The effects of industry specialization and client importance on audit quality for financially distressed clients

(1)

The effects of industry specialization and client

importance on audit quality for financially

distressed clients

Lucas Conijn, 10676376

18th of June, 2014

Supervisor: Jeroen van Raak

Master thesis Accountancy&Control

(2)

1

Abstract

This study investigates whether audit quality is higher for industry specialists and whether the relative client importance affects auditor independence using US data from the period 2008-2012. The sample was constructed using publicly available data on US-based auditors from AuditAnalytics and Compustat and contains 6,825 unique firm-year observations with an Altman Z-score of lower than 3. Audit quality is proxied by the propensity to issue a going-concern audit opinion. I find significant evidence that industry specialists provide higher audit quality than non-specialists. I find no evidence that client fee dependence influences the propensity to issue a going-concern opinion. These results hold when including non-audit fees as well.

1. Introduction

1.1 Background

One of the main assumptions underlying financial statements is the going-concern assumption. Under this assumption a company is expected to continue operations in the foreseeable future and not go out of business. This assumption is vital for the valuation of assets, as it means that assets can be valued upon their business value in use rather than their termination value, which is in general a lot lower. If a firm is not expected to continue to stay in business in the foreseeable future, the auditor can give an adverse opinion in the form of a going-concern opinion. If a going-concern opinion from the auditor is appropriate, but not issued, Francis (2004) considers this an audit failure. The going concern opinion is an important signal for investors as it is off course vital for them to know whether the firm they are investing in will continue its operations in the future. Altman (1982) finds that a going concern opinion is seen as a signal of potential bankruptcy.

A firm is required to present its current state in the financial statements, but obviously has incentives to present itself as good as possible to be able to attract cheap capital.

The value of an external audit is derived from the improved decision making investors can do on the basis of independently, professionally checked information, assuring that the financial information disclosed is correct and representative of the current state of business.

The party paying the audit firm is the audit client. The audit firm as a private firm, operating on a for-profit basis has its own incentives to be attractive to potential clients. This can create a conflict of interest for the auditor (Moore et. al, 2003). Especially firms performing less well may be inclined to overstate their performance and be scared off by too strict audits. Given the relation between auditor and audit client, the auditor is incentivized to

(4)

3 be biased in favor of their clients (Moore et al., 2003). The audit fees from clients provide commercial pressures on the auditor. On the other hand there are also pressures that incentivize the auditor to act independent and issue the going concern opinion when appropriate. The market and regulation provide these incentives as reputation and litigation effects from misreporting are detrimental to the auditor (DeAngelo, 1981a; Carcello&Palmrose, 1994). Auditors’ reports are valuable only if the market recognizes these as trustworthy and may be doubtful to invest in a firm if they believe that the financial information presented by a firm is not reliable. An auditor caught misreporting damages his reputation and thereby its attractiveness to potential clients. The larger the auditor, the larger his reputational losses will be when caught misreporting. Nelson et. al. (2002) find that not only auditor size may affect independence, but also that the size of the audit client is very relevant, as larger clients are found to have more bargaining power in persuading the auditor to engage in aggressive accounting practices. Contradictory to the findings by Nelson et al. (2002) are the findings by Crasswell (2002), who finds that there is no difference in the propensity to issue a qualified opinion on the basis of fee dependence. The findings by Reynolds&Francis (2001) contradict Nelson et al. (2002) even more as they find that large clients are treated even more conservatively by their auditor. The main driver for this conservatism was found to be the litigation risk, which is significantly greater for larger clients. It is difficult to draw a conclusion on the basis of these contradicting results, but it is a certainty that there are cases documented where large clients convinced auditors to acquiesce to aggressive accounting policies, Enron being the most famous one. Enron was reported to contribute 35% of the audit fees for the Arthur Andersen office in Houston (Francis, 2004). One could argue that such dependence on the fees of one client should always raise suspicion.

Another interesting development in the audit business is the increasing specialization of auditor along certain industry lines, as is found in the 2003 report by the general accounting office (GAO) (Cahan et. al., 2008). Auditors specializing in certain sectors can differentiate themselves from the unspecialized competitors and become the premium supplier of audits in their respective fields of specialization. Differentiation is interesting from a business perspective as Mayhew&Wilkins (2003) find a much stronger audit fee bargaining position for industry specialists, whereas expanding market share in without specialization leads to a lesser bargaining position, consistent with Porter’s theory (1980). A similar research conducted by Fung et al. (2012) confirmed these results at the office-level of analysis.

The increasing auditor specialization is an interesting phenomenon. As will be discussed in the literature section, there are strong research findings that indicate an improved

(5)

4 audit quality from industry specialized auditors (Carcello&Nagy, 2002; Balsam et al., 2003; Dunn& Mayhew, 2004; Carcello&Nagy, 2004; Romanus et al. 2008; Reichelt&Wang, 2010). When analyzing auditors, the researcher has the possibility to do research on different levels: firm-wide level, nation-wide level, office level and audit team level. Francis (1999) argues that the appropriate level of analysis is the office level as this is the level where audit contracting occurs. There are two other reasons for conducting the research at the local office level. Firstly, Francis (1999) finds that in the majority of the cases, the national industry leader is not the city-specific leader. So, if market share, which is the most common proxy for industry specialization, is to account for industry specialization, we may get much distorted results from selecting national industry leaders as opposed to local industry leaders. Secondly, fee dependence is much more present at the local level. Francis (2004), for instance, reports that, at the national level, Enron accounted for only 2% of Arthur Andersen’s revenues, a very small proportion when held against the 35% at the local level. However, the reputational risk suffered by Arthur Andersen did not pertain to only the Houston office, but actually caused the collapse of the entire firm. Independent behavior by the auditor is vital for its reputation and thereby its value-delivery, but may be jeopardized by the financial dependence on large clients.

1.2 Research question

On the basis of the background the following research question was developed: For financially distressed companies, how are industry specialization and client fee dependence affecting audit quality?

This research question is a twofold question where the effect of industry specialization on audit quality will be examined, and the effect of client importance on audit quality is examined. The logical two sub-questions are thus:

1. What is the effect of industry specialization on audit quality for financially distressed clients?

And,

2. What is the effect of client fee dependence on audit quality for financially distressed clients?

Answering these research questions using an empirical approach should provide useful insights in the effects these two variables have on the audit quality, and thereby on the decision usefulness of the information. The paper will look at these relations at the office-level following the argument of Francis (1999) stated in the introduction. The variables will

(6)

5 be explained further on in the paper. The academic contribution as well as the societal contribution is below.

