Managerial legal liability and Big 4 auditor choice

JBR-07989; No of Pages 13 Journal of Business Research xxx (2013) xxx–xxx

Contents lists available at ScienceDirect

Journal of Business Research

Managerial legal liability and Big 4 auditor choice Hsin-Yi Chi a, Tzu-Ching Weng b,⁎ a b

Department of Accounting, National Chung Hsing University, Taiwan Department of Accounting, Feng Chia University, Taiwan

a r t i c l e

i n f o

Article history: Received 1 August 2012 Received in revised form 28 November 2013 Accepted 5 December 2013 Available online xxxx Keywords: Managerial legal liability Big 4 auditors Going concern Earnings management

a b s t r a c t This study investigates the effect of directors' and officers' (hereafter D&O) liability insurance coverage on auditor choice. Based on a sample of 671 Taiwanese listed firms with D&O legal liability insurance data, our evidence shows that companies with excess D&O liability insurance coverage are less likely to appoint Big 4 auditors. Furthermore, we find that Big 4 auditors are more likely to issue unclean opinions and to constrain the abnormal accruals and ‘beating or meeting’ earnings benchmarks for their clients with excess D&O liability insurance coverage. The findings document that a higher level of D&O liability insurance coverage increases Big 4 auditors' concerns about the credibility of financial statements. Given this, Big 4 auditors have incentive to require more conservative accounting choices for these clients in order to minimize possible litigation risk and reputation damage. © 2013 Elsevier Inc. All rights reserved.

1. Introduction This study investigates whether director and officer (hereafter D&O) liability insurance coverage affects the choice of a Big 4 auditor. The existing literature on D&O liability insurance coverage is mainly based on data from the United States and Canada (Chalmers, Dann, & Harford, 2002; Chung & Wynn, 2008; Wynn, 2008). Very little has been done outside of North America, especially in terms of emerging markets. In contrast to the United States and Canada, high ownership concentration is a feature of listed companies in East Asia. Most controlling owners also hold management positions in their controlled companies (Classens, Djankov, & Lang, 2000; Yeh, Lee, & Woidtke, 2001). Prior research indicates that tight control allows controlling owners to make decisions on auditor choice in East Asia, including Taiwan (Fan & Wong, 2005; Hung, Lu, & Ou, 2008). D&O liability insurance policy is also affected by controlling owners (Chi, Weng, Chen, & Huang, 2013). Thus, we would like to investigate the association between D&O liability insurance and auditor choice in emerging markets. Taiwan provides an ideal setting for investigating the relationship between D&O liability insurance coverage and auditor choice. Due to severe information asymmetry in Taiwan, a number of mechanisms for reducing this information asymmetry are employed in practice. For example, an audit opinion is a commonly used tool for communicating credible information about a company's financial position to investors. While over 94% of listed companies in the United States and Canada

⁎ Corresponding author. Tel.: +886 442517250. E-mail address: [email protected] (T.-C. Weng).

(Chung, Hillegeist, & Wynn, 2012) employ Big 4 auditors, only 75% of listed companies in Taiwan appoint Big 4 auditors (Chen, Lin, & Lin, 2009). It is worth testing for the differences between Taiwanese companies audited by Big 4 and those audited by non-Big 4 firms. Moreover, because Taiwan is currently the only emerging country that mandates that publicly listed companies disclose the details of their D&O liability insurance policies, this information can help us to investigate whether managerial legal liability affects auditor choice in emerging markets. There are several arguments on the role of D&O liability insurance. One view is that D&O liability insurance plays a monitoring role, because D&O liability insurance insurers thoroughly scrutinize the ensured and coverage limits (Griffith, 2005; Holderness, 1990; O'Sullivan, 1997). Another view is that D&O liability insurance reduces the effectiveness of litigation as a device to monitor managers, since it insulates directors and officers from the threat of litigation (Chung & Wynn, 2008; Core, 1997; Wynn, 2008). Although the empirical evidence is mixed on this issue, D&O liability insurance still serves as an important protection for companies from the defense and settlement of lawsuits. Serious information asymmetry exists between companies and investors in emerging markets (e.g. Taiwan). In order to have symmetrical beliefs about the probability and distribution of director and officer losses, investors typically estimate a company's risk conditions by attempting to understand their financial positions. This creates an environment in which a high-quality auditor is able to play an information mediation role. Thus, companies with higher level of D&O liability insurance coverage might be more likely to appoint Big 4 auditors to eliminate agency conflicts. However, recent studies support the argument that excess D&O liability insurance coverage reflects ex ante managerial opportunism in accounting policy choice (Chalmers et al., 2002; Chung & Wynn, 2008; Core, 1997; Wynn, 2008). This argument for managerial

0148-2963/$ – see front matter © 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jbusres.2013.12.003

Please cite this article as: Chi, H.-Y., & Weng, T.-C., Managerial legal liability and Big 4 auditor choice, Journal of Business Research (2013), http:// dx.doi.org/10.1016/j.jbusres.2013.12.003

2

H.-Y. Chi, T.-C. Weng / Journal of Business Research xxx (2013) xxx–xxx

opportunism suggests that managers who carry abnormally high D&O liability insurance tend to behave opportunistically and create moral hazard problems (Lin, Officer, & Zou, 2011). As a result, managers have reduced incentive to act in the best interests of shareholders and the temptation to engage in opportunistic activities that benefit themselves. However, if opportunistic managers hire Big 4 auditors, they are less likely to achieve opportunistic goals, because Big 4 auditors are capable of thoroughly scrutinizing their earning quality. Given this, opportunistic managers may be less inclined to hire Big 4 auditors. Since empirical research investigating the relationship between managerial legal liability and auditor choice is lacking, exploring this relationship could potentially open up a new avenue of research on corporate governance. To address the issue of potential endogeneity, we employ a series of econometric analyses, an instrumental variables approach, a propensityscore matching method and the Heckman two-stage model. We find that companies with excess D&O liability insurance coverage are less likely to appoint Big 4 auditors, indicating that managerial opportunistic behavior is an important determinant of auditor choice. Furthermore, additional tests reveal that Big 4 auditors are more likely to issue unclean opinions, to constrain abnormal accruals and to exceed or meet earning benchmarks of clients with excessively high D&O liability insurance coverage. The results indicate that Big 4 auditors have a high incentive to provide conservative accounting. It is in their interest to prevent future litigation exposure and possible reputation damages where the protection of D&O insurance might induce moral hazard problems for wellprotected directors and officers. This study adds the following contributions to the extant literature. First, unlike prior studies, which focus primarily on whether D&O liability insurance policy is associated with disclosure behavior and corporate decisions (Chalmers et al., 2002; Chung & Wynn, 2008; Lin et al., 2011; Wynn, 2008), this study provides the first empirical evidence to show that D&O liability insurance coverage affects the appointment of auditors. We find a significantly negative association between excess D&O liability insurance and Big 4 auditor choice. Given the managerial opportunism hypothesis, we indicate that managerial opportunistic behavior plays an essential role in explaining auditor choice decisions. Second, we extend the literature on auditor characteristics by investigating the accounting policies of Big 4 auditors towards clients with excess D&O liability insurance coverage. Our findings suggest that Big 4 auditors tend to be more conservative with clients carrying abnormally high D&O liability insurance. Therefore, Big 4 auditors have incentive to require more conservative accounting choices for these clients in order to minimize possible litigation exposure and reputation damage. The remainder of this paper is organized into five sections. Section 2 describes the institutional background and reviews relevant literature. Section 3 shows our research design and data sample. Section 4 reveals the empirical results and findings, and Section 5 presents our conclusions.

Bureau (TSFB) promulgated the Securities Investors and Futures Trader Protection Act (SIFTP Act) and the Securities and Futures Investors Protection Center (hereafter SFIPC) under the Act. Since the legal environment has changed in the early 2000s, the Taiwan Insurance Institute reported that listed companies purchasing D&O liability insurance coverage increased from 8.5% in 2002 to 33% in 2006 (Taiwan Insurance Institute, 2010). There was a sharp increase in the demand for D&O liability insurance by listed companies in Taiwan. The Institute also reported that nearly 75% of claims against listed companies have been made by investors, 12% made by creditors and 13% made by others (e.g. customers, employees and other related parties). Board directors and officers have the following duties specified in the Company Law and Securities and Exchange Act in Taiwan. Pursuant to the Company Law, the board of directors and officers are responsible for their company's behavior and should fulfill their fiduciary obligations with the cane of a good administration to check the company's financial reports. If management negligence causes any loss on the part of the company, management should be held responsible for reimbursing investors for their covered losses2. Pursuant to the Securities and Exchange Act (which is similar to the Securities and Exchange acts of 1933 and 1934 in the United States), a company's board members should reimburse genuine investors as victims of the company's false financial reports3. Moreover, the government-supported organization “SFIPC” has begun to simulate American-style securities class actions; Article 28 of the SIFTP Act states that a class-action lawsuit can be filed when more than twenty securities investors authorize the SFIPC to apply for indemnification payments. Since this act was established, the SFIPC in Taiwan has dealt with 57 class-action cases and more than 60,300 plaintiffs required a total amount of indemnification of about US $0.9 billion as of the end of 2008 (SFIPC 2008 annual report). Under the supervision and guidance of competent authorities, the SFIPC has made significant progress in the fulfillment of class actions and in the protection of shareholders' equity. Furthermore, their success in acquiring reimbursement for investors in these cases marks a significant step in Taiwan's efforts to protect investors. 2.2. D&O liability insurance coverage and auditor choice

Before the end of the 1990s, most stock investors in Taiwan were individuals, who usually hesitated to take legal action when their rights were infringed upon, either because they lacked sufficient information on the provision or because they regarded the filing of a lawsuit as an action that consumes both time and money. Therefore, in order to improve the protection of investors1, the Taiwanese Securities and Futures

Big 4 auditors are organized as national partnerships with national administrative offices. They are more likely to invest in staff recruitment and training, information technology, standardized audit programs and knowledge-sharing practices. Numerous studies have investigated the notion that Big 4 auditors provide higher-quality services than nonBig 4 auditors. Theoretical support for such a quality differentiation is provided by DeAngelo (1981), who shows that larger audit firms have greater incentives to detect and reveal management misrepresentation in financial reporting. Several empirical studies provide evidence consistent with DeAngelo's result. DeFond and Jiambalvo (1993) find that clients of Big 4 auditors are less likely to have financial reporting errors or irregularities. Becker, DeFond, Jiambalvo, and Subramanyam (1998) also report lower discretionary accruals for Big-4-audited companies. Therefore, Big 4 auditors have brand-name reputations and are widely viewed as producing higher quality of audit service, which enables them to perform a stronger monitoring role than non-Big 4 auditors. There are a number of determinants of Big 4 auditor choice. First, as agency problems increase, Big 4 auditors are demanded because greater assurance is seen to reduce information asymmetry (Carey, Simnett, &

1 The Company Law allows derivative actions by shareholders owning 3% of a company continuously for a year, who may petition supervisors to sue directors and bring such suits directly if supervisors fail to do so (Article 214 of Company Law). Even so, class action litigation in Taiwan is more costly and unusual. There are several reasons for this. First, there is a serious, out-of-pocket economic disincentive to plaintiffs, and, second, there is no civil discovery in Taiwan. As a result, information costs to plaintiffs can be high. Third, securities class actions often involve some expertise without which judges find it difficult to examine the legal and factual issues.

