A political economic analysis of auditor reporting and auditor switches

Transcription

1 Rev Acc Stud (2006) 11:21 48 DOI /s z ORIGINAL ARTICLE A political economic analysis of auditor reporting and auditor switches K. Hung Chan Æ Kenny Z. Lin Æ Phyllis Lai-lan Mo Ó Springer Science+Business Media, Inc Abstract This study examines whether auditor opinions are affected by political and economic influences from governments. We use auditor locality (local versus non-local) to capture such influences from local governments in China. Based on data from China s stock markets for the period , we find that local auditors, who have greater economic dependence on local clients and are subject to more political influence from local governments than non-local auditors, are inclined to report favorably on local governmentowned companies to mitigate probable economic losses. Moreover, companies with qualified opinions are more likely to switch from a non-local auditor to a local auditor than companies with unqualified opinions. Contrary to some prior studies, we find that in China s political environment, local government-owned companies that switched from a non-local auditor to a local auditor after receiving a qualified opinion can succeed in opinion shopping. Keywords Auditor locality Æ Audit qualifications Æ Auditor switches Æ Economic dependence Æ Political influence JEL Classification M42 This paper examines how political and economic influences from governments affect auditor opinions. Specifically, we study whether local auditors in China, who have greater economic dependence on local clients and are more vulnerable to political influence from local governments than non-local auditors, are more likely to report K. H. Chan Æ K. Z. Lin Department of Accountancy, Lingnan University, Tuen Mun, Hong Kong P. L. Mo (&) School of Accounting and Finance, The Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong afphmo@inet.polyu.edu.hk

2 22 Rev Acc Stud (2006) 11:21 48 favorably on their local government-owned clients. We define an auditor to be local when it is located in the same province (or equivalent in China) as the client, and more than 50% of its clients total assets are owned by clients in the same province as the auditor. In China, listed companies are owned primarily by local governments, and the majority of local audit firms was previously affiliated with and had close connections with these governments (DeFond, Wong, & Li, 2000; Tang, 1999; Yang, Tang, Kilgore, & Hong, 2001). These institutional characteristics of heavy government influence on company management and audit firms are common in many transitional economies, and provide a unique setting for investigating how political influence affects auditor opinions. 1 While we expect that local auditors, under the significant political influence of local governments, attempt to avoid economic losses by giving clean opinions to local government-owned listed companies, they may be subject to other costs, such as litigation costs and political costs resulting from audit failures. As elaborated later, in China s political and legal environment, the probability of adverse public or legal action is quite small, but losing a client as a result of issuing a qualified opinion is quite likely. Therefore, we expect a local auditor in this circumstance will tend to be lenient to retain important clients. Whether this expectation is correct is an empirical question that we address in this paper. With this expectation, we then examine whether companies that received qualified opinions are likely to switch from non-local auditors to local auditors. Finally, we re-examine the main hypothesis of political influence by testing whether local auditors are indeed more likely than non-local auditors to issue subsequent unqualified opinions to local government-owned companies that previously received qualified opinions and switched auditors. In other words, we investigate whether management can succeed in opinion shopping by strategically switching auditors (i.e. switching from a non-local auditor to a local auditor). We test our hypotheses by analyzing all companies listed on the Shenzhen and Shanghai stock exchanges during the period Consistent with our expectations, we find that local auditors tend to issue more clean audit opinions to local government-owned companies than non-local auditors, after controlling for auditor size and other variables expected to affect auditor opinions. Moreover, companies with qualified opinions are more likely to switch from a non-local to a local auditor than are companies with unqualified opinions. We also find that compared to non-local auditors, local auditors are significantly more likely to issue a subsequent clean opinion to a previously qualified local governmentowned company that switched to them. This paper contributes to the existing literature on auditor reporting incentives and auditor switching. First, we use auditor locality as a measure of audit quality in a transitional economy, as auditor locality can capture both the economic dependence and the political influence of governments on auditor opinions. Prior studies suggest that an economic dependence on clients creates incentives for auditors to compromise their independence (e.g. DeAngelo, 1981). Most of the studies measure economic dependence 1 Privatization and capital market development are key aspects of economic reforms adopted by many transitional economies, including China and its neighbors like Russia and Vietnam. A common feature of corporate share ownership structure in transitional economies is the dominance of insider ownership and government control, particularly in strategic industries like defense, minerals, and precious stone extraction (Krivogorsky, 2000; Wright & Nguyen, 2000). Concentrated corporate ownership and a long history of heavy government influence on business activities reduce the demand for independent auditing in these economies (Sucher & Bychkova, 2001).

