Abstract
China Journal of Accounting Studies, 2015 Vol. 3, No. 4, 277–293, http://dx.doi.org/10.1080/21697213.2015.1100088 Audit firm governance of branches and office-level acceptance of initial engagement: Evidence from the CICPA survey data a a b Chunfei Wang ,XiWu * and Tiebing Zeng School of Accountancy, Central University of Finance and Economics, Beijing, China; Registration Department, Chinese Institute of Certified Public Accountants, Beijing, China This study examines the impact of an audit firm’s governance of branches on the initial engagement acceptance by branches. Using the survey data on audit firms’ governance of branches from the Chinese Institute of Certified Public Accountants (CICPA), we find that, compared with branches that are better governed by the audit firm, weakly-governed branches are more likely to accept initial engagement with a prior-year modified audit opinion. Further, for initial engagements with a prior-year modified audit opinion, the tendency of weakly governed branches to continuously issue a modified audit opinion is weaker than better governed branches. Our evidence is consistent with the notion that an audit firm’s internal governance has a positive effect on the effectiveness of audit firm’s quality control. Keywords: audit firm; branches; initial engagement; internal governance 1. Introduction How an audit firm’s internal governance affects the conduct of the auditor and the quality of the audit has been a topic of concern in academic and professional circles in recent years (DeFond & Francis, 2005; EC, 2010; PCAOB, 2011). In the past, there have been some theoretical discussions centring on a firm’s internal governance (Wu & Chen, 2012). There have been relatively few examples of empirical studies based on large samples of archived data, due to limited data (Maijoor & Vanstraelen, 2012). Among these few studies, Deumes, Schelleman, Bauwhede, and Vanstraelen (2012) examine the relationship between a firm’s disclosed internal governance and auditing quality, finding no significant correlation. This paper uses the reported data on audit firms’ internal governance from the Chinese Institute of Certified Public Accountants (CICPA), with the aim of examining the effects of the firms’ governance on auditing conduct and the audit outcome. In China, the audit firms have expanded by establishing branches in situations where there are variable degrees of control in the relationship between the central management of the firm and the operation of the branches. According to the regulation issued by China’s Ministry of Finance, an audit firm’s branch refers to a non-independent branch of a legal person established by the audit firm due to business development needs that uses the audit firm’s name to engage in the business of a certified public accountant. (The control relationships are explained in Section 2.) Specifically, this paper measures the firms’ level of governance over the branches (abbreviated as the ‘branch governance’) and examines the effects of this *Corresponding author. Email: wuxi@cufe.edu.cn Paper accepted by Kangtao Ye. © 2015 Accounting Society of China 278 Wang et al. governance on the initial auditing engagement acceptance and reporting conduct at the branch level. One important consideration for examining the firms’ governance over branches is that, driven by the influence of policy and by market forces in recent years, China’s audit firms have experienced a wave of mergers in which a number of firms have expanded rapidly in size and the number of branches has increased substantially. Within different firms, however, there are differences in the firms’ levels of governance over the branches, and the branches may lack effective constraints and supervision from the firms. For this reason, professional institutions and regulatory agencies have all regarded the governance over branches as an important component of the audit firms’ internal governance and as a key regulatory and supervisory target. At present, how- ever, research centring on the level of branch governance by Chinese firms and its results continues to be rare. Additionally, since 2011, the CICPA has required audit firms to report information related to the governance over branches, providing our study with a good basis of experimental variables. This paper uses the branches’ acceptance of an initial engagement as the research setting to analyse the effects of the governance over branches, considering engagement acceptance as one of the engagement quality control elements belonging to the audit firms and as the starting point of any study on engagement quality. Meanwhile, com- pared with a continuing engagement, an initial engagement acceptance has a greater degree of information asymmetry, and the auditors face higher initial auditing costs and business risks (DeAngelo, 1981). In such a case, the firms’ internal governance mecha- nism may play a more important role. Upon acceptance of an initial engagement, one possibility is that firms with better governance have a greater tendency to avoid high- risk clients; another possibility is that even if firms accept high-risk clients, they still adopt corresponding measures to respond to the risks (for example, by maintaining a strict threshold to issue clean audit opinions). From the perspective of a branch, if the firm has a better (worse) governance over the branch, then the branch may view the risk aversion or risk control of initial engagement acceptance with more (less) importance. We obtained two main findings using the 2011–2012 filing data from certified pub- lic accountants and the publicly available data on listed companies. First, branches that had worse governance were more likely to accept listed clients that were issued modi- fied audit opinions by a previous audit firm compared with branches that had better governance. Second, in terms of the initial acceptance of clients that were issued modi- fied audit opinions by a previous auditor, the tendency for branches that had worse governance to continue to issue modified audit opinions was weaker compared with branches that had better governance. This paper extends the previous literature in a number of ways. First, this paper measures the level of branch governance in audit firms by using the CICPA’s propri- etary data, examines the relationship between the level of branch governance and the auditing quality control results, and attempts to observe the ‘black box’ of firms’ gover- nance over branches, thus adding empirical evidence in the field of audit firms’ internal governance (Deumes et al., 2012). Second, this paper studies the characteristics of branches’ conduct and the factors influencing their daily operations and quality control to expand on the office-level auditing literature (e.g. Francis & Yu, 2009; Reynolds & Francis, 2001). Third, previous studies on the acceptance of an initial engagement by audit firms