Does client-specific litigation risk influence the audit quality?

Does client-specific litigation risk influence the audit

quality?

Huiqi Ni

10824251

Academic year 2014-2015

Program code E001.6314M0244.S23

2

Content

1. Introduction ... 3

1.1 Background ... 3

1.2 Research question ... 4

1.3 Motivation ... 5

2. Literature review and hypothesis ... 6

2.1 Audit quality ... 6

2.2 Auditor litigation ... 7

2.3 The relationship between audit quality and audit litigation ... 8

2.4 Hypothesis ... 10

3 Research design ... 11

3.1 Data collection ... 11

3.2 Measures of audit quality ... 11

3.3 Measures of litigation risk ... 13

3.4 Other variables ... 14

References ... 错误！未定义书签。

1. Introduction

1.1 Background

Shareholders and the other stakeholders make their investments based on the financial statements of the company. The financial statements can provide information on the performance of the company, so it is important that the information on the financial statements is reliable and relevant. Because of the importance, the investors demand that external auditors should audit the financial statements and give a true and fair view. Therefore, the accounting and auditing standards ask for auditors to be independent and focus on audit quality. Similar to the auditing standards, the courts believe that the auditors need to focus on substance over form. For example, the US Supreme Court points out that the auditors ARE liable for the misleading financial reports even when the financial reports comply with US GAAP. Therefore, the auditors should be concerned with how the financial reports reflect the underlying economics.

The legal view of auditing suggests audit quality is related with audit failure or no audit failure. If an auditor is not independent or an independent auditor issues an incorrect audit report because of the failure to collect sufficient evidence, then an audit failure occurs. Conversely, a good audit means that an auditor complies with the auditor standards and issues a true and fair audit opinion on the client’s financial statements. Audit failure has economic outcome on auditors, companies and the stakeholders. It can cause potential liability payment and lead to damage about audit firms’ reputation. Therefore, litigation risk encourages the auditors to be cared about financial reporting quality.

4

Recently, more and more literature begins to focus on the litigation risk and audit failure. Seetharaman, Gul and Lynn (2001) use evidence from UK firms cross-listed on US market to explore litigation risk and audit pricing. They find that audit fees are higher for UK companies which list on the US market, since litigation risk exposure is higher in US. The auditors ask for higher price because of the high liability exposure. Prior researches also focus on the relationship between litigation and audit quality. Simunic and Stein (1987) have recognized the potential usefulness of litigation in judging audit quality. And Palmrose (1988) compares litigation activities of independent auditors and finds that auditors with relative low litigation activities represent higher quality suppliers and the high litigation activities normally represent low audit quality. Recently, there are more specific researches about the litigation risk and audit quality. Khurand Raman (2004) and Francis and Wang(2008) compare the different litigation risk environment in various countries and document evidence on the positive impact of litigation risk on audit quality. However, Choi et al (2008) indicate that country-level litigation risks are not related with auditors and these findings are inconsistent with the findings of Khurand Raman (2004). Therefore, Jerry and Guoping (2011) investigate the impact of client-specific litigation risk on the audit quality. They use a large sample of client-year observation from 1988 to 2006 and find that the level of discretionary accruals of Big N auditors’ client is more likely to be lower than that of non Big N auditors when the clients have high litigation risk. And they conclude that the audit quality differentiation between Big N and non Big N is more evident in high risk. The relationship between the litigation risk and audit quality has caused more and more interest.

1.2 Research question

Prior literature has investigated the relationship between the litigation risk and audit quality. According to Jerry and Guoping (2011), when the litigation risk is high, the level of discretionary accruals is low. And this paper also examines the relationship between the client-specific litigation risk and audit quality. However, this paper

specifically uses the real activities manipulation as proxy for audit quality. According to prior literature (Kim et al 2002), real activities manipulation consists of managing cash flow from operation activities, discretionary expense and production costs. I investigate the relationship between litigation risk and the three individual components of real activities manipulation.

The primary objective of the paper is to explore whether a client with high litigation risk influences the audit quality in the U.S. And specifically examine the relationship between the earnings quality and the three component of real activities manipulation. Since real activities manipulation and accruals are two ways of earnings management, I also include the discretional accruals as another proxy for audit quality to investigate the relationship.

1.3 Motivation

Although there are literatures that investigate litigation risk among countries and through industries, the research on the impact of client litigation risk on the audit quality is few. Prior researches have explored the relationship between the litigation activities and the quality of the audit. Simunic and Stein (1987) recognize the usefulness of litigation in judging audit quality and Palmrose (1988) considers the litigation activities as an indicator of the audit quality. The analysis of the audit quality in industries with different litigation risk in this paper can provide a deep understanding of the relationship between the litigation and the audit quality.

The paper also extends the Khurand Raman (2004) and Francis and Wang (2008) which examine the litigation risk on country-level, since this paper focuses on the client litigation level and give a different investigation on the relationship between litigation risk and audit quality. Also by using the real activities manipulation as proxy for audit quality, it contributes to the literature of Jerry and Guoping (2011).

6

earnings.( Cohen et al 2008, Zang 2007). While Jerry and Guoping (2011) focus on discretionary accruals, this paper use real activities manipulation and give an

alternative view on the relationship between litigation risk and audit quality. Also, the proxy of discretionary accruals contributes to Jerry and Guoping (2011) and provides a more comprehensive view on the relationship between client litigation risk and audit quality.

Moreover, the litigation activities against audit firms occurred in recent years cause large amount fees to audit firms and the reputation becomes worse. This damages the economic efficiency of the audit industry. Understanding the deep influences of litigation can help chose the suitable client portfolio and reduce the likelihood of litigation activities.