3 Rev Acc Stud (2006) 11: based on client size relative to the total clientele of an audit firm (e.g. DeAngelo, 1981; Francis & Wilson, 1988; Lys & Watts, 1994; Stice, 1991). Reynolds and Francis (2001) examine how client size influences auditor reporting decisions at the local office level. In this study, we extend the idea of Reynolds and Francis (2001), that locality may be a factor affecting economic dependence. However, we examine the locality issue in a different context as most audit firms in China have only one office. In this paper, we study the reporting incentives and behavior of local versus non-local audit firms. In China s institutional setting, local auditors have greater economic dependence on local clients than do non-local auditors. They are also subject to more political influence from the local government, as maintaining a good relationship with the local government bureaucrats is often essential for their businesses. Thus, local auditors have strong incentives to be lenient and report favorably on clients that are owned by the local governments. Second, we provide evidence of opinion shopping by focusing on a transitional economy that has relatively few regulations and lower litigation risk. While opinion shopping is generally futile in the U.S. (Chow & Rice, 1982; Krishnan, 1994; Krishnan & Stephens, 1995), we find that in China, companies with qualified opinions can shop for unqualified opinions by switching from non-local to local auditors. Our paper extends and complements the DeFond et al. (2000) study on China s audit market and audit quality. Whereas the main theme of the DeFond et al. (2000) study is to test the effect of China s new auditing standards on audit quality and market concentration during the period , the main theme of our study is to test the effect of the political and economic influences from local governments in China on audit quality and auditor switches for the period To our knowledge, our study is the first that examines audit quality from a political perspective. DeFond et al. (2000) find that the new auditing standards lead to improvement in auditor independence proxied by the increase in modified audit opinions, and managers try to avoid qualifications by choosing smaller auditors in IPOs. We find that political and economic influences from local governments induce local auditors of the same jurisdiction to issue clean opinions to local government-owned companies, and managers of companies with qualified opinions try to avoid subsequent qualifications by switching from non-local to local auditors. We also provide a direct test to show that many of these managers succeed in obtaining subsequent clean opinions. To summarize, while the new auditing standards cause IPO clients to seek a more accommodating auditor by choosing smaller auditors, political and economic pressures induce clients to do the same by switching to politically vulnerable local auditors which include some large auditors in China. Given that China and other transitional economies wish to build a credible auditing profession and an efficient capital market, our study has policy implications for regulators and practitioners. For example, our results support the need for the recent regulations in China that seek to increase the demand for independent audits, such as selling off state shares in listed companies to mitigate government influence on company management. However, additional measures are necessary to curb opinion shopping, including the need to further develop regulations that require detailed public disclosure of auditor client disagreements and to have a more stringent auditor license renewal process. The introduction of such institutional arrangements should result in higher costs being imposed on companies that switch auditors to shop for opinions, and on auditors who issue inappropriate opinions. The remainder of the paper is organized as follows. Section 1 describes the relevant institutional features in China and Section 2 develops the research hypotheses.

4 24 Rev Acc Stud (2006) 11:21 48 Section 3 presents the research method. Section 4 discusses the empirical results and Section 5 concludes. 1. Institutional background 1.1. Characteristics of China s audit market Prior to 1998, a unique characteristic of China s audit market was the affiliation of audit firms with government institutions or government-controlled bodies. Following the adoption of the open door policy in 1979, there was a rapid growth in foreign investment which in turn created a demand for independent audit services for tax collection purposes (DeFond et al., 2000). In response to this, China re-established its auditing profession in the early 1980s to provide external audit services to foreign investors. At that time, due to the lack of capital, most audit firms were established and sponsored by local government agencies. Specifically, the government agencies owned the audit firms, transferred or seconded qualified personnel to work in the audit firms and controlled the operations of the firms. This affiliation of audit firms with local government agencies resulted in the lack of independence and regional protectionism. Sponsoring agencies such as the finance bureaus, the tax bureaus, the bureaus in charge of different industries or other local government agencies often demanded companies located within their administrative territory to be audited by their sponsored audit firms (Yang et al., 2001). At the same time, auditors judgments and the type of audit reports issued were often affected by the sponsoring local government agencies (Tang, 1999; Zhong, 1998). With the rapid development of capital markets, the Chinese government realizes the importance of building a credible independent auditing profession. The promulgation of independent auditing standards in 1995 signifies a major step to improve audit quality in China. However, the key factor impeding the supply of independent audits in China has been the influence and control of audit firms by the government agencies as discussed above. Therefore, in 1997, the Ministry of Finance and the China Securities Regulatory Commission issued regulations to disaffiliate the audit firms from their sponsoring government agencies. The reform, i.e. the disaffiliation, was officially completed in After the reform, although the audit firms severed their official ties with their government sponsors in the areas of finance and organizational linkage, their personnel (who were formerly government-affiliated auditors) still maintained close relationships with local governments for doing business. Many local audit firms are able to find new clients or retain existing clients because of their close relationship with local governments (MOF, 2000). Respect for and obedience to local authorities is important for maintaining good business relationships in China (Chan & Mo, 2002; Hofstede, 2001). This institutional relationship leads to an environment in which local governments still exert strong political influence over the activities of audit firms located in their jurisdictions, and in turn potentially influence auditor independence. This phenomenon is an example that policy implementation has not satisfactorily achieved policy objectives in China. In addition, for the majority of audit firms that are licensed to provide services to listed companies, their services tend to be locally oriented. The lack of mobility and a narrow geographical dispersion reduce the ability of these auditors to resist client pressure.