2 Article 23 of Company Law stipulates that board directors are jointly liable with the company to reimburse any person who suffers damages or losses resulting from his/her wrongful act which is within the scope of the company's business. 3 Article 20 of the Securities and Exchange Act stipulates that directors and officers, who violate the provision of financial reports or any other relevant financial or business documents filed or publicly disclosed by an issuer in accordance with the act containing no misrepresentations or nondisclosures, shall be held liable for damages sustained by bona fide purchasers or sellers of the securities.

2. Institutional background and literature 2.1. The legal system in Taiwan

Please cite this article as: Chi, H.-Y., & Weng, T.-C., Managerial legal liability and Big 4 auditor choice, Journal of Business Research (2013), http:// dx.doi.org/10.1016/j.jbusres.2013.12.003

H.-Y. Chi, T.-C. Weng / Journal of Business Research xxx (2013) xxx–xxx

Tanewski, 2000; DeFond, 1992; Knechel, Niemi, & Sundgren, 2008). Companies are more likely to appoint Big 4 auditors when their ownership structures indicate agency conflicts (Choi & Wong, 2007; Fan & Wong, 2005). For example, Hung et al. (2008) and Chen, Hsieh, Chen, and Hung (2010) use data from Taiwan and find that controlling owners may hire Big 4 auditors to assure minority shareholders that their interests are protected. Moreover, agency problems may increase in accordance with firm size and complexity (Francis, Maydew, & Sparks, 1999; Hay & Davis, 2004; Knechel et al., 2008). As a company becomes larger, it is more difficult for owners to control and oversee operations due to an increasing number of subordinates, locations and activities (Abdel-Khalik, 1993; Kinney & McDaniel, 1989). Growth in size and complexity of a company may reduce overall efficiency in the organization and result in higher levels of risk between the manager/owner and subordinates. Therefore, companies that are more complex tend to hire Big 4 auditors (Francis et al., 1999; Hay & Davis, 2004; Knechel et al., 2008; Shiue, Chang, & Kao, 2008). Second, a company that wants to obtain external funding needs credible financial statements; therefore, it is more likely to hire a Big 4 auditor (Blackwell, Noland, & Winters, 1998; Knechel et al., 2008; Pitman & Fortin, 2004). Chin, Lin, and Lin (2002) and Chen, Lin, and Zhou (2005) indicate that listed companies in Taiwan are inclined to hire Big 4 auditors for an IPO process. The success of new financing projects depends on higher disclosure quality which helps eliminate potential concerns and risks for shareholders and creditors (Das & Sengupta, 2001; Lambert, Leuz, & Verrecchia, 2007; Sengupta, 1998). Because of brand-name reputations and better technologies for detecting problems, Big 4 auditors are more likely to constrain aggressive or questionable accounting information (Becker et al., 1998). Hence, in order to increase the reliability of financial reporting, a company that is planning to obtain external funding tends to appoint a Big 4 auditor (Mansi, Maxwell, & Miller, 2004; Pitman & Fortin, 2004). However, managerial legal liability has two opposite effects on Big 4 auditor choice. According to agency theory, companies with excess D&O liability insurance coverage might be inclined to appoint Big 4 auditors. Due to more severe information asymmetry and weaker investor protection and legal enforcement in emerging markets (e.g. Taiwan), the possibility of manager opportunism creates uncertainty in outside investors with respect to company reported earnings when managers carry excessive D&O liability insurance coverage (Chung & Wynn, 2008; Wynn, 2008). Since prior research shows that managers protected by higher D&O liability insurance coverage tend to engage in potentially opportunistic activities (Chalmers et al., 2002; Chung & Wynn, 2008; Wynn, 2008), outside investors might require that managers seek credible ways to prove that they are not opportunistic. This creates an environment where high-quality auditors can contribute to eliminating agency conflicts. Fan and Wong (2005) find that East Asia firms subject to greater agency problems are more likely to hire Big 4 auditors than firms that are subject to smaller agency problems. To deal with outside investor concerns that managerial opportunism might be occurring, companies with excess D&O liability insurance coverage are more likely to appoint Big 4 auditors and thus send a credible signal to outside investors. On the other hand, the management opportunism argument suggests that companies with abnormally high D&O liability insurance coverage might be less likely to engage Big 4 auditors. Prior research finds that managerial opportunism is one of the determinants of D&O liability insurance decisions. Managers mitigate their personal legal liability against failure decisions with D&O liability insurance coverage, since D&O liability insurance policies provide litigation costs for claims made against individual directors and officers for wrongful acts where indemnification does not apply (Chung & Wynn, 2008; Wynn, 2008). Core (1997, 2000) uses Canadian companies to test for the motivation of D&O liability insurance purchases and suggests that entrenched (opportunistic) managers tend to purchase more insurance. Boyer and Delvaux-Derome (2002) find that companies with weaker governance

3

systems are more likely to purchase D&O liability insurance. Chung and Wynn (2008) suggest that managerial legal liability coverage reduces earnings conservatism. Therefore, if D&O liability insurance weakens the incentive of directors and officers by insulating them from the threat of litigation, these directors and officers are more likely to create moral hazard problems (Baker & Griffith, 2010). This moral hazard problem could be exacerbated as the managers perceive lower litigation risk. If these managers have greater incentive to behave opportunistically, such as arranging related party transactions, investing in the projects that increase their prestige or reporting more discretionary accruals (Aggarwal & Samwick, 2003; Denis, Denis, & Sarin, 1997), then they could be less likely to hire Big 4 auditors to monitor and constrain their aggressive and questionable behaviors. Since managerial legal liability coverage serves as two opposite directions in terms of Big 4 auditor choice, it does not support a directional prediction to the impact of companies with excess D&O liability insurance on Big 4 auditor choice. We thus hypothesize that decisions of D&O liability insurance coverage affect the appointment of Big 4 auditors. 3. Research design In this section, we discuss the measure of excess D&O liability insurance coverage and Big 4 auditor choice and report on the methodology, data and sample employed in this study. 3.1. Measure of excess D&O liability insurance coverage To capture excess (unexpected) D&O liability coverage, we follow prior literature (e.g., Chung & Wynn, 2008; Wynn, 2008) and define excess D&O liability insurance coverage, EXCOV, as the residual from the regression of D&O liability insurance coverage limit on its determinants. The estimated model is as follows: DOCOV ¼ α0 þ α1 SIZE þ α2 LEV þ α3 CROSS þ α4 OUTDIR þ α5 OUTOWN þα6 RETVOL þ α7 HIGHTECH þ α8 CASH þ αn ΣINDUSTRY

ð1Þ

þαm ΣYEAR þ εt where, DOCOV represents the D&O liability insurance coverage limit scaled by lagged total assets; SIZE is the natural logarithm of total assets; LEV is the debt ratio; CROSS is a dummy variable that takes the value of one if the firm is cross-listed in exchanges outside of Taiwan, and zero otherwise; OUTDIR is the percentage of outside director on the board; OUTOWN is the percentage of common shares held by outside stockholders; RETVOL is the volatility of stock returns measured by the natural logarithm of annualized variance of daily return over the current fiscal year; HIGHTECH is a dummy variable that takes the value of one if the firm is classified as a high-tech company, and zero otherwise; CASH is the sum of cash, cash equivalents and short term investments scaled by lagged total assets; and DYear and DIndustry are year and industry fixed effects, respectively. 3.2. Estimation equations In order to test our hypothesis, we estimate a model for estimating the appointments of Big 4 auditors on excess D&O liability insurance coverage and a set of control variables. However, Zou, Wong, Shum, Xiong, and Yan (2008) suggest that companies with greater controlling-minority shareholder incentive conflicts tend to purchase more D&O liability insurance. This implies that the purchase of excess D&O liability insurance is an agency problem, and the choice of Big 4 auditors might be an agency issue as well. Therefore, we perform a test to address a potential endogeneity concern by controlling for the decision to purchase excess D&O liability insurance. At the first stage, we utilize the instrumental variable approach, adopt a model to

Please cite this article as: Chi, H.-Y., & Weng, T.-C., Managerial legal liability and Big 4 auditor choice, Journal of Business Research (2013), http:// dx.doi.org/10.1016/j.jbusres.2013.12.003

4

H.-Y. Chi, T.-C. Weng / Journal of Business Research xxx (2013) xxx–xxx

estimate the probability of purchasing excess D&O liability insurance and include several company characteristics, as follows: EXCOV ¼ β0 þ β1 LEV þ β2 SIZE þ β3 SPREAD þ β4 COMPLEXITY þ β5 AGE þβ6 LOSS þ β7 VARCFO þ β8 PYRAMID þ β9 INBOARD þ β10 DUAL þβ11 VC þ β12 INWEALTH þ β13 EXPAND þ β14 CONTRACT þδ  INDUSTRY þ ϕ  YEAR

PYRAMID One if the ultimate owner exercises control through at least one publicly traded company, and zero otherwise; INDBOARD Percentage of independent directors on the board; DUAL One if the same person holds a dual position as the board chair and CEO of the company, and zero otherwise; VC The divergence between control rights and cash flow rights possessed by the largest ultimate owner of the firm.