2. Literature review and hypothesis

2.1 Audit quality

Audit quality and its determinants have been widely discussed. Audit standards point out that audit quality is achieved by the issuance of the appropriate audit report on whether the clients comply with generally accepted accounting principles. Palmrose(1988) states that audit quality is defined in terms of the level of assurances, which means the probability financial statements contain no material omissions or misstatements. According to Palmrose (1988), high levels of assurances correspond to higher quality service and audit failures become less likely with higher quality services. However, according to Francis (2011), audit quality is a complex concept and cannot be reduced to a simple definition. Francis (2011) points out that there are gradations of audit quality across a continuum from a low to high-quality audits and the quality is affected by factors including audit inputs, audit process, the characteristics of accounting firms and regulatory institutions. A widely used definition of audit quality is that of DeAngelo (1981). DeAngelo (1981) defines audit

quality as “the market assessed joint probability that a given auditor will both discover a breach in a client’s accounting system, and report the breach.” This definition has two components: the likelihood that auditors detect the misstatements or omission; the appropriate action on the discovery.

Prior papers also directly link earnings-quality to audit quality. Jones (1991) use abnormal accruals to investigate audit quality. They find that clients of Big 4 auditors have lower abnormal accruals which imply less earnings management. Therefore, they conclude Big 4 have higher audit quality. According to Nelson e al (2002), they find evidence from Big 4 accounting firms that auditors detect earnings management behaviors and require clients to make appropriate actions. Also Ramanis and Lennox (2008) measure audit quality using engagement hours and find that when auditor takes more effort, the earnings quality of client is higher. Gunny and Zhang (2009) use the PCAOB’s accounting firm inspection report to explore whether PCAOB inspections can find out real audit quality during the inspection period. The paper use abnormal accruals, restatements and the going concern opinion to proxy for actual audit quality. The data in the paper are from PCAOB website during 2005.1.31 to 2009.1.31. They find that deficient PCAOB inspection reports are associated with higher abnormal accruals which proxy for audit quality. They also conclude that the seriously deficient report is associated with lower audit quality whose proxy is propensity to issue going concern opinion. Therefore, Gunny and Zhang (2009) document the direct link between client earnings quality and audit quality.

2.2 Auditor litigation

The users of financial statements expect that auditors will detect the material misstatement or omissions. If the auditors fail to do so, they will suffer from litigation activities when investors have losses. According to Palmrose (1988), litigation against audit firms normally involves initial discovery of misleading financial statements, the filing of law suit and finally resolution of these suits. And litigation entails costs such

8

as loss of reputation, loss of time.

There is a stream research on the factors that affect the litigation risk. Pratt (1994) examines the association between audit firm characteristics and audit firm litigation risk. With a field experiment including 243 audit partners and managers of Big 6 firms in USA, Pratt (1994) finds that the clients’ poor financial condition, high levels of sales growth, high level of market value are associated with a higher probability of audit failure. The asset structure of clients is also related to litigation risk. Fuerman (1997) examines the determinants of litigation against auditors using the sample of 86 observations from 1992 to 1997. The findings indicate that the insurance of an AAER, client company bankruptcy, class period length and the restatement of previously issued audited annual reports are positively associated with naming the auditors a defendant. Casterella (2010) analyzes the effect of audit firm characteristics on litigation risk. The paper uses a continuous measure of the cost of litigation and select data from a large insurance company to exam the link between several audit firm characteristics and audit-related litigation. They find that larger firms, firms experiencing rapid growth, firms that sue their clients, firms with a history of regulatory problems and firms that choose smaller deductibles all face greater litigation risk. Schmidt (2012) examines the effect of auditors provided non-audit services on the litigation against auditors. He finds that the non-audit services are positively related with the lawsuits against auditors.

2.3

The relationship between audit quality and audit

litigation

There is research that examines the relationship between litigation and audit quality. Some research investigates the litigation activities as a measure of audit quality. Successful litigation against auditors can be a measure of an audit failure. According to Simunic and Stein (1987), they have recognized the potential usefulness of litigation in judging audit quality. And Palmrose (1988) compares litigation activities

of independent auditors and use the litigation activities as a mean to distinguish the audit quality. Palmrose (1988) use the empirical analysis and take the data of 472 legal cases and resolutions for those cases. Those cases include the suit against the Big 8 and non Big 8 audit firms from 1960 to 1985. The results suggest that the Big 8 auditors have lower litigation activities than the non-Big 8 auditors. Finally, the paper finds that auditors with relative low litigation activities represent higher quality suppliers and the high litigation activities normally represent low audit quality.

More research has also focused on the relationship between litigation risk and audit quality. Green (1999) study the litigation risk for auditors and the risk society. The paper points out that in Canda the audit firms tried to have the Provincial governments approve the establishment of Limited Liability Partnerships. Similar initiatives also happened in US. Therefore, the paper takes the “risk society” model to investigate the reason why the audit industry focuses on reduced liability exposure. Recent study also begin to investigate whether a country’s legal system influence the auditors’ behavior. Since the legal systems may propose the standard of care that auditors must meet to satisfy the legal responsibility. Failure to meet the legal responsibility may lead to the liability exposure.

Khurana and Raman (2004) investigate whether the perceived higher quality of a Big 4 audit is related to country-level litigation risk. The paper use the cost of equity capital as a proxy for financial reporting credibility and the paper indicates that the higher the audit quality, the higher the financial reporting credibility. It concludes that Big 4 auditees have a lower cost of equity capital than non Big4 auditees in US, but the situation is not in Australia, UK and Canada where litigation risk is relative lower than US. Francis and Wang (2004) also test whether Big4 auditors treat their clients consistently around the world and whether their treatment is related to the legal systems in different countries. Specifically, they use the total and abnormal accruals and they find that accruals of Big4 clients are small in countries with greater investor protection. The result shows that audit behavior is affected by legal incentives and

10

litigation risk. Furthermore, Francis and Wang（2008）investigate deeply into the topic. They examine whether earnings properties of Big 4 clients observe in US also existed in other countries. Using stock price and earnings per share data from 1994-2004, the paper finds that Big 4 auditors impose a high level of earnings quality on their clients in countries where there is high litigation risk. Beside the above papers that investigate litigation risk under country-level, Jerry and Guoping (2011) investigate the impact of client-specific litigation risk on the audit quality. They use a large sample of client-year observation from 1988 to 2006 and use the discretionary accruals as proxy for audit quality. Jerry and Guoping (2011) mainly focus on the audit quality differentiation and the client litigation risk. They find that the level of discretionary accruals of Big N auditors’ client is more likely to be lower than that of non Big N auditors when the clients have high litigation risk. These studies seem to show that the audit quality will be higher when the litigation risk is low.

