Do Individual Auditors Affect Audit Quality? Evidence from Archival Data *

Transcription

1 Do Individual Auditors Affect Audit Quality? Evidence from Archival Data * Ferdinand A. Gul Monash University Sunway campus Donghui Wu The Hong Kong Polytechnic University Zhifeng Yang City University of Hong Kong * Contact: Ferdinand A. Gul (Tel: , ferdinand.gul@buseco.monash.edu.my), Donghui Wu (Tel: , afdwu@polyu.edu.hk), Zhifeng Yang (Tel: , zhifeng@cityu.edu.hk). We thank Charles Chen, Shimin Chen, Zhihong Chen, Jong-Hag Choi, John Goodwin, Yuyan Guan, Bingbing Hu, Jong-Bon Kim, Yinghua Li, Nancy Su, Xijia Su, Yong Yu, Yuan Zhang, and the participants in the 2011 Joint Symposium by City University of Hong Kong, National Taiwan University, and Shanghai University of Finance and Economics for their helpful comments, and Joanna Chan, Yanan Wen, and Yuxiao Zhou for their able research assistance.

2 Do Individual Auditors Affect Audit Quality? Evidence from Archival Data Abstract This study examines the importance of individual auditors in determining audit quality using a large set of archival Chinese data. We find that there is a significant variation in audit quality across individual auditors. The effects that individual auditors have on audit quality are both economically and statistically significant, and stronger than the corresponding effects of audit firms. We also find that the individual auditor effects on audit quality can be partially explained by the auditors characteristics such as educational background, Big N audit firm experience, rank in the audit firm, and political background. Our work highlights the importance of analyzing audit quality at the individual auditor level. Keywords: individual auditor; audit quality; auditor characteristics; archival research. Data Availability: data used in this study are publicly available from the sources described in the paper. - i -

3 1. Introduction This study examines whether and how audit quality varies across individual auditors. Our work represents a response to the recent call from academics and policy-makers for more scrutiny and understanding of audit quality at the individual auditor level. The importance of auditor individual differences in the audit process has been articulated by several writers. For example, Nelson and Tan (2005, p.42) make the following point: Auditors need to perform a variety of tasks to form an overall assurance or attestation opinion. To do so, various personal attributes of the auditor (e.g., skills and personality) influence the outcome. As such it is likely that individual characteristics of the auditor could affect the quality of the audit being undertaken. However, prior archival research has largely conducted the audit quality analysis at the audit firm or individual office level (see Francis (2004) for a review). The importance of individual auditors in determining audit quality has received increasing attention from policy-makers and academics in recent years. For example, former SEC commissioner Steven Wallman (1996) suggests that in assessing auditor independence, the focus should be on the individual, office, and other unit of the firm making audit decisions with respect to a particular audit client (p.78, emphasis added). In a recent review paper, DeFond and Francis (2005) suggest that the audit quality analysis be push from the audit firm or office level down to the individual auditor level. Similarly, Church et al. (2008) advocate that researchers investigate whether there is a systematic relationship between individual characteristics and the quality of audit reporting. We analyze the variation in audit quality across individual auditors in the Chinese market, where the engagement auditors are required to sign the audit report. The role of the signing auditors in China is similar to that of engagement partners in other markets, in that the signing auditors lead the audit engagement team and are responsible for decision-making on significant matters in the audit process. Hence the audit reporting outcome and the client firm s financial statements could be significantly influenced by the signing auditors. The names of the signing auditors are disclosed, and their profile data are also publicly available. We therefore can match individual auditors with individual audit engagements, and obtain their personal data. These characteristics make the Chinese market a useful setting in which to investigate the effects of individual auditors on audit - 1 -

4 quality. Although individual auditors may influence the audit outcome with their characteristics, they are constrained by the quality control mechanisms within the audit firm. In the audit process, they must follow auditing standards and highly standardized audit procedures. Key decisions are often centralized at the audit firm level, and their work is subject to internal and external reviews. Thus it is not clear ex ante whether individual auditors can have significant effects on audit quality, and if so, how large their effects would be. The first part of our work attempts to address this question. In our research design, we assign an indicator variable to each individual auditor who meets our requirements, as detailed in Section 3. We then estimate an audit quality model by including these indicators and, also control for client firm, audit firm, and year effects, and time-varying client firm characteristics that could possibly affect audit quality. This research design allows us to separate the effects of individual auditors on audit quality from those of the client firms and the audit firms, and to assess, not only the presence, but also the magnitude and variation of the individual auditors effects on audit quality. By construction, the effects of individual auditors on audit quality estimated here capture their fixed effects. We use multiple audit quality measures, including audit reporting (AR) aggressiveness, the client firm s abnormal accruals and non-core earnings, and the presence of a small profit or not. We find that the effects of individual auditors are significant both statistically and economically for all quality measures. The inclusion of auditor indicators to the base model increases the explanatory power by 4.66% to 8.31%, or by 7.50% to 34.95% relative to the base model s adjusted R-square. The frequency of individual auditors who have significant effects on audit quality is much greater than what would be expected by chance. For example, the percentages of the individual auditors effects for AR aggressiveness that are significant at 5% and 10% levels are 12.9% and 18.0%, respectively. The variation in the effects of individual auditors is also considerable. For example, the abnormal accruals reported by client firms for an auditor at the 75 th percentile of the distribution of individual auditors effects would be 2.7% higher than the one at the 25 th percentile. Thus, individual auditors differ to a great extent in terms of audit quality or audit aggressiveness. Interestingly, the effects of individual auditors on audit quality turn out to be stronger than the effects of audit firms. Adding audit firm indicator variables to the base model increases the adjusted R-squares modestly by 0.63% to 1.77%

