Abstract
In this study, we examine whether Chinese Certified Public Accountant (CPA) firms with more partners available to perform audit engagements deliver higher quality services. Based on a sample of 2,990 company-year observations over the period of 2002 to 2015, we find that CPA firms with smaller staff–partner ratios are associated with a lower likelihood of restatements of their clients’ financial statements. Interestingly, we find that this association is less evident when engagement partners have excessive workloads, suggesting the need to balance workloads across partners within a firm. Overall, our results highlight the importance of considering partner availability and its impacts on audit quality, which echo regulators’ growing concern about the inadequate mix of audit experience, expertise, and supervision within CPA firms. Our study is timely and the results are informative to the recent policy debate. From a public policy perspective, our results suggest that CPA firms should be required to disclose information about the deployment of audit partners to help financial statement users make inferences about the quality of their work.
Introduction
In light of growing trends in auditor workloads and responsibilities, regulatory authorities worldwide are particularly concerned about the insufficient deployment of partners and imbalanced workload among audit partners due to their presumed effects on audit outcomes (Public Company Accounting Oversight Board [PCAOB], 2015). Using a sample of listed Chinese companies for the period of 2002 to 2015, we examine whether the availability of audit partners, as captured by staff–partner ratios (firm level) and partner workloads (partner level), affects audit quality, measured by the presence of restatement of the company’s prior financial statements. Although prior studies have examined how audit firm characteristics (e.g., size, tenure, industry specialization, and litigation risk) are associated with audit quality, we have limited understanding of whether and to what extent other characteristics, such as the division of human capital within the firm, affect audit production and audit outcomes. Our research aims at investigating the supply side of audit quality that considers how appropriate mix and deployment of audit personnel within Certified Public Accountant (CPA) firms affects audit quality, whereas prior evidence is sparse. 1
CPA firms trade on the expertise and knowledge of their human capital to provide quality intangible services for profit. It is a challenge for firms to maintain adequate staffing on an engagement team with the proper mixture of differing staff ranks and experience levels to achieve the preferred level of audit quality while also remaining sufficiently profitable. Human resources represent the most significant form of capital in CPA firms, where the division and utilization of labor within firms are vital for building efficient human capital (Kor & Leblebici, 2005). The PCAOB (2015) includes CPA firm staff–partner ratios (i.e., human capital leverage) as a potential audit quality indicator. This ratio represents the availability of audit partners at the firm level who could affect internal monitoring and the learning capabilities of audit personnel to ensure high-quality audits. It also reflects mentoring effectiveness and professional development among the audit personnel of a CPA firm, particularly in the transfer of knowledge and expertise from engagement partners to other audit staff. We assume that CPA firms with a smaller staff–partner ratio (e.g., a higher denominator with a constant numerator) are likely to provide higher quality services than other firms, ceteris paribus. 2
Apart from firm-level human capital attributes, individual auditor characteristics also play a role in the audit process (DeFond & Zhang, 2014; Gul et al., 2013; Knechel et al., 2015). Prior studies suggest that audit partners with heavy workloads have less time and effort to devote to supervision of staff auditors and review of work papers, which represent major activities that help ensure their clients’ financial statements are free from material errors (e.g., Gul et al., 2013). If lower staff leverage at the firm level leads to more effective audits, then partner availability at the individual auditor level is likely to moderate the leverage effect. It is particularly manifested when the audit partners are overloaded. Therefore, we further examine whether the association between audit quality and staff leverage hinges on the intensity of engagement partner workload. 3
China offers an excellent setting to test the relationship between CPA firms’ human capital and audit quality. First, the lack of market dominance by any of the Big 4 firms or large local firms provides an opportunity and the need to examine whether factors other than Big N reputation or auditor size drive audit quality. Second, in the absence of high litigation risk, CPA firms in China may not deploy labor resources at the optimal level to support delivery of quality audit work. Third, the number of audit personnel an engagement partner supervises ranges from 4 to 34 and, in light of rapid growth in initial public offerings (IPO) numbers in China, an average partner in our sample reviewed or signed 1 to 12 audit reports in 2015. Sufficient interfirm variation in the deployment of senior engagement executives increases the possibility of observing the audit quality effect. Finally, China requires that two or more engagement partners sign an audit report, allowing us to examine the workload effect at the micro-level and identify the workload of the more available partner. Although the setting of disclosing the identity of signatory audit partners provides a unique opportunity to examine auditors’ reporting behavior at the partner level (Lennox & Wu, 2018), little or no evidence exists on how internal labor resources and staff deployment affect audit practices in China and beyond. Our research takes a step in this direction and, in particular, examines the influence of audit partner availability on audit quality.