1.3 Motivation

1.3.1 Academic contribution

This paper attempts to contribute to the audit quality literature by investigating the relationship between auditor specialization and audit quality, and the relation between client-fee dependence and audit quality at the local-office level for financially distressed clients. Reichelt&Wang (2010) have conducted a similar research with regards to the relation between auditor specialization and audit quality, but have used a different method to distinguish their sample of financially distressed clients. Li (2009) has looked into the relation between relative client size and audit quality, but has made use of different methods. Carcello&Nagy (2004) investigated a similar research question looking at the interaction effect between auditor industry specialization and client size. Their research, however, was conducted at the national level. I hope to contribute to the literature by investigating the effects of the two independent variables on audit quality, at the office level, specifically for financially distressed clients. The focus on financially distressed clients is important as, according to the argument of Defond&Jiambalvo (1994), these companies are more inclined to engage in earnings management practices. A higher quality of the external audit would prevent these practices and keep the reliability of financial statements on an acceptable level.

1.3.2 Societal contribution

From a societal point of view I consider this research question to be very interesting. Given the aforementioned increasing industry specialization in the audit business, it is especially interesting to see whether the differentiation strategies pursued by audit firms actually lead to an increased audit quality in a situation where the client would desire a more lenient attitude from the auditor to remain attractive. From an investor perspective it would, of course, be absolutely undesirable if industry specialists would be less willing to issue a going-concern opinion than non-specialists. Ideally, the specialist auditor would have a higher propensity to issue a going-concern opinion as they should be more capable of detecting problems. The specialist auditor receives a fee premium over its competitors (Ferguson et al., 2003; Ferguson et al., 2006; Francis et al., 2005; Fung et al., 2012), so it should be capable of providing higher audit quality. In addition to this, the market also positively values the contracting of a specialist auditor as found in a study by Knechel (2007). This study would ideally confirm the perceived added value from contracting an industry specialist.

(7)

6 With regards to the second sub-question, it will be interesting to see whether in situations where the client contributes significantly to the total revenues of the office, the auditor is still, at least equally, willing to issue a going-concern opinion as it would be to smaller clients. If the research were to show that clients that contribute a larger share of the revenues also have a larger say in how the audit firm conducts its business, this could be a worrying indication that the incentives in the audit market are not strong enough to enforce independent behavior and conservatism to protect investors. In an ideal situation, the research would show that larger clients are not treated more lenient, but are treated equally conservative to smaller sized clients. Finding this result would imply that the results from previous studies (Li, 2009; Reynolds&Francis, 2001) hold for financially distressed clients as well.

2. Literature review and hypotheses

2.1 Introduction

In this section I will review the previous literature on audit quality and the factors influencing audit quality. On the basis of Deangelo’s (1981b) definition of audit quality this section will analyze the reasons for industry specialization, the valuation of the market of specialization, and the effects of specialization. Secondly, I will look at the motives for, and against disclosing a discovered breach and the findings of previous studies on this. The final section will discuss the influence of financial distress on the relations between auditor industry specialization and audit quality, and the influence of client importance on audit quality. From the analysis of prior literature I will derive my hypotheses, which will be tested empirically.

2.2 Audit quality

In a review of the literature on audit research Francis (2004) points out that there is a very small percentage (<1%) of outright failures, where outright failures are described as situations where either GAAP is not enforced by the auditor or when the auditor does not issue a modified or qualified report in the appropriate circumstances. Legally satisfying the requirement of having the financial statements audited thus seems not to be too big of a problem. Francis (2004) argues that audit quality is best conceptualized as a continuum as opposed to the dichotomy between failure and non-failure as there is a large difference in audit quality above the legal minimum. For this paper I will take the definition of Deangelo (1981b) as a starting point. Deangelo (1981a) describes audit quality as: 1. The probability that an auditor will detect a breach in the accounting system, and 2. The likelihood of

(8)

7 disclosing the detected breach. This definition of audit quality is especially useful as these two probabilities can easily be related to the main influential variables in this research. Firstly, the probability that an auditor will detect a breach in the accounting system depends, among other factors, on the skill of the auditor. Specializing in specific industries will result in increased attention for, and knowledge of the specific characteristics of these industries. By specializing the auditor differentiates itself from competition in terms of quality. Whether audit quality actually increases will be discussed further on in this section, but it is clear that specialization, in theory, should lead to an increased probability of detecting a breach in the accounting system, when present. Secondly, the likelihood of disclosing any detected breaches directly relates to the independence of the auditor. In this research I will see whether a more important client, in terms of revenues, will be treated more lenient. A more lenient treatment could indicate impaired independence and have negative effects on the reliability of financial information disclosed to investors.

2.3.1 Industry specialization

As mentioned in the previous section, audit quality is best conceptualized as a continuum (Francis, 2004). The GAO 2003 that was mentioned in the background section describes an increasing trend of industry specialization. The choice to specialize in certain industries can be seen as a differentiation strategy in terms of Porter’s generic strategies framework (1980). Differentiators provide a special product or a higher quality product that their competitors cannot deliver, in return for this, the price to be paid for this product is higher than with cost-leaders. In most situations, the consumer has the decision to either acquire a product from a cost-leader or a differentiator, where he has to make a trade-off between price and quality. In the situation of selecting an auditor, the situation is slightly different, as the party paying the bill is not the party that uses the acquired product. The users of financial statements, usually investors, are ensured that the quality of the financial statements is adequate and that these faithfully represent the financial position of the firm. Having a higher quality audit performed on the financial statements will give a higher level of assurance that the statements are correct. An auditor that provides this higher quality can distinguish itself from competitors and thereby become a preferred supplier of audit services. This illustration of the situation brings about a number of questions; Is specialization valued by investors? Is the quality of audit actually higher when done by a specialist? And finally, is the client willing to pay more for this product?

(9)

8

2.3.2 Market reactions

As described above, the beneficiaries of the external audit are the investors. Indirectly, the firm hopes to benefit from increased investor trust as a consequence of higher quality assurance on their financial statements. Knechel et al. (2007) show how the market values the selection of specific auditors by examining the stock market returns as a result of an auditor switch. They find the strongest effects when switching from a non-big 4 auditor to a big 4 auditor (consistent with Deangelo’s (1981a) theory on larger reputational losses) but also find that when firms switch from a non-specialist big-4 auditor to a specialist big-4 auditor, the market positively reacts to the change. Their findings provide strong incentives for auditors engage in specialization and for firms to hire specialist auditors.