2.4 Hypothesis

According to Caramanis and Lennox (2008) and Gunny and Zhang (2009), they document that audit quality and earning quality have direct link. They find that when the audit quality is low, the earning of the client is also of low quality. Henienger(2001) also indicates that higher earnings management leads to more auditor litigation Therefore, this paper focuses on earnings management to reflect the audit quality. Since real activities manipulation and accrual earnings management are two ways to manage earnings, this paper use the real activities manipulation to show the earnings management. Based on the prior literature ( Khurana and Raman 2004, Francis and Wang 2008, Jerry and Guoping 2011), I stimulate the following hypothesis:

H1. The real activities manipulation is lower when client specific litigation risk is

high.

3 Research design

3.1 Data collection

I collect the financial statement data of listed companies in US and U.S big 4 auditors from 2003 to 2012 through the Compustat database and AuditAnalytics. And the stock data is found through the CRSP database using the Ticker and PERMNO. When I select samples, I impose the criteria that the firm needs to have available fully consolidated financial statements covering 12-month period, , the necessary data for the explanatory variables included in the litigation risk model and other control variables in the Empirical model . Originally, I find 92,832 observations from the above database. After deleting the missing data, it left 8350 observations. Furthermore the data which are unable to calculate the variables is also deleted and finally there are 7,128 firm-year observations.

3.2 Measures of audit quality

3.2.1 Real activities manipulation

There are input-based measures and output-based measures. Input-based measures evaluate audit quality using observable inputs to the audit process. These measures such as audit size, audit fee have been used in prior literature as proxy for audit quality. However, since these measures are inputs into auditing process and are all based on actual observed characteristics, it is not guaranteed that they can lead to observable output and therefore are noisy proxies. (Defond; Zhang. 2014, pp. 289).

According to Defond and Zhang (2014), output-based measures are those which focus on the supply side effects on audit quality. Output-based measures include restatements, financial reporting quality, cost of capital etc. Prior literatures about the relationship between litigation risk and audit quality have mainly used the

12

output-based measures as proxies for audit quality. Khurana and Raman (2004) which examines the relationship between country-level litigation risk and audit quality use the cost of capital as proxy for audit quality. Jerry and Guoping (2010) investigate the impact of client-specific litigation risk on the audit quality and use the discretionary accruals as proxies. Defond and Zhang (2014) take the cost of capital as a very indirect measurement of actual audit quality, since the audit has a small impact on company valuations compared to company performance

Therefore, since the input-based measures and the output-based measures such as cost of capital have obvious limitations, I will use the real activities manipulation as proxy for audit quality. Real activities manipulation is defined as management actions that deviate from normal business practices undertaken for purpose of meeting or beating certain earnings thresholds. (Roychowdhury 2006). Based on the Cohen et al. 2008 and Chi et al. 2011, I use three individual proxies to develop a comprehensive measure of real activities manipulation. Specifically, I compute the RAM index using the following individual measures: (1) abnormal level of operating cash flow (Abn_CF); (2) abnormal production costs (Abn_Prod); (3) abnormal discretionary expenses (Abn_Exp). According to Cohen et al 2008, the RAM Index is calculated as

Abn_CF- Abn_Prod+ Abn_Exp. High levels of REM index indicates high level of real

activities manipulation. Since the three individual measures provide more information about the real earning management than the RAM index alone, we report the results about the REM index and also the three individual proxies (Abn_CF, Abn_Prod,

Abn_Exp). If the result is consistent with the hypothesis, then the litigation score

should be negatively related with REM index, Abn_CF, Abn_Exp and positively related with Abn_Prod

3.2.2

Discretionary accruals

Discretionary accruals are always used to measure earnings management in the prior studies (Klein, 2002; Kothari et al, 2005; Sun et al, 2011; Kim et al, 2012). In this paper, I also use discretionary accruals as the second proxy for earning management.

Based on Sun(2001), I use a cross-sectional version of the modified Jones model since there are less restrictive data requirements. Since Kothari et al(2005) suggests discretionary accruals adjusted for a performance-matched firm’s discretionary accruals are less misspecified , I calculate the performance –matched discretionary accruals. Because managers manipulate earnings in two ways, downwards or upwards (Klein 2002), I use the absolute value of discretionary accruals to make analysis. If the result is consistent with the hypothesis, then the litigation score should be negatively related with ABS_DA.