Using a sample of 2,990 observations that involved audit firm merger and acquisition (M&A) activities over the period of 2002 to 2015, 4 we find that CPA firms with a smaller staff–partner ratio are associated with higher audit quality, as measured by fewer future accounting restatements by clients. We further find that the effect of low staff leverage is less positive for partners whose total assets of audited listed clients are in the upper quartile of the sample distribution (i.e., excessively overloaded partners). These results are robust to controlling for various auditor and client characteristics. They also survive some additional tests, including the auditor size effect, client selection effect, restatement severity test, standard errors with or without clustering on client firms, and additional control variables. Overall, our results suggest that appropriate deployment of staff within a CPA firm, with particular consideration of partner availability, is an audit quality driver.
Our findings provide support for the PCAOB’s concern that high staff–partner ratios and excessive partner workload could be a root cause of audit ineffectiveness. Through an examination of the moderating effect of individual audit partner workload on the relation between partner availability and audit outcomes, we also shed light on a previously underexplored issue in the literature regarding the relevance of the human capital in the audit industry.
Our results have implications for policy and practice. From a policy viewpoint, they suggest that regulators should impose an upper limit on the number of clients that an engagement auditor can handle during busy periods, and require audit firms to maintain proper staffing levels with a balanced mix of talent and experience within an engagement team. Although clients can appoint/dismiss CPA firms with a desirable/undesirable mix of audit personnel for audit engagements, we believe that regulators’ participation can possibly be an additional means to strengthen the healthy development of the accounting profession. To enhance the transparency of auditing practices and provide useful information to stakeholders concerning audit quality, regulators should also consider requiring audit firms to disclose information about individual engagement staffing and workload allocation. Such disclosure enables regulators to identify and scrutinize companies audited by overloaded partners. From the practice perspective, our results suggest a need for CPA firms to balance workload among audit partners to facilitate greater partner involvement in each engagement. One way for firms to relieve existing partners’ work stress is to assign nonaudit activities such as engagement proposals, staff recruitment, performance reviews, fee negotiation, employee training, and community services to less busy partners. Our results could also benefit the audit committee’s effective evaluation of auditor performance and efficient oversight of the audit process.
Although based on China’s unique institutional setting, which makes it possible for us to link audit office traits to audit outcomes, our findings are generalizable to other economies having an audit environment similar to China’s (e.g., India). 5 The lack of stringent quality controls and extensive training of lower level employees in local, small audit firms allows audit partners to play more prominent roles in transferring their knowledge and expertise to nonexecutive audit staff.
We organize the remainder of this article as follows. The following section explains the institutional background and develops our research hypotheses. The third section describes data and research methodology, and the fourth section presents empirical results. We conclude in the final section.
Background and Research Hypotheses
The Audit Market in China
The role of auditing in China before the economic reforms during the early 1980s was to ensure the sufficiency of tax revenue collection for the government (Lin & Chan, 2000; Yang et al., 2001). The post-1980 economic reforms, which transformed the planned economy into a market economy, resulted in the corporatization of state-owned enterprises and boosting of foreign investments, thereby creating strong demand for independent and high-quality audits (DeFond et al., 2000). Since the reforms, there has been a general trend of improvement in audit quality in China (e.g., J. P. Chen et al., 2001; DeFond et al., 2000; Firth et al., 2005). Important milestones in this quality improvement include (a) the disaffiliation program to separate CPA firms from their sponsoring government agencies (Firth et al., 2012), (b) the gradual harmonization of domestic and international accounting and auditing standards (Firth et al., 2005; Lin & Chan, 2000; Yang et al., 2001), and (c) the removal of the cap on the liability exposure of negligent auditors (Firth et al., 2012).
The financial reporting landscape in China also changes over time in response to growing demands from financial statement users during the economic development phrases. In 1992, the government issued the Accounting Standards for Business Enterprises (ASBE), which provides the conceptual framework for financial reporting in China. The ASBE was revised in 1998 to allow companies for rooms in exercising fair judgment according to the reporting circumstances. The government further amended the ASBE in 2001 by extending the application of asset impairment to certain asset classes. In 2007, the government formally adopted International Financial Reporting Standards (IFRS)-equivalent ASBE that emphasizes the fair-value accounting concept. Liu et al. (2011) examine the impact of IFRS on accounting quality in China and find that the quality improved with decreased earnings management and increased value relevance of accounting measures since 2007. Such improvement seems to have translated into fewer restatements of corporate reports, as evident by a declining trend in the restatement rate from 8.16% during 2002 to 2006 to 5.71% during 2007 to 2015 in our sample.
With reference to the Sarbanes–Oxley Act of 2002 (Sections 302 and 404) in the United States that requires U.S. listed companies to disclose details of material weaknesses in internal control, China released the first Basic Standard of Enterprise Internal Control in 2008, followed by three supporting guidelines in 2010. Becoming mandatory from January 1, 2012, these regulations require listed companies to provide management evaluation report on the effectiveness of internal controls and an auditor’s report on the control effectiveness. Such requirement may create extra workload for auditors in the areas of internal control tests, document filing with securities regulator, and internal control report preparations.