2.3.3 Quality effects of specialization

The market thus seems to attach value to auditor industry specialization, but does specialization actually lead to higher quality? The study by Cenker&Nagy (2008) indicates that auditors take into account their firm’s industry specializations when evaluating their client portfolios. They describe that in the period following SOX implementation firms started re-evaluating their client portfolios as a result of increased risk related to auditing, in terms of reputation as well as litigation. Their findings show that specialization decreases litigation risk and clientele mismatch and thereby increases audit quality. The effects were only conclusively proven at the local level. Furthermore the study by Balsam et al. (2003) shows clients of industry specialists have higher earnings quality, proxied by the discretionary accruals of a client. Dunn&Mayhew (2004) find that clients audited by an industry specialist have higher disclosure quality. Their results were only valid in unregulated industries. This is explained by the fact that in regulated industries the legal minimum level of reporting is already very high, so no demand/possibility for differential audit quality over that exists. Reichelt&Wang (2010) study the effects of auditor industry specialization using two proxies for audit quality: earnings quality (abnormal accruals) and the propensity to issue a going concern report. They find lower abnormal accruals for clients of specialized auditors as well as a higher propensity to receive a going concern report. Carcello& Nagy (2002) and Carcello&Nagy (2004) test the relationship between auditor industry specialization and the likelihood of fraudulent financial reporting and find that this likelihood is significantly smaller for clients of industry specialists.

Romanus et al. (2008) use yet another measure to proxy for audit quality: accounting restatements. They find that clients of industry specialists require significantly less restatements than others. Also, their study shows that switching from a non-specialist to a

(10)

9 specialist auditor increases the likelihood of restatements in the core operating accounts, and vice-versa. Their findings show that especially in the accounts that contain sector specific information the audit quality of an industry specialist is significantly higher. This study very clearly shows that specializing enhances the industry specific knowledge needed to accurately assess the correctness of accounting figures produced. Besides the increased level of skill found with industry specialists, Francis, Crasswell&Taylor (1995) also find that the reputation of industry specialists also plays a larger role. The specialist auditor will be inclined to protect its reputation even more fiercely than other respected auditors to retain its position as the preferred supplier in the respective industry. Specialization is thus found to have positive effects on both probabilities identified by Deangelo (1981a); 1. The probability that an auditor will detect a breach in the accounting system; and 2. the likelihood of disclosing the detected breach.

No research was found indicating that auditor industry specialization has negative, or no effects on audit quality.

2.3.4 Pricing of specialist audits

Industry specialization is found to have a positive effect on audit quality and the market values this. The remaining question is whether audit clients consider these effects valuable enough to reward the auditor for differentiating. Much research has been done on this topic. The research by Basioudis&Francis (2007) finds that city-specific industry leadership commands a premium from clients. More studies in different settings show this effect, but also show that being a national industry leader alone does not lead to a premium (Ferguson et al. 2003; Ferguson et al., 2005; Ferguson et al. 2006), which is consistent with Francis’ (1999) and Carcello’s (1992) argument that clients value the attributes of the local office over the attributes of the auditor nation-wide. Mayhew&Wilkins (2003) provide evidence consistent with Porter’s (1980) theory on competitive strategy by showing that only industry specialists focusing on specific industries are able to bargain for a premium fee, whereas large players in the market, that do not differentiate will be pressured by the clients to pass on some of the scale benefits by lowering the audit premium. This is consistent with Porter’s (1980) theory, where the undifferentiated cost-leader is attractive through its low price, rather than its high quality. The price premium for specialist auditors makes the research question very relevant as the premium should be justified by a higher observed audit quality.

(11)

10

2.4.1 Auditor independence

The second dimension of audit quality: auditor independence will be discussed here.

Auditor independence is vital for audit quality, as a biased auditor’s reports will have little informational value for investors. DeAngelo (1981) explains the commercial incentives that would cause a bias with the auditor in favor of the client, as well as the market-, and regulation-based incentives that would push the auditor in the direction of more independent behavior. An auditor may acquiesce with faulty accounting in order to please the client, hoping to retain the client for coming years. However, when caught, an auditor will lose its credibility lowering its attractiveness to potential clients and lowering the value of its delivered product to existing clients, which could lead to a renegotiation of the audit fee or a switch by the client.

Given the importance of the matter, much attention has been given to the topic of auditor independence. Below is a review of the found literature that will aid in the development of my hypothesis.

2.4.2 Findings on auditor independence

A research by Crasswell (2002) finds that fee dependence does not affect auditors’ propensity to issue unqualified reports. The results are indicative that the market-, and regulation-based incentives are stronger than the commercial-client incentives. Their research was also conducted at the local office level, as well as the national level, with findings similar for both levels of analysis. The study by Li (2009) finds in a post-SOX setting that there is actually a positive relation between audit fees and the propensity to issue a going-concern opinion. These results may, however as Li indicates, be subject to some bias given the time-frame. 2003, the year where the positive relation was found, was off course a very turbulent time for auditors and may not be representative of ‘normal’ business. The results by Li (2009) were found to be driven mainly by Big 4 auditors. Reynolds&Francis (2001) test the relation between the propensity to issue a going concern report and client fee dependence and find that larger clients are treated more conservatively than smaller clients. Larger clients receiving a going concern report is consistent with market-, and regulation-based incentives.

A somewhat contradicting result was found by Nelson et al. (2002). In their study they tested whether client size affected the likelihood that an attempt at earnings management required adjustment. Their results showed that auditors were more likely to require smaller firms to adjust than larger firms. Their results are explained by the fact that larger clients have more resources to defend an aggressive accounting position.

(12)

11 Finally, Carcello&Nagy (2004) investigated the combined effects of auditor industry specialization and client size. Their results showed a weaker relation between auditor industry specialization and audit quality for larger clients. They used the natural log of the client’s assets as variable for size. This variable is not directly a measure of dependence, as were a percentage of the audit office’s revenues, but does give an indication of the revenues a client brings in, as audit fees are usually determined on the basis of size. The decreased strength of the relation could, according to the authors, be explained through either increased complexity of the client, or client pressure. Additional tests showed that the weaker relationship was not the consequence of increased client complexity, but was explained by the client size.

Previous literature is not entirely conclusive on the matter of independence as some find signs of impaired independence for larger clients, where other research finds no signs of this, or even increased scrutiny for these larger clients. It is difficult to formulate a hypothesis on the basis of prior literature here, but I will attempt to do so in a following section.

2.5 The effects of financial distress

Firms in financial distress run the risk of going out of business in the foreseeable future. This is obviously detrimental to the attractiveness of the firm for investors as they will be less likely to make a positive return on their investment. Firms in financial distress, on the other hand, will be keen on attracting capital to finance business activities to regain profitability or pay off debts. Carcello&Palmrose (1994) finds that financially distressed firms can potentially pose a threat for the auditor. Firms going bankrupt can result in lawsuits for the auditor, especially when the auditor did not issue a going concern report. The paper finds consistent evidence that modified reports can protect the auditor against litigation risk. An interesting finding is that the litigation risk is larger for auditors of large clients. Based on this, one would expect a stricter scrutiny on large clients as the risks are higher. Furthermore, Stice (1991) finds a more conservative accounting approach with auditors of firms close to violating debt covenants, which is a logical consequence of the findings of Carcello&Palmrose (1994). Defond&Jiambalvo (1994) find that firms violating debt covenants actively attempt to upward manipulate earnings. The need for higher quality audits, thus, does not only arise from the fear of litigation, but also from the increased incentives for firms to engage in aggressive accounting practices. From these researches one could conclude that financial distress places the auditor under a magnifying glass. The issuing of a going concern opinion will be very painful for a distressed firm, increasing the likelihood of an auditor switch, whilst at the same time the market will punish the auditor even more severe when it fails to warn for potential bankruptcy.