3.3 Measures of litigation risk

Prior literatures indicate that litigation risk is associated with lots of factors. (Carcello and Palmrose 1994; Shu 2000). According to Krishnan and Zhang (2005), they develop a comprehensive measure of litigation score based on the findings of Shu (2000). Since Shu (2000) incorporate the findings of prior study, the model that Krishnan and Zhang (2005) developed is more recent and convincing. The model only use the significant variables in Shu’s (2000) model and compute the variables for the one-year period ending with the last day of the quarter. Therefore, the model in this paper is based on the model of Krishnan and Zhang (2005), the litigation risk is calculated in the following equation:

LITSCORE= 0.276*SIZE + 1.153*INV + 2.075*REC + 1.251*ROA + 1.501*LEV + 0.301*GROWTH +1.060*DELIST + 0.928*TECH + 0.463*OPINION - 10.049

Where

LISCORE = litigation score

SIZE = natural log of total assets at the end of the year. INV = inventory divided by total assets at the end of the year REC = receivables divided by total assets at the end of the year ROA = net income in the year deflated by total assets

14

LEV = total liabilities divided by total assets at the end of the year GROWTH = change in sales from the previous year to the current year

divided by the previous year’s sales

DELIST = 1 if the company is delisted because of financial difficulties within the next year, and 0 otherwise

TECH = 1 if the company’s SIC code is in the 2830s, 3570s, 7370s,

8730s, and between 3825 and 3839, and 0 otherwise.

OPINION = 1 if the company received a going concern opinion in the

previous year, and 0 otherwise.

I use this model to calculate the litigation score of the companies. Those samples with high litigation score will be categorized into high litigation risk group. While the sample companies with low litigation score will be categorized into low litigation risk group. And I will use the median litigation score as a criterion to compare. Litigation score equals to one if the company belongs to high litigation risk group and zero if the company is low litigation risk group.

3.4 Empirical Models

To explore the relationship between the audit quality and litigation risk, I estimate the following model based on the Kim et al (2012) and Jerry and Guoping (2011):

RAM Index =

𝛼

+ 𝑎

𝐿𝐼𝑇𝑆𝐶𝑂𝑅𝐸

+ 𝑎

𝐴𝐵𝑆

+ 𝑎

𝑆𝐼𝑍𝐸

+

𝛼

𝑀𝐵

+ 𝛼

𝐴𝐷𝐽_𝑅𝑂𝐴

+ 𝛼

𝐵𝐼𝐺4

+ 𝛼

𝐿𝐸𝑉

+ 𝛼

𝑅𝐷_𝐼𝑁𝑇

+

𝛼

𝐴𝐷_𝐼𝑁𝐷_𝐼𝑁𝑇

+ 𝜀

(1)

𝐴𝐵𝑆_𝐷𝐴

=

𝛼

𝑀𝐵

+ 𝛼

𝐴𝐷𝐽_𝑅𝑂𝐴

+ 𝛼

𝐵𝐼𝐺4

+ 𝛼

𝐿𝐸𝑉

+ 𝛼

𝑅𝐷_𝐼𝑁𝑇

+

𝛼

𝐴𝐷_𝐼𝑁𝐷_𝐼𝑁𝑇

+ 𝜀

(2)

Where

ABS_DA = absolute value of the performance-matched discretionary accruals RAM Index ( or COMBINED_RAM) = Abn CF - Abn Prod + Abn Exp

Abn CF = the level of abnormal cash flow from operation

Abn Prod = the level of abnormal production costs, where production costs are defined as the sum of costs of goods sold and the changes in inventory.

Abn Exp = the level of abnormal discretionary expense, where discretionary expenses are the sum of R&D expenses, advertising expenses.

LITSCORE = Litigation risk, equal to 1 for high litigation risk company and 0 for low

litigation risk company.

Size = natural Log of the market value of equity

MB = market- to-book ratio, measured as MVE/ BVE , where BVE is the book value

of the equity.

ADJ_ROA = industry-adjusted ROA, where ROA is measured as income before

extraordinary items scaled by total assets

BIG4= big auditors, equal to 1 for Big 4 auditors and 0 for non-big 4 auditors. Lev = long term debt scaled by total assets

RD_INT = R&D expense scaled by net sales for the year

AD_IND_INT = advertising intensity for the four-digit SIC code industry for the year,

measured by the ration of advertising expenditures to total firm assets.

In equation (1) and (2), we use the multiple regressions. Prior literature suggests that accruals and real activities are the two ways to manage earnings. Firms are more likely to engage in the real earnings management when their ability to manage

16

accruals is constrained. ( Roychowdhury 2006; Cohen et al 2008; Zang 2007 ). Zang (2007) documents that accrual earnings management and real earnings management function as substitutes. To control for the substitutive effect of the two earnings management ways, I include the ABS_DA, a proxy for accrual earnings management, in the real activities management equation as control variable. And in the accrual earnings management equation, I include the COMBINED RAM, a proxy for real activities management, as a control variable.

I also include several other control variables in the regression model to control for the possible effect on the earnings management. Skinner et al (2002) suggest that when firms have high growth opportunities, the earnings management is higher. Roychowdhury (2006) have similar findings. They find that firms with high growth opportunities and bigger size have greater earnings management. Therefore, I include the size of the firms and the growth opportunities in the regression model, using the proxy SIZE and MB respectively. The ROA is also used to control for the effect of firm performance on the estimation of earnings management.

Further, prior research shows that Big N auditors do high-quality audit and the earning management maybe different for the firms audited by Big N. (Francis et al 1999, Sun et al 2010) Therefore, the regression model includes a control variable, Big

4 for the firms who use one of big 4 auditors as their auditor. The regression model

also use Lev as a variable since Klein (2002) suggests that firms with high leverage level are more likely to manage earnings.

Based on McWilliams and Siegel (2000) and Kim et al (2012), I also include the R&D intensity and industry advertising intensity as control variables in the regression model. Since when R&D expense and advertising intensity of the firm are higher, the earnings are also higher.

4. Results

4.1 Descriptive Statistics

Table 1 presents descriptive statistics. The mean and median of absolute value of the discretionary accruals are 0.969, 0.985 respectively. The mean and median of real activities management is 0.064 and 0.067. Specifically, the mean value of CFO abnormal level, abnormal Discretionary expense and combined RAM are 0.061, 0.023 and 0.064 respectively, suggesting that on average, the firms do not seem to engage in the real activities management such as sales management. However, the mean of abnormal production cost is positive (0.021), indicating that the sample firms may seem to engage in earning management through overproduction.