Chinese auditors face intense time pressure because they must finish all audit work and issue audit reports before April 30, which is the deadline for all Chinese listed companies to disclose their audited financial reports. They also need to evaluate enterprise internal control effectiveness and provide annual opinions since 2012. The time pressure for auditors is more obvious if their clients are larger and client operations are more complex. To establish a “check and balance” mechanism for quality assurance, China requires two signatory auditors to sign off on audit reports. One auditor must be either a partner of the audit firm or the chief CPA, and the other must be responsible for the fieldwork. This requirement increases engagement partners’ legal liabilities for issuing inappropriate audit opinions. In light of growing workloads and expectations, Chinese regulators have raised concerns about insufficient partner involvement in audit engagements. For example, the Chong Qing Institute of Certified Public Accountants published an investigative report on the overall practice quality of CPA firms in their 2016 annual audits, criticizing partners in some firms for devoting little effort to and showing limited involvement in engagement review, quality assurance, and staff supervision in their firms. 6 This increasing concern about audit partner involvement reveals an acute need to investigate how labor deployment affects auditor availability and, in turn, audit quality.
Research Hypotheses
Audit quality determinants are a focus of the accounting literature, as quality audits lend credibility to financial statements and increase investor confidence. DeAngelo (1981) defines audit quality as an auditor’s ability to detect (i.e., competence) and report (i.e., independence) errors and irregularities in financial statements. Audit firms with larger clientele are more independent and motivated to provide higher quality audits. They are also more competent because of their accumulated professional skills and knowledge and additional resources for staff training. This competence is associated with auditors’ internal strengths, such as educational levels, experience, and industry expertise (Lys & Watts, 1994), consistent with the notion that firms can improve audit quality by maintaining an appropriate mix of staff and skills.
Human resources represent the most significant form of capital in service-oriented firms, where the division of labor within firms is essential for building efficient human capital (Kor & Leblebici, 2005). The division of labor also affects the availability of senior staff for handling work-related tasks and supervising junior staff. However, prior studies have not thoroughly examined the role of the division of labor within CPA firms. In this study, we consider two aspects of labor division: staff–partner ratios and partner workloads. Because of the likely differences in level of expertise and responsibilities among the workforce, understanding the composition of audit personnel within CPA firms is important for evaluating whether they effectively and efficiently perform their work (Bedard & Chi, 1993).
Partner availability at firm level (staff–partner ratio)
We first consider whether staff leverage affects audit quality. Prior literature on organizational structure suggests that low staff leverage can help management maintain effective monitoring of subordinates and make efficient decisions for companies (Keren & Levhari, 1979; Qian, 1994; Williamson, 1967). For example, Qian (1994) suggests that management has limited ability and thus cannot effectively supervise all subordinates if staff leverage is too high. Keren and Levhari (1979) and Williamson (1967) argue that management makes decisions based on information received from subordinates. In a highly leveraged firm, management may lose control over subordinates, and thus these subordinates are less likely to satisfactorily complete their tasks. Because of the loss of information from subordinates, the costs of delays in decisions increase for firms with high staff leverage. We expect that these arguments on organizational structure also apply to accounting firms and that increasing the staff–partner ratio would reduce partner availability and, as a result, their control over staff. A reduction in partner monitoring would have a negative effect on the quality of audit procedures conducted by audit personnel. Furthermore, with extra responsibility for overseeing staff, partners would have less time and effort to devote to monitoring audit procedures, which would impair their audit decisions (PCAOB, 2013a). Although partners can delegate their audit jobs to managers, they cannot delegate their monitoring roles because they are personally liable for the audit reports they sign.
By contrast, smaller staff–partner ratios imply a closer review of staff performance and quality training (e.g., more frequent experience sharing from supervisors) within the firms. A smaller staff–partner ratio not only allows partners to review the progress and performance of individual staff on a regular basis, but also enhances the professional development of the staff by providing timely advice. In addition, with less staff to supervise, partners can develop closer relationships with them for more effective coaching. Specialist coaching assists staff in developing crucial client-specific knowledge and practical skills (Hunt & Weintraub, 2007). Therefore, we believe that effective staff deployment in the form of closer internal review and greater capability sharing among audit personnel leads to high-quality audits.
Low staff leverage also ensures that partners can mentor audit staff and take an active interest in their career progression (BDO Canada, 2015). Close relationships between staff and partners facilitate staff career planning and retention. Poor employee retention is a threat to the service quality of CPA firms (CPA Australia, 2004). High staff turnover makes it difficult for auditors to detect material misstatements during the first few years of an engagement because they lack client-specific knowledge that is acquired in the early stages of their careers (Johnson et al., 2002). CPA firm-specific knowledge also affects audit quality because new staff require an induction period to understand the firm’s people, clients, auditing systems, and procedures (CPA Australia, 2004).