5 After showing that audit aggressiveness varies across individual auditors, we next explore whether this variation can be explained by auditors personal characteristics. Research on audit judgment and decision-making (JDM) suggests that audit quality is affected by auditor JDM attributes such as expertise, ability, risk profile, cognitive style, and independence (see Nelson and Tan (2005) and Nelson (2009) for reviews of prior studies). Based on the prior literature, we consider the following personal characteristics: education, gender, birth cohort, Big N experience, rank, and political background, assuming that these characteristics are associated with one or more of the attributes relevant to audit JDM. We find that partners exhibit a relatively conservative style of audit reporting, consistent with prior findings that partners are more conservative and take a tougher stand in requesting accounting adjustments than non-partner auditors (Trotman et al. 2009; Miller 1992). Educational background also makes a difference. Auditors who hold a Master s degree or above tend to be more aggressive, perhaps because they are more confident and optimistic. Auditors who have been exposed to Western accounting systems and modern theories of the corporation in their college education are more conservative. This could be due to their exposure in their early education to the notion that financial statements are designed to solve the information asymmetry between insiders and outside shareholders. We also find that auditors who have worked in international Big N audit firms tend to be more conservative. This is understandable since Big N audit firms are found to be more conservative than other audit firms (Francis 2004). The generally conservative environment in Big N firms may have a significant influence on their auditors judgment and decisions. Alternatively, it is also possible that auditors recruited by Big N audit firms may be inherently more conservative than those recruited by other audit firms. Auditors who have political affiliation, proxied by membership of the Chinese Communist Party, are more likely to be associated with lower quality audits than others. A possible reason for this is that CCP membership may give individual auditors some protection from being penalized for audit failure, which induces them to behave more aggressively. In the final step, we show that the documented effects of individual auditors on audit reporting and clients earnings quality are positively correlated with the likelihood of being sanctioned by the regulatory authority. As regulatory sanction for audit failure is an ex post measure of audit quality, this finding suggests that the documented effects of individual auditors indeed capture the difference in audit quality across individual - 3 -

6 auditors. This study advances our understanding of audit quality in several important ways. First, we show that there is a large variation in audit quality across individual auditors, and thus demonstrate the importance of analyzing audit quality at the individual auditor level (DeFond and Francis 2005). To the best of our knowledge, we are the first to quantify the effects of individual auditors on audit quality. Our results are important, as they reveal that individual auditors play an active role in the financial reporting process. Second, although prior research has examined the influence of auditors personal attributes on the performance of individual audit tasks (Nelson and Tan 2005), we respond to Church et al. (2008) by systematically exploring the associations between a rich set of auditors demographic characteristics and the overall quality of audit engagements. We show that various auditor characteristics can influence audit reporting and client firms earnings quality. Equally important, we find that a large portion of the variation in the effects of individual auditors is not explained, which suggests that more research is required to explore this issue in the future. Third, previous findings on the importance of the effects of auditor characteristics on audit performance are mostly obtained in experimental settings, while our results are based on a large set of archival demographic data. In this sense, our work bridges the archival and behavioral accounting research. This study also links to a growing body of literature on the effects individuals have on business decisions. Recent studies in accounting and finance show that individual managers, such as CEOs and CFOs, matter for a wide range of corporate decisions and policies (Bertrand and Schoar 2003; Ge et al. 2010; Bamber et al. 2010; Dyreng et al. 2010). Although there are significant differences between corporate executives and individual auditors in terms of their influence or power over the underlying decisions, individual auditors still imprint their mark on the audit outcomes. Our evidence is consistent with the notion that people rather than business organizations make decisions and, thus, that the role of individuals in business decisions deserves more attention (Kachelmeier 2010)