do not differentiate between the firm level and the office level. This paper differentiates between them and focuses on studying the acceptance of an initial China Journal of Accounting Studies 279 engagement and reporting characteristics at the branch level to expand on the field of auditor change, especially in terms of the literature related to decision-making pertain- ing to acceptance of an initial engagement and response to risk (e.g. Johnstone, 2000; Johnstone & Bedard, 2003). This paper also has practical and policy implications. Regulators and researchers have expressed concerns about inferior audit quality by Chinese audit firms despite such forceful regulatory efforts as waves of audit firm mergers since 1999 (Ke, 2014; Liu, Liu, & Wang, 2010). Our evidence shows that when audit firms had worse gover- nance of the branches, the results of engagement quality control on branches were worse. This relationship implies that the increased management of branches by audit firms will reduce their overall risk and improve audit quality; simultaneously, this rela- tionship also implies that it is necessary for professional organisations and regulatory agencies to pay close attention to branch governance and branches’ audit quality. Section 2 develops the study’s hypotheses. Section 3 explains the original informa- tion on branch governance and the data processing methods used in this paper. Sec- tion 4 introduces the study design. Section 5 reports the results of the empirical analysis. Section 6 concludes. 2. Hypotheses development 2.1. The development of branches and the firms’ branch management status in China’s audit market Under the guidance and promotion of China’s certified public accountant professional organisations and governmental authorities, China’s audit market experienced a wave of audit firm mergers over the last decade. Many firms expanded rapidly in size by establishing branches or absorbing other audit firms to create branches. Table 1 lists the branch development situation of China’s major audit firms in recent years. In this table, the statistical data (Column A) based on the top 100 accounting firms in the country show that each firm had an average of 1.62 branches in 2003 but that each firm’s average number of branches increased to 5.18 in 2009. The analysis of the audit firms Table 1. The branch development situation of China’s major audit firms. Column A: Top 100 Column B: CSRC-licensed audit accounting firms firms Year N Average no. of branches N Average no. of branches 2003 100 1.62 68 1.76 2004 100 2.04 67 2.13 2005 NA NA 2006 100 3.38 59 3.53 2007 100 4.01 59 4.24 2008 100 4.38 56 5.45 2009 100 5.18 51 7.43 2010 NA 47 8.66 2011 NA 46 9.30 2012 NA 44 10.52 2013 NA 40 12.45 Notes: In Column A, the data are from the audit firms of the top 100 ranking publicly disclosed by the CICPA during 2004 and 2010. In Column B, the data of 2003–2009 are limited to the CSRC-licensed audit firms of the top 100 ranking by the CICPA and the data of 2010–2013 are from the comprehensive evaluation database of the CICPA. NA means that the data is not disclosed in a given year. 280 Wang et al. licensed by the China Securities Regulatory Commission (CSRC) (Column B) shows that the branches of the CSRC-licensed audit firms have a larger size, and the increase in their number of branches in recent years is also very apparent, increasing from an average of 1.76 branches in 2003 to an average of 12.45 branches in 2013. As the number of branches increased, the firms’ management and governance over branches also faced challenges from many perspectives. Wang, Liao, Liu, Zeng, and Yang (2007) noted that obtaining various domestic professional qualifications and bidding on major projects were linked to the size of the audit firm, leading firms to establish branches or absorb other audit firms to create branches to meet policy needs. Although this type of exogenous development model was favourable to the firms in terms of achieving size targets in the short term, it lacked the integration of ideas and resources between the firms and the branches, easily causing shock to the firms’ inter- nal governance principles (especially the principle of collective selection and arrange- ment) (Wu & Chen, 2012). Liu et al. (2010) indicated that the firms lacked effective control over their branches and basically had even lost the control rights. Although they might be one legal entity in form, they were discordant in reality, and the phe- nomenon in which each did as it pleased in its own manner was more prevalent. After examining the intermediary organisations in the securities market in 2013, the CSRC also noted that the first main problem of audit firms involved inadequate quality control systems and mechanisms in some of the firms, meaning that they failed to achieve inte- gration in the firms’ branch management (Accounting Messenger, 8 October 2014). Prior empirical evidence has also shown that the branches’ professional quality in auditing might be lower than that of the firms (Wang & Xin, 2010; Wu, Zeng, & Wang, 2014). 2.2. Institutions and regulations related to the branch governance The policymakers and regulators of China’s Certified Public Accountant profession have paid a high degree of attention to the governance of the branches. For example, the Ministry of Finance issued the ‘Interim Measures for the Examination, Approval, and Supervision of Accounting Firms’ (abbreviated below as ‘Interim Measures’)on15 January 2010, requiring audit firms to achieve substantive unity between the firms and the branches in terms of personnel, finance, business, technical standards, and informa- tion management. Moreover, Liu et al. (2010) elaborated the requirements, as sum- marised below: (1) Personnel. Human resources are a firm’s most important resource. A firm’s implementation of a unified personnel system helps to ensure that the level of professionalism and employee quality of the firm and the branches are roughly equal, thereby enhancing the professional competence of the branch employees and compensating for ‘innate deficiencies’. (2) Finance. The core of substantive unity lies in finance. Financial unity is the most effective measure for avoiding discordance; the firm and the branches divide revenue and expenditures and hold each responsible for balancing its budget, with each thus overseeing its own fiefdom. The unified management of the firm’s and the branches’ revenue and expenditures can greatly lessen the branches’‘free rider’ behaviour in reputation maintenance and reduce the con- flicts of interests and contradictions between the firm as a whole and its branches. China Journal of Accounting Studies 281 (3) Business and technical standards. Business management mainly involves engagement acceptance and project implementation, and technical standards are the norms of business implementation. A firm establishes unified business man- agement and