estimating Eq. (2) and Eq. (3). In column 1 of Table 5, the coefficient on Times is negative (β=-0.0010) and significant at 0.1 level. In column 2 of Table 5, the coefficient on Interview_1 is negative (β=-0.0040) but insignificant, while the coefficient on Interview_2 is negative (β=-0.0053) and significant at 0.05 level. Results suggest that the frequency of audit engagement interviews is positively associated with audit quality. The more times auditors are interviewed by CICPA, the higher their audit quality. Besides, compared with auditors who are not interviewed by CICPA, auditors who are interviewed by CICPA for the first time have no significantly higher audit quality. Only those interviewed by CICPA for more than once provide significantly higher-quality audit.

Table 4: The Frequency of Audit Engagement Interviews and Audit Quality

Note: ^*, ^*** and ^*** denote significance level of 10%, 5% and 1% respectively.

Table 5: Auditor Size, Audit Engagement Interviews and Audit Quality

Note: ^*, ^*** and ^*** denote significance level of 10%, 5% and 1% respectively.

4.4. Auditor Size, Audit Engagement Interviews and Audit Quality

In further research, we investigate the moderating role of auditor size in the relationship between audit engagement interviews and audit quality. We divide the total sample into Big 4 and non-Big 4 groups and run regressions separately.

Table 5 shows the grouped regression results. The coefficient of Interview is positive but insignificant in column 1 for the Big 4 subsample, indicating that audit engagement interviews have no significant effect on the audit quality of Big 4 auditors at audit firm level. The coefficient of Interview is negative and significant in column 2 for the non-Big 4 subsample, indicating that the interviewed local auditors responded to the audit interview by conducting a stricter audit. The spillover effect of audit engagement interviews mainly exists in local auditors. One possible explanation is that Big 4 auditors are more concerned about their reputation in the audit market and tend to provide higher quality audit service than domestic auditors no matter whether they are interviewed or not. They don’t have to improve their internal governance and quality control system after the audit interview. Thus, we observe no significant differences between the interviewed Big 4 auditors and the non-interviewed Big 4 auditors.

4.5. Audit Engagement Interviews and Audit Delay

According to the above analysis, after the audit engagement interview, audit partners are likely to improve the risk management system of the audit firm and urge all the auditors in the firm to conduct a stricter audit. Auditors in the interviewed audit firm may spend more time on substantive procedures to obtain more evidence (Seetharaman, Gul, & Lynn, 2002). Thus, their audit delays are expected to increase after the audit engagement interview. Table 6 reports the regression results of audit engagement interviews and audit delays. The results in column 1 of Table 6 reveals a significant and positive coefficient on Interview (β=0.0118). In column 2 of Table 6, the coefficient on Times is positive (β=0.0021) and significant at 0.01 level. Thus, the results in Table 6 support our prediction.

Table 6: Audit Engagement Interviews and Audit Delay

Note: ^*, ^*** and ^*** denote significance level of 10%, 5% and 1% respectively.

4.6. Robustness Tests

We conduct several sensitivity analyses to rule out alternative explanations of our results and confirm the reliability of the results.

1. Prior research finds a negative association between controlling ownership and earnings quality (La Porta et al., 1998; Lewellyn & Bao, 2017). To control the potential effect of ownership concentration on our model, we run robustness test by adding the shareholding ratio of the largest shareholder (Top 1) as a control variable. The results are reported in column 1 of Table 7. The coefficient of Top 1 is positive but insignificant, indicating that ownership concentration has no significant effect on earnings quality. The coefficient on Interview remains negative and significant at 0.05 level.

Table 7: Robustness Test

Note: ^*, ^*** and ^*** denote significance level of 10%, 5% and 1% respectively

2. Prior studies find a significant impact of managerial ownership on earnings quality (Warfield, Wild, & Wild, 1995; Gabrielsen, Gramlich, & Plenborg, 2002). To control the potential effect of managerial ownership on our model, we run additional test by adding the proportion of managerial ownership (MGT) as a control variable. As can be seen from column 2 of Table 7, the coefficient of MGT is negative (β=-0.0158) and significant at 0.1 level. The coefficient of MGT is negative (β=-0.0158) and significant at 0.1 level. Our inferences remain unchanged.