Notwithstanding the preceding discussions, one may argue that high staff leverage can have advantages over low leverage, implying a positive association between staff leverage and audit quality. Increasing staff numbers (for a given number of partners), hence more human resources, allows audit partners to delegate more work, thereby releasing part of their capacity for more important engagements. In addition, a ratio with a larger numerator means that partners have more audit personnel under their control. With sufficient workforce, auditors are less likely to take shortcuts and hence more likely to identify material misstatements. Consistent with these arguments, DeFond and Zhang (2014) and Kor and Leblebici (2005) argue that firms with larger staff bases achieve greater efficiency in leveraging human capital and monitoring quality. 7
Despite the rich discussion of staff leverage in recent literature, the empirical evidence of how the human deployment in auditing profession affects audit quality is rather limited. 8 Williamson (1967) suggests that whether high staff leverage is beneficial depends on the trade-off between benefits from having more resources under the command of the top executive and the costs of losing control over the transmission of information and instructions across successive levels in a hierarchical organization. Taking the viewpoint that the incremental effect of more audit partners on staff supervision and coaching outweighs the effect of more nonexecutive audit staff, we state our first hypothesis on a CPA firm’s staff–partner ratio as follows:
Partner availability at the individual auditor level (partner workload)
Audit firms consist of multiple offices where multiple audit partners serve several clients. Findings of prior auditing research suggest that the characteristics of the units at different levels (audit firm, branch office, and individual partner) matter for audit quality. Recent studies suggest that audit quality varies across individual signing auditors in China (Gul et al., 2013). Although we hypothesize that achieving low staff leverage (i.e., auditor availability at firm level) leads to high-quality audits, we also consider the necessity and importance of examining whether partner availability at the individual auditor level is likely to moderate the leverage effect. As detailed information of the staff–partner ratio for a particular audit is not available, we use engagement partner workload as a surrogate measure.
We argue that engagement partners with lower workloads are more available to mentor and coach employees and monitor audit work, whereas partners with excessive workloads have insufficient time to devote to supervising audit procedures and reviewing audit work papers and documentation, 9 consistent with the busyness effect that auditing multiple clients dissipates auditor effort and, thus, reduces audit quality (Gul et al., 2017; Hermanson et al., 2007; PCAOB, 2012, 2013a; Sundgren & Svanstrom, 2014). Goodwin and Wu (2016) argue that, due to significant unexpected extra workloads, partner busyness negatively affects audit quality at the disequilibrium point. 10 Examining a different outcome, Habib et al. (2018) find a positive association between partner busyness and cost of equity capital, and they attribute the findings to the low processing accuracy and reduced professional skepticism associated with busy partner workload that leads to increased information risk and thus increased cost of equity. If lower staff leverage at the firm level leads to more effective audits, then excessive workload at the partner level is likely to mitigate the leverage effect. Therefore, we further examine whether the relation between audit quality and staff leverage hinges on the amount of partner workload. We state our hypothesis as follows:
Research Methodology
Data Collection
We collect the auditor and auditee information from various sources: auditor reports, audited financial statements, the China Stock Market and Accounting Research (CSMAR) Database, and the Chinese Institute of Certified Public Accountants (CICPA) annual reports. 11 We start with an initial sample of 22,287 Chinese A-share company-year observations for the period of 2002 to 2015. We exclude the following observations: (a) 295 company-years in the financial industry; (2) 1,547 company-years with IPO; (c) 1,446 company-years with missing information about signatory audit partners and other auditor characteristics; and (d) 2,361 company-years with missing data for client characteristics. This data selection process yields a sample of 16,638 company-year observations.
It is possible that accounting restatements and partner availability are endogenously determined; in other words, companies anticipating a future restatement may be motivated to choose a CPA firm with more staff per partner for unknown reasons. To deal with this potential endogeneity issue, among the 16,638 observations identified above, we focus on analyzing those audited by CPA firms in the M&A period. 12 China’s accounting profession underwent significant development in the late 1990s when the government pushed local CPA firms to grow through M&A activities to better compete with international firms. Such M&A activities represent a shock and have immediate and significant impacts on CPA firm structures (Chan & Wu, 2011). These structural changes, which likely impact staff–partner ratios, are independent of CPA firm clients. Therefore, by analyzing listed companies audited by CPA firms in the M&A period, we can investigate the association between staff–partner ratios and audit quality with minimal client selection bias. Our final sample is thus composed of 2,990 company-year observations where the auditing firm engaged in M&A activities over the period of 2002 to 2015. Panel A of Table 1 describes our sample selection procedures.
Sample Selection Procedures.
Note. CPA = Certified Public Accountant; CSRC = China Securities Regulatory Commission.