(13)

12

2.6 Research level

The research will be conducted at the local office level, as previous research (Francis, 1999) finds this to be the appropriate level for auditing research. Francis argues that this is the case, as audit contracting and decision making occurs at the local level. The paper also documents that auditor reputations vary significantly from city to city. The example of Enron mentioned in the introduction is illustrative of the fact that client fee dependence at a national level is much less present than at the local level, hence its effect on auditor independence is less visible. Furthermore, Ferguson et al. (2003), Ferguson et al. (2006), Francis (2005), and Reichelt&Wang (2010) finds the local level to be the appropriate level for auditing research as national industry leaders, in general, do not provide the industry leader for each city. Hence, looking at national market shares and client importance will give a much distorted image of reality.

2.7 Hypotheses

On the basis of the prior literature I decided to test the following hypotheses:

H1: Industry specialization has a positive effect on audit quality for financially distressed clients

H2: Client fee dependence has a positive effect on audit quality for financially distressed clients

The first hypothesis is based on a rather straightforward message from prior literature, as all research (Balsam et. al., 2003; Carcello&Nagy, 2004; Cenker&Nagy, 2008; Dunn&Mayhew, 2004; Reichelt&Wang, 2010; Romanus et. al., 2008) pointed towards increased audit quality as a result of industry specialization. I found no reason why these findings would not apply in a setting of only financially distressed clients. Finding support for H1 would indicate that industry specialist auditors provide investors with a higher level of reliability of the financial statements issued by financially distressed firms. Investors would have a better idea of whether the firm will stay in business long enough to earn back their investment.

The second hypothesis is more difficult to formulate as the previous literature (Carcello&Nagy, 2004; Li, 2009; Crasswell, 2002; Nelson et. al., 2002; Reynolds&Francis, 2001) found no conclusive evidence on the relation between fee dependence and audit quality. On the basis of the findings on financially distressed firms (Carcello&Palmrose, 1994; Defond&Jiambalvo, 1994) I expect the audit quality to be higher for larger clients in terms of audit fee contribution.

Finding support for H2 would imply that auditors are taking into account the litigation risk and anticipate market reactions to an audit failure, and therefore are inclined to enforce more

(14)

13 conservative accounting upon its client. One could argue that this is desirable as, firstly, a firm should not be able to pressure the auditor, reducing the reliability of financial statements, and secondly, the consequences of an audit failure on a large client going bankrupt are likely greater than those of a smaller client.

Figure 1: theoretical model

3. Methodology

3.1 Sample selection

The sample selected exists of all firm-year observations of financially distressed listed firms in the US from 2007 to 2012. Due to the economic crisis, a large number of firms found itself financial distress; therefore I expect this period to yield a sufficient sample to test the model. The sample will consist firm-year observations with valid audit fee data with an Altman Z-score of lower than 3. I will apply the appropriate formula to compute the Altman Z-Z-score for manufacturing and non-manufacturing firms. The cut-off score of 3 was found appropriate to identify firms potentially heading for bankruptcy (Altman, 2000). As a starting point I downloaded all audit fee data of publicly traded companies from AuditAnalytics for the period 2007-2012 rendering a sample of 94,025 unique firm-year observations. Deleting all non-US auditors reduced the sample to 85,477 audit opinions. Removing all audits with no fee data or a fee of $0,- brought the sample brought the sample down to 66,476 observations. The sample was matched to the US Census file that places cities in their respective MSA. Due to incomplete city data in the original sample, 10,922 observations were eliminated. An additional 1,961 observations were removed where no industry data was included. The removal of another 415 duplicate firm-year observations rendered a sample of 53,180. This sample was used to compute the market shares of auditors in local industries, as well as the computation of client fee dependence. The sample was matched to Compustat firm financial data on the basis of CIK company identifiers. This rendered a sample of 31,176 firm-year observations, of which 21,315 had all data required to compute the Altman Z-score. Subsequently, all financial institutions were filtered from the sample, which was reduced by

(15)

14 743 to a sample of 20,572. From this sample all firm-year observations with an Altman Z-score of more than 3 were removed rendering a sample of 13,353 observations. Dropping all first firm-year observations brought the sample back by 2,905 observations, to a total of 10,448. The first firm-year observations were removed, as the study only focuses on first-time concern opinions. For the first observation it is not possible to find whether a going-concern opinion was issued previously. Deleting all first firm-year observations and all firms that had received a going-concern opinion in a prior year rendered a sample of 8,897 observations. Data trimming on all continuous variables to filter out outliers reduced the sample to 7,902 observations. Finally, all city-industry-year combinations with only one observation were removed, leading to a final testing sample of 6,825 audits. The removal of single city-industry-year observations is consistent with Reichelt&Wang (2010) and prevents bias as a single observation would always make the auditor in question a specialist under both definitions used in this study. The testing sample contained 243 first-time going-concern opinions, 70 different MSAs, 53 industries, 612 city-industry combinations over the period 2008-2012.

Table 1 Sample selection

AuditAnalytics 2007-2012 94,025

- Non-US auditors (- 8,548) 85,477

- Invalid fee data (- 19,001) 66,476

-No city-data (- 10,922) 55,556

- No SIC data (- 1,961) 53,595

- Duplicate firm-year observations (- 415)

Market share & Fee dependence computation sample

53,180

- Missing Altman data (- 31,865) 21,315

- Financial institutions (- 743) 20,572

- Non-distressed firms (- 7,219 13,353

- First-time firm-year obs. (- 2,905) 10,448

- Non-first-time GCs (- 1,551) 8,897 - Outliers (- 995) 7,902 - Monopolists (- 1,077) Final sample 6,825 Nr of first-time GCs 243 Nr of MSAs 70 Nr of industries 53 Nr of city-industry combinations 612

(16)

15

3.2Method

This section introduces the model used to test the hypotheses developed in section 2.7, explains the variables, their measurement, and the motivation behind the inclusion of chosen control variables.