7 2. Literature Review and Research Questions 2.1. Literature Review The Audit Quality Analysis Audit quality is determined by an auditor s ability to discover breaches of accounting standards and their incentives to report such breaches, i.e., audit quality is a product of auditor competence and independence. DeAngelo (1981) argues that large audit firms are associated with higher audit quality because they are more independent. For large auditors, such as the Big N international audit firms, no single client is of more importance and there is more to lose if they misreport. Furthermore, Big N firms have established brand name reputations and, thus, have incentives to protect their reputation by providing high-quality audits (Simunic and Stein 1987; Francis and Wilson 1988). Motivated by these arguments, early studies use the dichotomy between Big N and non-big N audit firms, and show that Big N audit firms are of higher quality and are more conservative (Becker et al. 1998; Francis and Krishnan 1999; Teoh and Wong 1993). Big audit firms consist of many city-based practice offices. DeAngelo s (1981) argument regarding audit quality and firm size can be applied to the office level. In terms of economic importance, for example, a client that is not big relative to a Big N firm could be very important to one of the firm s offices. Accordingly, more recent studies shift the audit quality analysis from the firm level to the office level (Reynolds and Francis 2000; Krishnan 2005; Francis and Yu 2009). For example, Francis and Yu (2009) show that the bigger offices of Big 4 firms are of higher quality because bigger offices have more in-house expertise. A natural extension of this literature is to move the audit quality analysis further down, from the level of the office to that of the individual auditor, because individual auditors may differ in regard to both dimensions of audit quality, i.e., independence and competence (DeFond and Francis 2005). Accounting scholars have recently begun to investigate the role of individual auditors in determining audit quality. For example, Chen et al. (2010) perform one of the first analyses of how economic dependence affects audit quality at the individual auditor level using Chinese data, and find that the impact of client importance on the independence of individual auditors is conditional on the legal environment

8 2.1.2 How Do Individual Auditors Influence Audit Quality? Individual auditors differ not only in terms of their incentives, but also in regard to attributes such as risk preference, expertise, ability, and cognitive style. Knechel (2000) argues that audit quality is ultimately dependent on an auditor s JDM qualities as auditing is inherently a JDM process. An auditor s JDM qualities are, in turn, determined by the auditor s personal characteristics. Indeed, research on auditors personal attributes has a long tradition in auditing. Prior research (see Nelson and Tan (2005) and Nelson (2009) for reviews of relevant papers) shows that auditor knowledge, expertise, personality, and cognitive style each play a significant role in determining the performance of audit tasks. The extent to which auditor characteristics can influence the audit outcome is likely to be affected by the quality control mechanisms within the audit firm. In fact, audit firms try to maintain consistency in audit quality through control mechanisms, including standardization of work procedures, centralized control with risk and materiality decisions, and socialization precisely because of the idiosyncrasies of individual auditors (Jeppesen 2007) The Managerial Fixed Effect Literature A recent stream of the literature has documented that individual executives exert significant influence over a wide range of firm policies. In a seminal paper, Bertrand and Schoar (2003) show that a significant portion of the heterogeneity in firms investment, financial, and organizational practices can be explained by the presence of executive fixed effects. Following the methodology developed by Bertrand and Schoar (2003), Dyreng et al. (2010) show that top executives have incremental effects on their firms tax avoidance. Ge et al. (2010) find that CFO-specific factors explain a significant portion of the heterogeneity in financial reporting practices. Top executives not only affect corporate strategic decisions, such as M&As, but also those seemingly second-order decisions, such as voluntary disclosures. For example, Bamber et al. (2010) find that top executives exert economically significant individual effects on five aspects of management forecast: forecast frequency, forecast precision, the news conveyed by the forecast, and the bias in, and accuracy of the forecast. Prior literature also examines whether observable executive characteristics, such as gender, education, and experience, can explain manager fixed effects. Overall, the findings suggest that these observable characteristics, at best, partially explain the managerial fixed - 6 -

9 effects on firm decisions. However, the lack of a strong association between observable attributes and manager effects does not lessen the main conclusion that individual managers matter (Dyreng et al. 2010). Instead, it suggests that certain unidentified factors are important in explaining these fixed effects and, thus, highlights the importance of first quantifying the effects of managers characteristics Research Questions Kachelmeier (2010) points out that the managerial effect research shows that people rather than business organizations make decisions and, thus, bridges the archival and behavioral accounting research. Similarly, in the audit process, individual auditors may play an important role in making decisions. The personal attributes of individual auditors, e.g., their risk preferences, experiences, and incentives, may have a significant impact on audit quality (Nelson and Tan 2005). However, the importance of individual auditors in determining audit outcomes has not been widely examined in the existing archival accounting research. This could be due to the lack of data on individual auditors in the United States. DeFond and Francis (2005) thus suggest that scholars analyze audit quality at the individual auditor level in those markets where data are available. We examine the importance of individual auditors in determining audit quality using a large set of archival data related to the audits of public firms in China. We seek to answer two related questions. First, is there a significant variation in audit quality across individual auditors? Second, if so, can observable demographic characteristics of individual auditors, such as educational background, experiences, and gender, explain the variation in audit quality across individual auditors? To answer the first question, we adopt the methodology developed by Bertrand and Schoar (2003). This approach allows us to determine the presence, magnitude, and variation of the individual auditor effects on audit quality. This step is important for several reasons. First, while individual auditors may influence the audit outcome with their characteristics, ex ante it is not clear how large their effects would be. Unlike corporate executives such as CEOs who are very powerful and may dictate corporate decisions, auditors may not have such discretion in their judgment and decision making. Individual auditors must comply with the auditing standards promulgated by the profession body or the regulatory authority, and follow standardized audit procedures to perform their work. The material decisions are also controlled by the audit firm or the - 7 -