ð2Þ We use three instrumental variables. INWEALTH, calculated as the percentage of shares held by insiders who sit on the board, EXPAND, which equals one if the book value of total assets at the end of the fiscal year increases by more than 25% from the beginning of the fiscal year and zero otherwise and CONTRACT, which equals one if the book value of total assets at the end of the fiscal year decreases by more than 25% from the beginning of the fiscal year, are used as proxies for litigation risk. The choice of the instrumental variable INWEALTH is justified because D&O liability insurance could be opportunistically purchased to protect insiders against litigation risk arising from the expropriation of minority interests (Chalmers et al., 2002; Core, 1997; Zou et al., 2008). In addition, as mentioned by Core (1997), since claims made against directors and officers often arise due to mergers, acquisitions, divestitures and dispositions, litigation risk is expected to be positively related to whether firms merge with or acquire another company (EXPAND) and to whether the firm divested a business or sold or disposed of substantial assets (CONTRACT). In both cases, because these companies can expect to face greater risk of lawsuits, they are inclined to purchase more D&O litigation insurance coverage. Therefore, we include EXPAND and CONTRACT as the other instrumental variables and expect the coefficients to be positive in Model (2). Our first stage regression also controls for a series of firm characteristics. We use the instrumental variable approach by incorporating the fitted value of excess D&O liability insurance in the following probit regression: ProbðBIG4 ¼ 1Þ ¼ Κ 0 þ Κ 1 FittedEXCOV þ Κ 2 ATURN þ Κ 3 CURR þ Κ 4 ROA þΚ 5 LEV þ Κ 6 SIZE þ Κ 7 SPREAD þ Κ 8 COMPLEXITY þΚ 9 AGE þ Κ 10 LOSS þ Κ 11 VARCFO þ Κ 12 PYRAMID ð3Þ þΚ 13 INDBOARD þ Κ 14 DUAL þ Κ 15 VC þ δINDUSTRY þϕ  YEAR

For the control variables, we follow Lawrence, Minutti-Meza, and Zhang (2011) and control for several company characteristics, such as profitability effect (AUTRN, CURR and ROA), financial leverage (LEV) and company size and age (SIZE and AGE). We also control for variables capturing uncertainty, such as information asymmetry (SPREAD), complexity of company operations (COMPLEXITY), financial distress (LOSS) and variability (VARCFO) (Hakim & Omri, 2010; Willenborg, 1999). Finally, because Fan and Wong (2005) suggest that agency conflicts might affect the choice of Big 4 auditors, we include several variables relating to agency problems in our model. PYRAMID is a proxy for a pyramidal ownership structure (Classens et al., 2000). INDBOARD is used to present lower agency costs when the percentage of independent directors on the board is higher (Sharma, Naiker, & Lee, 2009). DUAL represents greater agency costs when the CEO also serves as the board chairman (Armstrong, Guay, & Weber, 2010; Yermack, 1996). VC is included as a proxy for entrenchment effects that might allow controlling shareholders to expropriate minority shareholders (Fan & Wong, 2005). Since our hypothesis is multidirectional, all significance levels are based on two-tailed tests. 3.3. Data and sample Our sample consists of Taiwanese listed companies for the period of 2008–2010 on the Taiwan Stock Exchange (TSE). The D&O liability insurance data are publicly available in a proxy statement because the TSFB has required companies to disclose the existence of D&O liability insurance policies since the end of 2007. Data for the firm level information and ownership structure, including financial statement data and voting rights and cash flow rights, is obtained from the Taiwan Economics Journal (TEJ) database. 4. Empirical analysis 4.1. Sample selection and univariate analyses

The definition of variables in Model (3) is as follows: One when the auditor is a Big 4 auditor at the end of the fiscal year, and zero otherwise; Fitted_EXCOV The fitted value from the regression of excess D&O liability insurance coverage on determinants of D&O liability insurance in Model (2); AUTRN Net sales divided by total assets at the end of the previous year; CURR Current assets divided by current liabilities; ROA Net income divided by total assets at the end of the previous year; LEV Long term debt plus debt in current liabilities divided by total assets at the end of the previous year; SIZE Natural logarithm of total assets at the end of the fiscal year; SPREAD Mean bid/ask spread over one year starting from 3 months after the end of fiscal year t; COMPLEXITY Natural logarithm of number of segments; AGE The number of years since the company was established; LOSS One if the firm reports a negative net income, and zero otherwise; VARCFO Volatility of cash flows, the variation of net cash flows over the previous 3 years; BIG 4

The initial sample for this study consists of all companies listed on the Taiwan Stock Exchange. Because the insurance data had not been officially and fully disclosed until 2008, the sample period runs from 2008 to 2010. We obtain the insurance data, financial data and auditor data from the Taiwan Economics Journal (TEJ) database. Panel A of Table 1 summarizes our sample selection procedure. As outlined in panel A, we indentify a total of 4086 firm-years over the period of 2008–2010. We delete 153 firm-years in financial institutions and regulated utilities from the sample to avoid the confounding effects of regulation. We then delete 12 and 42 firm-year observations with insufficient auditor and financial data. We also remove 2013 firm-year observations4 without purchasing D&O liability insurance. Finally, to mitigate the effect of outliers, we winsorize observations that fall in the top 1% and bottom 1% of the empirical distribution for each variable. The final sample5 consists of 1866 firm-year observations6. 4 The deleted sample of 2013 firm-year observations for the period 2008 through 2010 includes 733 firms in 2008, 660 firms in 2009 and 620 firms in 2010. 5 There are 73 companies purchasing D&O liability insurance at the first time during 2008–2010, which is nearly 4% of the total purchased sample. In addition, there are 10 companies canceling D&O liability insurance during 2008–2010, which is only 0.5% of the non-purchased sample. 6 The sample of 1866 firm-year observations for the period 2008 through 2010 includes 571 firms in 2008, 631 firms in 2009 and 664 firms in 2010.

Please cite this article as: Chi, H.-Y., & Weng, T.-C., Managerial legal liability and Big 4 auditor choice, Journal of Business Research (2013), http:// dx.doi.org/10.1016/j.jbusres.2013.12.003

H.-Y. Chi, T.-C. Weng / Journal of Business Research xxx (2013) xxx–xxx

5

Table 1 Sample selection, descriptive statistics and correlation coefficients of variables. Panel A: Sample selection Number of company-years from 2008 to 2010 Firms in financial and regulated industries Unavailable audit firm data Unavailable financial data Number of company-years in the full sample No purchase D&O liability insurance firm Number of purchase D&O liability insurance sample

4086 (153) (12) (42) 3879 (2013) 1866

Panel B: Descriptive statistics for purchase D&O liability insurance sample Variables

Big 4 audited firm

Non-Big 4 audited firm

(n = 1663)

LIMIT(million) LIMITA LNLIMIT EXCOV SIZE AUTRN CURR LEV ROA SPREAD COMPLEXITY AGE LOSS VARCFO PYRAMID INDBOARD DUAL VC

t-value

Wilcoxon Z

4.126*** 2.244** 5.739*** −2.950*** 6.831*** 1.615* 2.174** −2.349*** −8.744*** −3.986*** 1.086 −4.237*** −7.379*** −1.037 −1.765* 3.241*** −4.677*** 1.938**

5.504*** 3.304*** 5.506*** −4.137*** 6.252*** 2.259** 3.782*** −2.559*** −8.160*** −2.463** 2.224** −5.402*** −7.277*** −0.542 −1.605* 2.852*** −4.651*** 2.150**

(n = 203)

Mean

Median

Mean

Median

309.2 0.177 5.254 −0.002 15.315 0.971 2.535 0.339 0.051 0.071 0.612 22.801 0.198 0.009 0.234 0.212 0.272 6.737

163.8 0.061 5.216 0.011 15.129 0.787 1.886 0.313 0.054 0.063 0.693 21.000 0.000 0.004 0.000 0.286 0.000 2.390

172.8 0.089 5.008 0.016 14.582 0.896 2.347 0.374 −0.021 0.083 0.608 26.325 0.424 0.011 0.266 0.172 0.429 5.250

131.3 0.036 5.118 0.029 14.526 0.708 1.595 0.363 0.007 0.068 0.689 24.000 0.000 0.004 0.000 0.222 0.000 1.420

a. Variable definitions: BIG 4 One when the auditor is a Big 4 auditor at the end of the fiscal year, and zero otherwise; EXCOV The residual from the regression of adjusted D&O liability insurance coverage on determinants of D&O liability insurance; LIMIT The D&O liability insurance coverage limit at the end of the fiscal year; LIMITA The D&O liability insurance coverage limit divided by total assets at the end of the previous year; LNLIMIT The natural log of D&O liability insurance coverage limit at the end of the fiscal year; SIZE Natural logarithm of total asset at the end of the fiscal year; AUTRN Net sales divided by total assets at the end of the previous year; CURR Current assets divided by current liabilities; LEV Long term debt plus debt in current liabilities divided by total assets at the end of the previous year; ROA Net income divided by total assets at the end of the previous year; SPREAD Mean bid/ask spread over 1 year starting from 3 months after the end of fiscal year t; COMPLEXITY Natural logarithm of number of segments; AGE The number of years since the company was established; LOSS One if the firm reports a negative net income, 0 otherwise; VARCFO Volatility of cash flows, the variation of net cash flows over the previous three years; PYRAMID One if the ultimate owner exercises control through at least one publicly traded company, and zero otherwise; INDBOARD Percentage of independent directors on the board; DUAL One if the same person holds a dual position as the board chair and CEO of the company; VC The divergence between control rights and cash flow rights possessed by the largest ultimate owner of the firm; b. Statistical significance of the difference in the means and medians is based on a two-tailed test. ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels, respectively.

Panel B of Table 1 shows the descriptive statistics for the two subsamples when the sample is partitioned by Big 4 auditors and non-Big 4 auditors. The table also presents the values from T-tests and Wilcoxon Z-tests for differences in means and medians between the two types of companies. Panel B shows that the companies audited by Big 4 auditors have higher D&O liability insurance limits, indicating that Big 4 auditors tend to accept the engagement of large firms. However, we find that the clients of Big 4 auditors have significantly lower levels of excess D&O liability insurance than those audited by non-Big 4, which suggests that companies carrying excess D&O liability insurance are less likely to hire Big 4 auditors. In addition, we find that companies audited by Big 4 auditors have greater deviation in cash rights and control rights, supporting the evidence from Fan and Wong (2005). Finally, the comparisons of firm characteristics in Panel B shows that D&O liability insurance purchasers audited by Big 4 auditors have significantly

(i) larger firm size, (ii) less financial leverage, (iii) more profitability, (vi) equity issues in overseas, (v) lower information asymmetry and volatility. In Table 2, the upper triangle shows Spearman rank correlation coefficients while the lower triangle reports Pearson correlation coefficients. The results reveal that the dummy variable for Big 4 auditors (BIG4) is negatively and significantly associated with excess D&O liability insurance (EXCOV), suggesting that firms with abnormally high D&O liability insurance are less likely to appoint Big 4 auditors. We also find that companies with larger size, lower financial leverage, higher profitability, less information asymmetry and higher divergence in voting rights from cash flow rights tend to hire Big 4 auditors. In summary, these results preliminary support the management opportunism hypothesis according to which managers with excess D&O liability insurance have less incentive to appoint Big 4 auditors.