For the control variables, the ROA is -0.255 and this negative number indicates that the sample firms are less profitable than the industry. Also, about 65.1% of our sample firms are audited by Big 4 accounting firms. The Advertising expense is on average 0.032, which is about 3.2% of the total assets.

TABLE1 :Descriptive statistics of selected firms Panel A : full

sample

n Mean Median Std.Dev

25th Percentile 75th Percentile Dependent variables ABS_DA 7128.000 0.969 0.985 0.131 0.962 1.000 CFO abnormal level 7128.000 0.061 0.064 0.487 -0.015 0.141

DIS exp Abnomal 7128.000 0.023 -0.024 0.282 -0.065 0.049

Prod Abnormal 7128.000 0.021 0.012 0.674 -0.149 0.190

Combined RAM 7128.000 0.064 0.027 0.989 -0.226 0.305

Variable of Interest

18 LITSCORE 7128.000 0.500 0.000 0.500 0.000 1.000 Control variables SIZE 7128.000 5.951 6.043 2.537 4.082 7.680 M/B 7128.000 2.651 2.022 27.141 1.081 3.576 ROA 7128.000 -0.255 0.040 5.365 -0.045 0.090 BIG 4 7128.000 0.651 1.000 0.477 0.000 1.000 LEV 7128.000 0.190 0.050 1.237 0.000 0.222 RD 7128.000 -3.154 0.000 318.791 -0.067 0.530 AD 7128.000 0.032 0.010 0.069 0.003 0.032

Panel B: Descriptive Statistics by High litigation risk firm VS low litigation risk firm

High litigation risk firm Low litigation risk firm

n Mean Median n Mean Median

Dependent variables

ABS_DA 7128.000 0.964 0.983 7128.000 0.974 0.987

CFO abnormal level 7128.000 0.035 0.047 7128.000 0.088 0.081

DIS exp Abnomal 7128.000 0.046 -0.022 7128.000 0.001 _-0.026

Prod Abnormal 7128.000 0.038 0.018 7128.000 0.004 0.005 Combined ram 7128.000 0.043 0.006 7128.000 0.085 0.045 Variable of Interest LITSCORE 7128.000 1.000 1.000 7128.000 0.000 0.000 Control variables SIZE 7128.000 5.077 4.972 7128.000 6.824 6.872 M/B 7128.000 2.461 1.845 7128.000 2.842 2.180 ROA 7128.000 -0.525 0.019 7128.000 0.014 0.052 BIG 4 7128.000 0.514 1.000 7128.000 0.788 1.000 LEV 7128.000 0.225 0.020 7128.000 0.154 0.087

RD 7128.000 -7.909 0.000 7128.000 1.599 0.090

AD 7128.000 0.034 0.011 7128.000 0.029 0.010

Panel B of table 1 compares descriptive statistics of variables between low litigation risk firms and high litigation risk firms. Using the litigation score formula above, we calculate the litigation score for every firm observation. Then the median of these litigation scores is used as criteria to divide the firms. The firms whose litigation score higher than the median are divided into high litigation risk group. Otherwise, the firms are low litigation risk group.

The mean of abnormal discretionary accruals for high litigation risk firms is 0.964, which is lower 0.974 of low litigation risk firms. Also, for real activities management, I find lower mean values of CFO abnormal level and combined RAM for high litigation risk firms than for low litigation risk firms. These results indicate that when the litigation risk is high, there are lower real activities manipulation and lower abnormal accruals. This is consistent with our hypothesis which real activities manipulation and discretionary accruals are lower when client specific litigation risk is high.

Moreover, I observe that high litigation risk firms are smaller than low litigation risk firms, since the mean value of Size for high litigation risk firms and low litigation risk firms are 5.077 and 6.824 respectively. I also find that the high litigation risk firms have smaller mean M/B and ROA than the low litigation risk firms. This implies that the earnings performance of high litigation risk firms is worse than low litigation risk firms. The mean leverage level for high litigation risk firms is 0.225, which is larger than low the 0,154 of low litigation risk firms.

20

litigation risk

Table 2 reports the results of multiple regression analyses which use the measures of real activities manipulation. Table 2 mainly shows the Combined RAM and also the specific three components of RAM, which are abnormal CFO, abnormal discretionary expense and abnormal production cost. The estimation of the abnormal CFO, abnormal discretionary expense and abnormal production costs can be found in Appendix. The results of Table 2 have the coefficient and the P-value which indicate the significant level. And it is mainly focus on the high litigation risk group.

For the regressions of CFO abnormal level, the estimated coefficient for litigation score is negative and significant since the p-value is 0.0126. Also, the Litigation score is negatively related with the Combined RAM. These results support with my hypothesis. Given the low real activities manipulation and high CFO abnormal level indicate high audit quality. The results suggest that when the client litigation risk is high, the audit quality is higher and therefore there is fewer real activities manipulation. The result is consistent with Sun and Guo (2011), who finds that audit quality is negatively associated with client litigation risk. However, for the regression of abnormal Discretionary Expense, the estimated coefficient for litigation score is positive. In addition, I find that the abnormal production cost is positively related with the litigation score. According to Kim et al (2002)， the higher abnormal CFO, higher discretionary expense and lower production cost indicate real activities manipulation. The results of abnormal production cost regression model and discretionary expense regression model suggest that the earnings management is positively associated with litigation risk. This suggests that the high litigation clients’ incentive to manage earnings has significant influence. Specifically, the effect of high litigation risk clients’ incentive to manage earnings may dominant the effect of the auditors’ incentive to constrain earnings management.

coefficient on the ABS_DA is positive for CFO abnormal level, DIS expense abnormal level and Combined RAM. And the coefficient on the ABS_DA is negative for production abnormal level. These are consistent with Cohen et al (2008). About the other control variables in the Combined RAM regression model, the coefficient on