professional standards according to its own service capabilities and risk tolerance, which can effectively prevent branches from exceeding their authorisation in accepting engagements or independently undertaking engage- ments that are clearly beyond their professional capabilities, thereby lowering the overall audit risk. (4) Information management. Information management comprehensively supports the above and is an important means for substantive unity. In terms of government regulation, in 2009, the Ministry of Finance Supervision and Inspection Bureau issued a ‘Notice from the Ministry of Finance on Further Improving the Administrative Supervision Work at Securities Qualified Accounting Firms’ (Cai Jian [2009] No. 6) to strengthen the supervision of all branches of securi- ties qualified audit firms. In terms of administration of the profession, since 2011, the CICPA has attempted to refine the ‘substantive unity’ between firms and branches stip- ulated in the ‘Interim Measures’ through the audit firm Comprehensive Excellence Evaluation System, which requires firms to fill out information on relevant indices as an important basis for ranking the top 100 firms in the nation. 2.3. Hypotheses of the study Jensen and Meckling (1992) indicate that the optimal level of centralisation/decentrali- sation can be determined by the organisation’s trade-off of costs due to the lack of information and costs due to inconsistent objectives. In other words, the optimal alloca- tion of decision-making power within an organisation balances information and agency costs. Considering the issue of a firm’s branch management in centralised/decentralised decision-making from the perspective of information costs, given that the branches enjoy special knowledge of clients and market segments, the firm needs to consider granting some autonomy to the branches in order to lower the firm’s information costs. Next, the issue of agency between the firms and the branches is discussed. An audit firm’s branch, as a non-independent branch of a legal person, joins with the firm’s name to contract externally and perform auditing; that is, the branch and the firm share the overall reputation of the firm. Meanwhile, as a non-independent branch of a legal person, the branch cannot independently bear external legal liability; therefore, the entire firm ultimately bears the branch’s professional potential for reputation loss and legal liability. Moreover, given that the branch conducts business in the name of the firm, the implications is that the branch also belongs to the firm as an agent established to conduct business; therefore, the branch has a considerable degree of autonomy in accepting clients, conducting business, and submitting reports (Choi, Kim, Kim, & Zang, 2010; Reynolds & Francis, 2001) and would thus produce a considerable degree of information asymmetry in relation to the firm. A branch may enjoy most of the rev- enue from its clients, but the entire firm bears the potential for reputation loss and legal liability; thus, the branch has the motivation to compromise with clients to earn revenue from them and sacrifice the reputation of the entire firm. In particular, the establishment of branches in China’s securities audit market is, to a considerable degree, utilised for expanding the firm’s size in the short term and improving the firm’s industry ranking (Wang et al., 2007). Many firms that originally did not have professional qualifications 282 Wang et al. in securities have also entered the listed company auditing market through mergers; they have shortcomings in both professional competence and independence. In reality, a firm’s brand reputation is, to a very great degree, accumulated over many years by the firm. Once a problem occurs, the loss of reputation suffered by the firm is far greater than that suffered by the branch, and the branch’s resources can join other firms relatively easily. Accordingly, a firm may adopt various types of governance measures that target branches to reduce the firm’s internal agency costs and to ensure, as much as possible, that branches’ business risks and quality are controllable. The stricter a firm’s manage- ment and supervision over a branch, the more difficult it is for the branch to simply consider its own economic interests; thus, the branch may be more likely to show the characteristics of risk aversion in terms of engagement acceptance. The previous litera- ture finds that a modified audit opinion can reflect a higher risk in surplus management (e.g. Francis & Krishnan, 1999) and indicates that the company may have a higher existing risk in continuing operations (e.g. Butler, Leone, & Willenborg, 2004); the audit client is often motivated to purchase audit opinions (Lennox, 2000). In practice, professional organisations and regulatory agencies have long been concerned with the purchase of audit opinions by changing the auditor, and they have made a listed com- pany’s change in audit firm a key target of regulatory inspection. Accordingly, if a firm’s branch accepts listed clients from other firms, then the probability that regulators will check the firm is increasingly higher. In particular, when the previous auditor has issued modified audit opinions to the listed company, the risk of supervision is especially prominent. Therefore, we propose hypothesis H1: H1: Branches that have worse governance are more likely to accept listed clients that were issued modified audit opinions by a previous auditor compared with branches that have better governance. The previous literature has found that if the audit firm itself intends to adopt an appro- priate risk response plan, accepting high-risk clients remains possible (Johnstone, 2000; Johnstone & Bedard, 2003). Given that different branches have already accepted clients with prior-year modified audit opinions, with all other conditions being equal, when the branch governance is better, the branches’ attitude to audit reporting may be more cau- tious. Moreover, branches with worse governance may not do as well as branches with better governance in terms of the willingness and capabilities for quality control, and they also cannot adopt appropriate risk response plans (including being less prone to issuing modified audit opinions to this type of client). Accordingly, we propose hypothesis H2: H2: In terms of the initial acceptance of listed clients, branches that have worse gover- nance are less likely to continue to issue modified audit opinions on clients with prior-year modified audit opinions compared with branches that have better governance. 