3. The China Securities Regulatory Commission (CSRC) is the regulatory body in China. Any listed firms in China that make false and misleading presentation will be punished by CSRC. Prior research investigates the impact of disciplinary sanction imposed by CSRC on audit quality and finds that sanctioned auditors tend to provide higher-quality audit (Chen, Firth, Gao, & Rui, 2005). To control the potential effect of disciplinary sanction on audit quality, we delete observations that audited by sanctioned auditors. The results are presented in column 3 of Table 7. The coefficient on Interview remains significantly negative.

4. The Shanghai Stock Exchange (SHSE) and the Shenzhen Stock Exchange (SZSE) began to make their comment-letter reviews publicly available in 2015. Since then, a lot of research emerges in the field of the economic consequences of comment-letter reviews. Most research finds a significant increase in the audit quality of the auditors who received a comment letter. To alleviate the impact of the change of policies, we exclude observations before the year 2014. The sample is reduced to 12660 firm-years. As can be seen from column 4 of Table 7, the coefficient of Interview is significant (p<0.05). Our inferences remain unchanged.

5. Conclusions

This study examines the impact of audit engagement interviews on audit quality. We expect and find a positive association between audit engagement interviews and audit quality. Several sensitivity tests are performed to confirm the robustness of our results. The empirical results based on these robustness checks support our hypothesis. Further research indicates that the positive association between audit engagement interviews and audit quality only exists in firms audited by domestic auditors. There is an overall increase in audit quality for audit firms other than Big 4 auditors. The frequency of audit engagement interviews has a significant effect on audit quality. The more times auditors are interviewed by CICPA, the higher their audit quality. Besides, only those auditors interviewed by CICPA for more than once conduct a higher quality audit.

Our results provide evidence for the spillover effects of audit regulation enforcement, which is helpful to understand the role of audit regulation on audit firms and individual CPAs. At present, China's legal environment is weak, which causes the ineffective restriction on market participant. The occurrence of a series of financial frauds in China’s capital market has sparked a widespread concern for auditor independence and audit quality. Government supervision became a target of public criticism. The audit engagement interview program adopted by CICPA is an effective means to cope with audit market failure. The interviews can be transmitted to all the staff members of the interviewed audit firm, which can influence the behavior of auditors, prompt them to correct misconducts and improve audit quality. To enhance the influence of the audit engagement interviews, CICPA should improve the timeliness of the audit interview, optimize the topics of the interview, and properly select audit firms to interview.

The audit interviews are always arranged in the period from March to April, when the audit procedures are nearly completed. What the interviewed auditor can do is quite limited after the interview. If the interviews are conducted in advance, the interviewed auditors can make time for adjusting the audit plan and audit procedures. CICPA can make full use of AI technology and Big Data to implement real-time and dynamic regulation on potential risks of listed firms, and remind audit firms to take countermeasures as soon as possible in case of high risks. Besides, CICPA should improve the intensity of the audit interviews and keep closely watch on the countermeasures of the interviewed auditors. As for auditors, they should give full consideration to the risk warnings given by CICPA in the audit interview and adjust their audit plans if necessary. CICPA has an information advantage and can evaluate the macroeconomic risks better than audit firms. Auditors should attach importance to the audit interviews, enhance risk awareness and risk response capacity.

Our research is subject to several limitations. First, our research only examines the impact of whether the audit firm is interviewed or not on auditors’ behavior. We do not distinguish different types of risk tips given by CICPA. CICPA interviewed auditors for a number of reasons. For example, On April 10th, 2019, CICPA interviewed an audit firm to warn against the audit risks of its client which had changed its audit firm near the disclosure date of the annual report. On February 24th, 2018, CICPA interviewed another audit firms to warn against the internal control risks of firms with multiple mergers and acquisitions. We do not investigate whether the types of risk tips have an impact on the spillover effect on the audit interviews. Future research can subdivide the types of risk tips and examine the impact of various risk tips on auditors’ behavior and audit quality.

Second, our research only examines the impact of audit engagement interviews on audit quality of the current period. We do not investigate the interviewed auditors’ behavior and audit quality one year or two years subsequent to the interviews. Future research can examine the impact of the audit interviews from a longitudinal perspective so as to comprehensively understand the long term effect of this audit regulation policy.

Third, our research has limitation in the measurement of audit engagement interviews. CICPA hasn’t disclosed the listed firms involved in audit engagement interviews. Due to the unavailability of the data, we only investigate the impact of audit engagement interviews at audit firm level. Future research can attempt to transform the risk tips disclosed in CICPA’s announcements into quantitative data, and investigate the effect of audit engagement interviews at individual auditor level or client firm level.