To test H1 and H2 I will be estimating a logistic model:

FTGC= β1SPEC+ β2CFD+ β3COSTAT+ β4CURR+ β5LEV+β6ALTMAN+ β7RLAG+ β8SECTIER+ β9OPLOSS+ β10LOSS+ β11ROA + β13LNSIZE+ β14LIT+ β15BIG4 Where:

FTGC=1, if the auditor issued a first-time going-concern audit opinion, 0 if not SPEC= 1, if the auditor is an industry specialist, 0 if not

CFD= continuous measure of the fee as total percentage of audit fees COSTAT=0, if the firm is still active, 1 if it is out of business

CURR= current ratio of a firm – current assets/current liabilities LEV= control measure for leverage – liabilities/assets

ALTMAN=Altman bankruptcy predictor as control measure for the severity of financial distress

RLAG= control factor for reporting lag BIG4= 1 if Big 4, 0 if not

SECTIER= 1, if audited by Grant Thornton or BDO

OPLOSS= 1, if the firm reported negative operating cash flows LOSS=1, if the firm reported negative net income

ROA= return on assets – net income/assets LNSIZE= natural logarithm of assets

LIT=1, if the firm is in a high litigation industry1

3.3 Variable measurement

3.3.1 Industry specialist definition

As industry specialization is not directly observable, market share will proxy for this. This proxy is widely accepted as a measure of industry specialization (Dunn&Mayhew, 2004). To define an auditor as an industry specialist this paper will use the market share of the local office. Industries are defined on the basis of 2-digit SIC codes. The total market will be defined as the sum of total audit fees paid by the respective industry, where the percentage a single auditor receives represent its market share. As Defond (2002) found no relation between audit quality and non-audit fees, the paper will only take audit fees into account. An auditor can be an industry specialist if it is the largest in terms of market share, consistent with Carcello&Nagy (2004), or if the auditor has a market share over 50% (Reichelt&Wang, 2010). Following Cenker&Nagy (2008), the local level will be represented by MSA2s, thus

1

Industries with SIC codes: 2833-2836, 3570-3577, 3600-3674, 5200-5961, and 7370 were defined as high litigation industries (Reichelt&Wang, 2008)

2

Forbes glossary defines an MSA as follows: “Metropolitan Statistical Area is a geographic entity designated by OMB for use by federal statistical agencies. A metropolitan statistical area consists of at least one urbanized

(17)

16 the market share will be based on the percentage of the market an auditor audits in the specific MSA, excluding local industries with less than two firms, and excluding financial services industry due to the fundamentally different nature of their business. Market share calculations are based on the largest sample available with valid audit fee-, and industry data (53,180). Separate regressions will be done to test whether having the largest market share, being the market leader, is associated with higher audit quality, or that a sufficiently large market share has to be held to be associated with higher audit quality. The second definition is thus more restrictive than the first as a highly fragmented market will always have a market leader, but not necessarily be dominated by one party.

3.3.2 Client importance

Client importance is proxied by fee dependence, which represents the financial importance, or relative size, of an audit client for the auditor. Consistent with Li (2009) the measure fee dependence will be calculated simply be dividing the client’s audit fee by the total audit fees an accounting office makes. The ratio audit fee divided by total revenues of the local office in a particular year is calculated using the largest available sample with valid audit fee data (53,180).

3.3.3 Control factors

I will be controlling for a number of factors that are found to be of influence on the likelihood to receive a going-concern opinion. I will discuss these factors below, including their expected sign.

The first control variable is company status. This is a dichotomous variable indicating whether a business is currently still in business or is out of business. Although it does not indicate the reason why a firm went out of business, I consider it a proxy for bankruptcy, as it is highly likely that a financially distressed firm going out of business did so because of bankruptcy. The expected sign for this variable is positive as it would be well justifiable to issue a going concern opinion to a firm that eventually goes bankrupt.

The second control variable is the firm’s current ratio (Li, 2009), computed as the firm’s current assets divided by the firm’s current liabilities. This is a common indicator of a firm’s liquidity as current liabilities exceeding current assets may lead to difficulties in meeting financial obligations. The expected sign is negative.

area with a population of 50,000 or more, along with adjacent territory with a high degree of social and

(18)

17 The third control variable is financial leverage, computed as a firm’s liabilities divided by its assets. Highly leveraged firms are considered more risky, therefore an auditor will be more likely to issue a going-concern to these firms (Li, 2009). The expected sign is therefore positive. Despite selecting the sample based on a firm’s Altman Z-score (Reichtelt&Wang, 2010), I do still control for the variable as firms with a Z-score at the lower end of the spectrum are still more likely to receive a going-concern opinion ceteris paribus. I expect a negative relationship between Z-score and GCAO. The fifth control variable that is included in the model is reporting lag, representing the number of days between the fiscal year end and the reporting date. This measure is included as firms with long reporting lag are associated with going concern opinions (Li, 2009). The expected sign is positive.

Big N auditors are found to provide higher quality audits (DeAngelo, 1981), therefore I will control for the fact whether an auditor is a big 4 auditor, or not, as is consistent with previous research (Carcello&Nagy, 2004; Reichelt& Wang, 2010). I expect a positive relationship between BIG4 and GCAO. I also control for second-tier auditors, being Grant Thornton and BDO, consistent with Reichelt&Wang (2010). Included in the model are also indicator variables for operating loss and loss. Operating loss means that the firms report negative operating activities cash flow. Loss means the firm represents negative net income. Both variables are expected to be positively associated with receiving a going-concern (Reichelt&Wang, 2010; Li, 2009). In a similar fashion I expect ROA (return on assets) to be negatively associated with the likelihood of receiving a going concern opinion, as these variable indicates the ability of a firm to generate income (Li, 2009). I expect a negative association between the log of size, as Defond (2002) explains that large firms have more resources to negotiate in the threat of a potential bankruptcy. The last variable that is controlled for is the sensitivity of the industry to litigation. LIT represents a dichotomous variable, being 1 if the firm is in a high litigation industry. I expect a positive association between LIT and FTGC, as an auditor will likely be more conservative in fear of possible lawsuits.

4. Results

4.1 Descriptive statistics

4.1.1 Full sample

This section gives the descriptive statistics of the final sample. The final sample consists of 6,825 unique, complete firm-year observations. The sample contains 243 first-time going-concern receivers, a percentage of 3.5%, which is relatively low compared to the 6% in the

(19)

18 sample of Li (2009). The average market share of (active) auditors in a local industry is 37.6%. On average, observed audit fees represent 9.7% of an office its revenues, which is comparable to Li (2009). In total, 3379 of the 6825 (36%) audits in the sample were performed by the market leader. In 2176 of the 6825 (31.9%) of the cases the audit was performed by an auditor with a market share over 50%, these figures are very comparable to the 35% and 32.7% found in Reichelt&Wang (2010). A group of 1111 firms (16.3%) is no longer in business. The big 4 audited 4663 (68.3%) of the observations, an additional 557 (8.2%) were audited by Grant Thornton or BDO. High litigation industries were found in 1482 (21.7%) of the observations. The average reporting lag is 69 days. The Altman Z-scores observed in the sample have a mean of negative 1.9, indicating a high degree of financial distress over all. The average leverage ratio in the sample is 0.63.