3. Measurement of the firms’ branch governance levels 3.1. Description of the raw data In 2010, the CICPA established a Comprehensive Excellence Evaluation System, and the firms have filled out the reports on the branches’ management indices since 2011. In reference to the ‘Interim Measures’ issued by the Ministry of Finance, the system is established with the five first-order indices shown in Table 2: financial unity, personnel China Journal of Accounting Studies 283 unity, technical standard unity, business unity, and comprehensive examination. Under some circumstances, the second-order indices under the first-order indices are designed to be two either-or reverse propositions. For example, the expenditure index under the financial unity index includes ‘the branch’s expenditures are arranged by the branch itself’ and ‘the branch’s expenditures are under unified arrangement by the firm’.To avoid double counting, we only select indices that reflect looser branch governance, that is, ‘the branch’s expenditures are arranged by the branch itself’. If a certain firm reports the index to be 1, this value indicates that the firm has less managerial and supervisory authority in this area and that the firm’s level of substantive unity manage- ment over the branch is relatively weak. The system also has three second-order indices that jointly measure the conditions of matters in the same dimension. For example, ‘the branch’s major management sys- tems are thoroughly investigated and approved by the firm’, ‘the branch’s major man- agement systems are formulated by the branch itself’, and ‘the branch’s major management systems are under unified formulation by the firm’ jointly reflect the degree of the firm’s intervention into the branch’s major management systems; we select the first two indices. Similarly, with regard to the firm’s frequency of examina- tion of the branches, the system has also established three propositions (once a year, once every 2 years, or once every 3 years); we select the last two indices. In addition, for some indices established by the system, the second-order index values observed in the sample firm reports are all 1 or 0, lacking differences; these values are excluded. Because the system requires the firms to self-report relevant information, it is easy for self-overestimation tendencies to exist. The implication is that firms are aware of the close attention that professional organisations and regulatory agencies pay to the governance over branches and that the firms have given compliance replies to certain proposition presentations designed by the system. Concerned with the reliability of the survey data from professional organisations, we conducted a confidence test. The results indicate that the Cronbach’s α reliability coefficient was 0.87, which shows that the measurement results are credible. Table 2 indicates that between 2011 and 2012, 26.6% (28.7%) of the branches arranged their own financial expenditures (revenues). In terms of personnel, 35.1% of the branches recruited their own staff, and 2.1% of the heads of branches were desig- nated by the branches themselves and reported to the firms for approval. In terms of technical standards, the major management systems were thoroughly investigated and approved (under unified formulation) by the firm in 19.1% (78.8%) of the branches and were formulated by the branches themselves in 2.1% of the branches. In terms of busi- ness, 2.1% of the branches accepted major businesses (such as listed company busi- nesses) themselves. In terms of comprehensive examination, 87.2% of the firms examined branches once per year (with 8.5% every two years, and 4.3% every three years). Despite the small reported differences in many second-order indices, the afore- mentioned statistics still show some differences in the governance over branches. 3.2. Further processing of the raw data on the governance over branches Considering the relatively numerous items of governance over branch information col- lected by the CICPA system and the high level of conceptual relevance, we used factor analysis to reduce the dimensions and build the indices for the governance over branches. Prior to factor analysis, we first performed the Kaiser-Meyer-Olkin (KMO) test and Bartlett’s sphericity test. The results showed that the coefficient of the KMO 284 Wang et al. Table 2. The index of the audit firm governance over branches. First-level Index Second-level index Variables N % Financial unity Branch revenue controlled by individual branches X1 94 26.6 Branch expenses controlled by headquarters – Branch revenue controlled by individual branches X2 94 28.7 Branch revenue controlled by headquarters – Branches pay management fees X3 94 20.2 Personnel unity Branch staff recruited by individual branches X4 94 35.1 Branch staff recruited by headquarters – Branch chief executive officers approved by X5 94 2.1 headquarters Branch chief executive officers appointed by – headquarters Technical Major branch management systems approved by X6 94 19.1 standard unity headquarters Major branch management systems established by X7 94 2.1 individual branches Major branch management system imposed by – headquarters Business unity Acceptance of major engagements (such as the X8 94 2.1 engagements of the listed company) undertaken by individual branches Acceptance of major engagements (such as the – engagements of the listed company) undertaken by headquarters Comprehensive Branches inspected by headquarters once a year – examination Branches inspected by headquarters once every two X9 94 8.5 years Branches inspected by headquarters once every three X10 94 4.3 years test was 0.54 (>0.5) and the Bartlett’s sphericity test Chi statistic was 3627.03 (p = 0.000), indicating that the data reported to the CICPA are suitable for use in factor analysis. Subsequent steps are as follows. First, we calculate the correlation matrix eigenvalues and factor contribution rate and determine the number of common factors. The factor contribution rate represents the ratio of each factor’s level of variability accounting for the level of variability for all factors. When the cumulative contribution rate reaches 85% or higher or when the eigenvalue is less than 1, then the number of common factors has been determined. Panel A in Table 3 shows that the first three factors’ eigenvalues are greater than 1 and that the cumulative contribution rate is 94.22%; these values are the basis for selecting the first three factors as the common factors. Second, we rotate the factors and load the matrix. We use principal component analysis to solve the corresponding estimation matrix and use variance maximisation rotation in the orthogonal rotation. Panel B in Table 3 shows that Factor 1 mainly cor- relates with the finance and personnel indices (X1-X4); Factor 2 mainly correlates with the technical standard and business indices (X7-X8); and Factor 3 mainly correlates with the comprehensive examination indices (X9-X10). Third, we calculate the branch governance indices based on the weighted factor scores. We use the load matrix in the second step to estimate the score of each China Journal of Accounting Studies 285 Table 3. Factor analysis results (N = 94). Panel A: Eigenvalue and variance