Size, M/B and AD is positive, especially the coefficient on the M/B is significant. These results suggest that the real activities manipulation is higher for firms who have big size, large M/B ration and advertising expense. However, the coefficient on the ROA in the Combined RAM model is negative, indicating that when the firm has good earnings performance, the firm is less likely to engage in real earnings

TABLE 2 :Multiple regression of Real Activities Manipulation and Litigation Score

Coefficient P-Value

Coefficient

P-Value Coefficient P-Value Coefficient P-Value

LITSCORE -0.290

0.0126 * 2.41E-02

 0.008 **

-0.005

0.623

-2.94E-04

0.94603

ABS_DA

0.368

< 2e-16 *** 0.527

< 2e-16 *** -0.609

< 2e-16 ***0.413

0.0138 *

SIZE

-0.005

0.366

-0.003

0.200

-0.010

0.040 *

0.006

0.672

M/B

0.0004

0.272

0.000

0.010

0.001

0.00069 *** 0.004

1.06e-05 ***

ROA

-0.004

0.005** -0.004

5.7e-07 *** -0.008

2.53e-09 ***-0.002

0.521

BIG 4

0.062

0.018 *

-0.043

0.00023 ***-0.001

0.971 -0.079

0.261

LEV

-0.0008

0.893

-0.006

0.114

0.007

0.214 -0.033

0.0377 *

RD

2.612e-07

0.990

1.308e-06

0.921

1.320e-06 0.950 0.000

0.9292

AD

-0.712

7.41e-09 *** 0.657

< 2e-16 *** 0.005

0.961

2.068

1.51e-10 ***

CFO abnormal level

DIS exp Abnomal

Prod Abnormal

Combined ram

Panel C:High litigation risk group

22

management. Also, the coefficient on the Big 4 in the Combined RAM model is -0.097, which is negative. This suggests that the firms who have Big 4 as their auditors may engage in less real activities manipulation. This finding is consistent with Khurana and Roman’s（2004） and Francis and Wang’s (2009) and Sun’s et al (2011) findings that the Big 4 auditors have higher audit quality. I also find that the firms with higher leverage have lower level of real activities manipulation. Since the coefficient on the LEV in the Combined regression model is negative (-0.033). This finding is similar with Sun and Guo’s (2011) finding that the discretionary accruals are lower for firms have lager debt.

Taken the above results together, the main findings suggest that the litigation risk is negatively associated with the real earnings management. This finding supports the first hypothesis in the paper that real activities manipulation is lower when the litigation risk is higher. Also, the findings are consistent with Sun and Guo’s（2011） that audit quality is higher when the litigation risk is high. The findings in the above analyses support in Khurana and Roman’s（2004） and Francis and Wang’s (2009) another aspect. Since Khurana and Roman（2004） and Francis and Wang (2009) based their findings on country-level litigation explosure, the findings in above analyses based on the client litigation risk. All these papers find that the audit quality is higher when the litigation risk is high.

4.3

The

relationship between

Accrual-Based

earnings

management and litigation risk

Table 3 presents the results of regression analysis of discretionary accruals. Because the residuals can be related over time, the results also include the P value which suggests the significant level. The discretionary accruals can be positive or negative because managers are likely to manage earnings upward or downward. Therefore, the results below mainly show the regression analysis between ABS_DA which is the absolute value of discretionary accruals and the litigation risk. The ABS_DA

regression model is used to analyze the two group clients: the high litigation risk clients and the low litigation risk clients. The Table 3 shows the results based on two groups. The estimation of the abnormal discretionary accruals is in APPENDIX.

TABLE 3 :Multiple regression of Accrual-Based earnings management and Litigation Score

High litigation risk Low litigation risk

Coefficient P-Value Coefficient P-Value

LITSCORE -0.042 0.206355 -0.028 0.112 Combined RAM 0.007 0.013 * -0.029 <2e-16 *** SIZE 0.005 0.002 ** -0.001 0.413 M/B -0.0001 0.274 0.000 0.796 ROA 0.002 1.48e-08 *** 0.009 0.155 BIG 4 0.015 0.088 0.011 0.048* LEV -0.002 0.258 0.024 0.010 * RD -1.012e-07 0.989 0.000 0.540 AD 0.088 0.039 * -0.001 0.972

For the high litigation risk group and low litigation group, the coefficient on the litigation score are both negative. This means that the litigation risk and the discretionary accruals are negatively associated. When the litigation risk is high, there are fewer discretionary accruals. This result supports the second hypothesis that discretionary accruals are lower when the litigation risk is high. Compare the high litigation risk groups and the low litigation risk groups, the results show that the coefficient on the litigation score is -0.042 and -0.028 respectively. The comparison presents that there are fewer discretionary accruals in high litigation risk environment than in low litigation risk environment. This suggests that the audit quality is higher when the litigation risk is high, since the discretionary accrual is the proxy for audit quality. This finding consistent with the Sun and Guo’s (2011) findings that the audit

24

quality differentiation is more evident in high litigation risk environment.

Furthermore, I control for the real earnings management in the regression model. The coefficient on the Combined RAM of the low litigation risk group is negatively related with the ABS_DA. This suggests that the firms who choose to manage earnings through the discretionary accruals are less likely to engage the real activities manipulation. It is consistent with Chi et al (2011) which presents that discretional accruals and real earnings management are two ways to manage earnings and they are substitute with each other. However, in the high litigation risk group, the coefficient on the RAM is positive. This may indicate that the effect of auditors’ inventive to detect the earnings management fails to dominant the effect of managers’ incentive to manage earnings. Therefore, the high litigation risk firms engage in both accrual earnings management and real activities manipulation. About the other control variables for the high litigation risk groups, the coefficient on Size, ROA is positive and significant, which indicating that the firms with big size and good earnings performance may be more likely to engage the accrual-based earnings management. The LEV is negatively associated with the ABS_DA, suggesting that the firms who have higher leverage level engage in lower accrual-based earnings management.