5 Rev Acc Stud (2006) 11: Characteristics of China s stock markets Since the privatization processes began in the early 1990s, China s stock markets have been increasing in importance. The total number of listed firms increased from 50 in 1992 to 1208 in The total market value of publicly listed shares reached RMB3,833 billion (approximately US$ 479 billion) at the end of 2002, which exceeded that of Hong Kong (approximately US$ 464 billion) and was ranked second in Asia (Hong Kong Economic Journal Monthly, 2003). One institutional characteristic of the Chinese stock markets is that listed companies are sponsored and controlled by government-related entities. Most listed companies are business units carved out of state-owned enterprises by a local government agency, which then serves as the parent holding company of that listed company. A typical Chinese listed company has a mixed ownership structure with three predominant groups of shareholders, namely the state, legal persons, and individuals, with the state being the single largest shareholder. 2 The composition of the boards and supervisory committees of listed companies are also characterized by a very high state presence. In addition, general managers are appointed by their government owners. In other words, governments, mostly local governments, essentially control listed companies. This concentration of government ownership provides the government owners with both the incentive and the ability to control a firm s accounting information and its reporting policies. The government owners also have special access to the firm s financial data and usually do not need to rely on publicly released financial information for performance evaluation (Klassen, 1997), and this adversely affects the demand for independent audits. According to Aharony, Lee and Wong (2000), the incentives for earnings management are greater for local government-owned companies than those under the central government because many central government-owned companies belong to protected industries such as defense and natural resources and their profitability is, to a certain degree, protected. Thus local government-owned companies have a greater need for manipulating their reported earnings. 3 Though minority shareholders may doubt the truthfulness of a firm s accounting information, they lack the ability to monitor the management, or to pursue legal action. 4 The litigation costs involved are relatively high, because there is no contingent fee 2 State shares are held by the state, with the local government being the controlling owner in most of the listed companies. State shares are non-tradable, but are transferable to government related institutions. Legal person shares are held by state-owned institutions, including securities firms, trust and investment companies, mutual funds, and state-owned enterprises that have at least one non-state owner, and can only be traded in blocks in a designated market. Tradable A shares and B shares are held and traded mostly by individuals in stock exchanges. On average, these tradable shares comprise of about one third of all the shares of Chinese listed companies. 3 Managers can manipulate earnings by modifying how they interpret accounting standards, or by timing or structuring transactions (Healy & Wahlen, 1999). Nelson, Elliott and Tarpley (2002, 2003) define earnings management as broadly including earnings management that is (1) consistent with GAAP, (2) difficult to distinguish from GAAP, and (3) clearly not GAAP. They explain that while auditors are most likely to require the client to make an adjustment in situation (3), they are less likely to do so in the other two situations. However, in any case, auditors are only likely to require adjustment of earnings management attempts that they identify as material. Failure to make a required adjustment could result in qualified opinions. In this paper, we are mainly concerned with earnings management that will lead to qualified reports. 4 In China, it is very difficult for individual shareholders to pursue lawsuits against the controlling government owners. For example, recently two groups of residents in Shanghai took legal actions against the government for corrupt dealings in the allocation of land to a property developer. Not only did the residents lose the cases, the lawyer who represented the residents was also sentenced to 3 years in jail (SCMP, 2003).

6 26 Rev Acc Stud (2006) 11:21 48 system for litigation in China. Moreover, it is traditional in the Chinese culture to place harmony at the pinnacle of all social principles (Hofstede, 2001, p. 354). Courtroom confrontations disrupt harmony and are discouraged. In China, there is an intensive competition for companies to be listed on stock exchanges as the government implements a planned quota system for IPOs. The National Planning Commission first decides on the total amount of money to be raised each year and then distributes this quota to national ministries, commissions and local governments. Therefore, local governments consider having a listed company in their region as a symbol of wealth and prestige. However, shares of companies reporting losses for three consecutive years will be suspended from trading on the stock exchanges. Being de-listed brings about not only economic losses to the region and industries, but also disgrace to the local government (Chen, Su, & Zhao, 2000). In addition, the government-entity owners (i.e. the unlisted parent companies) often rely on the listed companies as a source of capital. However, the listed companies must meet the requirement of having an average return on equity of at least 6% for three consecutive years to be able to raise additional capital (CSRC, 2001a). To the extent that a firm is subject to the pressure of achieving the target profitability, its management is motivated to manipulate earnings. Therefore, local governments have strong incentives to direct management toward reporting earnings that meet the profitability criteria and selecting an auditor that will not challenge such a reporting. As a result, there is little demand for independent auditing (Aharony et al., 2000; DeFond et al., 2000). 2. Research hypotheses In this section, we develop the research hypotheses based on a political economic model to explain how local governments influence auditor opinions in China. We expect that local auditors report favorably on local government-owned companies to mitigate economic losses under the current political environment. We also expect that companies with qualified opinions are more likely to switch from a non-local auditor to a local auditor to increase the probability of receiving a subsequent clean opinion. Figure 1 depicts the political economic model for auditor reporting and auditor switches. It also summarizes the research hypotheses to be discussed below Political and economic influences on auditor opinions In China, local governments not only can allocate the businesses they control to their favorable auditors, but can also provide administrative conveniences to their auditors via either government agencies or the public utilities they control. Because the local government has concentrated political and economic power in a province (locality), local auditors in the same province are eager to maintain a good relationship with the government bureaucrats. Furthermore, since the primary political institutions and administrative structure are the same in each province, such political and economic influences are prevalent regardless of the size of the province. Therefore, local auditors with concentrated business in a province have incentives to report in harmony with the desires of the bureaucrats to reduce economic losses. This incentive is further increased by the rather small probability of legal action against auditors for issuing inappropriate audit opinions in China (DeFond et al., 2000), and the fact that most audit firms in China are limited liability companies. This phenomenon is generally compatible with the predictions of political cost theory, that political pressure influences an entity s reporting behavior and that the entity will tend to manipulate financial reports to reduce political costs or economic losses