10 practice office. Moreover, their work is subject to the internal and external review. Because of these quality control mechanisms, there is a relatively small room for individual auditors to exercise their discretion. Second, we cannot conclude from prior archival auditing research that individual auditors affect audit quality. The archival auditing research has just begun to look at the role of individual auditors. This literature typically examines the association between a single aspect of auditor characteristics such as gender (Hardies et al. 2010) and experience (Chi et al. 2008), and finds no or weak evidence that individual auditor characteristics matter for audit quality. Although prior research controls for time-varying client firm characteristics, the time-invariant client specific factors that could impact audit quality and correlate with the effects of individual auditors are omitted. The research design of our study, outlined in Section 3, allows us to isolate the effects of individual auditors on audit quality after controlling for time-varying client firm determinants, client firm fixed effects, audit firm fixed effects, and year effects. This research design can generate cleaner evidence of the individual auditor effects on audit quality. Third, although the audit JDM research acknowledges the heterogeneity in individual auditors and its impact on audit performance, prior studies in this literature are carried out in the laboratory setting, making it difficult to evaluate the external validity of their findings. Moreover, they typically focus on the effects of auditors attributes on the performance of individual audit tasks instead of the overall quality of an audit engagement, which is the subject studied by us. Fourth, individual auditors differ in numerous aspects and thus focusing solely on a limited set of observable characteristics may seriously underestimate their effects on audit quality. Indeed, the managerial fixed effect literature has shown that unidentified or unobservable factors are much more important than the observable characteristics in explaining the influence of individuals on decisions. This suggests that focusing on the observable characteristics only may lead to incorrect inference that individual auditors have little or no impact on their practice. Hence, to demonstrate the importance of individual auditors on audit quality, it is necessary to estimate the overall effects, which capture the influences of both observable and unobservable or unidentified characteristics, of individual auditors on audit quality. After estimating the effects of individual auditors on audit quality, we then explore - 8 -

11 whether the variation of these effects on audit quality across individual auditors can be explained by their demographic characteristics. Church et al. (2008) suggest that researchers investigate whether there is a systematic relationship between individual characteristics and the overall audit quality. Our second step represents such an attempt. 3. Research Design 3.1. Empirical Models To effectively estimate the effects of individual auditors on audit quality, it is important to control for all other factors that could impact audit quality. We follow the methodology developed by Bertrand and Schoar (2003) to construct the individual auditor sample and estimate the auditors effects on audit quality. For each audit quality measure, we estimate the following ordinary least-square model: y it = βx it + α t YEAR t + γ i CLIENT i + κ j AUDFIRM j + λ k AUDITOR k + ε it, (1) where y it are the audit quality measures, which will be defined below, X it is a vector of time-varying client firm variables that may affect audit quality, YEAR t is a set of year indicators, CLIENT i is a set of client firm indicators, AUDFIRM j is a set of audit firm indicators, AUDITOR k is a set of individual auditor indicator variables, and ε it is the regression error term. 1 The coefficient on the auditor indicator, λ k, captures the fixed effect of individual auditor k on audit quality. For brevity, we use auditor effects or individual auditor effects to refer to the fixed effects of individual auditors on audit quality. Client firm and audit firm effects are included to mitigate the concern that the results are driven by time-invariant client firm and audit firm characteristics. As will be explained later, we define audit quality proxies so that higher values indicate more aggressive audits. A significantly positive value of λ k suggests that individual auditor k is relatively aggressive, 1 We do not control for the fixed effects of audit offices in the model. Many audit firms in China do not have multiple offices. During our sample period, the mean (median) number of city-based offices per audit firm is 1.42 (1.00). The fixed effects of office would be highly correlated with those of audit firm, if both are included in the model and the results could be biased. Moreover, in studying the impacts of client importance on audit quality, Chen et al. (2010) show that results are similar whether the client importance variable is measured at the audit office or firm level. We therefore only include audit firm fixed effects in the model