In summary, based on the above analysis of regression model, the findings that the accrual-based earnings management is lower when the litigation risk is high are consistent with the findings of Sun and Guo (2011). Also, the results of the ABS_DA regression model support the second hypothesis in this paper.

5. Conclusion

This paper examines the relationship between the client litigation risk and audit quality. Specifically, I use the real activities manipulation and accrual-based earnings as proxies for audit quality. When the audit quality is high, the level of earnings management should be lower. Therefore, I hypothesize that the real activities

manipulation is lower when the litigation risk is high and the accrual-based earnings management is lower when the litigation risk is high.

The findings in this paper support the hypothesis. I find that generally the real activities manipulation is negatively associated with the litigation risk. Specifically, the results in the abnormal CFO level regression model and the combined RAM regression model indicate the negative relationship between the real activities manipulation and litigation risk. However, the results in the abnormal production cost regression model indicate that the relationship between the real activities manipulation and litigation risk is positive. The reason for this maybe is that the managers’ incentive to manage the earnings outweighs the auditors’ incentives to detect the earnings management. For the relationship between the accrual-based earnings management, I find that the discretionary accrual is negatively related with litigation risk and there are fewer discretionary accruals when the client litigation risk is high. Taken together, the evidence generally presents that there are fewer real activities manipulation when the client litigation risk is high and there are fewer discretionary accruals when the litigation risk is high. This means that the audit quality is high when the client litigation risk is high.

I obtain the results after controlling for the other determinants of the earnings management and the substitution between the real activities manipulation and the accrual-based earnings management. Since prior literature focus on the association between the higher audit quality and the country-level litigation risk. Khurana and Raman (2004) finds out that Big 4 auditees have a lower cost of capital than non-Big 4 auditees in US, but not in UK where litigation risk is low. Francis and Wang (2004) study whether Big 4 auditors treat their clients around the world is associated with the legal system of the countries. Both of these two papers conclude that audit quality differentiation between the Big N and the non-Big N is positively related with country-level legal system. Therefore, the research which focuses on the relationship between the audit quality and litigation risk on the client level is limited. Sun and Guo

26

(2011) investigate the association between the audit quality differentiation and the client litigation risk. So, this study mitigates the limited literature problem and provides more evidence on the relationship between the audit quality and client litigation risk.

Furthermore, this study makes contribution especially to Sun and Guo (2011) because of the proxies in the paper. The literature which focuses on the audit quality and client litigation risk is limited. Sun and Guo (2011) use the abnormal discretionary accruals as the proxy for audit quality and investigate the relationship between audit quality differentiation and client litigation risk. Since accrual-based earnings management and real activities management are two ways to manipulate the earnings, this paper uses both of the discretionary accruals and real activities manipulation as proxies for audit quality. Using the proxy of discretionary accruals, the findings about the relationship between the accrual-based earnings management and the client litigation risk in this paper can be used to evaluate the findings of Sun and Guo (2011) and maybe provide evidence to Sun and Guo’s (2011) findings. Also, because Sun and Guo’s (2011) use the sample from 1988 to 2006, my paper which use samples from 2003 to 2012 can help investigate whether the findings of Sun and Guo (2011) still valid in recent time period. Moreover, this study uses the real activities manipulation as proxy for audit quality. The findings about the real activities manipulation provide a more comprehensive view on the relationship between the audit quality and client litigation risk. This study complements the literature which simply focuses on the country-level litigation risk. And this study also contributes to the papers which investigate the relationship between audit quality and client litigation risk but does not use the real activities manipulation as proxy for audit quality. Since the literature which investigates the audit quality and client litigation risk is limited, I look forward to more future research which investigates more on client litigation risk and audit quality.

Reference

1. Ahsan Habib, Haiyan Jing, Md. Borhan Uddin Bhuiyan, Ainul Islam (2014). Litigation risk, financial reporting and auditing: A survey of the literature.

Research in accounting regulation 26(2014) 145-163.

2. Boon, J.P., Khurana, I. K., & Raman, K. (2011). Litigation risk and abnormal accruals. AUDITING: A Journal of Practice Theory, 30(2), 231-256.

3. Christensen, B.E, Omer, T.C, Sharp, N.Y, &Shelley, M.K. (2013). Pork bellies and public company audits: Have audits once again become just another commodity? Working paper, Texas A&M University.

4. Casterella, J.R., K.L. Jensen and W.R. Knechel (2011). Litigation Risk and Audit Firm Characteristics. Auditing: A journal of practice&Theory.29,(2),pp.71-82 5. Carcello,J.V. and Palmrose, Z.-V.(1994), “Auditor litigation and modified

reporting on bankrupt clients”, Journal of Accounting Research, Vol. 32, pp.1-30. 6. Cohen, D., A, Dey, and T. Lys.2008. Real and accrual-based earnings management in the pre- and post-Sarbanes-Oxley periods. The Accounting Review 83 (3): 757-787.

7. Chi, W., Lisic, L., Pevzner, M. (2011). Is Enhanced Audit Quality Associated with Greater Real Earnings Management? Accounting Horizons: pp.315-375

8. DeAngelo, L., (1981). Auditor Size and Audit Quality. Journal of Accounting &

Economics, 3: pp.183-199.

9. Defond. M. and Zhang .J. (2014). A review of archival auditing research. Journal

of Accounting and Economics 58: pp. 275-326.

10. Francis. J. and Wang. D. (2008). “The joint effect of investor protection and Big 4 audits on earnings quality around the world”, Contemporary Accounting Research, Vol.25, pp. 1-39.