Please cite this article as: Chi, H.-Y., & Weng, T.-C., Managerial legal liability and Big 4 auditor choice, Journal of Business Research (2013), http:// dx.doi.org/10.1016/j.jbusres.2013.12.003

6

H.-Y. Chi, T.-C. Weng / Journal of Business Research xxx (2013) xxx–xxx

Table 2 Correlation matrix. Variable

(1)

(2)EXCOV (3)SIZE (4)ATURN (5)CURR (6)LEV (7)ROA (8)SPREAD (9)COMPLEXITY (10)AGE (11)LOSS (12)VARCF (13)PRYMID (14)INDBOARD (15)DUAL (16)VC

(2) −0.098 (0.000)

(1)BIG4 −0.068 (0.003) 0.156 (0.000) 0.030 (0.189) 0.027 (0.241) −0.054 (0.019) 0.199 (0.000) −0.092 (0.000) 0.048 (0.037) −0.098 (0.000) −0.169 (0.000) −0.024 (0.300) −0.025 (0.287) 0.075 (0.001) −0.108 (0.000) 0.042 (0.069)

0.157 (0.000) −0.031 (0.181) −0.033 (0.175) 0.072 (0.002) 0.026 (0.261) 0.050 (0.020) 0.062 (0.007) −0.098 (0.000) 0.052 (0.024) 0.166 (0.000) 0.165 (0.000) −0.032 (0.165) 0.093 (0.000) 0.224 (0.000)

(3) 0.145 (0.000) 0.214 (0.000)

0.058 (0.013) −0.249 (0.000) 0.205 (0.000) 0.213 (0.000) −0.255 (0.000) 0.240 (0.000) 0.247 (0.000) −0.181 (0.000) 0.123 (0.000) 0.242 (0.000) −0.234 (0.000) −0.158 (0.000) 0.037 (0.109)

(4)

(5)

(6)

(7)

(8)

0.052 (0.024) −0.038 (0.151) 0.004 (0.859)

0.088 (0.000) −0.033 (0.174) −0.289 (0.000) 0.069 (0.003)

−0.059 (0.010) 0.046 (0.042) 0.119 (0.000) 0.165 (0.000) −0.268 (0.000)

0.189 (0.000) 0.073 (0.002) 0.202 (0.000) 0.329 (0.000) 0.244 (0.000) −0.024 (0.297)

−0.057 (0.014) 0.048 (0.022) −0.240 (0.000) −0.053 (0.023) 0.025 (0.276) −0.003 (0.887) −0.204 (0.000)

−0.114 (0.000) 0.199 (0.000) 0.308 (0.000) −0.017 (0.467) −0.282 (0.000) −0.144 (0.000) −0.152 (0.000) 0.173 (0.000) 0.021 (0.369) 0.043 (0.065) 0.002 (0.935) 0.046 (0.046)

−0.283 (0.000) 0.101 (0.000) 0.031 (0.182) −0.243 (0.000) −0.187 (0.000) −0.059 (0.011) 0.087 (0.000) −0.013 (0.588) 0.068 (0.003) −0.020 (0.398) −0.052 (0.025)

−0.002 (0.949) 0.048 (0.037) −0.201 (0.000) −0.051 (0.000) 0.107 (0.000) 0.129 (0.000) 0.021 (0.360) −0.022 (0.349) 0.013 (0.588) 0.082 (0.000)

−0.218 (0.000) 0.055 (0.018) −0.033 (0.149) −0.363 (0.000) −0.083 (0.000) −0.007 (0.778) 0.011 (0.622) −0.112 (0.000) −0.057 (0.014)

−0.094 (0.000) −0.186 (0.000) 0.199 (0.000) 0.177 (0.000) −0.044 (0.058) 0.118 (0.000) 0.078 (0.000) 0.029 (0.199)

a. The correlation coefficients of variables are for Spearman correlation in upper triangle and for Pearson correlation in lower triangle; p-values, presented in parentheses below the coefficients, are based on a two-tailed test. See Table 1 for variable definitions.

4.2. Multivariate analysis 4.2.1. Excess D&O liability insurance coverage and Big 4 auditor choice Table 3 reveals the results of testing for the relationship between excess D&O liability insurance and Big 4 auditor choice. In Column (1), we perform the instrumental variable approach by incorporating the fitted value of excess D&O liability insurance into our probit regression. In the first stage, we find that firms with a greater number of shares held by insiders (INWEALTH) and higher liability risk from a merger or divestiture (EXPAND and CONTRACT) tend to carry excess D&O liability insurance. The results also indicate that firms with higher financial leverage, larger size, more information asymmetry, previous loss and severe agency problems are more likely to hold excess D&O liability insurance. The results at the first stage are generally similar to those from previous research (Chung & Wynn, 2008; Core, 1997; Lin et al., 2011; Wynn, 2008). In our second stage of Column (1), we find that the coefficient of Fitted_EXCOV (coef. = −6.649, z = −2.07) is significantly negative at the 5% level and the marginal effect7 is −0.1738, which suggests that a unit change in excess D&O liability coverage decreases the likelihood of hiring a Big 4 auditor by 17.38%. This finding supports the argument for the managerial opportunism hypothesis according to which excess D&O liability insurance coverage reflects managerial opportunism (Chalmers et al., 2002; Chung & Wynn, 2008; Core, 1997; Wynn, 2008). This implies that opportunistic companies are less likely to appoint credible auditors (e.g. Big 4 auditors) to monitor and constrain

7 The marginal effects in this table are computed as f(α + β′X)β, where β′X is computed at the mean values of the independent variables (Greene, 2011). The marginal effect is the difference in probability when the variable equals one versus when it equals zero, evaluated at the means of the other variables.

their aggressive or questionable behavior. On the contrary, they tend to hire non-Big 4 auditors to mitigate a monitoring mechanism in safeguarding accounting information which auditors play. Regarding the control variables, the firms with more profitability (ROA) and larger size (SIZE) tend to hire Big 4 auditors, while firms with greater information asymmetry (SPREAD), greater numbers of years (AGE), and higher pyramidal ownership structures (PYRAMID) are less likely to hire Big 4 auditors. The coefficient on VC is significantly positive, suggesting that if companies have more controlling-minority shareholder incentive conflicts, then they are more likely to appoint Big 4 auditors. This result is consistent with Fan and Wong's study (2005), which documents that companies having agency problems embedded in the ownership structures prefer to appoint high-quality auditors. In addition, because the sample used in our analysis is based on firmlevel data for the years 2008 to 2010, we have unbalanced panel data from 1866 firm-year observations. However, given that most of the determinants on auditor choice are firm-specific, the unobserved heterogeneity might lead to spurious correlations. In order to control for the problem of serial correlation, we create mean variables for each company. Because each company will only be in the sample once (Greene, 2011), 696 firm observations are included in the analysis. Column (2) in Table 3 shows that the coefficient of Fitted_EXCOV is negatively related to the appointment of a Big 4 auditor (coef. = −0.522, z = − 2.14). Therefore, the results remain qualitatively unchanged after controlling for unobserved heterogeneity. Moreover, Core (2000) suggests that a propensity-score matching model has more advantages than an instrument variable method for a solution to the endogeneity problem. It does not require the identification of instrumental variables that could be subjective judgments of reliability. Also, it is robust to misspecification of the functional form, since it is not subject to the parametric assumptions of the instrument

Please cite this article as: Chi, H.-Y., & Weng, T.-C., Managerial legal liability and Big 4 auditor choice, Journal of Business Research (2013), http:// dx.doi.org/10.1016/j.jbusres.2013.12.003

H.-Y. Chi, T.-C. Weng / Journal of Business Research xxx (2013) xxx–xxx

7

Table 2 Correlation matrix. (9)

(10)

(11)

(12)

(13)

(14)

(15)

(16)

0.052 (0.026) 0.060 (0.008) 0.281 (0.000) −0.248 (0.000) −0.334 (0.000) −0.139 (0.000) 0.031 (0.179) −0.086 (0.000)

−0.125 (0.000) −0.092 (0.000) 0.230 (0.000) −0.116 (0.000) −0.269 (0.000) 0.039 (0.092) −0.071 (0.002) −0.205 (0.000) 0.382 (0.000)

−0.168 (0.000) 0.033 (0.151) −0.196 (0.000) −0.214 (0.000) −0.187 (0.000) 0.096 (0.000) −0.350 (0.000) 0.173 (0.000) −0.082 (0.000) 0.027 (0.244)

−0.013 (0.588) 0.119 (0.000) −0.277 (0.000) 0.264 (0.000) 0.106 (0.000) 0.134 (0.000) −0.039 (0.090) 0.166 (0.000) −0.306 (0.000) −0.383 (0.000) 0.065 (0.005)

−0.025 (0.286) 0.156 (0.000) 0.227 (0.000) 0.008 (0.734) 0.055 (0.019) 0.031 (0.180) −0.008 (0.744) −0.043 (0.064) 0.021 (0.362) −0.040 (0.082) 0.002 (0.923) −0.018 (0.450)

0.066 (0.004) −0.035 (0.135) −0.269 (0.000) 0.051 (0.029) 0.104 (0.000) −0.024 (0.291) 0.013 (0.564) 0.126 (0.000) −0.141 (0.000) −0.260 (0.000) 0.009 (0.714) 0.267 (0.000) −0.130 (0.000)

−0.108 (0.000) 0.044 (0.059) −0.162 (0.000) 0.006 (0.808) −0.034 (0.148) 0.009 (0.699) −0.091 (0.000) 0.056 (0.016) −0.040 (0.085) −0.017 (0.468) 0.090 (0.000) −0.016 (0.493) −0.157 (0.000) 0.011 (0.651)

0.080 (0.000) 0.109 (0.000) 0.015 (0.522) 0.098 (0.000) 0.006 (0.806) 0.040 (0.083) 0.043 (0.066) 0.039 (0.088) −0.022 (0.334) −0.152 (0.000) −0.052 (0.025) 0.083 (0.000) 0.395 (0.000) 0.046 (0.049) −0.171 (0.000)

0.239 (0.000) −0.081 (0.000) −0.238 (0.000) 0.039 (0.094) −0.143 (0.000) −0.043 (0.063) −0.041 (0.075)

0.018 (0.445) −0.262 (0.000) 0.002 (0.994) −0.299 (0.000) −0.031 (0.183) −0.149 (0.000)

0.106 (0.000) 0.002 (0.923) 0.003 (0.898) 0.090 (0.000) 0.067 (0.004)

0.013 (0.577) 0.139 (0.000) 0.020 (0.384) 0.084 (0.000)

variable method. Therefore, we calculate the propensity scores for Model (3) and impose a caliper distance of 2%. We obtain a propensity-score matched sample of 394 firm-years including 197 Big 4 clients and 197 non-Big 4 clients8. As shown in Column (3) of Table 3, the coefficient of EXCOV is significantly negative. Even after propensity score matching, a significant difference is still found in the EXCOV coefficient. However, most control variables become statistically insignificant, implying effective propensity score matching. Multicollinearity issues also need to be considered at this stage. We calculate variable inflation factors (VIF), which measure the inflation in parameter estimates due to collinearities amongst independent variables. We find that the largest VIF is less than 1.63, which means that multicollinearity did not have an undue influence on our model.