12 e.g., more tolerant of clients aggressive accounting or maintains higher thresholds for issuing modified audit opinions. 2 We then link the coefficients on the individual auditor indicators obtained from Model (1) to the characteristics of individual auditors that relate to expertise, experience, risk profile, cognitive style, and incentives by the following model: λ k = α + δ k Z k + ε k, (2) where λ k are the estimated fixed effects of individual auditors on audit quality, Z k is a vector of auditor characteristics, and ε is the regression error term. The estimated fixed effects are used as the dependent variables in Model (2). Considering the possible measurement errors in the estimates, we use the Least Trimmed Squares (LTS) method (Rousseeuw 1984) in fitting the regressions. The LTS method generates more stable results in the presence of outliers The Construction of the Individual Auditor Sample Bertrand and Schoar (2003) and other studies in the manager fixed effect literature require that a top manager should work in at least two different firms and the manager should have been in each firm for at least three years. We construct our individual auditor sample in similar fashion. To be assigned an indicator variable, an auditor must meet two conditions: (1) the individual auditor has audited a client for at least three years and there are at least three years in which she/he does not audit this client; and (2) the individual auditor has audited at least two such unique clients. An auditor must audit a client for a few years so that she/he has a chance to imprint her/his mark on the client s earnings quality. We thus require that an auditor has audited a 2 More precisely, a positive value of λ k suggests that the audit outcomes of an individual auditor are relatively aggressive. The aggressive outcome could be due to the auditor being inherently more aggressive, i.e., uses a higher threshold for issuing modified opinions or delineating material and immaterial misstatements. It could also be due to the fact an auditor is unable to detect misstatements because she/he lacks knowledge, ability, and/or expertise and, thus, does not request accounting adjustments, or she/he waives accounting adjustments because she/he is persuaded by invalid evidence presented by client firms or compromises her/his independence in the face of economic incentives. Although the underlying reasons for having aggressive outcomes are different, the results are the same. For convenience, we say an auditor is more aggressive than another if the former s fixed effect on audit quality (λ k ) is larger than the latter s

13 client for at least three years. The other criteria are imposed so that we can separate individual auditor effects from client firm fixed effects. The importance of these criteria can be illustrated by the following extreme example. Suppose that an auditor has only one client and she/he has been the only auditor for that client in the firm s history. In this case, the auditor and the client firm indicator variables are perfectly correlated and it is impossible to separate the auditor effect from the client firm fixed effect. We thus require that the auditor must audit at least two clients and there are at least three years in which she/he does not audit each of these two clients. 3 Under this method, we estimate the incremental effect of auditor, k, on audit outcomes from the multiple clients she/he audits over time as the fixed-effect coefficient, λ k. It is worth noting that this method also mitigates the correlated-omitted-variables problem. After controlling for client firm fixed effects and time-varying characteristics in the regressions, the unobservable and thus omitted variables do not bias the auditor fixed-effect coefficients unless such variables change over time and across firms in the same pattern as audits performed by individual auditors over time and across firms Audit Quality Measures Audit reports and audited financial statements are two observable audit outcomes. Accordingly, in prior research, audit quality is measured by determining auditors thresholds for issuing modified audit opinions (MAOs), and the quality of clients reported earnings. The underlying assumption is that high-quality auditors maintain lower thresholds for issuing MAOs and constrain aggressive earnings management. To obtain convincing evidence of auditor effects, we employ four quality proxies, discussed follows. Audit reporting aggressiveness. Following prior studies (e.g., Francis and Krishnan 1999; DeFond et al. 2000), we define an indicator variable MAO, which equals 0 if a client receives a clean audit opinion and 1 otherwise. We then estimate the predicted probability of issuing MAOs by running a logistic model, with MAO as the dependent variable and a set of firm characteristics as explanatory variables. Our audit reporting aggressiveness measure (ARAgg) is the predicted probability minus the actual value of MAO. A higher value in ARAgg means that an auditor s propensity to issue MAOs is lower than what 3 Although these requirements are used in prior research on managerial fixed effects, they are admittedly arbitrary. We alter the number of years/clients in these requirements and examine whether our main findings are sensitive to such requirements in Section

14 would be predicted from the whole sample. The details about how we measure ARAgg are described in Section A.1 of Appendix A. Abnormal accruals. Following Dechow and Dichev (2002) and McNichols (2002), we estimate abnormal accruals (AbAcc) as the difference between working capital accruals and the fitted values from the accrual model. Section A.2 of Appendix A provides the details of the model for estimating abnormal accruals. Consistent with prior studies (Becker et al. 1998; Francis and Krishnan 1999; among others), higher abnormal accruals indicate more aggressive earnings and, thus, lower quality auditing. Below-the-line items. The adoption of below-the-line items or non-core earnings as another proxy for earnings quality is motivated by previous studies which find that Chinese firms tend to inflate earnings by timing the execution of transactions pertaining to below-the-line items (Chen and Yuan 2004; Haw et al. 2005). These transactions are often dubious related-party transactions and attract much attention from regulators and investors. Consistent with these studies, we define variable BL as the sum of investment net income, profits from other operations, and non-operating net income, scaled by year-ending total assets. BL thus measures the effect of these items on pre-tax ROA. Small profits. The presence of a small profit is interpreted as evidence of income-increasing earnings management (Burgstahler and Dichev 1997; Francis and Wang 2008). In this study, a client firm is considered to have a small profit if its ROA is between 0 and 1%, and is indicated by the variable SP. Auditor quality decreases with the likelihood of SP in audited financial reports. 4 Admittedly, earnings management does not necessarily violate Generally Accepted Accounting Principles, and is usually not outright fraud. However, aggressive earnings are often perceived to be of low quality, and can mislead financial statement users. The ambiguous nature of these financial reporting practices provide auditors with considerable 4 Note that SP is a dichotomous variable. While it is theoretically appealing to estimate a logistic model for a dichotomous dependent variable, here, we still apply the OLS method. This is because the complete or quasi-complete separation problem in the logistic fixed effect model occurs in our data, as some auditors clients never take a value of 1 in SP and it is impossible to compute the maximum likelihood values of the fixed effect coefficients for such auditors. Nevertheless, for dichotomous dependent variables, OLS coefficient estimates still remain unbiased, especially in large samples, and can be interpreted as usual (Wooldridge 2005, Ch.7). Bamber et al. (2010) and Ge et al. (2010) also apply the OLS method to models where the dependent variables are discrete