contribution ratio Factor Eigenvalue Proportion (%) Cumulative (%) 1 2.346 35.75 35.75 2 2.166 33.00 68.75 3 1.672 25.46 94.22 4 0.364 5.54 99.76 5 0.360 5.49 105.25 Panel B：Factor loadings (pattern matrix) and unique variances Variables Factor 1 Factor 2 Factor 3 X1 0.936 0.132 0.117 X2 0.922 0.137 0.046 X3 0.499 −0.115 −0.070 X4 0.524 0.145 −0.059 X5 0.225 −0.055 0.194 X6 0.117 −0.082 −0.038 X7 0.106 0.994 0.062 X8 0.106 0.994 0.062 X9 0.056 −0.102 0.916 X10 −0.082 −0.318 −0.873 Note: bold indicates the larger loadings for each factor. observed principal factor, and we then use the proportion of the factor contribution rate to the cumulative contribution rate as a weight when calculating the factor scores, thereby obtaining each observed branch governance index. The firms and branches from firms with higher branch governance indices have worse substantive unity condi- tions in terms of finance, personnel, business, and technical standards and have worse branch governance levels. The previous literature has typically used the Big Four international audit firms or China’s Big 10 firms to distinguish the firms’ reputation and size. Untabulated analysis shows that the correlation coefficients of the firms’ revenue size index for the Big Four and Big 10 firms are 0.83 and 0.71, respectively, and that the correlation coefficient between the branch governance index and the firms’ revenue is only –0.25. The impli- cation is that a significant difference exists between the level of branch governance that we measured and the firms’ size. It is possible that incremental information from different dimensions was measured. 4. Research design 4.1. Test model for H1 We used the following multinomial regression model to test our hypothesis H1: INITIAL ¼ d þ d MOD þ Controls þ e (1) tþ1 0 1 t When the dependent variable of Model (1), INITIAL , is 0, this value indicates that t+1 an audit service undertaken by a branch undergoes continuous auditing; when it is 1, this value indicates that a client changed auditors during period t + 1 and that a branch from a firm with a better governance over its branch has taken over the client; and 286 Wang et al. when it is 2, this value indicates that a client changed auditors during period t + 1 and that a branch from a firm with a worse governance over its branch has taken over the client. The traditional literature on auditor changes typically uses a dichotomous vari- able (i.e. whether a change in auditor occurred) as the dependent variable of an auditor change decision-making model; however, we take a multiple non-sequencing-dependent variable design approach, primarily to take into account our expectation in examining the audit clients’ change in auditors that when they made the decision, they tended to hire a branch with comparatively better or worse branch governance. Thus, there were three possible decision-making outcomes (i.e. no change in auditor, a change to a branch with a better form of branch governance, or a change to a branch with a worse form of branch governance). Given that the objective of this paper’s analysis is limited to the auditing undertaken by branches, we must identify the audit engagements undertaken by the branches. We used the certified public accountant information in the CICPA public searching system to determine whether the certified public accountant signing the annual report belonged to a branch or to a firm, and we identified the following four separate conditions: (1) the signing certified public accountants were all from a firm; (2) the signing certified public accountants were all from a branch; (3) the two signing certified public accountants were separately from a firm and a branch; and (4) the two signing certified public accountants were from different branches. When any of the signing certified public accountants was from a branch, we identi- fied the corresponding listed company’s annual audit report as auditing undertaken by a branch. When designing the sample, we required the listed company to have a branch undertake the service in period t and in period t + 1. When the listed company had changed the branch of the responsible auditor in period t + 1, we then identified the branch responsible for the audit in period t + 1 as belonging to an entity with better branch governance (i.e. the branch governance index is less than or equal to the median at the firm level, at this time INITIAL = 1) or worse branch governance (i.e. the t+1 branch governance index is greater than the median at the firm level, at this time INITIAL = 2). When performing multiple non-sequencing regression, we used those t+1 companies observed to have used the same branch for auditing during both period t and period t + 1 (INITIAL = 0) as the reference group. t+1 The experimental variable of this paper is MOD . When it is 1, this value indicates that the client received a modified audit opinion during period t; otherwise, this value is 0. This differentiation has already been discussed above. If a listed company received a modified opinion during period t and changed the presiding audit firm, then this action in itself implies a higher risk for the purchase of an audit opinion and creates a higher regulatory risk for the subsequent firm. According to hypothesis H1, compared with branches with better governance, branches with worse governance are more likely to undertake auditing for clients that received a prior-year modified opinion. Therefore, we expect the coefficient of MOD when INITIAL = 2 to be significantly greater t t+1 than the coefficient of MOD when INITIAL =1. t t+1 The control variables of Model (1) reference the previous literature related to audi- tor change (Huang & Zhang, 2010; Lennox, 2000), including the pre-change auditor’s characteristics and the audit client’s characteristics. The auditor’s characteristics include the overall size of the firm (when BIG10 is 1, this value indicates that the audit was China Journal of Accounting Studies 287 performed by a branch of the top 10 audit firms, as comprehensively ranked by CICPA annually; otherwise, this value is 0) and the size of the branch (OFFICESIZE = natural logarithm of the branch’s revenue). The audit client’s characteristics include the pre-change return on assets (ROA = net profit/net assets), whether there were losses (when LOSS is 1, this value indicates that there was loss that year; otherwise, this value is 0), asset-liability ratio (LEV = total liabilities/total year-end assets), company size (LTA = natural logarithm of total year-end assets), and the nature of the ultimate controller (when STATE is 1, this value indicates that the ultimate controller is state-owned; otherwise, this value is 0). In addition, we also controlled for annual and industry fixed effects. 