Panel B of Table 1 shows the industry distribution of our sample observations. By and large, this distribution is consistent with the overall industry distribution in the Chinese stock market. The majority of the sample companies are from manufacturing industries (around 62%), followed by wholesale and retail trades (around 6%) and conglomerates industries (around 5%).
Regression Model
We use accounting restatements identified by the Chinese stock market regulator as the proxy for audit quality. The proxy is the inverse measure of audit quality, and thus financial restatements indicate poor auditor performance. To test the two hypotheses, we employ the following multivariate regression:
The dependent variable, Restateit+1, is an audit quality measure based on company i’s financial restatements in year t+1 (characterized by RESTATEt+1 and RESTATEt+1_Clean). RESTATEt+1, is a dummy variable that equals 1 if company i’s current year audited financial statements are restated in a subsequent period (year t+1) because they contain a material error, and 0 otherwise. 13 We also use RESTATEt+1_Clean as an alternative measure of audit quality, which is defined as a dummy variable that equals 1 if the signature auditor issues an unqualified opinion on company i’s annual financial report, which is subsequently restated in year t+1 due to a material error, and 0 otherwise. Independent and competent auditors should be able to detect and report material misstatements or issue an appropriate audit opinion; otherwise, the quality of their work is doubtful (Chan et al., 2006; Chan & Wu, 2011; DeFond & Zhang, 2014; Ke et al., 2015).
SP_Ratioit and SP_Ratioit×High_Loadit are the main variables of interest in the model. SP_Ratioit equals the total number of CPAs, regardless of their position, over the total number of signatory partners of a CPA firm auditing company i in year t. 14 High_Loadit is an indicator variable that equals 1 if the audit workload of an engagement partner auditing company i is in the highest quartile of the sample distribution in year t, and 0 otherwise. Engagement partners’ audit workloads are measured by the natural logarithm of the total client assets of the more available engagement partner. In China, two partners (chief and engagement) are responsible for client audits, and we treat a partner as more available if his or her audit workload is lower than that of the other partner. 15 The interaction of High_Loadit and SP_Ratioit thus aims to test whether the slope coefficient on SP_Ratioit varies across engagement partners with high versus low audit workloads (High_Loadit).
We interpret the coefficients on the variables of interest as follows. The coefficient on SP_Ratioit itself captures the effect of variation in staff–partner ratios on audit quality when the more available partner’s audited client assets (in log form) are below the top quartile (High_Loadit = 0). The coefficient on High_Loadit alone captures the effect of high versus low partner workloads for CPA firms with an average SP_Ratioit. The coefficient on SP_Ratioit×High_Loadit measures the incremental effect of staff leverage when the engagement partner’s audit workload is heavy (High_Loadit = 1), relative to when it is light (High_Loadit = 0). The primary coefficients of interest are α1 and α3. As public clients audited by CPA firms with a smaller staff–partner ratio are less likely to have their accounting numbers restated than other clients (Hypothesis 1), we expect the coefficient on SP_Ratioit to be positive. A negative coefficient on the interaction term SP_Ratioit×High_Loadit would be interpreted as engagement partner workloads (auditor level) attenuating the effect of the availability of audit partners (firm level) to oversee the audit process (Hypothesis 2). In other words, more partners at the firm level contribute to audit quality; however, such quality would deteriorate if some audit partners’ workloads are excessively high.
For all regression models,
Our second set of control variables relates to client attributes, as suggested by prior studies (e.g., Chan et al., 2012; DeFond & Zhang, 2014; Firth et al., 2012; Ke et al., 2015). It includes Age (the number of years since the year of going public), Leverage (the ratio of total liabilities to total assets), Current_Ratio (current assets over current liabilities), Loss (nonpositive net profit), Client_Size (the natural logarithm of total client assets), ROE (net income over total equity), CFO (cash flow from operations over lagged total assets), MAO (equals 1 for modified audit opinion in the current year and 0 otherwise), Equity_Issue (equals 1 for equity issuance in subsequent years and 0 otherwise), Distress (equals 1 for reporting negative net income, negative working capital, or negative stockholders’ equity and 0 otherwise), SOE (equals 1 for state-owned enterprises and 0 otherwise), Cross_List (equals 1 for firms also listed in overseas stock markets or that issue B-shares and 0 otherwise), IND_DIR (the percentage of independent directors in the client’s board of directors), Meeting (the number of board of director meetings held during the year), Audit_COM (equals 1 if the client firm has audit committee and 0 otherwise), and IC_Index (the client firm’s internal control quality index). 18 Table 2 provides a summary of the variable definitions.
Variable Definitions.
Note. CPA = Certified Public Accountant; CICPA = Chinese Institute of Certified Public Accountants; MAO = modified audit opinion.