7 Rev Acc Stud (2006) 11: Political and Economic Relationships Auditor Opinions H1: Compared to non-local auditors, local auditors are Local Governments more likely to issue clean audit opinions to local government-owned listed Auditor Switches by Listed Companies H2: Companies with qualified opinions are more likely to switch from a non-local auditor to a local auditor than companies with unqualified opinions. Switch to local auditors Switch to non-local auditors Subsequent Opinions by Auditors H3: Compared to non-local auditors, local auditors are more likely to issue a subsequent clean opinion to local government-owned companies that were previously qualified and switched to them. Political Influence Political Influence companies. Listed Companies (mostly owned by local governments) Local Auditors Unqualified Significant Economic Dependence for Auditors Qualified Normal Auditor-Client Relationship Non-local Auditors Fig. 1 A political economic model for auditor reporting and auditor switches Unqualified Qualified

8 28 Rev Acc Stud (2006) 11:21 48 (e.g. Cahan, 1992; Cahan, Chavis & Elemendorf, 1997; Han & Wang, 1998; Key, 1997; Watts & Zimmerman, 1986). Because local auditors face greater political pressure from local governments and have greater economic dependence on local clients than non-local auditors, we expect that local auditors are more likely to report leniently on local government-owned companies in order to reduce probable economic loss resulting from the government owners unfavorable decisions on the choice of auditor and possibly similar action on other businesses as well. Hence, our first hypothesis is as follows: H1: Compared to non-local auditors, local auditors are more likely to issue clean audit opinions to local government-owned companies Qualified opinions and auditor switches Prior studies find that the receipt of a qualified opinion has a significant effect on a company s propensity to switch auditors (e.g., Chow & Rice, 1982; Craswell, 1988; Citron & Taffler, 1992; Krishnan, Kirshnan, & Stephens, 1996; Vanstraelen, 2003), and that the direction of switch is from a larger to a smaller auditor (e.g., Johnson & Lys, 1990). In China, DeFond et al. (2000) provide evidence that a general increase in qualified reports is followed by a decline in audit market share among large auditors as IPO clients tend to avoid qualifications by selecting non-top ten government-affiliated auditors during the period According to the Chinese regulations, companies that receive a qualified opinion or an unqualified opinion with an explanatory paragraph are required to publish their audit report in selected securities newspapers and to file with the China Securities Regulatory Commission (CSRC) (Chen et al., 2000). The CSRC recently imposed further requirement that a company s shares will be suspended from trading if the company receives a qualified opinion due to a GAAP violation and fails to make adjustments required by the auditors (CSRC, 2001b). Furthermore, the CSRC focuses its investigations on qualified opinions, but there have been no investigations on auditor switches. Thus, the cost of switching is expected to be lower than that of qualification. Following Hypothesis 1, we expect that companies with qualified opinions tend to switch from non-local to local auditors to increase the likelihood of receiving a subsequent clean opinion. 5 Hence, our second hypothesis is as follows: H2: Companies with qualified opinions are more likely to switch from a non-local auditor to a local auditor than are companies with unqualified opinions. 5 We do not expect audit fees to be a significant factor affecting auditor switches in China. The Chinese Institute of Certified Public Accountants and the Pricing Bureau set a minimum limit on the audit fee which is based on a percentage of a client s total assets (Beijing Pricing Bureau, 2001; Li & Jin, 2003). With the exception of the Big Five (Four) audit firms, most audit firms in China do not charge significantly above the limit due to competition. In addition, with the rapidly increasing workload due to new regulations, audit firms are unlikely to use low-balling strategies to entice companies to switch auditors. For example, in this study, for companies that switched auditors in 2001, 85 companies voluntarily disclosed their audit fees for 2000 and We find that these switching companies paid comparatively higher fees in 2001, by an average of RMB83,800 (approximately US$ 10,000) per engagement, or by 0.014% of total assets. Similarly, for companies that switched auditors in 2002, 56 companies disclosed their audit fees for both 2001 and The mean audit fee paid by these switching companies in 2002 is also higher than that paid in 2001 by RMB49,142 (approximately US$ 5,920) per engagement, or by 0.003% of total assets. While the Big Five firms do charge more than others, there were only a few switches away from the Big Five firms in our study period. As explained in our sensitivity analysis later, including or excluding the Big Five auditors does not change any of the conclusions of our analysis.