4.3. Additional tests 4.3.1. Excess D&O liability insurance coverage, Big 4 auditors and audit opinion Moreover, we examine the association between excess D&O liability insurance coverage, Big 4 auditors and audit opinion. Audit quality may be impaired if a non-Big 4 auditor is or expects to be negotiating an employment contract with his/her clients carrying abnormally high D&O liability insurance coverage. The concern is that collusion may occur explicitly or implicitly between non-Big 4 auditors and their opportunistic clients, since opportunistic companies may offer a lucrative engagement 8

Why did we not match a non-purchasing company with a purchasing company using the propensity-score matching model? Because the companies purchasing D&O liability insurance are nearly half of the total sample, and will yield 3732 matched pairs, who's matched sample size is similar with our original sample size, 4086. Therefore, many non-purchasing companies could not match purchasing companies in the same firm characteristics.

−0.131 (0.000) −0.157 (0.000) 0.405 (0.000)

0.003 (0.911) 0.067 (0.004)

−0.156 (0.000)

contract to non-Big 4 auditors, and as a result non-Big 4 auditors may be less likely to confront management with problems uncovered during the audit. Managerial opportunism might impair audit quality by requiring or negotiating with auditors to decrease the probability of problem discovery when they are intended to use aggressive accounting policies. However, because Big 4 auditors have larger audit firm size, their larger client base implies that Big 4 auditors have more to lose in the event of a loss of reputation. This larger potential loss results in a relatively greater incentive to be independent in comparison with non-Big 4 auditors who have a much smaller client base (Becker et al., 1998). Therefore, we expect that Big 4 auditors would have less incentive to issue clean opinions for clients with possible moral hazard. We use a probit model to test our predictions. Because most unfavorable audit opinions are issued for going concerns or problems, we use going concern opinions as our dependent variable. In addition, we also use modified audit opinions as a proxy for unclean opinions. Since the control variables largely capture the financial health of companies, we control for company size (SIZE), return on assets (ROA), leverage (LEV), cash ratio (CA), number of years (AGE) and audit opinions from the previous year (LAGMAO) (e.g., Chi & Chin, 2011; Dopuch, Holthausen, & Leftwich, 1987; Monroe & Teh, 1993; Mutchler, 1985). The model is as follows: ProbðGCO ¼ 1 or MO ¼ 1Þ ¼ γ0 þ γ1 BIG4 þ γ2 EXCOV þ γ3 BIG4 EXCOV þ γ4 SIZE þ γ5 ROA þ γ6 LEV

ð4Þ þγ7 CA þ γ8 AGE þ γ9 LAGGCO þ δ  YEAR þϕ  INDUSTRY;

where GCO equals one if a company receives a going concern opinion and zero otherwise; MO equals one if a company receives a modified

Please cite this article as: Chi, H.-Y., & Weng, T.-C., Managerial legal liability and Big 4 auditor choice, Journal of Business Research (2013), http:// dx.doi.org/10.1016/j.jbusres.2013.12.003

8

H.-Y. Chi, T.-C. Weng / Journal of Business Research xxx (2013) xxx–xxx

Table 3 Excess D&O liability insurance coverage and Big 4 auditor choice. Variable

Fitted_EXCOV

Predicted sign

?

Column (1)

Column (2)

Column (3)

Excess D&O liability insurance and Big 4 auditor

Controlling for heterogeneity bias

Propensity-score match sample

Second stage

First stage

Second stage

First stage

BIG4

EXCOV

BIG4

EXCOV

−6.649** (−2.07)

EXCOV ATURN

+

CURR

+

ROA

+

LEV

−/+

SIZE

+/+

SPREAD

−/+

COMPLEXITY

+/+

AGE

−/−

LOSS

−/+

VARCFO

−/+

PYRAMID

−/+

INDBOARD

+/+

DUAL

−/+

VC

+/+

INWEALTH

+

EXPAND

+

CONTRACT

+

Intercept

?

YEAR INDUSTRY wald χ2 Adj R2 N

0.066* (1.69) 0.007** (2.30) 1.869*** (4.38) −0.415 (−1.14) 0.219*** (3.47) −1.600* (−1.74) 0.121 (0.54) −0.021*** (−3.84) 0.075 (0.63) 0.265* (1.95) −0.131* (−1.87) 0.208 (1.62) −0.177 (−1.59) 1.197** (2.31)

−1.446** (−2.21) Included Included 149.11***

Big4

−0.522** (−2.14)

0.014* (1.99) 0.001* (1.73) 0.056* (1.76) 0.013** (2.32) −0.000*** (−2.28) 0.010*** (2.03) 0.682*** (5.88) 0.009* (1.79) 0.003 (0.66) 0.006** (1.67) 0.119*** (6.75) 0.044*** (4.01) 0.015*** (2.23) 0.016*** (2.05) 0.015 (0.56) Included Included

0.001 (0.21) 0.021 (0.99) 0.828*** (3.32) −0.019*** (−2.40) 0.000 (0.68) −0.012 (−0.94) 0.497*** (6.34). −0.019*** (−21.54) 0.074 (1.32) 3.443*** (3.00) −0.023 (0.45) 0.136 (1.09) −0.037 (−0.84) 0.095* (1.86)

0.175*** (5.03) Included Included 650.61***

0.198 1866

0.000* (1.76) 0.001*** (2.59) 0.000* (1.72) 0.023 (1.50) −0.000 (−0.83) 0.026*** (2.47) 0.817*** (5.78) −0.013 (−1.16) −0.007 (−0.27) 0.009 (1.02) 0.214*** (6.28) 0.123*** (3.51) 0.002* (1.93) 0.030* (1.70) 0.000 (0.01) Included Included 0.145

696

−2.085** (−2.40) 0.089 (0.82) 0.031 (0.88) 0.269 (0.40) −0.543 (−1.21) 0.367 (0.54) −0.356 (−0.24) 0.071 (0.18) −0.038 (−0.68) −0.377 (−0.25) 1.343 (0.33) −0.286 (−1.49) 1.093 (0.37) −0.236 (−1.50) 0.594 (1.06)

−1.256 (−1.19) Included Included −254.105*** 0.085 406

a. INWEALTH is the total number of shares held by insiders who sit on the board divided by the total number of the outstanding shares; EXPAND is one if the book value of total assets at the end of the fiscal year increases by more than 25% from the beginning of the fiscal year, and zero otherwise; CONTRACT is one if the book value of total assets at the end of the fiscal year decreases by more than 25% from the beginning of the fiscal year, and zero otherwise. See Table 1 for other variable definitions. b. Coefficient estimates are presented with t-statistics at first stage and with z-statistics at second stage and Column (3). c. Statistical significance is based on two-tailed test. ***, **, and ⁎ indicate statistical significance at the 1%, 5%, and 10% levels, respectively.

audit opinion9 and zero otherwise; BIG4 equals one if a company is audited by a Big 4 auditor and zero otherwise, and EXCOV is defined as excess D&O liability insurance coverage. For control variables, CA is measured as cash plus short-term investments and then divided by total assets; LAGGCO equals one if a company has received a going concern opinion in the previous year and zero otherwise. All the other variables are as previously defined. In Model (4), we employ the term, γ2EXCOV + γ3Big4*EXCOV to investigate the association between audit opinions and managerial opportunism. Moreover, we expect to find a positive sign for the coefficient of the interaction BIG4*EXCOV. The interaction terms in Model (4) allow us to test whether excess 9 Consistent with prior studies that use Taiwanese samples (e.g., Chi & Chin, 2011), we classify qualified opinions, adverse opinions, and disclaimers as modified audit opinions.

D&O liability affects Big 4 auditor reporting decisions. From a Big 4 auditor's perspective, moral hazard problems are important because they affect a client's inherent risk (Chung et al., 2012). Since companies with excess D&O insurance coverage could exacerbate moral hazard problems resulting in more lawsuits from their shareholders (Chung et al., 2012), Big 4 auditors issue more unclean opinions for clients with excess D&O insurance coverage. That is, auditors have a strong incentive to be more careful when the expected litigation risk is high. Table 4 reports the results for Model (4). We test Model (4) for the full sample and for the propensity-score matched sample. We use a propensity-score matching model in an attempt to control for differences in company characteristics. We estimate the propensity of the probit model, which calculates the probability of choosing Big 4 auditors in Model (3) and then matches a non-Big 4 client with a Big 4 client with

Please cite this article as: Chi, H.-Y., & Weng, T.-C., Managerial legal liability and Big 4 auditor choice, Journal of Business Research (2013), http:// dx.doi.org/10.1016/j.jbusres.2013.12.003

H.-Y. Chi, T.-C. Weng / Journal of Business Research xxx (2013) xxx–xxx

9

Table 4 Excess D&O liability insurance coverage, Big4 and audit opinion. Variable

Predicted sign

BIG4

+

EXCOV

+

BIG4*EXCOV

+

SIZE

−

ROA

−

LEV

+

CA

−

AGE

−

LAGGCO

+

Intercept

?

YEAR INDUSTRY Log likelihood Pseudo R2 N

Column (1)

Column (2)

GCO

MO

Full sample

Propensity-score matched sample

Full sample

Propensity-score matched sample

1.149* (1.83) 2.128** (2.10) 1.183** (2.07) −0.035 (−0.54) −5.797*** (−6.04) 0.939* (1.84) −0.406*** (−2.33) −0.000 (−0.02) 0.639*** (2.40) −6.709 (−0.04) Included Included −83.832 0.548 1866

0.598** (2.22) 1.677* (1.85) 0.853* (1.95) −0.090 (−0.84) −6.966*** (−5.64) 1.615*** (2.49) −1.844 (−1.50) 0.014 (0.93) 0.386* (1.89) −1.214 (−0.74) Included Included −60.302 0.423 394

0.221 (1.40) 2.895* (1.83) 1.605*** (2.60) −0.307 (−0.93) −1.137*** (−4.85) 0.121** (2.13) −1.045 (−1.07) −0.001 (−0.54) 2.869*** (14.41) −0.350 (−0.04) Included Included −115.082 0.508 1866

0.370 (1.60) 1.938* (1.71) 1.568** (2.02) −0.124 (−1.30) −1.867* (−1.86) 0.146** (2.25) −4.665*** (−3.04) 0.013 (1.13) 1.932*** (7.40) −1.498 (−0.76) Included Included −78.467 0.428 394

a. GCO is one if the company receives a going concern audit opinion in the current year and zero otherwise; MO is one if the company receives a modified opinion in the current year and zero otherwise; SIZE is natural logarithm of total asset at the end of the fiscal year; CA is cash plus short-term investments and then divided by total assets; LAGGCO is one if the company receive a going concern audit opinion in prior year and zero otherwise. Other variables see Table 1. Results for the dummy variables representing year and industry are not reported. Standard errors are adjusted for within-firm clustering. See Table 1 for other variable definitions. b. Coefficient estimates are presented with z-statistics in parentheses. c. Statistical significance is based on two-tailed test. ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels, respectively.