15 latitude to influence the audit outcomes; the extent that auditors may use this latitude could be affected by their individual personal characteristics. Firm characteristics such as profitability, growth, and leverage may affect earnings quality (Dechow et al. 2010) or auditors propensity to issue MAOs (e.g., Francis and Yu 2009). Moreover, prior research using Chinese data shows that the economic importance of client firms to specific auditors, and the ownership of clients may have a bearing on auditors decisions (Chen et al. 2010; Wang et al. 2008; Chan et al. 2006), We therefore control for the following time-varying firm characteristics when estimating the auditor fixed effects on the above audit quality measures: return on assets (ROA), the ratio of sales to assets (Turnover), the presence of loss or not (Loss), the log value of total assets (Size), book-to-market ratio (B/M), leverage ratio (Leverage), firm s listing age (Age), client importance (CI), and a variable to indicate whether a client firm is ultimately controlled by a local government in China (LGOV) Determinants of Auditor Effects The audit JDM literature suggests that audit quality is affected by auditors personal attributes, such as expertise (knowledge), experience, problem-solving ability, risk preference, cognitive style, and incentives (Nelson and Tan 2005; Nelson 2009). Our choice of individual characteristics is based on prior research and the unique characteristics of institutions in China; some of the variables selected are exploratory. We consider the following demographic characteristics of auditors that may relate to the above attributes: educational background, birth cohort, Big N work experience, gender, rank (partner or not), and political affiliation. Education. An auditor s educational background may affect their knowledge, risk preference, and values. The first variable in this category measures whether an auditor obtained a master s degree or above. The holders of master s degrees or above receive more education than others. They also command more job opportunities, higher salaries, and a greater likelihood of being promoted in China. 5 Hence, auditors who hold a 5 For example, a recent survey by MyCOS Inc., a leading education data provider in China, shows that in 2011 the starting salary for bachelor s degree holders is about RMB 2,400 per month, while that for Master s degree holders is about RMB 4,000 per month (an introduction to the report is available on

16 master s degree or above are likely to be more optimistic and aggressive than others. 6 The second variable in the educational category is to indicate an auditor s education cohort. We classify auditors into two groups based on whether they were born in 1971 or later. The idea is that those who were born in 1971 or later are likely to have entered university in 1990 or later, because the typical age for Chinese students to enter university is 19. Thus, this group is likely to have been exposed to Western accounting systems in their college education. In 1990, the first Chinese stock exchange was established in Shenzhen and the Chinese government announced that it would encourage the corporatization of state-owned enterprises. To accommodate these reforms, China decided to abandon the former Soviet accounting system, which was designed to assist planning in a command economy rather than to solve the information asymmetry between managers and outside investors, and stipulated relatively uniform and stringent accounting methods/procedures. Since then, Western accounting systems have been introduced and dominated university accounting education. To determine whether the exposure to the modern principles of financial reporting and concepts of corporate governance through university education influences auditors aggressiveness, we include a variable to indicate when an auditor received his/her college education. Specifically, the education cohort equals 1 if an auditor was born in or after 1971 and 0 otherwise. The third educational variable indicates whether an auditor majored in accounting in her/his college education. It is possible that accounting students behave differently from non-accounting ones in their subsequent careers. Gender. Females and males are arguably different in terms of problem-solving ability, risk preference, and cognitive style (Hardies et al. 2010). For example, Gold et al. (2009) find that female auditors are more influenced by male clients and less influenced by female clients than male auditors. Furthermore, the psychology literature suggests that females are more risk averse and more conservative in finance related matters than males (Fellner and Maciejovsky 2007; Eckel and Grossman 2002; Lundeberg et al. 1994). More recently, Srinidhi et al. (forthcoming), using US data find that firms with female directors 6 A number of extant studies suggest that MBA degree holders could have different styles from other executives. For example, Bertrand and Schoar (2003) show that MBA degree holders are relatively more aggressive than others. Unfortunately, our data only tell us whether the highest degree an auditor obtained is a doctorate, Master s, bachelor s, or below. Hence, in the education category, we consider whether an individual auditor has obtained a Master s degree or above