11. Fuerman, R.D. (1997). Naming auditor defendants in securities class actions.

Journal of Legal Economics, 7(1), 72-91.

12. Gunny ,K., and T. Zhang ,(2013). PCAOB inspection reports and audit quality.

28

13. Jere R. Francis (2004). What do we know about audit quality? The British

Accounting Review 36(2004) 345-368.

14. Jere R. Francis (2011). A framework for understanding and researching audit quality. Auditing: A journal of practice& theory Vol.30, No2, pp. 125-152.

15. Jerry Sun, Guoping Liu (2010). Client-specific litigation risk and audit quality differentiation. Managerial Auditing Journal.

16. Jong-Hag Choi, Rajib K. Doogar and Ananda R. Ganguly (2010). The riskiness large audit firm client portfolios and changes in audit liability regimes: evidence from the U.S. audit market. Contemporary Accounting Research Vol.21 pp. 747-785.

17. Jones, J., 1991. Earnings management during import relief investigations. Journal

of Accounting Research (Autumn), 193-228

18. Khurana, I.K. and Raman, K.K. (2004), “Litigation risk and the financial reporting credibility of Big 4 versus non-Big 4 audits: evidence from Anglo-American countries”, The Accounting Review, Vol.79, pp. 473-98.

19. Knechel, W.R, Krishnan, G.V, Pevzner, M.,Shefchik, L.B., and Velury, U.K.(2013). Audit Quality: Insights from the Academic Literature. Auditing: A journal of

Practice & Theory, Vol.32,pp.385-421

20. Krishnan, J. and Zhang, Y. (2005), “Auditor litigation risk and corporate disclosure of quarterly review report”, Auditing: A journal of Practice & Theory, Vol.24, pp.115-38.

21. Klein, A., 2002. Audit committee, board of director characteristics, and earnings management. Journal of Accounting and Economics (August), 375-400

22. Kothari, S. P.,A, Leone, and C. Wasley. 2005. Performance matched discretionary accrual measures. Journal of Accounting and Economics 39(1):163-197

23. Pratt, J and J.D.Stice (1994). The effects of Client Characteristics on Auditor Litigation Risk and Judgements, Required Audit Evidence, and Recommended Audit Fees. The Accounting review, 73, (4),pp. 639-656

24. Palmrose , Z.V. (1988). “An analysis of auditor litigation and audit service quality”, The Accounting Review, Vol.63, pp.55-73.

25. Schmidt, J.J. (2012). Perceived auditor independence and audit litigation: The role of non-audit services fees. The accounting review, 87(3), 1033-1065.

26. Shu, S. (2000), “Auditor resignations: clientele effects and legal liability”, Journal

of Accounting and Economics, Vol. 29, pp. 173-205.

27. Sotheara Riel, Carl Tano (2014). The impact of the global financial crisis on audit quality: A study of publicly listed Swedish firms. Uppsala University Library. 28. Shu, S. (2000), “ Auditor resignations: clientele effects and legal liability”,

Journal of Accounting and Economics, Vol.29,173-205

29. Yang Xu, Elizabeth, C., Neil, F. and Liwei, J. (2013). Responses by Australian auditors to the global financial crisis. Accounting and Finance, 53(2013), 301-338.

30

APPENDIX: Measurement of earnings management

proxy

1. Real Activities Manipulation

Based on the prior literature (Cohen et al 2008), when the managers use real activities manipulation such as sales management, the current operating cash flow is likely to be lower. I use the Jones model in Roychowdhury (2006) to estimate the normal level of operating cash flows as following:

CFO_t At−1 = α + α1 1 At−1 + β₁ St At−1 + β₂ ∆St At−1 + εt Where

CFO_t= Cash flow from operations in year t A= Total assets

S= Net sales ∆S_t= S_t − S_t−1

For every firm-year, the abnormal level of operating cash flow is the residual which is the ε_t form the model.

Abnormal production cost is another measure of real activities manipulation. According to Roychowdhury (2006), the production costs are the sum of COGS and changes in inventory in that year. Therefore, based on Roychowdhury (2006), I estimate the normal level of COGD using following model:

COGS_t = the costs of goods sold in year t A= Total assets

∆S_t= S_t − S_t−1

I estimate the normal level of changes in inventory using the following model:

∆INV_t A_t−1 = α + α1 1 A_t−1 + β1 ∆S_t A_t−1 + β2 ∆S_t−1 A_t−1 + εt Where

∆INV_t = the changes in inventory in year t A= Total assets

∆S_t= S_t − S_t−1

∆St−1= the changes in sales of previous year

Since the production cost is defined as the sum of COGS and changes in inventory in that year. Use the above two equations, I estimate the normal level of production cost using the following equation:

PRODt At−1 = α + α1 1 At−1 + β1 St At−1 + β2 ∆St At−1 + B3 ∆St−1 At−1 + εt Where

PROD_t= the sum of COGD and the inventory changes in year t

A= Total assets S= Net sales ∆S_t= S_t − S_t−1

∆S_t−1= the changes in sales of previous year

The third measure of real activities manipulation is the abnormal discretionary expense. Based on the Roychowdhury (2006), I estimate the normal level of discretionary expense as following:

DISEXPt At−1 = α + α1 1 At−1 + β St−1 At−1 + εt Where

32

A= Total assets S= Net sales

2. Discretionary Accruals

Based on Kothari et al.(2005) and Sun and Guo (2011), I use the residuals from cross industry regression model to estimate the firm’s discretionary accruals. The following model is used to estimate the normal level of discretionary accruals.

𝐴𝐶𝐶 A_t−1 = α1∗ 1 A_t−1 + β1 ∆St A_t−1 + β2∗ ( PPE A_t−1)εt Where

ACC = total accruals, measured as the difference between earnings before extraordinary items and discontinued operations and cash flows from operations.

A= Total assets ∆S_t= S_t − S_t−1