the closest predicted value within a maximum distance of 2%. Based on this method, we match the pairs with the replacement, which yields 394 matched pairs. As shown in Column (1) and Column (2), the coefficients of Big4*EXCOV are both significantly positive, supporting our prediction that companies audited by Big 4 auditors receive going concern opinions or modified opinions more often when managers are overly covered by D&O liability insurance. This suggests that since the penalties for issuing improper audit opinions are severe in Taiwan (Guan & Chang, 2010), auditors tend to issue more unclean opinions for companies with higher litigation risk. We find that a unit change in a Big 4 client with excess D&O liability increases the likelihood of receiving a going concern opinion and a modified audit opinion by 7.52% and 9.22% (or 5.97% and 6.75% for the propensity-score matched sample). In addition, the largest VIF in this model is 1.50, suggesting that multicollinearity did not have an undue influence. The coefficients on the control variables are mostly consistent with prior expectations (Chi & Chin, 2011; Dopuch et al., 1987; Monroe & Teh, 1993; Mutchler, 1985). Both in Column (1) and Column (2), the coefficients on ROA and CA are negative, implying that companies with higher profitability and more cash holdings are less likely to receive going concern opinions and modified opinions. The coefficients on LEV and LAGGCO are positive, showing that companies are more likely to receive unclean opinions when they have more long-term debts or unfavorable opinions in the previous year. 4.3.2. Excess D&O liability insurance coverage, Big 4 auditors and earnings management We further examine whether Big 4 auditors require more conservative accounting policies when they accept clients with excess D&O liability insurance coverage. Chung and Wynn (2008) suggest that managers have an incentive to report earnings aggressively when their expected legal liability is reduced given higher excess D&O liability

insurance coverage. Wynn (2008) also finds that managers with excess coverage are less likely to provide bad news forecasts or disclose bad news timely. The results document that managers' liability protection through D&O liability insurance coverage tends to affect managers' accounting policies in an opportunistic manner. From the auditors' perspective, DeFond and Subramanyam (1998) suggest that litigation risk motivates auditors to prefer income-reducing accounting choices. Kim, Chung, and Firth (2003) argue that Big 4 auditors tend to constrain income-increasing abnormal accruals due to higher potential litigation costs. Therefore, because Big 4 auditors attract more attention and their litigation is likely to damage the reputation of the Big 4 brand, we expect that Big 4 auditors have incentive to require more conservative accounting policies for clients who reveal signals of managerial opportunism. We examine the association between excess D&O liability insurance coverage and discretionary accruals for the clients of Big 4 auditors while controlling for other litigation risk factors (Cahan & Zhang, 2006). Discretionary accruals are measured by Jones model (1991). Specifically, we compute discretionary accruals (DA) as follows: h i ^ ð1=A Þ þ ϕ ^ ðDSALES −DAR Þ=A ^ DAt ¼ TAt =At−1 – ϕ 1 t−1 2 t t t−1 þ ϕ3 ðPPEt =At−1 Þ

ð5Þ where TA is total accruals (earnings before extraordinary items minus net cash flows from operations); A is total assets; ΔSALES is change in net sales; ΔAR is change in net account receivables; and PPE is gross ^ j (j = 0, …, 4) are property, plant, and equipment. The coefficients φ parameters for estimating the following equation:

TAt =At−1 ¼ ϕ1 ð1=At−1 Þ þ ϕ2 ðDSALESt =At−1 Þ þ ϕ3 ðPPEt =At−1 Þ þ et

ð6Þ

Please cite this article as: Chi, H.-Y., & Weng, T.-C., Managerial legal liability and Big 4 auditor choice, Journal of Business Research (2013), http:// dx.doi.org/10.1016/j.jbusres.2013.12.003

10

H.-Y. Chi, T.-C. Weng / Journal of Business Research xxx (2013) xxx–xxx

We then include an interaction of Big4 and EXCOV in the following OLS regression. In addition, a truncated regression is used for the analysis of positive DA, since the dependent variable is left-truncated at zero.

Table 5 (continued). Panel C: Analysis of reporting small increases in earnings Variable

Predicted sign

DAor positiveDA ¼ λ0 þ λ1 BIG4 þ λ2 EXCOV þ λ3 BIG4  EXCOV þ λ4 MB ð7Þ þ λ5 LEV þ λ6 SIZE þ λ5 OCF þ λ11 ACCR þ δ  INDUSTRY þ ϕ  YEAR

Table 5 Excess D&O liability insurance coverage, Big4 and earnings management. Panel A: Discretionary accruals Variable

Predicted sign

BIG4

−

EXCOV

+

BIG4*EXCOV

−

MB

+

LEV

−

SIZE

−

OCF

−

ACCR

−

Intercept

?

Adj R2 N

Column (1)

Column (2)

Full sample

Propensity-score matched sample

DA

DA

−0.018* (−1.85) 0.059 (1.60) −0.125** (−2.07) 0.015*** (7.65) −0.062*** (−4.44) −0.009*** (−5.16) −0.082*** (−3.21) −0.657*** (−34.68) −0.168*** (−6.02) 0.4748 1866

−0.021* (1.69) 0.102 (1.57) −0.299* (−1.93) 0.023*** (4.36) −0.094*** (−2.60) −0.018*** (−3.19) −0.054 (−0.86) −0.515*** (−10.42) −0.289*** (−3.53) 0.368 394

Panel B: Analysis of reporting small positive earnings Variable

Predicted sign

BIG4

−

EXCOV

+

BIG4*EXCOV

−

MB

+

LEV

−

SIZE

−

OCF

−

INDBOARD

−

DUAL

+

VC

+

FOREIGN

−

Intercept

?

YEAR INDUSTRY Log likelihood Pseudo R2 N

Column (1)

Column (2)

Full sample

Propensity-score matched sample

Small positive earnings

Small positive earnings

−0.279** (−2.02) 0.092 (1.44) −0.055* (−1.93) 0.361*** (6.70) −0.235*** (−3.11) −0.012*** (−2.73) −0.586 (−1.47) −0.213* (−1.85) 0.076 (0.90) 0.616* (1.81) −1.093*** (−2.53) −0.513 (−1.00) Included Included −621.658 0.076 1866

−0.097* (−1.77) 0.975 (0.98) −1.745 (−0.76) 0.320*** (3.14) −0.099** (−2.25) −0.098 (−1.25) −0.386 (−1.53) −0.103* (−1.70) 0.115* (1.70) 0.319 (1.33) −2.552* (1.96) −1.725 (−1.51) Included Included −170.908 0.053 398

BIG4

−

EXCOV

+

BIG4*EXCOV

−

MB

+

LEV

−

SIZE

−

OCF

−

INDBOARD

−

DUAL

+

VC

+

FOREIGN

−

Intercept

?

YEAR INDUSTRY Log likelihood Pseudo R2 N

Column (1)

Column (2)

Full sample

Propensity-score matched sample

Small increases in earnings

Small increases in earnings

−0.048** (−2.38) 0.685* (1.93) −1.195* (1.86) 0.016 (1.60) −0.248*** (−2.84) −0.061** (2.03) −0.367 (−1.03) −0.674*** (−3.00) −0.018 (−0.22) 0.328 (1.07) −0.176 (−0.58) −1.936*** (−4.42) Included Included −773.193 0.043 1866

−0.021 (−1.50) 0.540* (1.78) −1.415* (1.69) 0.077 (1.29) −0.365*** (−2.94) −0.049 (0.70) −0.079 (−0.11) −1.015* (−1.98) −0.116 (−0.69) 0.211** (2.25) −0.059 (−0.06) −0.513 (−1.00) Included Included −169.613 0.038 398

a. DA Discretionary accruals estimated using Jones (1991) model and controlling for company performance. MB is the ratio of market value of book value; OCF is net cash flows from operations divided by total assets; ACCR is lagged total accruals divided by lagged total assets. SMALL_EARNING is one if a firm reports small positive earnings (net income deflated by lagged total assets is between 0 and 2%), and zero otherwise; SMALL_INCREASE is one if a firm reports small earnings increases (change in net income deflated by lagged total assets is between 0 and 2%), and zero otherwise; FOREIGN is the percentage of shares held by foreign shareholders; Results for the dummy variables representing year and industry are not reported. Standard errors are adjusted for within-firm clustering. Other variables see Table 1. Results for the dummy variables representing year and industry are not reported. Standard errors are adjusted for within-firm clustering. See Table 1 for other variable definitions. b. Coefficient estimates are presented with t-statistics at Panel A and with z-statistics at Panels B and C. c. Statistical significance is based on two-tailed test. ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels, respectively.

where BIG4 equals one if a company is audited by Big 4 auditors and zero otherwise; and EXCOV is defined as excess D&O liability insurance coverage. We include an interaction of Big4 and EXCOV in Model (7) and expect a negative sign of the coefficient if Big 4 auditors are related to accounting conservatism for opportunistic clients. For control variables, MB is the ratio of market value to book value; OCF is net cash flows from operations divided by total assets; and ACCR is lagged total accruals divided by lagged total assets. The other control variables are defined as in Model (3). In addition, because earnings distribution is also used to test aggressive accounting decisions, we use a probit model to test two common benchmarks: companies reporting small positive profits (avoiding earnings loss), and companies reporting small positive earnings increases (avoiding earnings declines). Earnings are assumed to be of a higher quality (less subject to earnings management) if a company does not systematically meet benchmark earnings targets. The prediction is that Big 4 auditors are more likely to constrain clients who are covered

Please cite this article as: Chi, H.-Y., & Weng, T.-C., Managerial legal liability and Big 4 auditor choice, Journal of Business Research (2013), http:// dx.doi.org/10.1016/j.jbusres.2013.12.003

H.-Y. Chi, T.-C. Weng / Journal of Business Research xxx (2013) xxx–xxx

11

by excess liability insurance to meet or beat benchmark targets; thus a negative coefficient on EXCOV is anticipated. Model (8) is as follows:

applied to predict the likelihood of D&O liability insurance purchase:

ProbðBENCHMARK ¼ 1Þ ¼ σ 0 þ σ 1 BIG4 þ σ 2 EXCOV þ σ 3 BIG4

P ðPURCHASE ¼ 1Þ ¼ η0 þ η1 LEV þ η2 SIZE þ η3 MB þ η4 HIGHTECH

 EXCOV þ σ 4 MB þ σ 5 LEV þ σ 6 SIZE þ σ 7 OCF þ σ 8 OUTDIR þ σ 9 DUAL þ σ 10 VC

þ η5 ACQUIRER þ η6 CONTRACT þ η7 OWN

ð8Þ

ð9Þ

þ η8 OUTBLOCK þ η9 CASH þ ε

þ σ 11 FOREIGN þ δ  INDUSTRY þ ϕ  YEAR where BENCHMARK is coded as one if a company reports small positive earnings (or a small increase in earnings), and zero otherwise. The independent variables are the same as in previous models except for the percentage of shares held by foreign shareholders (FOREIGN). To test the reporting of small profits, we follow the approach recommended by Frankel, Frankel, Johnson, and Nelson (2002) and Carey and Simnett (2006); a company is classified as reporting small positive earnings if its net income deflated by lagged total assets is between 0% and 2%. To test small earnings increases, we follow the approach of Frankel et al. (2002), Ashbaugh, Fond, and Mayhew (2003), and Carey and Simnett (2006); a company is classified as reporting a small earnings increase if the change in its net income deflated by lagged total assets is between 0% and 2%. Panel A of Table 5 reports the results from the regression of discretionary accruals and small positive (or increases in) earnings on excess D&O liability insurance for clients of Big 4 auditors10. As shown in Columns (1) and (2), we find that the coefficients on the interaction of BIG4 and EXCOV are significantly negative, indicating that Big 4 auditors require more income-reducing abnormal accruals for clients who carry abnormally high D&O liability insurance coverage. The control variables are also consistent with our predictions. Panels B and C show the results from the regression of “meeting or beating” benchmarks on excess D&O liability insurance for clients of Big 4 auditors. We find that most of the coefficients on the interaction of BIG4 and EXCOV have a negatively and significantly predicted sign, which suggests that Big 4 auditors tend to avoid these clients to meet or beat their small positive earnings and small increases in earnings. In addition, because the largest VIF in these models is less than 1.69, multicollinearity did not have an undue influence. These findings document that Big 4 auditors did more conservative accounting by applying more stringent thresholds for clients who carry excess D&O liability insurance coverage, because Big 4 auditors might view companies with higher levels of D&O liability insurance coverage as a unique source of risk. This will raise Big 4 auditors' concerns about the credibility of earnings quality and so they tend to require more income-reducing accruals for these clients. In sum, a negative interaction between Big4 and EXCOV indicates that although companies with excess D&O liability insurance might create moral hazard problems, the appointment of Big 4 auditors is able to mitigate managers' incentives to manipulate earnings. Therefore, our results show that Big 4 auditors may substitute for the impact of excess D&O liability insurance on possible accounting problems.

where PURCHASE denotes the ex ante probability of a firm purchasing D&O liability insurance that is ex post coded one if a firm purchase D&O liability insurance and zero otherwise. At the first stage, we include the following variables in Model (9): the financial leverage (LEV); the company's total assets (SIZE); the firm's growth opportunity (MB); high technology (HIGHTECH); increasing in book value of total assets (EXPAND); decreasing in book value of total assets (CONTRACT); the percentage of shares held by the CEO (OWN); the percentage of shares held outside shareholders (OUTBLOCK); and the cash holding (CASH). We then obtain fitted values from the logistical regression and calculate the inverse Mills ratio (Heckman, 1979). The inverse Mills ratio (λ) is then used as an additional explanatory variable in Model (3) to correct for potential self-selection bias. Table 6 presents the regression result for Model (3) with an inverse Mills ratio (λ) as an additional explanatory variable. Column (2) shows the results of the first stage. We find that when a company suffers from higher litigation risk (LEV, SIZE, MB and HIGHTECH) and a greater possibility of a merger or divestiture (EXPAND and CONTRACT), it is more likely to purchase D&O liability insurance to reduce directors' and officers' personal liability. Moreover, we find that if managers are inclined to be more opportunistic (OWN and OUTBLOCK), then they tend to purchase D&O liability insurance. Finally, we find that a company with higher cash holdings (CASH) is less likely to purchase D&O liability insurance, since cash indemnification provides more protection for directors and officers. After we control for selection bias, it can be seen that the coefficient of EXCOV (coef. = −1.862 z = −3.15) is positive and significant at the second stage in Column (1), consistent with our results in Table 3. In addition, the variableλ is insignificant, indicating that our model did not suffer from selection bias and the results remain consistent. In summary, the greater D&O liability insurance coverage in mitigating the probability of engaging in Big 4 auditors is robust in several treatments that address endogeneity concerns. 4.4.2. By-year analysis Our sample period from 2008 to 2010 exists within or after the global financial crisis. In order to control for the influence of the financial crisis, we perform our model using year by year regressions. Table 7 presents the regression results. Consistent with previous findings, the coefficients on Fitted_EXCOV are significantly and negatively related to Big 4 auditor choice. In addition, we find that companies with a duel position of board chairman and CEO are less likely to hire Big 4 auditors in 2008. This may suggest that companies revealing greater agency costs are less inclined to appoint high-quality auditors as a monitoring role in the period of global financial crisis.

4.4. Robustness tests 5. Conclusion 4.4.1. Controlling for self-selection bias Another important concern for our model is self-selection bias. Prior research suggests that companies with several corporate characteristics are likely to purchase D&O liability insurance to reduce litigation risk (Chung & Wynn, 2008; Core, 1997). In order to avoid this potential validity treat, we further use a 2-stage treatment effect model in our regression to control for potential selfselection bias related to a company's demand for D&O liability insurance. Specifically, following the determinants of D&O liability insurance purchase from Chung and Wynn (2008), Model (9) is

At the end of 2007, in order to enhance the effectiveness of corporate governance, TSFB required listed companies to disclose available information on D&O liability insurance purchases. Prior studies conducted on D&O liability insurance suggest that the decision to purchase excess D&O liability insurance signals opportunistic behavior on the part of managers (Chalmers et al., 2002; Chung & Wynn, 2008; Core, 1997; Wynn, 2008). This publicly available data provides us with a setting with which to investigate whether D&O liability insurance policies are associated with Big 4 auditor choice.

Please cite this article as: Chi, H.-Y., & Weng, T.-C., Managerial legal liability and Big 4 auditor choice, Journal of Business Research (2013), http:// dx.doi.org/10.1016/j.jbusres.2013.12.003

12

H.-Y. Chi, T.-C. Weng / Journal of Business Research xxx (2013) xxx–xxx

Table 6 Excess D&O liability insurance coverage and Big 4 auditor choice: controlling for selfselection bias. Variable

Predicted sign

EXCOV

?

ATURN

+

CURR

+

ROA

+

LEV

−/+

SIZE

+/+

SPRED

−

COMPLEXITY

+

AGE

−

LOSS

+

VARCFO

+

PYRAMID

−

INDBOARD

+

DUAL

−

VC

+

LAMDA

?

MB

+

HIGHTECH

+

EXPAND

+

CONTRACT

+

OWN

+

OUTBLOCK

−

CASH

−

Intercept

?

YEAR INDUSTRY Pseudo R2 N

Column (1)

Column (2)

Second stage

First stage

BIG4

PURCHASE

−1.862*** (−3.15) 0.029* (1.74) 0.029 (1.26) 1.818*** (4.02) −0.814*** (−2.91) 0.214*** (4.61) −1.098 (−1.08) 0.128 (0.52) −0.020*** (−2.82) 0.194* (1.82) 0.237* (1.79) −0.147 (−1.22) 0.766*** (2.70) −0.248*** (−2.72) 0.515 (1.11) −0.293 (−1.62)

−1.446** (−2.21) Included Included 0.148 1866

0.221* (1.66) 0.279*** (15.50)

0.099*** (5.38) 0.672*** (13.31) 0.169*** (2.25) 0.579*** (5.63) 3.047*** (3.32) −0.465*** (−2.30) −1.475*** (−7.52) −5.105*** (17.36) Included Included 0.113 3879

We find a negative relationship between excess D&O liability insurance coverage and the selection of Big 4 auditors, which supports the management opportunism argument that excessive D&O liability insurance coverage could exacerbate moral hazard problems and lead to more opportunistic management behavior. This suggests that managers who carry abnormally high litigation coverage are less likely to appoint higher-quality auditors. Moreover, because excess D&O liability insurance coverage increases auditor concerns over client earning information, we find that to avoid future litigation risk and possible reputation damage, Big 4 auditors have an incentive to issue unclean opinions and require more income-reducing accounting choices for these clients. Therefore, the overall evidence supports the management opportunism argument for the demand of excess D&O liability insurance

Table 7 Excess D&O liability insurance coverage and Big 4 auditor choice: year by year regressions. Variable

Predicted sign

Fitted_EXCOV

?

ATURN

+

CURR

+

ROA

+

LEV

−

SIZE

+

SPREAD

−

COMPLEXITY

+

AGE

−

LOSS

+

VARCFO

+

PYRAMID

−

INDBOARD

+

DUAL

−

VC

+

Intercept

?

INDUSTRY wald χ2 N

Column (1)

Column (2)

Column (3)

2008 year

2009 year

2010 year

−6.241** (−2.35) 0.015 (0.95) 0.029** (2.18) 1.249*** (3.97) −1.091** (−2.20) 0.182** (2.01) −1.178 (0.63) 0.048 (0.12) −0.018* (−1.82) 0.238 (1.13) 0.502* (1.77) −0.156 (−0.82) 1.059** (2.19) −0.331** (−2.11) 0.934** (2.06) −0.532 (−0.46) Included 93.43*** 573

−4.324** (−2.00) 0.008* (1.86) 0.029* (1.94) 0.606*** (2.82) −0.059 (−0.76) 0.161 (0.94) −3.076* (−1.70) 0.270 (0.75) −0.022** (−2.35) 0.331 (0.91) 0.547** (2.00) −0.028 (−0.11) 0.419*** (2.57) −0.119 (−1.46) 1.494** (1.97) −0.840 (−0.56) Included 81.22*** 631

−7.232*** (3.30) 0.055* (1.73) 0.015** (2.49) 1.697* (1.89) −0.463* (−1.67) 0.159*** (2.79) −0.458 (−0.29) 0.046 (0.15) −0.016* (−1.90) 0.188 (1.36) 0.722** (2.07) −0.019 (−0.11) 0.203** (2.40) −0.056 (−0.46) 2.533*** (3.97) −1.381 (1.36) Included 231.65*** 662

a. Coefficient estimates are presented with z-statistics in parentheses. See Table 1 for variable definitions. b. Statistical significance is based on two-tailed test. ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels, respectively.

and indicates that managers with abnormally high litigation coverage are less likely to hire Big 4 auditors and then have a higher likelihood of receiving unfavorable audit opinions.