4.2. Test model for H2 We used the following binary logistic model to test this study’s hypothesis H2: MOD ¼d þd MOD þd MOD WEAKGOV þd WEAKGOV þControlsþe tþ1 0 1 t 2 t tþ1 3 tþ1 (2) To test H2, the scope of analysis for Model (2) is limited to the branches’ initial engage- ment acceptance. When the dependent variable MOD is 1, this value indicates that t+1 the listed client that changed firms received a modified audit opinion from the subse- quent firm in the initial audit; when it is 0, this value indicates that the listed client that changed firms received an unqualified audit opinion from the subsequent firm in the ini- tial audit. In accordance with hypothesis H2, we pay close attention to the initial audit after receiving a prior-year modified opinion and whether the branches with worse gov- ernance show a looser audit attitude, i.e. the coefficient δ of MOD *WEAKGOV , 2 t t+1 wherein when WEAKGOV is 1, this value indicates that the branch responsible for t+1 the audit in period t + 1 belongs to a firm with worse governance over branch (i.e. the branch governance index is greater than the median). If δ is significantly negative, the implication is that after a branch with worse governance accepts a high-risk client, not only did it fail to provide more adequate accompanying risk response measures but it also showed a looser audit reporting standard to such a client. Conversely, if there is no significant difference between δ and 0, or even if δ is significantly greater than 0, the 2 2 implication is that although a branch with worse governance over branch has a greater tendency to accept a high-risk client, the branch shows a similar or even stricter audit reporting standard than does a branch with better governance over branch after it accepts the engagement; therefore, it may not bring higher risk to the firm overall. 4.3. Sample selection Considering that, currently, we can only obtain the 2011 and 2012 governance over branch-related variables on audit firms, the scope of our sample is limited to the observations for the 2011–2012 listed companies audited by the firms’ branches. In addition to the branch governance data, the other financial data were from the China Stock Market and Accounting Research (CSMAR) database. We also excluded finan- cial companies and sample observations that had missing data on relevant variables. In the end, 2,016 sample observations were obtained between 2011 and 2012. We winsorized all continuous variables at the 1% and 99% percentiles to mitigate the effect of outliers. 288 Wang et al. 5. Empirical results 5.1. Descriptive statistics Table 4 lists the descriptive statistics for the model variables. During the sample period, 123 companies (accounting for 6.1%) changed auditors in period t + 1 and were audited by better governed branches, while 46 companies (accounting for 2.3%) changed audi- tors in period t + 1 and were audited by worse governed branches. In the sample, 7.4% of the observed companies had losses in period t, 5.1% of the companies were issued modified audit opinions, the ultimate controller of 39.2% of the companies belonged to state-owned, and 42.4% of the companies were audited by the Big 10 firms. Table 5 lists the univariate analysis results of Model (1). Table 5 shows that in the observation group prior to the branch change (INITIAL = 0), the observed proportion t+1 of companies that received a prior-year modified audit opinion (MOD = 1) is 4.6%; when there was a change in branch to a branch with better governance over branch (INITIAL = 1), the observed proportion of companies that received a prior-year mod- t+1 ified audit opinion is 7.3%; and when there was a change in branch to a branch with worse governance over branch (INITIAL = 2), the observed proportion of companies t+1 that received a prior-year modified audit opinion is 17.3%. The statistical test indicates that the proportion of the INITIAL = 2 observation group that received a prior-year t+1 modified audit opinion is significantly higher than those of the INITIAL = 0 observa- t+1 tion group (p < 0.01) and the INITIAL = 1 observation group (p = 0.053), and there t+1 is no significant difference (p = 0.172) for the above between the INITIAL =0 t+1 observation group and the INITIAL = 1 observation group. The univariate analysis t+1 results in Table 5 are consistent with hypothesis H1. Table 4. The descriptive statistics for the model variables. Variables N Mean Sd. Min Median Max INITIAL 2016 0.107 0.375 0.000 0.000 2.000 t+1 MOD 2016 0.051 0.219 0.000 0.000 1.000 BIG10 2016 0.424 0.494 0.000 0.000 1.000 OFFICESIZE 2016 19.874 1.046 16.365 20.004 21.809 ROA 2016 0.045 0.054 −0.174 0.044 0.201 LOSS 2016 0.074 0.262 0.000 0.000 1.000 LEV 2016 0.447 0.245 0.031 0.446 1.143 LTA 2016 21.666 1.281 19.082 21.480 25.735 STATE 2016 0.392 0.488 0.000 0.000 1.000 Variable definitions： INITIAL = 0 if an audit service undertaken by a branch undergoes continuous auditing; t+1 1 if a client changed auditors during period t+1 and a branch from a firm with a better governance over branch has undertaken the audit; and 2 if a client changed auditors during period t+1 and a branch from a firm with a worse governance over branch has undertaken the audit. MOD = 1 indicates that the client received a modified audit opinion during period t; and 0 otherwise. BIG10 = 1 if the audit is conducted by a Big 10 audit branch and 0 otherwise. OFFICESIZE = the natural logarithm of branch’s revenue. ROA = net income for year t / total assets at year t. LOSS = 1 indicates that there was loss that year; otherwise, this value is 0. LEV = year-end total liabilities / year-end total assets. LTA = the natural logarithm of total assets at the year t. STATE = 1 if the client is a state-owned enterprise, and 0 otherwise. t China Journal of Accounting Studies 289 Table 5. Univariate results of Model (1). N % (MOD =1) INITIAL = 0 1847 4.6% t+1 INITIAL = 1 123 7.3% t+1 INITIAL = 2 46 17.3% t+1 z-stat. p-value INITIAL = 2 vs. INITIAL = 0 3.963 0.000 t+1 t+1 INITIAL = 2 vs. INITIAL = 1 1.932 0.053 t+1 t+1 INITIAL = 1 vs. INITIAL = 0 1.367 0.172 t+1 t+1 See Table 4 for variable definitions. 5.2. Test results of hypothesis H1 Table 6 lists the multinomial regression results of Model (1). We used the companies that did not change auditors as a frame of reference. The test results indicate that when the dependent variable INITIAL = 1, the coefficient of MOD is –0.040, which is t+1 t not significantly different from 0 (p = 0.930), and when the dependent variable INITIAL = 2, the coefficient of MOD is 1.290, which is significantly greater than 0 t+1 t Table 6. The audit firm governance over branches and acceptance of initial engagement. Dep. Var: INITIAL = 1 INITIAL =2 t+1 t+1 Coeff. Coeff. (z-stat.) (z-stat.) Experimental variable MOD −0.040 1.290** (–0.09) (2.15) Control variables BIG10 0.576** −0.016 (2.05) (–0.04) OFFICESIZE −0.321*** −0.324** (–2.65) (–2.12) ROA −0.780 −1.269 (–0.31) (–0.33) LOSS 1.213*** 0.272 (3.08) (0.38) LEV −0.433 −0.055 (–0.85) (–0.07) LTA 0.025 0.006 (0.24) (0.03) STATE −0.049 −0.464 (–0.23) (–1.50) Constant 3.018 2.670 (1.08) (0.57) Year-fixed effects Included Included Industry-fixed effects Included Included N 2,016 Pseudo R 0.053 ***,**, and *represent 0.01, 0.05, and 0.10 significance levels, respectively (two-tailed). See Table 4 for variable definitions. 