3484 firms reported a net operating loss and 1864 reported a negative net income. On average a negative return on assets of 12.1% was found. The current ratio of the observations was 2.05. The descriptive statistics on the control variables roughly resemble those found in previous research (Li, 2009; Reichelt&Wang, 2010).

4.1.2 First-time going-concern sample

As mentioned previously, the number of first- time going-concern opinions issued was 243. The market leader issued 58 of these going-concern opinions. In 50 cases the auditor held a market share of over 50%. The average fee dependence of these observations is 10.6%, against 9.7% in the total sample. The number of firms that went out of business in this subsample is 54 (22.2%). The mean ROA is significantly lower at negative 63.2%. Almost all of the firms, 229 (94.2%), of the firms reported a loss, where 184 (75.7%) reported a net operating loss. Big 4 auditors issued 90 (37.0%) of the going-concern opinions, making their presence in this subsample significantly lower than in the full sample of distressed firms. The Altman score averages negative 11.84, which indicates that the financial distress in the subsample is significantly higher than in the regular test sample. The reporting lag in this subsample is substantially higher than in the regular sample with an average of 92 days versus 69 in the total sample. The firms in this subsample are also significantly less liquid with a current ratio of 1.46 versus 2.05 in the total sample, however indicating that current assets are still, on average, larger than current liabilities. High litigation industries are more present in the subsample, with 29% of the observations being from high litigation industries, versus 22% of the observations in the general sample being from high litigation industries.

(20)

19

4.1.3 Correlation among variables

Table 3 presents the correlations among variables. Some of the variables show rather high correlation coefficients, such as the correlation between RLAG and BIG4 at negative -0.49 is also rather high, indicating that the reporting lag is lower in big-4 firms. Furthermore I observe a strong negative correlation between LNSIZE and RLAG (-0.62), LNSIZE and ALTMAN (0.44), LNSIZE and BIG4 (0.62), LNSIZE and OPLOSS (-0.51), LNSIZE and LOSS (-0.44), LNSIZE and ROA (0.46). For obvious reasons the variables LOSS and OPLOSS are rather strongly correlated (0.49). Despite these high correlations, none of the correlations cause multicollinearity as all the variance inflation factors are only 3.22 and lower, which is below the suggested problematic threshold for multicollinearity of 10 (Li, 2009). Consistent with Li (2009), I excluded some of the variables with higher variance inflation factors to observe differences in coefficients, but found no significant changes in coefficients, suggesting multicollinearity is not a problem.

(21)

(22)

(23)

6,825 observations. Table 4 shows this for the definition of specialist with a minimum of 50% of the market, table 5 for the definition of specialist when being market leader.

Table 4 Specialist at +50% market share

The first definition of specialist is more restrictive as it requires a majority of the market to be in the hands of one auditor, whereas the second definition allows for a specialist even in a highly fragmented market, where the market leader may hold only e.g. 10% of the market. In 2008 the sample contains 474 unique city industry combinations, representing 50 industries in 65 different MSAs. PWC holds a market share over 50% in 36 city-industry combinations, EY in 41, Deloitte in 29 and KPMG in 25. Grant Thornton and BDO both dominate 7 local industries. In 2009 the sample contains 469 unique city industry combinations, representing 50 industries in 61 different MSAs. PWC is specialist in 34 cases, EY in 36, DT in 32, KPMG in 24. Grant Thornton is specialist in 9 local industries, BDO in 6. Smaller audit firms dominate 23 local industries. In 2010 the sample contains 453 unique city industry combinations, representing 55 MSAs and 51 different industries. PWC, EY and Deloitte are found to be specialists in 33 cases, KPMG in 26, GT and BDO in 7 and 11, respectively. Non-big 6 firms dominate 22 local industries. In 2011 the sample contains 422 unique city industry

Auditors/Fiscal Years 2008 2009 2010 2011 2012 Average

PWC 36 34 33 31 30 33 EY 41 36 33 36 36 36 DT 29 32 33 29 34 31 KPMG 25 24 26 22 24 24 Grant Thornton 7 9 11 8 7 8 BDO 7 6 7 7 7 7 Others 25 23 22 18 21 22

Total city-industry specialists 170 164 165 151 159 162

Total cities 65 61 55 56 57 59

Total industries 50 50 51 51 52 51

Total city-industry combinations 474 469 453 422 426 449

Total city-auditor combinations 355 339 341 323 326 337

(24)

23 combinations, 56 different MSAs and 51 industries. PWC is the dominant party in 31 of these local industries, EY in 36, Deloitte in 29, KPMG in 22, GT in 8, BDO in 7, and smaller auditors in 18. In 2012 the sample contains 426, representing 52 industries and 57 MSAs. PWC is dominant in 30 of the cases, EY in 36, DT in 34, and KPMG in 24. BDO and GT both dominate 7 industries. Smaller firms dominate 21 local industries. In this sample of financially distressed firm-year-observations on average 162 specialists are observed in 449 local industries.

Table 5 Market leader definition of specialist

*

the number of city-industry combinations does not match the number of city-industry specialists. All industries have a market leader, but the market leader does not necessarily audit an included observation, as the sample only includes distressed firms (altman<3)

Under the market leader definition of specialist PWC is found to be a specialist in 37 of the local industries in 2008. EY is specialist in 41 industries, DT in 30, KPMG in 28, GT in 10, BDO in 8 and others in 28. In 2009 PWC leads 36 local industries, EY 37, DT 35, KPMG 28, GT 12, BDO 6 and others 24. In 2010 PWC leads 34 markets, EY 35, DT 35, KPMG 30, GT 12, BDO 7 and others 24. In 2011 PWC is the leader in 32 local industries, EY in 38, DT in 32, KPMG in 26, GT in 9, BDO in 7, and others in 22. In the final year PWC leads 31 local markets, EY 37, Deloitte 34, KPMG 27, GT 8, BDO 7 and smaller firms 24. The sample includes an average of 174 industry specialists from a total of 449 unique industry city combinations.

Auditors/Fiscal Years 2008 2009 2010 2011 2012 Average

PWC 37 36 34 32 31 34 EY 41 37 35 38 37 38 DT 30 35 35 32 34 33 KPMG 28 28 30 26 27 28 Grant Thornton 10 12 12 9 8 10 BDO 8 6 7 7 7 7 Others 28 24 24 22 24 24

Total city-industry specialists 183 178 134 166 168 174

Total cities 65 61 55 56 57 59

Total industries 50 50 51 51 52 51

Total city-industry combinations 474 469 453 422 426 449

Total city-auditor combinations 355 339 341 323 326 337

(25)

24

4.3 Test results

This section discusses the results of the logit regressions that are done to test the hypotheses. The first hypothesis expects that industry specialists deliver higher audit quality. Specialist auditors are expected to have a more conservative attitude and therefore issue a going-concern opinion faster than their competitors. The logic underlying this hypothesis is twofold, as explained in the previous sections. Firstly, the auditor has a more extensive knowledge of the industry and is therefore able to better comprehend the financial figures of clients and identify potential attempts of earnings management. Secondly, the auditor has to protect its reputation as specialist to remain the preferred supplier of external audit services and justify any price premiums it charges over competitors. The second hypothesis expects that the larger the fee dependence of an auditor on a specific client is, the more likely this auditor is to issue a going-concern opinion. The reasoning behind this is that relatively large clients pose a larger litigation threat, and that it would be even more important to demonstrate independence in act, as financial independence is impaired.