9 Rev Acc Stud (2006) 11: Subsequent audit opinions in the post-switch period Policy makers in the West have expressed continuing concerns about opinion shopping and its effect on auditor independence. However, only a few studies focus specifically on examining the relationship between auditor switching and subsequent audit opinions (Chow & Rice, 1982; Craswell, 1988; Smith, 1986). While empirical studies have found little evidence of success at opinion shopping because auditors tend to treat switchers more conservatively than non-switchers in their opinion decisions (Krishnan, 1994), analytical studies suggest that a switch could reduce qualification probability. For example, Magee and Tseng (1990) find that auditors have incentives to compromise their independence if a client can benefit from the preferred reporting strategy after an auditor switch. Teoh (1992) suggests that a strategic auditor engages in a cost-benefit analysis by balancing the costs of losing the client against the costs of litigation associated with an incorrect opinion. Following this, our study is an empirical test of such a trade-off in costs in China s political and business environment. In China, though problematic auditor reporting is beginning to attract the attention of the public and the regulators, there has been only one case of shareholder litigation against auditors since the establishment of stock exchanges in the early 1990s. With more stringent regulations being imposed on listed companies with qualified opinions, management has enhanced incentives to shop for clean opinions. Therefore, we further analyze the local government s influence on auditor opinion decisions by examining whether listed companies in China are successful in opinion shopping. We focus our attention on companies with qualified opinions (i.e. problematic companies) to see if they succeed in obtaining a subsequent clean opinion after they switch to a local auditor. Consistent with Hypothesis 1, we expect that local auditors who are subject to greater political pressures from the local governments are more likely to issue subsequent favorable audit reports to local government-owned companies that were previously qualified and switched to them. H3: Compared to non-local auditors, local auditors are more likely to issue a subsequent clean opinion to local government-owned companies that were previously qualified and switched to them. 3. Research method 3.1. Data collection We collected auditor identity, audit opinion types, and client firm characteristics, including the top ten shareholders, from the 1996 to 2002 annual reports of all companies that were listed on the Shanghai and Shenzhen stock exchanges. This information is available in the China Infobank on-line database and the CD-ROM database of Chinese listed companies that is produced by the Securities Times and the Shanghai Securities News. The period was chosen because China s independent auditing standards came into effect in January 1996, audit firms officially disaffiliated from government sponsors in 1998 and the relative frequency of companies switching auditors increased substantially after Therefore, this period allows us to capture the relationship between auditor s reporting behavior and auditor switches. Panel A of Table 1 reports that, of the 6,531 firm years that comprise the population of listed companies during the period under study, our data consists of 6,229 firm-year

10 30 Rev Acc Stud (2006) 11:21 48 observations after excluding firms with missing data and firms which have a diffused ownership structure (i.e. none of the firm s shareholder holds 20% or more of the shares). Alternative controlling ownership thresholds are considered in the sensitivity analysis. Following previous empirical studies (Chen et al., 2000; Chen, Chen, & Su, 2001; DeFond et al., 2000), we classify opinions that are unqualified with an explanatory paragraph, qualified, disclaimer, and adverse, as qualified opinions. According to the official guidelines, unqualified opinions with explanatory paragraphs should only be issued for events and transactions that do not have a direct influence on the financial statements, but are important to financial statement users. However, Chinese auditors often employ unqualified opinions with explanatory paragraphs as an alternative to qualified opinions to reduce the probability of losing their clients (Chen et al., 2000). This type of opinion differs in form rather than in substance from a qualified opinion, and is considered a quasi-qualification (Xu, 1998). The China Securities Regulatory Commission also treats this type of opinion the same as a qualified opinion in terms of disclosure requirements (CSRC, 2001b). Therefore, we consider unqualified opinions with explanatory paragraphs as qualified opinions. An examination of audit reports for the sample companies indicates that typical events for unqualified opinions with explanatory paragraphs include related party transactions, uncertainty about asset value, and financial distress or going concern issues. Typical events for qualified opinions are scope limitations, uncertainty about asset value, and financial distress or going concern issues. The reasons for disclaimer or adverse opinions are similar to those of qualified opinions but the effects are more pervasive and material. Panel B of Table 1 presents descriptive statistics for the types of audit opinions issued over the period of study. We identify auditor switches by ascertaining whether the same auditor audited the company in successive years. This excludes involuntary auditor switches resulting from the cessation of business of the previous auditor or from the merger of the incumbent auditor with other auditors. Auditor switches also exclude switches to new entrant auditors, as those switches are often involuntary in China (DeFond et al., 2000). 6 Panel C of Table 1 provides statistics for voluntary auditor switches during the period of We further identify the nature of share ownership of the listed companies from a tailormade database. 7 The database provides detailed information about the top ten shareholders 6 The number of audit firms licensed to audit listed companies was quite stable for the period but has decreased significantly from 106 in 1999 to 78 in 2000 and 72 in The major reason for the decrease is the mergers of audit firms. In this study, we also excluded companies that switched to new entrant auditors. The new entrant auditor might have been an audit firm without a license to audit listed companies but was engaged to conduct the audit on behalf of the predecessor. When the non-licensed firm was granted its own license, it might keep the client and sign the audit report. Although the auditor identity is different, it is not really an auditor switch. If the new entrant auditor happens to be a local auditor, then this could potentially bias the findings in this study. However, there were only a few new entrant auditors subsequent to 1996 (five in 1998 and two in 2002) and most of them did not participate in the audit of listed companies until several years later. Specifically, for the five new entrant auditors in 1998, only one engaged in auditing the financial statements of listed companies in that year, whereas for the two new entrant auditors in 2002, none participated in the audit of listed companies in that year. 7 The database was prepared by Shenzhen GTA Information Technology Company Limited based on the detailed information about the top ten shareholders disclosed in the annual reports of listed companies and other relevant publications such as company prospectuses for the period To ensure the accuracy of the database, we have cross checked the data for a sample of 100 randomly selected companies against their annual reports, and if necessary, prospectuses and websites. We also confirmed the reliability of the data with partners of two CPA firms, who have audited listed companies in China.