References Abdel-Khalik, A.R. (1993). Why do private companies demand auditing? A case for organizational loss of control. Journal of Accounting, Auditing and Finance, 8(1), 31–52. Aggarwal, R. K., & Samwick, A. A. (2003). Why do managers diversify their firms? Agency reconsidered. The Journal of Finance, 58(1), 71–118. Armstrong, C. S., Guay, W. R., & Weber, J. P. (2010). The role of information and financial reporting in corporate governance. Journal of Accounting and Economics, 50(2), 179–234. Ashbaugh, H., Fond, R. L., & Mayhew, B. W. (2003). Do non-audit services compromise independence? Further evidence. The Accounting Review, 78(3), 611–639. Baker, T., & Griffith, S. J. (2010). Ensuring corporate misconduct: How liability insurance undermines shareholder litigation? Chicago: The University of Chicago Press. Becker, C. L., DeFond, M. L., Jiambalvo, J., & Subramanyam, K. R. (1998). The effect of audit quality on earnings management. Contemporary Accounting Research, 15(1), 1–24. Blackwell, D. W., Noland, T. R., & Winters, D. B. (1998). The value of auditor assurance: Evidence from loan pricing. Journal of Accounting Research, 36(1), 57–70. Boyer, M., & Delvaux-Derome, M. (2002). The demand for directors' and officers' insurance in Canada. Working paper. : University of Montreal. Cahan, S. F., & Zhang, W. (2006). After Enron: Auditor conservatism and ex-Andersen clients. The Accounting Review, 81(1), 49–82. Carey, P., & Simnett, R. (2006). Audit partner tenure and audit quality. The Accounting Review, 81(3), 653–676. Carey, P., Simnett, R., & Tanewski, G. (2000). Voluntary demand for internal and external auditing by family businesses. Auditing: A Journal of Practice & Theory, 19, 37–51 (Supplement). Chalmers, J. M., Dann, L. Y., & Harford, J. (2002). Managerial opportunism? Evidence from directors' and officers': insurance purchases. The Journal of Finance, 57(2), 609–636.

Please cite this article as: Chi, H.-Y., & Weng, T.-C., Managerial legal liability and Big 4 auditor choice, Journal of Business Research (2013), http:// dx.doi.org/10.1016/j.jbusres.2013.12.003

H.-Y. Chi, T.-C. Weng / Journal of Business Research xxx (2013) xxx–xxx Chen, Y. J., Hsieh, T. J., Chen, S.C., & Hung, D. K. (2010). Corporate governance and auditor choice. Providence Journal of Accounting, 15(1), 21–32. Chen, C. Y., Lin, C. J., & Lin, Y. C. (2009). Audit partner tenure, audit firm tenure, and discretionary accruals: Does long auditor tenure impair earnings quality? Contemporary Auditing Research, 25(2), 415–445. Chen, K. Y., Lin, K. L., & Zhou, J. (2005). Audit quality and earnings management for Taiwan IPO firms. Managerial Auditing Journal, 20(1), 86–104. Chi, H. Y., & Chin, C. L. (2011). Firm versus partner measures of auditor industry expertise and effects on auditor quality. Auditing: A Journal of Practice & Theory, 30(2), 201–229. Chi, H. Y., Weng, T. C., Chen, G. Z., & Huang, C. H. (2013). Ownership structure and managerial liability coverage. American Accounting Association Annual Meeting, Anaheim, CA. Chin, C. L., Lin, H. W., & Lin, M. F. (2002). On the association between IPO anomalies and management forecasts. International Journal of Accounting Studies, 34, 31–56. Choi, J. H., & Wong, T. J. (2007). Auditors' governance functions and legal environments: An international investigation. Contemporary Accounting Research, 24(1), 13–46. Chung, H. H., Hillegeist, S. A., & Wynn, J. P. (2012). Directors' and officers' legal liability insurance, audit pricing and litigation risk. SSRN working paper. Chung, H. H., & Wynn, J. P. (2008). Managerial legal liability coverage and earnings conservatism. Journal of Accounting and Economics, 46(1), 135–153. Classens, S., Djankov, S., & Lang, L. H. P. (2000). The separation of ownership and control in East Asian corporations. Journal of Financial Economics, 58(1), 81–112. Core, J. E. (1997). On the corporate demand for directors' and officers' insurance. The Journal of Risk and Insurance, 64(1), 1636–1669. Core, J. E. (2000). The director's and officers' insurance premium: An outside assessment of the quality of corporate governance. Journal of Low Economics and Organization, 16(2), 449–477. Das, S., & Sengupta, S. (2001). Asymmetric information, bargaining, and international mergers. Journal of Economics and Management Strategy, 10(4), 565–590. DeAngelo, L. E. (1981). Auditor size and audit quality. Journal of Accounting and Economics, 3(3), 183–199. DeFond, M. (1992). The association between changes in client firm agency costs anduditor switching. Auditing: A Journal of Practice & Theory, 11(1), 16–31. DeFond, M., & Jiambalvo, J. (1993). Factors related to auditor–client disagreements over income-increasing accounting methods. Contemporary Accounting Research, 9(2), 415–431. DeFond, M., & Subramanyam, K. R. (1998). Auditor changes and discretionary accruals. Journal of Accounting and Economics, 25(1), 35–67. Denis, D. J., Denis, D. K., & Sarin, A. (1997). Agency problems, equity ownership, and corporate diversification. The Journal of Finance, 52(1), 135–160. Dopuch, N., Holthausen, R., & Leftwich, R. (1987). Predicting audit qualifications with financial and market variables. The Accounting Review, 62(3), 431–454. Fan, J. P. H., & Wong, T. J. (2005). Do external auditors perform a corporate governance role in emerging markets? Evidence from East Asia. Journal of Accounting Research, 43(1), 35–72. Francis, J. R., Maydew, E. L., & Sparks, H. C. (1999). The role of Big 6 auditors in the credible reporting of accruals. Auditing: A Journal of Practice & Theory, 18(2), 17–34. Frankel, R. M., Johnson, M. F., & Nelson, K. K. (2002). The relation between auditors' fees for non-audit services and earnings management. The Accounting Review, 77, 71–105 (Supplement). Greene, W. H. (2011). Econometric analysis (7th ed.): Prentice Hall. Griffith, S. J. (2005). Uncovering a gatekeeper: Why the SEC should mandate disclosure of details concerning directors' and officers' liability insurance policies. University of Pennsylvania Law Review, 154(5), 1147–1208. Guan, Y. D., & Chang, C. H. (2010). An empirical study on the impact of the Rebar event on the audit clients' stock price. Journal of Accounting and Corporate Governance, 7(2), 19–46. Hakim, F., & Omri, M.A. (2010). Quality of the external auditor, information asymmetry and bid–ask spread: Case of the listed Tunisian firms. International Journal of Accounting and Information Management, 18(1), 5–18.

13

Hay, D., & Davis, D. (2004). The voluntary choice of an audit of any level of quality. Auditing: A Journal of Practice & Theory, 23(3), 37–53. Heckman, J. J. (1979). Sample selection bias as a specification error. Econometrica, 47(1), 153–161. Holderness, C. (1990). Liability insurers as corporate monitors. International Review of Law and Economics, 10(2), 115–129. Hung, C. S., Lu, Y. H., & Ou, C. S. (2008). The critical deciding factors of auditor choice decision: An application of the decision tree technique. Journal of Accounting and Corporate Governance, 5(2), 55–78. Jones, J. (1991). Earnings management during import relief investigations. Journal of Accounting Research, 29(2), 193–228. Kim, J. B., Chung, R., & Firth, M. (2003). Auditor conservatism, asymmetric monitoring and earnings management. Contemporary Accounting Research, 20(2), 323–359. Kinney, J. R., & McDaniel, L. S. (1989). Characteristics of firms correcting previously reported quarterly earnings. Journal of Accounting and Economics, 11(1), 71–93. Knechel, W. R., Niemi, L., & Sundgren, S. (2008). Determinants of auditor choice: Evidence from a small client market. International Journal of Auditing, 12(1), 65–88. Kothari, S. P., Leone, A. J., & Wasley, C. E. (2005). Performance matched discretionary accrual measures. Journal of Accounting and Economics, 39(1), 163–197. Lambert, R., Leuz, C., & Verrecchia, R. (2007). Accounting information, disclosure and the cost of capital. Journal of Accounting Research, 45(2), 385–420. Lawrence, A., Minutti-Meza, M., & Zhang, P. (2011). Can Big 4 versus non-Big 4 differences in audit-quality proxies be attributed to client characteristics? The Accounting Review, 86(1), 259–286. Lin, C., Officer, M. S., & Zou, H. (2011). Directors' and officers' liability insurance and acquisition outcomes. Journal of Financial Economics, 102(3), 507–525. Mansi, S., Maxwell, W., & Miller, D. (2004). Does auditor quality and tenure matter to investors? Evidence from the bond market. Journal of Accounting Research, 42(4), 755–793. Monroe, G. S., & Teh, S. T. (1993). Predicting uncertainty audit qualifications in Australia using public available information. Accounting and Finance, 33(2), 79–106. Mutchler, J. F. (1985). A multivariate analysis of the auditor's going-concern opinion decision. Journal Accounting Research, 23(2), 668–682. O'Sullivan, N. (1997). Insuring the agents: The role of director's and officers' insurance in corporate governance. The Journal of Risk and Insurance, 64(3), 545–556. Pitman, J. A., & Fortin, S. (2004). Auditor choice and the cost of debt capital for newly public firms. Journal of Accounting and Economics, 37(1), 113–136. Sengupta, P. (1998). Corporate disclosure quality and the cost of debt. The Accounting Review, 73(4), 459–474. Sharma, V., Naiker, V., & Lee, B. (2009). Determinants of audit committee meeting fequency: Evidence from a voluntary governance system. Accounting Horizon, 23(3), 245–263. Shiue, M. J., Chang, Y. S., & Kao, C. T. (2008). Self-selection of auditors and audit fee. Journal of Contemporary Accounting, 9(2), 167–200. Taiwan Insurance Institute (2010). Liability Insurance Research Report. https://fsr.tii.org. tw/iiroc/fcontent/research/research03.asp Willenborg, M. (1999). Empirical analysis of the economic demand for auditing in the initial public offerings market. Journal of Accounting Research, 37(1), 228–239. Wynn, J. P. (2008). Legal liability coverage and voluntary disclosure. The Accounting Review, 83(6), 1639–1669. Yeh, Y. H., Lee, T. S., & Woidtke, T. (2001). Family control and corporate governance: Evidence for Taiwan. International Review of Finance, 2(1/2), 21–48. Yermack, D. (1996). Higher market valuation for firms with a small board of directors. Journal of Financial Economics, 40(2), 185–211. Zou, H., Wong, S., Shum, C., Xiong, J., & Yan, J. (2008). Controlling-minority shareholder incentive conflicts and directors' and officers' liability insurance: Evidence from China. Journal of Banking and Finance, 32(12), 2636–2645.

Please cite this article as: Chi, H.-Y., & Weng, T.-C., Managerial legal liability and Big 4 auditor choice, Journal of Business Research (2013), http:// dx.doi.org/10.1016/j.jbusres.2013.12.003