290 Wang et al. (p = 0.031); there is also a significant difference (p = 0.068) between the coefficient of MOD when the dependent variable INITIAL = 1 and that when INITIAL =2. t t+1 t+1 This result implies that compared with branches with better governance, branches with worse governance are more likely to accept listed clients that were issued modified audit opinions by previous auditors, supporting hypothesis H1. With regard to the control variables, we find that the coefficient of OFFICESIZE is significantly negative (p < 0.01), implying that the larger the size of the branch, the less likely it is that the audit clients will change auditors. When the dependent variable INITIAL = 1, the coefficients of BIG10 and LOSS are significantly greater than 0, t+1 t t showing that when the overall size of the firm hired by the audit client is larger or when the audit client suffers losses, the client is more likely change to a branch with a better governance. 5.3. Test results of hypothesis H2 Table 7 lists the regression results of Model (2). Table 7 shows that the coefficient of MOD is significantly positive (p < 0.01). This result implies that the modified audit opinion for high-risk clients initially accepted by branches with better governance would retain significant persistence. The coefficient of the interaction term, MOD *WEAKGOV , is significantly negative (p < 0.05), showing that after branches t t+1 with worse governance accepted high-risk clients, they did not display a stricter audit reporting standard than branches with better governance but instead were looser. Regarding the control variables, they are consistent with the previous literature. The regression results show that when the listed companies have higher profitability levels and a larger size and when the actual controller is state-owned in nature, the companies are less likely to receive modified opinions, while companies with higher debt levels are more likely to receive modified opinions. Table 7. Regression results of Model (2). Dep. Var: MOD Coeff. z-stat. t+1 Experimental variables MOD 8.146*** 4.31 MOD *WEAKGOV −3.396** −2.02 t t+1 WEAKGOV 2.117 1.59 t+1 Control variables BIG10 0.576 0.27 t+1 OFFICESIZE −0.106 −0.23 t+1 LOSS −0.704 −0.41 t+1 ROA −37.829*** −3.19 t+1 LEV 6.698*** 2.96 t+1 LTA −1.080*** −3.89 t+1 STATE −3.842* −1.95 t+1 AR_INV −7.940*** −2.79 t+1 Constant 17.359* 1.85 Year-fixed effects Included Industry-fixed effects Included N 272 Pseudo R 0.729 ***,**, and *represent 0.01, 0.05, and 0.10 significance levels, respectively (two-tailed). WEAKGOV = the governance over branch index is greater than the median at the audit firm level. AR_INV = The sum of accounts receivable and inventory divided by total assets at the fiscal year end. See Table 4 for the definition of the other variables. China Journal of Accounting Studies 291 5.4. Robustness analysis 5.4.1. Consideration of mandatory changes The new clients accepted by audit firms may also derive from mandatory rotation by the State-owned Assets Supervision and Administration Commission of the State Coun- cil or from newly appointed audit firms. To avoid the possible interference caused by such situations, we excluded mandatory firm change situations. The untabulated test shows that the main results remain qualitatively unchanged. 5.4.2. Exclusion of Big Four observations from the sample In recent years, the highlighted wave of accounting firm mergers in China occurred domestically within a non-Big Four scope. Therefore, there may be relatively large dif- ferences in the governance over branches between Chinese local audit firms and the Big Four international firms (Ernst & Young, KPMG, Deloitte Touche Tohmatsu, and PricewaterhouseCoopers). In the robustness test, we excluded the Big Four audits from the sample. The main conclusions remain qualitatively similar. 5.4.3. Different definition of branch auditing As noted above, in determining whether a certain listed client’s auditing belongs to a firm or a branch, we define branch auditing as when any of the signing certified public accountants is from a branch. In the robustness test, we require that the signing certified public accountants all come from branches. The main conclusions did not change significantly. 6. Conclusion This paper used the branch governance survey data reported to the CICPA and studied the relationship between the governance over branch and the branches’ acceptance of initial audit engagements. We found that compared with branches with better gover- nance, branches with worse governance were more likely to accept listed clients that had been issued modified audit opinions by predecessor audit firms. Furthermore, the tendency of branches with worse governance to continue to issue modified audit opin- ions on listed clients that received pre-change modified audit opinions was weaker than that of branches with better governance. The evidence implies that the risk prevention awareness of branches with worse governance is weaker than that of branches with better governance. The evidence pre- sented in this paper shows the effect of an audit firm’s level of branch governance on the branches in terms of engagement acceptance and implementation and also supports the high level of attention paid to the internal governance of audit firms by regulatory agencies and professional organisations. Acknowledgements We are grateful for helpful comments from two reviewers, Qingquan Xin (discussant), Kangtao Ye (Associate editor), Liansheng Wu (Joint-editor), Pauline Weetman (language editor), Jason Xiao (Joint-editor), and the CJAS conference participants at Central University of Finance and Economics (June 2015). The names of authors are ordered alphabetically. We acknowledge financial support from the National Natural Science Foundation of China (71302131), the MOE 292 Wang et al. Project of Humanities and Social Sciences (13YJC630160), the Postdoctoral foundation (2015M570125), the Beijing Municipal Commission of Education ‘Joint Construction Project’, the Beijing Municipal Commission of Education ‘Pilot Reform of Accounting Discipline Cluster- ing’, and CUFE Innovative Research Team Project ‘Empirical Accounting and Auditing’. Part of the data used in this study is taken from the Accounting Firm Assessment Database of the CICPA. The third author, Tiebing Zeng, declares that the opinions expressed in this paper are his personal views and do not represent any view or opinion of the CICPA. Disclosure statement No potential conflict of interest was reported by the authors. Funding National Natural Science Foundation of China10.13039/501100001809 [grant number 71302131]; China Education Ministry Social Science Research Project [grant number 13YJC630160]. Notes 1. In 2009, the General Office of the State Council forwarded the Ministry of Finance’s ‘Notice and a Number of Opinions on Accelerating the Development of China’s Certified Public Accountant Profession’ (Guo Ban Fa [2009] No. 56), encouraging domestic firms to be ‘bigger, stronger’. 