17 are associated with higher earnings quality. These findings and suggested differences between female and male auditors motivate us to consider gender. Big N experience. An auditor s experience may affect her/his judgment and actions. We use a variable to indicate whether an auditor has worked in one of the international Big N audit firms. Big N international audit firms as a group are more independent and provide higher quality audits than other audit firms. To achieve high and consistent audit quality, Big N firms tend to recruit those who are more sociable and adaptable to bureaucratic systems and the culture, values, and goals of the Big N firms (Jeppeson 2007). The work experience in Big N audit firms thus is likely to mold auditors who end up being different from auditors in a non Big N firm. Alternatively, those who tend to be recruited by Big N audit firms may have relatively more conservative personalities, which also is likely to lead to conservative audit outcomes. Birth cohort. Important events that happen during childhood or youth could have a significant impact on an individual s risk attitude, personality, values, and cognitive base. Hence, auditors of the same birth cohort may have some similarities in their judgment and decision-making because they are likely to be affected by the same important events in their early life. We thus include auditors birth year. 7 Rank. We consider whether a signing auditor is a partner or not. The auditing literature shows that auditors who are partners act differently from other auditors. Audit partners own and manage audit firms, thus the goal congruence of partners and the audit firm should be greater than that of non-partner auditors and the audit firm. From this perspective, Miller (1992) argues that partner auditors should be more conservative than non-partner auditors. Partners also have more authority, both within the audit firm and as perceived by the clients, and can take a harder stand than other auditors in requesting for accounting adjustments. This conjecture is borne out by Trotman et al. (2009) who provide evidence which shows that partners request higher initial proposed write-down than non-partner auditors. Political affiliation. We include a variable to indicate whether an auditor is a member of the Chinese Communist Party (CCP). This is motivated by prior research in accounting and finance which finds that political factors may influence business decisions. 7 Bertrand and Schoar (2003), Bamber et al. (2010), and Dyreng et al. (2010) include birth cohort as a determinant of manager effects for similar reasons

18 Li et al. (2008), for example, suggests that private entrepreneurs in China have incentives to joint CCP. One important benefit brought by political participation is relaxed regulatory oversight (Faccio 2006). It is possible that CCP membership may provide some protection for auditors in case of audit failure, e.g., auditors who are CCP members may receive lighter penalty than others if both are similarly responsible for audit failure. The insurance effect brought by CCP membership may induce auditors with CCP membership to behave relatively aggressively. Hence, we include CCP membership as a proxy for auditors political affiliation and participation. 4. Empirical Results 4.1. Sample and Data Our data are obtained from the following sources. We obtain accounting and stock return data from the China Stock Market and Accounting Research Data Base (CSMAR). Audit opinions and the identities of audit firms and individual signing auditors are manually collected from firms annual reports. We crosscheck the identities of individual signing auditors against the enquiry system complied by the Chinese Institute of Certified Public Accountants (CICPA) at Data for the individual auditors demographic information is also obtained from this database. We manually input each individual auditor s full name into the relevant search fields and match the search results with the audit firm and individual auditor data we collected from companies annual reports. We begin with a population of 15,571 non-financial firm-year observations listed on the Shanghai and Shenzhen stock exchanges between 1998 and We drop 770 observations without the required accounting, stock return, or signatory auditor identity data, resulting in a total of 14,801 firm-year observations in our full sample. Normally, two individual auditors are required to sign an audit report for a client 8 In early years, most audit firms in China were affiliated with the government or with publicly-funded universities. Later, the Chinese government required audit firms to disaffiliate from the government and universities. This disaffiliation program was largely complete by We start our sample period from fiscal 1998 (the actual audits for that year s financial statements were conducted in 1999, immediately after the completion of the disaffiliation program) to mitigate the possible effects of organizational change on audit firms

19 firm. We identify a total of 3,726 unique individual auditors who sign audit reports for client firms. Among them, 878 auditors meet the requirements specified in Section 3. However, when two auditors work as a relatively stable team over time, the client portfolios of these pairs of auditors tend to be almost the same. This leads to high correlations between the two auditor indicator variables, making it difficult to differentiate the effect of one auditor from that of the other. To mitigate this problem, we examine the correlations between the auditor indicator variables. If the correlation coefficients are higher than 0.70, we discard the auditor with the smaller client portfolio. In case two auditors have identically sized client portfolios, we randomly discard one of the two in the pair. After this procedure, we have 861 individual auditors for the fixed effect estimations. 9 Table 1 shows the descriptive statistics of the dependent (Panel A) and independent variables (Panel B) used in the model for estimating auditor fixed effects. To mitigate the undue influence of outliers, we winsorize all of the continuous variables at the bottom and top yearly percentiles throughout the paper. For variables ARAgg and AbAcc, the means are close to zero because both variables are essentially the regression residual values. However, both variables show considerable variation in the data. The mean of BL is 0.012, suggesting that, on average, the use of below-the-line items increases firms pre-tax ROA by 1.2%. Approximately 11.6% of firms report a ROA between 0 and 1% during the sample period. Panel B reports the time-varying firm characteristics. The values of these variables are reasonably distributed with some degrees of variation. The average ROA is 2.9%, and about 13% of our sample firms have losses. The turnover ratio (sales/total assets) is about 62%. The average leverage is 49%, and the average listing age is slightly below 7 years. The client importance measure for each client firm (CI) is determined by the ratio of the log value of the client s assets to the log value of the total assets of all clients audited by its signing auditors. The mean CI is 0.273, with a standard deviation of Finally, about 52% of client firms are ultimately controlled by local governments in China. (Insert Table 1 here) 9 Our main results are not sensitive to the dropping of auditor pairs with high correlations, which is not surprising given the small number of individual auditors dropped from the sample