11 Rev Acc Stud (2006) 11: Table 1 Descriptive information on data selection, audit opinions and auditor switches Panel A: Data selection Total number of listed firms ,088 1,160 1,208 6,531 Less: Missing data Firms with diffused ownership Total used ,038 1,104 1,151 6, (%) 1997 (%) 1998 (%) 1999 (%) 2000 (%) 2001 (%) 2002 (%) (%) Panel B: Types of audit opinion Clean unqualified 465 (93) 612 (87) 682 (83) 737 (81) 81 (85) 961 (87) 996 (86) 5,334 (86) Unqualified with explanation 24 (5) 52 (7) 87 (11) 109 (12) 97 (9) 92 (8) 99 (9) 560 (9) Qualified 11 (2) 39 (6) 39 (5) 56 (6) 49 (5) 38 (4) 41 (4) 273 (4) Disclaimer & adverse 2 (0) 10 (1) 11 (1) 11 (1) 13 (1) 15 (1) 62 (1) Total 500 (100) 705 (100) 818 (100) 913 (100) 1,038 (100) 1,104 (100) 1,151 (100) 6,229 (100) 1997 (%) 1998 (%) 1999 (%) 2000 (%) 2001 (%) 2002 (%) (%) Panel C: Voluntary auditor switches Voluntary switch 31 (6) 62 (9) 59 (7) 104 (11) 119 (11) 93 (8) 468 (9) No auditor switch 469 (94) 643 (91) 759 (93) 809 (89) 919 (89) 1,011 (92) 4,610 (91) Total 500 (100) 705 (100) 818 (100) 913 (100) 1,038 (100) 1,104 (100) 5,078 (100)

12 32 Rev Acc Stud (2006) 11:21 48 of each of the listed companies, including the name, the number and the percentage of shares held, and the ultimate owner of each of the shareholders. We classify a listed company as a local government-owned company if the largest shareholder is a local government entity who owns at least 20% of the shares. 8 The largest shareholder has the largest influence in the company. The 20% cutoff is considered large enough to have effective control of the company (La Porta, Lopez-de Silanos, Shleifer, & Vishny, 1999). This is especially true when the largest shareholder is a government entity. In China s political environment, it is unlikely that other shareholders will challenge government control. The 20% cutoff is also the threshold for companies to use the equity method to account for long-term investments in China as it is considered that significant influence is present (MOF, 2002). Alternative cutoffs are used in a sensitivity test. Appendix A summarizes the client ownership and auditor locale based on audits of 2002 financial reports of listed companies in China Specification of logistic regression models To test the hypotheses, we use three logistic regression models. While Models (1) and (3) test auditor opinion decisions for H1 and H3, Model (2) tests management s strategic auditor switching behavior for H2. Opinion it ¼ b 0 þ b 1 Locality it þ b 2 LocGov it þ b 3 Locality it LocGov it þ b 4 Auditor it þ b 5 Client it þ b 6 ROE it þ b 7 Leverage it þ b 8 Current ratio it þ b 9 Receivable it þ b 10 Inventory it þ b 11 Age it ð1þ þ b 12 Loss iðt 1Þ þ b 13 Bshare it þ b 14 After98 þ e it Switch it ¼ b 0 þ b 1 Opinion iðt 1Þ þ b 2 LocGov it þ b 3 DClient it þ b 4 DROE it þ b 5 Management it þ b 6 Loss iðt 1Þ þ b 7 New issue it þ b 8 Bshare it ð2þ þ b 9 After98 þ e it Subsequent Opinion it ¼ b 0 þb 1 Switch it þb 2 LocGov it þb 3 Switch it LocGov it þb 4 Auditor it þ b 5 Client it þ b 6 ROE it þ b 7 Leverage it þ b 8 Current ratio it þ b 9 Receivable it þ b 10 Inventory it þ b 11 Age it þ b 12 Loss iðt 1Þ þ b 13 Bshare it þ b 14 After98 þ e it ð3þ where i denotes the sample firm and t denotes the year in the sample period. 8 For example, the largest shareholder of Xinjiang International Industry (XII) Co. Ltd (stock code: ) is Xinjiang Foreign Trade (Group) Co. Ltd., which holds 53.43% of XII s total stock. According to the background information on the controlling shareholder (i.e. Xinjiang Foreign Trade (Group) Co. Ltd.) disclosed in XII s annual report, the ultimate owner of Xinjiang Foreign Trade (Group) Co. Ltd. is Xinjiang Tongbao Asset Management Co. Ltd., a wholly owned company of Xinjiang provincial government. Thus, XII Co. Ltd. is classified as a local government-owned company. On the other hand, the largest shareholder of Sinochem International Company Ltd (stock code: ) is China Chemical Import & Export Corporation which holds 64.4% of the total shares outstanding. As disclosed in the annual report of Sinochem, China Chemical Import & Export Corporation is a state-owned company established in 1950 and directly controlled by the State Council of the central government. Therefore, Sinochem International Company Limited is classified as a non-local government-owned company.