Empirical Findings
Descriptive Statistics and Correlations
Table 3 presents the descriptive statistics and correlation coefficients. Panel A shows that the mean restatement rate (RESTATEt+1) is 0.056, suggesting that 5.6% of the clients’ financial statement numbers are subsequently restated due to material errors. 19 On average, 3.1% of the financial statements with unqualified audit opinions are restated in the following year (RESTATEt+1_Clean), indicating that an inappropriate opinion is issued by the auditor. The mean staff–partner ratio (SP_Ratioit) is 7.553, which implies that one partner supervises, on average, seven audit staff working on audits of financial statements. 20 The mean High_Loadit is 0.263, which represents companies being audited by partners in the top quartile of workload (with the average total client assets of RMB 39 billion and the average total number of clients of 3.4 per partner). Over the sample period, the top 10 CPA firms (BIG10) audited approximately 55.7% of the companies listed in China; 21 approximately 15.1% of CPA firms (Specialist_F) and 1.1% of the signatory partners (Specialist_P) are classified as industry specialists; and listed clients in China stay with the same CPA firm for approximately 6.122 years (Firm_Tenure) and with the same engagement partners for approximately 2.748 years (Partner_Tenure).
Descriptive Statistics and Correlations.
Note. Correlation coefficients that are significant at the 5% level or better are in bold. See Table 2 for variable definitions.
For client characteristics, it can be seen from Table 3 that 11.6% of the sample companies report losses (Loss), whereas the mean Client_Size is RMB 2.86 billion (raw values), the mean Leverage is 0.512, and the mean ROE is 0.064. These characteristics are similar to those reported in prior studies (e.g., Gul et al., 2013). Panel B shows the correlations of the variables that capture auditor and client characteristics. We note that SP_Ratioit is positively correlated with RESTATEt+1 and RESTATEt+1_Clean at the 1% significance level. On average, the top 10 CPA firms have smaller staff–partner ratios and higher partner workloads and their clients are associated with fewer incidences of restatement (RESTATEt+1 and RESTATEt+1_Clean). 22 The correlation analyses provide preliminary evidence that smaller staff–partner ratios are associated with fewer restatements in subsequent years. Next, we run multivariate analyses to determine whether this result holds after controlling for auditor and client characteristics.
Main Regression Results
Table 4 reports the main results of the regression model. 23 To examine the main effects of SP_Ratioit and High_Loadit, we run the regression excluding their interaction and report the results in Columns 1 and 2. Consistent with Hypothesis 1, we find a positive and highly significant main effect for SP_Ratioit, but not for High_Loadit, regardless of whether we use RESTATEt+1 (Column 1) or RESTATEt+1_Clean (Column 2) as the dependent variable. 24 These results are consistent with our argument and support Hypothesis 1 that CPA firms with lower staff leverage are associated with higher audit quality (i.e., have a lower propensity for client restatements in subsequent years). The marginal effect of SP_Ratioit is 0.004, which suggests that a one standard deviation decrease in SP_Ratioit (3.603; Table 3, Panel A) decreases the probability of restatement by around 1.4%. This impact is economically significant, given that the average probability of issuing a restatement is just 5.6%.
Probit Regression Results (Sample Size = 2,990).
Note. See Table 2 for variable definitions.
, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively. Significances are based on two-tailed test.
To the extent audit quality differs across CPA firms as a function of staff leverage, we next examine whether there is a mediating effect caused by engagement partners’ workload. As hypothesized, we expect the coefficient estimated in the regression for the interaction of SP_Ratioit and High_Loadit to be negative, controlling for the separate main effects of each of these variables. Results (Columns 3 and 4) show that the interaction variable, SP_Ratioit×High_Loadit, is negatively associated with RESTATEt+1 and RESTATEt+1_Clean (p < .10), suggesting that the effect of low staff leverage is mitigated when partner workloads are heavier. We also partition our sample into two subsamples at the top quartile of partner workload and then compare the coefficients between subsamples with high and low workloads. Although we do not tabulate the results in the interest of brevity, we find that SP_Ratioit is significantly smaller (at the 10% level) for the former subsample than for the latter. Taken together, these results provide solid evidence supporting Hypothesis 2 that, although audit quality can be enhanced by greater partner involvement at the firm level, it deteriorates when some audit partners’ workloads are excessively high. Our results provide support for the PCAOB’s call for a balanced workload allocation among audit partners to facilitate greater partner involvement in each audit engagement.
Regarding the control variables, four control variables whose coefficients showing consistent statistical significance in all models are Distress, Cross_List, IND_DIR, and IC_Index. Specifically, the results in Table 4 suggest that cross-listed companies and companies with more independent directors on boards and better internal controls are less likely, whereas financially distressed companies are more likely, to have a future restatement than companies with opposing characteristics. These results are similar to prior research findings (e.g., Bell & Carcello, 2000; Francis et al., 2013; Gul et al., 2013; Ke et al., 2015). 25
The largest variance inflation factor (VIF) for the independent variables is 6.512. A VIF below 10 is not regarded as high in accounting research (Lennox et al., 2012). According to Gujarati (1999) and Greene (2008), multicollinearity is unlikely to be problematic in our regressions because all VIFs are lower than 10. Following Subramanyam and Venkatachalam’s (2007) use of Vuong’s (1989) test to compare the R2 value of the two regression models, we find (untabulated) that the R2 value of the regression models is significantly larger than that of the baseline models that exclude SP_Ratioit and High_Loadit (Z = −1.612, p = .052 for the model using RESTATEt+1 as the dependent variable; Z = −1.494, p = .068 for the model using RESTATEt+1_Clean as the dependent variable).