20 4.2. Individual Auditor Fixed Effects Main Results Table 2 contains the results of the regressions for estimating Model (1), based on the four audit quality measures presented in Columns (1) to (4). In Panel A, we report the coefficients and t-statistics of the control variables. In all regressions, we include year, client firm, audit firm, and individual auditor indicators. The adjusted R 2 s range between 32.08% (ARAgg regression) and 66.80% (AbAcc regression). This suggests that the control variables and the fixed-effect indicators explain a substantial portion of the variation in the dependent variables. In Panels B to D, we assess the significance of client firm, audit firm, and individual auditor fixed effects, respectively. In addition to the F-statistics that evaluate the joint significance of these fixed-effect indicators, we also examine how these indicators improve the models explanatory power. Specifically, following Collins et al. (1997), we calculate the incremental R 2 that can be attributed to each set of fixed-effect indicators as: R 2 CF = R 2 Full R 2 w/o CF, R 2 AF = R 2 Full R 2 w/o AF, R 2 IA = R 2 Full R 2 w/o IA, (3a) (3b) (3c) where R 2 Full is the adjusted R 2 of the full model including all fixed effects, R 2 w/o CF is the adjusted R 2 of the model that excludes client firm fixed effects and, analogously, R 2 w/o AF (R 2 w/o IA) is the adjusted R 2 of the model without the audit firm (individual auditor) fixed effects. R 2 CF thus represents the incremental explanatory power provided by the set of client firm indicators, while R 2 AF and R 2 IA represent the incremental explanatory power contributed by audit firm and individual auditor fixed effects, respectively. We also scale each R 2 statistic by the adjusted R 2 of the base model to determine the relative percentage increase in R 2 : % R 2 CF = (R 2 Full R 2 w/o CF)/R 2 w/o CF, % R 2 AF = (R 2 Full R 2 w/o AF)/R 2 w/o AF, % R 2 IA = (R 2 Full R 2 w/o IA)/R 2 w/o IA. (4a) (4b) (4c) The F-statistics over the panels suggest that all three sets of fixed-effect indicators are highly significant for all regressions across the columns, except the audit firm fixed

21 effects in the small profit regression (Column (4) of Panel C). As for explanatory power, the client firm fixed effects substantially increase the models R 2 s: R 2 CF ranges from 12.50% in the AbAcc regression to 18.46% in the ARAgg regression, which can be translated into % R 2 CF from 23.01% to %. This suggests that audit reporting decisions or earnings quality measures as proxies for audit quality vary considerably across client firms. It is therefore important to control for firm-specific differences in these measures to provide a stronger test of audit firm/auditor effects. 10 In Panel B, we observe that the inclusion of the audit firm indicators modestly improves the explanatory power of the audit quality models. R 2 AF ranges from 0.63% to 1.77%, and % R 2 AF is between 0.96% and 5.83%. As shown in Panel C, adding auditor effects also improves the explanatory power of the model: R 2 IA ranges from 4.56% to 8.31%, translating to 7.50% to 34.95% in % R 2 IA. Taken together, the results suggest that the client firm, audit firm, and individual auditor fixed effects on audit outcomes co-exist. Although individual auditor effects are not as strong as client firm effects, they appear to be stronger than audit firm effects in terms of explanatory power. We next examine the fixed effects of individual auditors, the crux of our analysis, in more detail. (Insert Table 2 here) While the F-statistics suggest that auditor effects are jointly significant, it is possible that the results are driven by a small number of significant coefficients. We therefore examine the frequency of auditor effects that are significant at the conventional levels. The results are reported in Panel A of Table 3. Under the null hypothesis that individual auditors have no effect incremental to the other variables considered in the regressions, one would expect about 5% (10%) of auditors to have coefficients significant at the 5% (10%) level. The results reveal that the actual percentages of auditors with significant coefficients are much greater than the expected ones. For example, in the case of ARAgg, the percentage of auditor effects that are significant at the 5% (10%) level is 12.9% (18.0%) Indeed, in untabulated analysis, we find that the audit firm/auditor effects are seriously overstated if the client firm fixed effects are not controlled for. 11 Dyreng et al. (2010) report that approximately 12% (17%) of top executives fixed effects are significant at the 5% (10%) level in explaining tax avoidance. In Ge et al. (2010), the actual