13 Rev Acc Stud (2006) 11: The dependent variable for Model (1), Opinion it, is classified as either qualified (= 1) or unqualified (= 0) as explained earlier. Locality it =1 represents a company being audited by a local auditor. We consider an audit firm to be a local firm when the firm is located in the same jurisdiction (province or equivalent in China) as the client, and more than 50% of its clients total assets come from the same jurisdiction. 9 In other words, an audit firm is classified as non-local if the firm and the client are located in different jurisdictions, or if they are in the same jurisdiction, the audit firm has a diversified clientele (i.e. no more than 50% of its clients total assets come from the same jurisdiction). An audit firm in these circumstances is less likely to be influenced by the local government. On the other hand, an audit firm that has clients concentrated in one jurisdiction and is located in the same jurisdiction of a local government-owned client is most vulnerable to the political influence of the local government. Alternative definitions of local auditor are tested in a sensitivity analysis. As explained earlier, we classify a listed company as a local government-owned company (LocGov it =1) if the local government (or the entities it controls) is the largest shareholder and holds at least 20% of the shares. Aharony et al. (2000) find that companies under the control of provincial governments have greater incentives to manipulate earnings than those under the direct supervision of the central government. As earnings management often leads to audit qualification, it is thus in the self-interest of local governments to direct management to hire an auditor who is willing to participate in the financial packaging of companies that are under their control. To test the effect of political influence of the local government on auditor opinions, we use the interaction term, Locality it _LocGov it, to examine whether auditors facing differential political cost exposure tend to issue a particular type of opinion to companies that are owned by the local government. Specifically, Locality it _LocGov it =1 means a local government-owned company that is audited by a local auditor in period t. The significance of this interaction term will suggest that the combined effect of auditor locality and government ownership affects the type of audit opinion issued. The dependent variable for Model (2), Switch it, is a dichotomous measure for an auditor switch based on the local and non-local categories. Switch it =1 represents a switch from a non-local to a local auditor. The focus of this model is to examine whether companies that received a qualified opinion in the preceding year (i.e. Opinion i(t-1) =1) are more likely to switch from a non-local to a local auditor. 9 Before 2000, virtually no audit firm in China had multiple offices in different provinces. The number of audit firms with multiple offices has increased subsequent to As at the end of 2002, of the 72 audit firms licensed to audit listed companies (including the Big Four), 17 firms have two offices and three firms have three offices (including the head office) located in different provinces in China. For those audit firms that have offices in different provinces, the 50% requirement for local auditor status is satisfied only if every office has local clients comprising over 50% of the office s clientele, i.e. over 50% of the clients total assets come from clients in the same province where the office is located. This implies that the firm as a whole serves mainly local clients. This definition for multi-office local audit firms is rather conservative. In a few cases, a firm has one office serving mainly local clients and one office serving mainly non-local clients. Such firms were classified as non-local. Alternatively, we classify an audit firm with multiple offices as local if the combined local clients of the offices are more than 50% of the total clientele of the entire audit firm (regardless of the distribution in any one office). This classification measures the extent of local services for the firm as a whole. Our sensitivity test results indicate that the main results are insensitive to this alternative definition of local auditor for multi-office firms. Moreover, we also conducted a sensitivity test on the three regression models by excluding audit firms with multiple offices from the data. The results are substantially identical to our main analyses discussed later. The significance of all the main test variables remains unchanged.