Cross-Sectional Analysis
Intuitively, the role of partners would be perceived as substantially more important in small CPA firms than that in large firms due to the possible differences in available resources and training. To address this possible size-varying effect on audit quality, we rerun the regressions for the subsamples of audits conducted by Big 10 auditors and non-Big 10 auditors. Results in Table 5 show that SP_Ratioit continues to be positively associated with RESTATEt+1 for both subsamples of firms, but the mediating engagement partner effect exists only for the Big 10 subsample. These findings not only reinforce our main argument of achieving low staff leverage in enhancing high audit quality, but also highlight the importance of balancing partner workload particularly in the bigger firms with a generally larger clientele (where clients receiving more auditor attention produce higher quality reports).
Probit Regression Results for Clients of Big 10 Auditors Versus Clients of Non-Big 10 Auditors (Dependent Variable = RESTATEt + 1; Sample Size = 2,990).
Note. See Table 2 for variable definitions.
, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively. Significances are based on two-tailed test.
Robustness Checks
We first check whether our results are robust to clustering standard errors by firm and year and to excluding control variables that are highly correlated (CPA_No, Partner_Tenure, and Client_Size). Although not tabulated for brevity, the results of these robustness tests remain qualitatively similar to those reported in Table 4.
Second, we extend our sample to include all listed companies in China. For the main test, to deal with potential selection effects of CPA firms on audit quality, we limit our sample to those observations whose CPA firms had M&A activities in year t. To check whether this limitation has a significant impact on the results, we rerun our regressions using firm observations whose CPA firms had no M&A activities in year t (Column 1) and using all firm-years audited by CPA firms, regardless of their M&A activities (Column 2). As reported in Table 6, we find that SP_Ratioit and SP_Ratioit×High_Loadit continue to be loaded with significantly positive and negative coefficients in the respective column.
Probit Regression Results for Full Sample and Observations Without M&A (Dependent Variable = RESTATEt + 1).
Note. See Table 2 for variable definitions.
, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively. Significances are based on two-tailed test.
Third, we consider the effect of low-balling on audit quality. H. W. Huang et al. (2015) examine the association between audit firm changes, low-balling of initial audit fees, and audit quality in China, and find that the low-balling effect on audit quality varies depending on whether the same or new partners audit the client. To rule out that, on average, our observed effects of staff leverage and partner availability are simply due to initial year audit fee discount following an audit firm change, we replicate our Table 4 analysis using sample observations without the auditor changes. As shown in Table 7, we continue to find a positive (negative) and significant coefficient on SP_Ratioit (SP_Ratioit×High_Loadit), which are consistent with our main findings. 26
Probit Regression Results for the Sample Excluding Companies With Auditor Change (Sample Size = 2,687).
Note. See Table 2 for variable definitions.
, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively. Significances are based on two-tailed test.
Fourth, we consider the severity of subsequent restatements as suggested by prior studies. Dao et al. (2014) argue that restatements initiated by external parties are more severe than those prompted by the companies themselves. Similarly, Hennes et al. (2008) find that irregularities (intentional misstatements) are more likely than errors (unintentional misstatements) to be followed by fraud-related class action lawsuits. In our sample, 60% of the restatements are initiated by companies themselves, 7% are initiated by external auditors, and 33% are initiated by government agencies (i.e., the securities regulator, the tax authority, and the Ministry of Finance). As an additional analysis, we focus exclusively on restatements initiated by the government agencies and auditors, and regress these restatements on partner availability. As shown in Table 8, the regression results continue to indicate expected directional effects of staff leverage and partner availability.
Probit Regression Results for Restatements Initiated by the Auditor and Government (Sample Size = 2,990).
Note. RESTATE_AG equals 1 if a company’s current year audited financial statements need to be restated in a subsequent period and the restatement is initiated by the auditor or the government, and 0 otherwise. See Table 2 for other variable definitions.
, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively. Significances are based on two-tailed test.
Finally, we measure audit quality at the CPA firm level. RESTATEt+1_Ratio represents the percentage of a CPA firm’s clients that experience a future restatement. The ordinary least squares (OLS) regression results reported in Table 9 confirm that our inferences about the effects of staff leverage and partner workload are materially insensitive to estimating our model with an alternative definition of the dependent variable.