Two logit regressions are run. The outcome of the logit regressions is presented in in table 6. Both regressions include the client fee dependence variable and all control variables. The first regression is done using the market leader definition of an industry specialist, the second one using the 50% percent minimum market share cut-off to define an industry specialist. The model specified in the methodology section is used to test the hypotheses. Variables are explained in the methodology section.

Under the market leader definition of industry specialist the association between receiving a first-time going-concern and being audited by a market leader is positive (coefficient 0.07), but not significant at any level (p=0.36). The model in its entirety is significant at the p<0.001 level with a pseudo Rsquared of 0.3191. The client fee dependence variable has a negative sign, indicative of a negative association between the relative size of a client and the propensity to issue a going-concern opinion, however the coefficient of -0.43 is not significant (p=0.359).

Under the second definition of specialist, requiring a minimum of 50% market share the reported outcome is a coefficient of 0.34, which is significant at the 0.05 level, one-tailed (p=0.074). The difference in outcomes between the two regressions is indicative of the fact that the largest participant in the market is not necessarily the best, but that controlling the majority of the market is indicative of specialization and higher audit quality. I expect the difference to be driven by highly fragmented markets, where the market leader is not able to gain diversification advantages.

(26)

25 Under both models all the control variables have the predicted signs, with exception of SECTIER under the first model. COSTAT is significant at the 0.05 level, showing a strong association between receiving a first-time going-concern and potentially going out of business. The current ratio is also significant at the 0.01 level, indicating that reduced liquidity is an important sign for auditors in their decision to issue a going-concern opinion. Reporting lag is found to be positively associated with the issuance of a first-time going-concern opinion, the association is significant at the 0.01 level. Operating loss and loss indicators both have a significant impact on the dependent variable. This effect is pretty clear cut, as profitable organizations or organizations with profitable operations are not likely heading for financial problems. ROA is also significant at the 0.01 level. The log of size is found to be negatively associated with receiving a going-concern, Defond (2002) explains that larger firms have more negotiating power to avoid potential bankruptcy, but this effect is not significant. The Big 6 are, unlike expected, not significantly associated with the likeliness of receiving a going-concern opinion. Li (2009) explains this outcome through the clientele differences between big N auditors and non-big N auditors. I also find an expected positive association between leverage and the propensity to issue a going-concern opinion, however this result is not significantly different from zero. Furthermore the test does not find a significant negative association between the Altman Z-score and the likelihood of receiving a going-concern. It could be possible that, because the sample is already selected on the basis of Z-score indicating severe financial distress, there is no further distinction between the firms on this characteristic. Litigation risk is positively associated with the issuance of going-concern opinions (0.014), but insignificant.

(27)

26

Table 6 Logit regression results

The dependent variable is the likelihood of issuing a first-time GC ( n=6825) Industry specialist at market

share>50%

Industry specialist at market leader Expected sign Coefficients Ρ-value Coefficients P-value

SPEC + 0.340 0.074 0.067 0.36 CFD + -0.430 0.359 -0.382 0.414 COSTAT + 0.320 0.079 0.320 0.079 CURR - -0.435 0.000 -0.433 0.000 LEV + 0.071 0.706 0.083 0.66 ALTMAN - -0.003 0.573 -0.003 0.606 RLAG + 0.038 0.000 0.038 0.000 BIG4 + 0.052 0.823 0.123 0.596 SECTIER + -0.002 0.994 0.009 0.973 OPLOSS + 1.026 0.000 1.028 0.000 LOSS + 1.309 0.000 1.292 0.000 ROA - -1.211 0.000 -1.197 0.000 LNSIZE - -0.068 0.266 -0.069 0.254 LIT + 0.014 0.936 0.008 0.964 Pseudo R2 0.3191 0.3177 Prob > chi2 0 0 4.4 Sensitivity analysis

To test sensitivity of my results I rerun the logit regression using a 30% cut-off to define a specialist, consistent with Romanus (2008). I find no significant association between

SPEC30P and FTGC, indicating that a specialist has to be dominant in the market in order to be associated with higher audit quality. Furthermore, I test whether including non-audit fees to compute client fee-dependence influences the results, which would be indicative that non-audit fees play a role in the independence of non-auditors. I find no significant association when including the non-audit fees in the fee-dependence computations. Results are presented on the next page in table 7.

(28)

27

Table 7 sensitivity analysis

The dependent variable is the likelihood of issuing a first-time GC ( n=6825)

Industry specialist at market share>30% Coefficients Ρ-value SPEC30P 0.079 0.659 CFDTOT -0.411 0.381 COSTAT 0.320 0.079 CURR -0.434 0.000 LEV 0.084 0.655 ALTMAN -0.003 0.612 RLAG 0.038 0.000 BIG4 0.109 0.642 SECTIER 0.006 0.983 OPLOSS 1.030 0.000 LOSS 1.291 0.000 ROA -1.199 0.000 LNSIZE -0.069 0.258 LIT 0.013 0.944 Pseudo R2 0.3178 Prob > chi2 0 5. Conclusion

This study investigates the effect of industry specialization and the effects of relative client size on audit quality. Audit quality is proxied by the propensity to issue a going-concern opinion. Industry specialization is defined in two ways. The first definition of an industry specialist is that the auditor in question is the market leader, whereas the second definition requires a minimum market share of 50%. The first definition is looser, as satisfying the second definition automatically implies being the market leader, but this does not work vice versa. The final sample consists of 6,825 unique firm-year observations that reported an Altman Z-score lower than 3.00. The sample contained only publicly available data from AuditAnalytics and Compustat from the period 2008-2012.

The results show a significant association between industry specialism when it is coupled with a minimum market share of 50% and audit quality. I find no significant results when the market leader is by default defined as specialist, without requiring a minimum market share percentage. From these results I conclude that industry specialization does lead to higher audit quality. These findings are highly relevant in the current audit market as the descriptive

(29)

28 statistics show that approximately one third of the local industries are dominated by one auditor, holding more than 50% of the market. The financial statements of firms audited by a specialist auditor thereby have a significantly higher degree of reliability than those of non-specialists. An important note is that we can assume that all CPA firms auditing publicly traded companies have to, and still do, adhere to a legal minimum audit quality. This study only looks at the differential quality over this minimum standard set by the SEC to protect investors and ensure proper quality audits.