14 34 Rev Acc Stud (2006) 11:21 48 Model (3) is used to test whether companies with a prior qualification successfully receive a subsequent clean opinion by switching auditors. Similar to Model (1), the dependent variable, Subsequent_Opinion it, takes the value of one if a switching company receives a subsequent qualified opinion. The key independent variable, Switch it _LocGov it, equals one if a switching company is owned by a local government and audited by a local auditor in period t. We expect this interaction term to be negatively associated with the dependent variable Control variables for Models (1) and (3) Prior studies often use auditor size to proxy for independence and audit quality as larger audit firms have more clients and are in a better position to withstand client pressure than smaller firms (e.g. DeAngelo, 1981). Following DeFond et al. (2000) and Francis and Wilson (1988), we use a dichotomous variable, Auditor it, to control for the auditor size effect on audit opinions. Auditor it =1, if an audit firm is a top ten firm based on the total client assets. Based on the results of DeFond et al. (2000), we expect companies audited by top ten audit firms to be more likely to receive qualified opinions. We include a number of client characteristic variables that were considered in prior literature (e.g. Chen et al., 2001; DeFond et al., 2000) to affect the likelihood of receiving qualified opinions as control variables in Models (1) and (3). We use Client it to proxy for client size by taking the logarithm of the client s year-end total assets and ROE it to measure profitability by taking net income over year-end total owners equity. As larger companies tend to be financially more stable than smaller companies and profitable companies are less likely to manipulate accounting numbers than unprofitable companies (Chen et al., 2001; Schwartz & Menon, 1985), we expect Client it and ROE it to be negatively associated with the probability of receiving qualified opinions. To control for the financial liquidity of companies, we include the debt to equity ratio (Leverage it ) and the current ratio (Current_ratio it ) in the audit opinion models and expect these two variables to have opposite signs for their coefficients. That is, companies with higher leverage ratios will be more likely, while companies with higher current ratios will be less likely, to receive qualified opinions (DeFond et al., 2000; Wilkins, 1997). In addition, the ratios of accounts receivable and inventory to total assets, Receivable it and Inventory it, are included to control for audit risk and complexity of clients. We expect these two variables to be positively correlated with the issuance of qualified opinions. The unique institutional environment in China induces many listed companies to engage in earnings management which in turn leads to the receipt of qualified opinions. Chinese regulations stipulate that listed companies must meet a target profitability level to raise additional capital by a rights issue, and they will be de-listed if they report losses for three consecutive years. Chen et al. (2001) find that companies after listing for some years will engage in earnings management to meet the regulatory profitability requirement and thus receive a qualified opinion. We include a dummy variable, Age it (=1, if company i has been listed for 3 years or more by period t), to capture the impact of this phenomenon and expect this variable to have a positive coefficient in the logistic models. Similarly, we include another dummy variable, Loss i(t-1), to control for the effect of delisting avoidance incentives on the probability of receiving qualifications. We expect companies that have reported losses for two consecutive years preceding the year of audit (i.e. Loss i(t-1) =1)to be more likely to manage earnings upwards, which in turn could lead to the receipt of qualified reports (Chen et al., 2001).

15 Rev Acc Stud (2006) 11: Since 1992, some listed companies in China have been authorized to issue B shares to foreign investors that are denominated in foreign currencies (U.S. dollars in the Shanghai Stock Exchange, and Hong Kong dollars in the Shenzhen Stock Exchange). These companies are required to prepare an additional set of financial statements for foreign investors according to International Accounting Standards, and to be audited by an international audit firm. In other words, companies with B shares are required to have one domestically licensed auditor and one international auditor. It is common that the domestic and the international auditors for B-Share companies are affiliated and they do coordinate when issuing audit opinions. As such, their audit opinions are mostly identical (Li & Wu, 2003). Companies with foreign ownership are generally larger and have greater demand for quality financial reports from foreign owners, and they are less likely to be qualified (Chen et al., 2001; DeFond et al., 2000). Thus, we include the variable, Bshare it, to control for the confounding effect of differential demand for quality financial information. Finally, given the changes in the audit environment over the period studied (in particular the audit firms officially separated from the government after 1998), we include a year dummy variable, After98, to control for the political influence of government sponsors on auditor opinions before and after the disaffiliation of audit firms Control variables for Model (2) As discussed earlier, local government-owned companies (i.e. LocGov it = 1) under government influence may have greater incentive to switch to a local auditor. In addition, changes in client size are likely to be associated with the direction of auditor changes as, for example, a significant expansion may lead to the change to a non-local auditor. In this study, DClient it is measured by the percentage change in assets between years t and t)1 and is used to control for the effect of change in client size on auditor switching decisions. Similarly, DROE it represents the percentage change in the return on equity between two periods and is used to control for the effect of changes in profitability on auditor switches. These two control variables are expected to have a negative correlation with the independent variable, Switch it. Previous studies find that changes in management are a significant factor affecting auditor switches (e.g. Beattie & Fearnley, 1995; Woo & Koh, 2001). In a society like China, where relationships are especially important in doing business, a change in management (often caused by a change in the board of directors and major shareholders) may trigger alterations in the auditor client relationship. A new board of directors will probably prefer an auditor with whom it has had some previous association. Alternatively, the new board can be displeased with the incumbent auditor s reluctance to support the company s accounting choices. In this study, we use a dummy variable, Management it, by assigning a value of one if there is a change in the board chairman in year t for reasons other than normal retirement, and zero otherwise. Similar to Model (1) and Model (3), we expect companies that reported losses for two consecutive years (Loss i(t-1) = 1) prior to the year of audit to be more likely to make income increasing manipulations and then switch to an accommodating auditor to mask these manipulations. Prior studies also suggest that, when raising capital, companies are more likely to change from a lower-quality to a higher-quality auditor to increase the marketability of new securities (DeFond, 1992; Francis & Wilson, 1988). We employ a dummy variable, New_Issue it, to reflect this notion and expect the coefficient to be negative. In addition, as companies with foreign ownership are required to prepare two sets of financial statements audited by a domestic as well as an international auditor, a change in