OLS Regression Results for Full Sample and Observations With and Without M&A (Dependent Variable = RESTATEt + 1_Ratio).
Note. RESTATEt+1_Ratio is the percentage of a CPA firm’s clients that experience a future restatement. See Table 2 for other variable definitions.
, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively. Significances are based on two-tailed test.
Self-Selection Bias
Our results, based on different subsamples and specifications, show that the positive association between the presence of restatements and the level of staff–partner ratios is moderated by audit partner availability. Arguably, partner availability is less of a self-selection concern in a sense that, unlike auditor choice where the client self-selects its auditor, partner availability is endogenously determined by the audit firm and thus exogenous to the client. Nevertheless, as we cannot completely rule out the possibility of clients selecting their audit firms based on the staff–partner ratio, we run the following additional tests.
It is possible that companies anticipating a future restatement select an audit firm with more staff per partner. It is also possible that audit firms with less staff per partner are more adept at screening out the companies that are most likely to have a financial restatement. The existence of both possibilities results in a positive (negative) association between future restatements and staff–partner ratios (partner availability). Therefore, such association is likely to be affected by the selection or screening bias associated with the client and the audit firm.
To rule out alternative explanations, we examine auditor changes, thereby causing a change in the staff–partner ratio, before and during the restatement years. Our results fail to support either of these alternative explanations. Specifically, we find no evidence (untabulated) that restated companies are more likely to change to audit firms with a higher staff–partner ratio, which is inconsistent with the client’s selection argument. Similarly, the results do not suggest that audit firms in the lowest quartile of the SP_Ratio’s distribution are more likely to quit from companies with restated financials, which is inconsistent with the audit firm’s screening argument.
Finally, if the possibility that nonrestatement companies choose to hire an audit firm with more partners available is true, we would expect a stronger staff leverage effect for these companies. To assess this possibility, we label audit firms in the lower quartile of staff–partner ratios in the year High_Availability and regress High_Availability on RESTATEt+1. The empirical results show no significant difference between the two groups of companies in their propensity to hire an audit firm with highly available partners. Thus, we rule out the possibility that nonrestated companies end up affiliating themselves with an audit firm with less staff per partner. These results provide some assurance that our core results are unlikely to suffer from a self-selection bias.
Conclusion
In this study, we examine whether audit quality, as measured by the presence of a future restatement, varies across audit firms as a function of audit partner availability and whether there is a mediating engagement partner effect. We argue that more partners available at the firm level (i.e., low staff leverage) have a positive effect on audit quality; however, this effect is less evident if some audit partners’ workloads are unduly high. Our empirical results support this argument, implying that audit quality is not merely a reflection of auditor size/reputation and auditor specialization (as found in previous literature), but also a reflection of audit-firm-specific characteristics such as human capital sufficiency and staff leverage.
We contribute to the literature by suggesting another proxy of audit quality to supplement the widely used measures of auditor size and specialization. Maintenance of an appropriate staff–partner ratio is one of the main policy initiatives that PCAOB proposes to deal with concerns about audit quality. Our study is timely and the results are informative to the recent policy debate. From a public policy perspective, our results suggest that CPA firms should be required to disclose information about the deployment of audit partners to help financial statement users make inferences about the quality of their work, a requirement similar to audit partner identity disclosure in U.S. public audits from 2017 (PCAOB, 2013b). From a practice viewpoint, our study highlights the importance of considering individual audit partner workload capacity in addition to audit-firm-level staff leverage when evaluating audit quality. Most importantly, our findings suggest that balancing workload among audit partners is crucial to ensure sufficient partner involvement for each audit engagement.
Although this study is based on the unique institutional setting in China, its findings apply to other parts of the world, especially emerging economies, where the institutional environment is weak and market mechanisms that protect against opportunistic practices are immature. Like most studies, ours has some limitations that also serve as opportunities for future research. First, despite the positive effect of having a smaller staff–partner ratio, we are unable to identify the “right” mix of audit workforce. Second, the validity of our results depends on whether the proxies for audit quality, the measurements of which are not conclusive, are appropriate. Finally, if data are available, future research could examine other indicators of audit quality such as chargeable hours per professional, chargeable hours managed per partner, percentage of audit personnel turnover, and training hours per audit professional.
Footnotes
Acknowledgements
The authors thank the Editor, Professor Bharat Sarath, the Associate Editor, Professor Kannan Raghunandan, and the anonymous referee for their insightful comments and constructive suggestions. They also thank Michael Firth, Jeong-Bon Kim, and Clive Lennox for their helpful advice and discussions on earlier drafts of the paper. In addition, they acknowledge the constructive feedback received from conference participants at the 2017 International Conference on Business and Information and the International Conference on Accounting and Finance 2017.
Data Availability
Data are available from the public sources cited in the text.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: A.W.Y.L. acknowledges the financial support from Lingnan University, Hong Kong.