2. For example, on 15 April 2002, Article II of the ‘Urgent Notice from the Chinese Institute of Certified Public Accountants on Improving the 2001 Annual Auditing Report on Listed Companies’ stipulated that the audit firms changed by listed companies should report the circumstances of the change in writing to the CICPA before the end of April and add relevant content to the subsequent guideline formulation and revision. 3. In the ‘Professional Ethical Code of Chinese Certified Public Accountants Guiding Opinion’ released on 25 June 2002, changes in audit firms were regulated: Prior to accepting an audit commission, the successor audit firm should inquire of the predecessor audit firm the reason for the audit client’s change in audit firm and pay close attention to the possible existing dis- agreements on major accounting and auditing issues between the predecessor auditor and the client. In ‘Independent Auditing Specific Standard No. 28 – Communication with Predeces- sor Auditors’, the content of the successor auditor’s inquiry to the predecessor auditor should be rational and specific and typically include the reasons that the predecessor auditor thinks might make the client change audit firms. Article IX in the ‘Chinese Certified Public Accountants Auditing Standard No. 1152 – Communication with Predecessor Auditors’, revised on 15 February 2006, still retains the same requirement. 4. The system does not provide indices related to information management unity. However, the system provides comprehensive examination indices on each matter under unified management. 5. The paired indices with all responses as ‘yes’ or ‘no’ include the following: (1) the firm can perform a unified allocation of the branch’s staff (yes), or the firm cannot perform a unified allocation of the branch’s staff (no); (2) the branch’s quality control is under unified imple- mentation by the firm (yes), or the branch’s quality control is implemented by the branch itself (no); and (3) the quality control review and report issuance on the auditing of listed companies are implemented by the firm (yes), or the quality control review and report issu- ance on the auditing of listed companies are implemented by the branch (no). 6. For example, the first factor’s factor contribution rate is 0.3575, and the total cumulative con- tribution rate is 0.9422; thus, the first factor’s weight is 37.94% (=0.3575/0.9422). 7. The website: http://cmis.cicpa.org.cn/cicpa2_web/public/query0/2/1.shtml. 8. Most of the control variables of Model (2) are the same as those of Model (1). In addition, the accounts receivable and inventory ratio (AR_INV = (accounts receivable + inventory)/ total year-end assets) is further controlled based on the previous literature. If this variable is uncontrolled, then our main results are unaffected. China Journal of Accounting Studies 293 References Accounting Messenger. (2014, October 8). Agency is frequently fined: Why is it so difficult to practise independently? http://ckxb.cn/bznr/Bd/20140929a5.html (In Chinese.) Butler, M., Leone, A., & Willenborg, M. (2004). An empirical analysis of auditor reporting and its association with abnormal accruals. Journal of Accounting and Economics, 37, 139–165. Choi, J.-H., Kim, C., Kim, J.-B., & Zang, Y. (2010). Audit office size, audit quality, and audit pricing. Auditing: A Journal of Practice & Theory, 29:73–97. DeAngelo, L. (1981). Auditor size and audit quality. Journal of Accounting and Economics, 3, 183–199. DeFond, M. L., & Francis, J. R. (2005). Audit research after Sarbanes-Oxley. Auditing: A Journal of Practice & Theory, 24 (Supplement): 5–30. Deumes, R., Schelleman, C., Bauwhede, H. V., & Vanstraelen, A. (2012). Audit firm governance: Do transparency reports reveal audit quality? Auditing: A Journal of Practice & Theory, 31, 193–214. European, Commission (EC). (2010). Audit policy: Lessons from the Crisis. Green Paper. Final. Brussels. Francis, J. R., & Krishnan, J. (1999). Accounting accruals and auditor reporting conservatism. Contemporary Accounting Research, 16, 135–165. Francis, J. R., & Yu, M. (2009). Big4 office size and audit quality. The Accounting Review, 84, 1521–1552. Huang, K., & Zhang, L. (2010). Audit regulation, auditor switching and successor auditor conser- vatism. Auditing Research, (1), 65–74. (In Chinese.) Jensen, M. C., & Meckling, W. H. (1992). Specific and general knowledge, and organizational structure. In L. Werin & H. Wijkander (Eds.), Contract economics (pp. 251–274). Oxford: Blackwell. Johnstone, K. M. (2000). Client-acceptance decisions: Simultaneous effects of client business risk, audit risk, auditor business risk and risk adaptation. Auditing: A Journal of Practice and Theory, 19 (1): 1–25. Johnstone, K. M., & Bedard, J. C. (2003). Risk management in client acceptance decisions. The Accounting Review, 78, 1003–1025. Ke, B. (2014). A personal perspective on protecting investors of publicly listed Chinese firms. China Journal of Accounting Studies, 2, 253–263. Lennox, C. (2000). Do companies successfully engage in opinion-shopping? The UK experience. Journal of Accounting and Economics, 29, 321–337. Liu, Y., Liu, G., & Wang, H. (2010). The reform and development of the CPA Profession in China. Beijing: People’s Publishing House. (In Chinese.) Maijoor, S., & Vanstraelen, A. (2012). Research opportunities in auditing in the EU, revisited. Auditing: A Journal of Practice & Theory, 31,115–126. Public Company Accounting Oversight Board (PCAOB). (2011). Improving the Transparency of Release No.2011-007, PCAOB Rulemaking Docket Matter No. 029,11. Reynolds, J. K., & Francis, J. R. (2001). Does size matter? The influence of large clients on office-level auditor reporting decisions. Journal of Accounting and Economics, 30, 375–400. Wang, N., Liao, J., Liu, E., Zeng, X., & Yang, T. (2007). A study on headquarter-branch admin- istration by accounting firms. The Chinese Certified Public Accountants Journal, (8), 50–53. (In Chinese.) Wang, B., & Xin, Q. (2010). Do office practice of audit firms influence audit quality and fees? Auditing Research, (2), 70–76. (In Chinese.) Wu, X., & Chen, M. (2012). Internal governance of public accounting firms: Theory, principles, and implications for organizational strategies. Auditing Research, (3), 76–82. (In Chinese.) Wu, X., Zeng, T., & Wang, C. (2014). Client importance: A measurement bias and its impact. China Accounting and Finance Review, 16,72–94.
Journal
China Journal of Accounting Studies
– Taylor & Francis
Published: Oct 2, 2015
Keywords: audit firm; branches; initial engagement; internal governance