The results show no significant association between the relative client size and audit quality. These results hold when taking into account non-audit fees as well. These findings are consistent with previous research (Crasswell, 2002) and imply that auditors are not significantly influenced in their decision to issue a going-concern audit opinion by the commercial importance of the respective audit client.

The main limitation of this study is the use of many proxies. Firstly, the propensity to issue going-concern opinion proxies for audit quality, as audit quality is not directly observable. Secondly, the use of market share to define specialization may be a limitation as it may not necessarily indicate specialist expertise. The final limitation arises from the dataset. The final dataset contained (only) 6,825 observations and 243 first-time going-concern opinions. A larger dataset may potentially find stronger results.

(30)

29

7. References

Altman, E. (1982). Accounting implications of failure prediction models. Journal of Accounting, Auditing, and Finance 6 (Fall): 4-19.

Altman, E. Corporate Financial Distress: A Complete Guide to Predicting, Avoiding, and Dealing with Bankruptcy. New York: John Wiley and Sons, 1983.

Altman, E. (2000). Predicting financial distress of companies: revisiting the Z-score and ZETA models. Stern School of Business, New York University, 9-12.

Cahan, S. F., Godfrey, J. M., Hamilton, J., & Jeter, D. C. (2008). Auditor specialization, auditor dominance, and audit fees: The role of investment opportunities. The Accounting

Review, 83(6), 1393-1423.

Carcello, J., Hermanson, R. and McGrath, N. (1992), “Audit quality attributes: the

perceptions of audit partners, preparers, and financial statement users”, Auditing: A Journal of Practice & Theory, Vol. 11 No. 1, pp. 1-15

Carcello, J. V., & Palmrose, Z. V. (1994). Auditor litigation and modified reporting on bankrupt clients. Journal of Accounting Research, 1-30.

Carcello, J. V., & Nagy, A. L. (2002). Auditor industry specialization and fraudulent financial reporting. In Symposium on Auditing Problems.

Carcello, J. V., & Nagy, A. L. (2004). Client size, auditor specialization and fraudulent financial reporting. Managerial Auditing Journal, 19(5), 651-668.

Choi, J. H., Kim, C., Kim, J. B., & Zang, Y. (2010). Audit office size, audit quality, and audit pricing. Auditing: A Journal of Practice & Theory, 29(1), 73-97.

Craswell, A. T., Francis, J. R., & Taylor, S. L. (1995). Auditor brand name reputations and industry specializations. Journal of accounting and economics,20(3), 297-322.

Craswell, A., Stokes, D. J., & Laughton, J. (2002). Auditor independence and fee dependence. Journal of Accounting and Economics, 33(2), 253-275.

DeAngelo, L. E. (1981). Auditor size and audit quality. Journal of accounting and

economics, 3(3), 183-199..

DeAngelo, L. E. (1981). Auditor independence, ‘low balling’, and disclosure regulation. Journal of accounting and Economics, 3(2), 113-127.

DeFond, M., 1992. The association between changes in client firm agency costs and auditor switching. Auditing: A Journal of Practice and Theory (Spring), 16–31.

DeFond, M. L., & Jiambalvo, J. (1994). Debt covenant violation and manipulation of accruals. Journal of accounting and economics, 17(1), 145-176.

(31)

30 DeFond, M. L., Raghunandan, K., & Subramanyam, K. R. (2002). Do non–audit service fees impair auditor independence? Evidence from going concern audit opinions. Journal of

accounting research, 40(4), 1247-1274.

Dunn, K. A., & Mayhew, B. W. (2004). Audit firm industry specialization and client disclosure quality. Review of Accounting Studies, 9(1), 35-58.

Ferguson, A. C., Francis, J. R., & Stokes, D. J. (2006). What matters in audit pricing: industry specialization or overall market leadership?. Accounting & Finance, 46(1), 97-106.

Francis, J. R., & Wilson, E. R. (1988). Auditor changes: A joint test of theories relating to agency costs and auditor differentiation. Accounting Review, 663-682.

Francis, J. R., Stokes, D. J., & Anderson, D. (1999). City Markets as a Unit of Analysis in Audit Research and the Re‐Examination of Big 6 Market Shares. Abacus, 35(2), 185-206. Francis, J. R. (2004). What do we know about audit quality?. The British accounting

review, 36(4), 345-368.

Francis, J. R., Reichelt, K., & Wang, D. (2005). The pricing of national and city-specific reputations for industry expertise in the US audit market. The accounting review, 80(1), 113-136.

Francis, J. R., & Yu, M. D. (2009). Big 4 office size and audit quality. The Accounting

Review, 84(5), 1521-1552.

Fung, S. Y. K., Gul, F. A., & Krishnan, J. (2012). City-level auditor industry specialization, economies of scale, and audit pricing. The accounting review,87(4), 1281-1307.

General Accounting Office (GAO). 2003. Public accounting firms: Mandated study on

consolidation and competition. GAO Report 03-864. Washington, D.C.: Government Printing Office.

Knechel, W. R., Naiker, V., & Pacheco, G. (2007). Does auditor industry specialization matter? Evidence from market reaction to auditor switches .Auditing: A Journal of Practice &

Theory, 26(1), 19-45.

Li, C. (2009). Does Client Importance Affect Auditor Independence at the Office Level? Empirical Evidence from Going‐Concern Opinions*. Contemporary Accounting

Research, 26(1), 201-230.

Mayhew, B. W., & Wilkins, M. S. (2003). Audit firm industry specialization as a differentiation strategy: Evidence from fees charged to firms going public. Auditing: A

Journal of Practice & Theory, 22(2), 33-52.

Moore, D. A., Loewenstein, G., Tanlu, L., & Bazerman, M. H. (2003). Auditor independence,

conflict of interest, and the unconscious intrusion of bias. Division of Research, Harvard

(32)

31 Nelson, M. W., Elliott, J. A., & Tarpley, R. L. (2002). Evidence from auditors about

managers' and auditors' earnings management decisions. The Accounting Review, 77(s-1), 175-202.

Porter, M. E. Competitive strategy. New York: Free Press, 1980.

Reichelt, K. J., & Wang, D. (2010). National and office‐specific measures of auditor industry expertise and effects on audit quality. Journal of Accounting Research, 48(3), 647-686. Reynolds, J. K., & Francis, J. R. (2000). Does size matter? The influence of large clients on office-level auditor reporting decisions. Journal of Accounting and Economics, 30(3), 375-400.

Romanus, R. N., Maher, J. J., & Fleming, D. M. (2008). Auditor industry specialization, auditor changes, and accounting restatements. Accounting Horizons, 22(4), 389-413.

Stice, J. D. (1991). Using financial and market information to identify pre-engagement factors associated with lawsuits against auditors. Accounting Review, 516-533.

The effects of industry specialization and client importance on audit quality for financially distressed clients